Mode of antimicrobial action of vanillin against Escherichia coli, Lactobacillus plantarum and Listeria innocua

Transcription

1 Journal of Applied Microbiology 2004, 97, doi: /j x Mode of antimicrobial action of vanillin against Escherichia coli, Lactobacillus plantarum and Listeria innocua D.J. Fitzgerald 1, M. Stratford 2, M.J. Gasson 1, J. Ueckert 3, A. Bos 3 and A. Narbad 1 1 Food Safety Science Division, Institute of Food Research, Norwich Research Park, Colney Lane, Norwich, Norfolk, UK, 2 Unilever R&D, Colworth House, Sharnbrook, Bedford, UK, and 3 Unilever R&D, Vlaardingen, Olivier Van Noortlaan, Vlaardingen, the Netherlands 2003/1072: received 24 November 2003, revised 1 March 2004 and accepted 1 March 2004 ABSTRACT D. J. F I T Z G E R A L D, M. S T R A T F O R D, M. J. G A S S O N, J. U E C K E R T, A. B O S A N D A. N A R B A D Aims: To investigate the mode of action of vanillin, the principle flavour component of vanilla, with regard to its antimicrobial activity against Escherichia coli, Lactobacillus plantarum and Listeria innocua. Methods and Results: In laboratory media, MICs of 15, 75 and 35 mmol l )1 vanillin were established for E. coli, Lact. plantarum and L. innocua, respectively. The observed inhibition was found to be bacteriostatic. Exposure to mmol l )1 vanillin inhibited respiration of E. coli and L. innocua. Addition of mmol l )1 vanillin to bacterial cell suspensions of the three organisms led to an increase in the uptake of the nucleic acid stain propidium iodide; however a significant proportion of cells still remained unstained indicating their cytoplasmic membranes were largely intact. Exposure to 50 mmol l )1 vanillin completely dissipated potassium ion gradients in cultures of Lact. plantarum within 40 min, while partial potassium gradients remained in cultures of E. coli and L. innocua. Furthermore, the addition of 100 mmol l )1 vanillin to cultures of Lact. plantarum resulted in the loss of ph homeostasis. However, intracellular ATP pools were largely unaffected in E. coli and L. innocua cultures upon exposure to 50 mmol l )1 vanillin, while ATP production was stimulated in Lact. plantarum cultures. In contrast to the more potent activity of carvacrol, a well studied phenolic flavour compound, the extent of membrane damage caused by vanillin is less severe. Conclusions: Vanillin is primarily a membrane-active compound, resulting in the dissipation of ion gradients and the inhibition of respiration, the extent to which is species-specific. These effects initially do not halt the production of ATP. Significance and Impact of the Study: Understanding the mode of action of natural antimicrobials may facilitate their application as natural food preservatives, particularly for their potential use in preservation systems employing multiple hurdles. Keywords: carvacrol, cytoplasmic membrane, Escherichia coli, Lactobacillus plantarum, Listeria innocua, mode of action, natural antimicrobial, vanillin. INTRODUCTION Many herbs and plant extracts possess antimicrobial activities against a wide range of bacteria, yeasts and moulds (Kim et al. 1995; Aziz et al. 1998; Ultee et al. 1998; Friedman Correspondence to: Arjan Narbad, Food Safety Science Division, Institute of Food Research, Norwich Research Park, Colney Lane, Norwich, Norfolk, NR4 7UA, UK ( arjan.narbad@bbsrc.ac.uk). et al. 2002); thus they provide a potentially rich source of novel biocides and preservatives. The antimicrobial activity of these plants can sometimes be attributed to the lowmolecular weight phenolic compounds that are present within them (Gould 1996; Davidson and Naidu 2000). Vanillin (4-hydroxy-3-methoxybenzaldehyde) is the major constituent of vanilla beans and is produced naturally via a multi-step curing process, however 90% of vanillin ª 2004 The Society for Applied Microbiology

2 MODE OF ANTIMICROBIAL ACTION OF VANILLIN 105 currently in use is synthetically produced (nature identical) from lignin, eugenol or guaiacol (Hocking 1997; Ramachandra Roa and Ravishankar 2000). Vanillin has generally recognized as safe (GRAS) status and is used as a flavouring/ aroma compound in foods and fragrance industries. Synthetic vanillin is also used as an intermediate in the chemical and pharmaceutical industries for the synthesis of herbicides and drugs (Walton et al. 2003). Recent reports have shown that vanillin can act as an antioxidant improving the keeping quality of precooked dried cereal flakes (Burri et al. 1989) and afforded significant protection against protein oxidation and lipid peroxidation in rat liver mitochondria (Kamat et al. 2000). Moreover, vanillin exhibits strong antimicrobial properties with activity demonstrated against a number of yeast and mould strains in laboratory media, fruit-based agar systems, fruit purees and fruit juices (López-Malo et al. 1995; Cerrutti and Alzamora 1996; Fitzgerald et al. 2004). However, to our knowledge no reports have yet detailed the mode of action of vanillin inhibition. Understanding of how antimicrobial compounds such as vanillin function is desirable if they are to find commercial application as natural preservatives. The mode of action of nearly all antimicrobials can be classed into one or more of the following groups: (a) reaction with the cell membrane, (b) inactivation of essential enzymes, or (c) destruction or inactivation of genetic material (Davidson 1993). Phenolic compounds are hydrophobic in nature and are therefore regarded as membrane active (Sikkema et al. 1994, 1995). Recent mode of action studies using plant essential oils (oregano, thyme and tea tree) or some of their phenolic constituents (carvacrol, eugenol and thymol) against several pathogenic bacteria and yeasts have shown that their activity resides in their ability to perturb the cell membrane resulting in the loss of chemiosmotic control leading to cell death (Ultee et al. 1999; Cox et al. 2000; Lambert et al. 2001; Carson et al. 2002; Burt and Reinders 2003; Walsh et al. 2003). Our objective was to determine the mode of action of vanillin inhibition of several food-related bacteria, namely E. coli, Lactobacillus plantarum and Listeria innocua. The effect of vanillin addition on respiration, membrane integrity, the potassium gradient, ph homeostasis and ATP pools was investigated. MATERIALS AND METHODS Bacterial strains and culture conditions E. coli MC1022 and L. innocua (ATCC 33090) were obtained from IFR stock cultures (Norwich, UK). Lact. plantarum was obtained from Unilever R&D stock cultures (Bedford, UK). E. coli cultures were grown in L-broth (10 g bacteriological tryptone (Becton Dickinson, Oxford, UK), 5 g yeast extract (Becton Dickinson), 5 g NaCl and 1 g glucose) at 37 C with shaking (200 rev min )1 ). Lact. plantarum cultures were grown statically at 30 C in M17 broth (Oxoid, Basingstoke, UK) supplemented with 0Æ5% (w/v) glucose (GM17). L. innocua cultures were grown in BHI broth (Oxoid) at 35 C with shaking (200 rev min )1 ). For solid media 1Æ5% (w/v) agar (Difco) was added. Chemical preparation Stock solutions of 2Æ5 mol l )1 vanillin (Sigma, Poole, Dorset, UK) and 1 mol l )1 carvacrol (Sigma) were prepared in 99Æ7 100% (v/v) ethanol. These stocks were stored at )20 C in the dark until required. Fluorescent probes propidium iodide (PI) and carboxyfluorescein diacetate succinimidyl ester (CFDA-SE) were employed in this study to analyse the membrane integrity and intracellular ph, respectively (Molecular Probes, Leiden, The Netherlands). PI has an excitation maximum (k ex ) of 536 nm and an emission maximum (k em ) of 617 nm with a fluorescence enhancement of 20- to 30-fold upon binding to DNA. A stock solution of 1 mg ml )1 was prepared in Milli Q water and stored at 2 C. CFDA-SE is cleaved by esterases upon entering the cytosol to yield fluorescent carboxyfluorescein (CF) that has a k ex of 494 nm and a k em of 518 nm. A stock solution of 1 mg ml )1 was prepared in dimethyl sulphoxide (DMSO; Sigma) and stored at 2 C. A 200 lmol l )1 nigericin (Sigma) stock was prepared in 99Æ7 100% (v/v) ethanol and stored at 2 C. Susceptibility testing Susceptibility to vanillin is expressed as minimal inhibitory concentration (MIC). Experiments were conducted in 10 ml volumes of media in 30 ml screw-capped bottles supplemented with vanillin in duplicate. Cultures grown in the presence of maximum ethanol levels only, acted as controls. Cultures were incubated under optimal conditions for a period of 96 h, after which the bottles were examined for growth or absence of growth. MIC was defined as the lowest concentration of vanillin required to completely inhibit growth and was determined visually. To establish the nature of the inhibitory activity of vanillin, 100 ll samples were taken and surface-plated onto agar and incubated under optimal conditions for up to 48 h. Determination of the effect of vanillin on respiration Overnight (16 h) cell cultures were subcultured (200 ll) into 10 ml fresh media and incubated for a further 3 h

3 106 D.J. FITZGERALD ET AL. (E. coli) or 5 h (L. innocua). Cells were harvested by centrifugation at 4 C (1580 g for 10 min) and washed once with 50 mmol l )1 potassium phosphate buffer [ph 7Æ0/ 0Æ1% (w/v) glucose for E. coli or ph 7Æ4/0Æ2% (w/v) glucose for L. innocua]. Cell suspensions were adjusted to a density of 0Æ2 atano.d. 600nm in the same buffers and kept at 2 C throughout the experiments. Oxygen consumption of cell suspensions was measured using a calibrated Clark-type oxygen electrode (Digital model 10, Rank Brothers Ltd, Cambridge, UK) linked to Model 1325 Econo chart recorder (Bio-Rad, Hercules, CA, USA). The chamber temperature was maintained via a circulating water bath and cell suspensions were preheated (5 min), then an initial respiration rate was quantified before the addition of any components. Vanillin was added to final concentrations of 0 40 mmol l )1 and the addition of 2% (v/v) ethanol alone acted as the control. All values were recorded from triplicate measurements and suitable controls were included to ensure that none of the test elements interfered with the assay. Flow cytometry analysis Fluorescent measurements were carried out using a calibrated Coulter EPICS ELITE flow cytometer (Beckman- Coulter, High Wycombe, Bucks, UK) with 488 nm excitation from an argon-ion laser. The sample analysis rate was kept below 1000 events per second and ended after events had been acquired. Red fluorescence (PI, membrane integrity analysis) was detected through a 610 nm filter, green and orange fluorescence (CFDA-SE, ph i analysis) were detected through 525 and 575 nm filters, respectively. In membrane integrity experiments, vanillin was added to a final concentration of 50 mmol l )1 (E. coli and L. innocua), or 70 mmol l )1 (Lact. plantarum) to cells in exponential phase of growth (ca 10 6 cells ml )1 ) and incubated under optimal conditions. Samples (20 ll) were taken after 15 min or 1 h and added to 1 ml of 50 mmol l )1 potassium phosphate buffer (ph 6Æ0) containing PI at a working concentration of 5 lg ml )1, mixed and incubated at room temperature for 5 min prior to flow cytometry analysis. A final sample was also taken following overnight incubation (18 h). Cells grown in the absence of vanillin acted as negative controls and heat-treated (80 C for 4 min) cells were used as positive controls. The effect of ethanol only addition was also examined. Intracellular ph measurements were performed with Lact. plantarum using a modified method from Breeuwer et al. (1996). Cells (ca 10 6 cells ml )1 ) were exposed to 20 or 100 mmol l )1 vanillin, 2 lmol l )1 nigericin or 4% (v/v) ethanol plus CFDA-SE (20 lg ml )1 ) and incubated for 10 min at room temperature. Samples (10 ll) were then added to 1 ml potassium phosphate buffer (50 mmol l )1 ) varying in ph from 4Æ5 to 7Æ5 and then introduced into the flow cytometer. Flow cytometry data was analysed using WinMDI v.2æ8, available as freeware from Determination of intra- and extracellular potassium levels Intra- and extracellular potassium levels were determined using a modified method from Ultee et al. (1999). Overnight (16 h) cell cultures were centrifuged at 4 C (1580 g for 10 min) and the cell pellets washed twice in either 50 mmol l )1 sodium HEPES buffer (ph 7Æ0 for E. coli, ph 7Æ4 for L. innocua) or 50 mmol l )1 sodium phosphate buffer (ph 6Æ5) for Lact. plantarum. Cell suspensions were adjusted to a density of 1Æ0 at an O.D. 600nm. After an initial incubation period of 10 min under optimal conditions, the experiments were started by energizing the cells with glucose to final concentrations of 0Æ1% (w/v) for E. coli, 0Æ2% (w/v) for L. innocua and 0Æ5% (w/v) for Lact. plantarum. At 10 min intervals, samples (1 ml) were removed and added to sterile Eppendorf tubes and centrifuged immediately (2300 g for 2 min). The extracellular supernatants were collected (900 ll) and immediately frozen using dry ice. The remaining supernatant was carefully removed and discarded. To the remaining cell pellet 1 ml TCA-EDTA buffer was added [10% (v/v) trichloroacetic acid and 2 mmol l )1 EDTA]. The cells were resuspended and then centrifuged immediately ( g for 2 min); the resulting intracellular supernatants were collected (900 ll) and immediately frozen using dry ice. Vanillin, carvacrol or ethanol (control) were added after the 20 min sampling point. The potassium concentrations were measured using a Flame Photometer (Atomic Absorption Spectrometer 3300, Perkin-Elmer Ltd, Bucks, UK) set to read absorbance at 766Æ5 nm with a slit width of 0Æ1 mm. Values were extrapolated against a standard calibration curve of KCl after dilution (1 : 10) in HCl [final concentration 5% (v/v) in Milli Q water]. All values were recorded from triplicate measurements and suitable controls were included to ensure that none of the test elements interfered with the assay. Determination of intra- and extracellular ATP levels Cell preparation and sampling procedure were as described above with the following amendments: 50 mmol l )1 potassium phosphate buffers were used and the extracellular supernatants collected (500 ll) were diluted 1 : 2 with 2 mmol l )1 EDTA before being frozen immediately using dry ice. ATP levels were determined using an Enliten Ò rluciferin/luciferase ATP assay kit (Promega, Southampton, UK) with 10 ll sample and 100 ll reagent volumes per assay. Luminescence (560 nm) was recorded with a Lumat LB 6501 luminometer (Berthold UK Ltd, St Albans, Herts,

4 MODE OF ANTIMICROBIAL ACTION OF VANILLIN 107 UK). All values were recorded from triplicate measurements and suitable controls were included to ensure that none of the test elements interfered with the assay. Viability assays Cells were prepared as described for the potassium determination assay in duplicate. Samples (20 ll) were taken at 0 and 1 h and a series of decimal dilutions were made in 1/4 strength Ringers solution ph 7Æ0 (Oxoid). The dilutions were surface plated in triplicate onto the appropriate agar, and the viable cell counts were enumerated after incubation under optimal conditions for h. tions of vanillin resulted in the ÔinhibitionÕ of respiration; however the addition of a minimum concentration of 30 mmol l )1 was required before the inhibitory effect overcame the stimulatory effect of ethanol in all cases. Ethanol was added due to the requirement for a solvent in the preparation of vanillin solutions. Maximum inhibition levels of 19 and 52% were achieved with the addition of 40 mmol l )1 vanillin to E. coli and L. innocua cell suspensions, respectively. A linear relationship existed between the inhibition of respiration and vanillin concentration in E. coli cell suspensions with an R 2 value of 0Æ97. A weaker linear relationship existed for cell suspensions of L. innocua with an R 2 value of 0Æ83. RESULTS Antimicrobial activity of vanillin The antimicrobial activity of vanillin was investigated against E. coli, Lact. plantarum, andl. innocua in laboratory media. After incubation for 96 h, MICs of 15, 75 and 35 mmol l )1 were established for the three strains respectively. Colony formation on agar plates inoculated with cells from cultures exposed to MIC levels of vanillin indicated that the inhibitory action of vanillin was bacteriostatic rather than bactericidal. Effect of vanillin on membrane integrity Flow cytometry was used in tandem with the nucleic acid stain propedium iodide (PI) to measure the effect of vanillin on membrane integrity. PI cannot enter cells with intact membranes resulting in low fluorescence intensity in the red spectrum, as shown by untreated E. coli cells (Fig. 2a). Cells that have damaged membranes allow the stain to enter the cell and bind to the DNA resulting in a fold increase in fluorescence intensity (Fig. 2b). A strong shift in intensity was also observed with Lact. plantarum and L. innocua cultures treated similarly (data not shown), however heat Effect of vanillin on respiration The addition of 2% (v/v) ethanol stimulated respiration in glucose-energized cells of E. coli and L. innocua to 114 and 79% above the normal respiration rate of untreated cells, respectively (Fig. 1). The addition of increasing concentra- (a) 75 Events I D Relative change in respiration rate (%) (a) R 2 = (b) R 2 = (b) 40 0 (a) Vanillin (mmol l 1 ) Fig. 1 Relative change* in the respiration rates of cell suspensions of Listeria innocua (d) and Escherichia coli (m) in the presence of vanillin. The values represent the mean of triplicate measurements. In the absence of vanillin 2% (v/v) ethanol acted as a control. Linear trendlines and R 2 values for L. innocua (a) and E. coli (b) are indicated *Compared with the respiration rates of untreated cell suspensions (0Æ0) 0 (b) 75 Events Log red Fig. 2 Flow cytometry histograms of red fluorescence (610 nm) of untreated (a) or heat-treated (b) Escherichia coli cultures stained with PI (5 lg ml )1 ). Gated zones show intact cells (I) and damaged cells (D)

5 108 D.J. FITZGERALD ET AL. treatment of these cultures proved to be less effective (Table 1). Escherichia coli cultures exposed to 50 mmol l )1 vanillin showed a 29% increase in the number of cells with damaged membranes within 60 min. After a longer incubation (overnight), some cells appeared to recover and this level was reduced to 13%. Lactobacillus plantarum cultures exposed to 70 mmol l )1 vanillin had only 10% of cells with damaged membranes after 60 min, this level increased to 55% after overnight incubation. Similarly L. innocua cultures exposed to 50 mmol l )1 vanillin had only a small proportion of cells with damaged membranes (4% after 60 min), however unlike Lact. plantarum cultures this level did not increase overnight. The addition of maximum ethanol concentrations was found not to affect the membrane integrity of any of the strains studied (data not shown). Effect of vanillin on intra- and extracellular potassium levels Disruption of the cytoplasmic membrane by vanillin would also be indicated by an effect on ion gradients between the cell and the external environment. Therefore intra (K þ in )- and extra (K þ ext )-cellular potassium levels were measured. Carvacrol, a membrane-active phenolic compound (Ultee et al. 1999; Lambert et al. 2001), was used as a positive control. Addition of 50 mmol l )1 vanillin to E. coli cell Table 1 Flow cytometry measurements using propidium iodide for identification of damaged cell membranes of Escherichia coli, Lactobacillus plantarum and Listeria innocua treated with vanillin. Data from control and heat-treated cells are also shown Bacterial strain Treatment Sampling time (min) E. coli Control* mmol l )1 vanillin Overnight 13 Heat 88 Lact. plantarum Control* mmol l )1 vanillin Overnight 55 Heat 53 L. innocua Control* mmol l )1 vanillin 60 4 Overnight 2 Heat 46 Membrane damage (%) *Ethanol was added to 2 or 2Æ8% (v/v) as required for the addition of 50 or 70 mmol l )1 vanillin, respectively. 80 C for 4 min. suspensions resulted in a rapid decrease of K þ in (6Æ3 2Æ1 ppm ml )1 ) within 10 min (Fig. 3a). A corresponding increase of K þ ext from 0Æ04 to 3Æ4 ppm ml )1 was recorded. Over the next 30 min only a slight reduction of K þ in was observed indicating that a K + gradient was still being maintained although to a lesser degree. Addition of 3Æ3 mmol l )1 carvacrol resulted in the sudden and complete (a) K+ (ppm ml 1 ) (b) 1 K+ (ppm ml 1 ) (c) 1 K+ (ppm ml 1 ) Time (min) Fig. 3 Intracellular (closed symbols, solid lines) and extracellular (open symbols, dotted lines) potassium levels of glucose-energized Escherichia coli (a), Lactobacillus plantarum (b) and Listeria innocua (c) exposed to 50 mmol l )1 vanillin (j() or3æ3 mmol l )1 carvacrol (mn). Exposure to 2% (v/v) ethanol (r)) acted as a control. Compounds were added at 20 min (indicated by arrow). The values represent the mean of triplicate measurements

6 MODE OF ANTIMICROBIAL ACTION OF VANILLIN 109 Table 2 Viable counts (CFU ml )1 )ofescherichia coli, Lactobacillus plantarum and Listeria innocua cells after the addition of either 50 mmol l )1 vanillin or 3Æ3 mmol l )1 carvacrol. Exposure to 2% (v/v) ethanol acted as a control. Values represent the mean of duplicate measurements Bacterial strain Treatment Sampling point (h) 0 1 E. coli Control 9Æ Æ mmol l )1 vanillin 6Æ Æ Æ3 mmol l )1 carvacrol 7Æ Lact. plantarum Control 4Æ Æ mmol l )1 vanillin 3Æ Æ Æ3 mmol l )1 carvacrol 4Æ Æ L. innocua Control 1Æ Æ mmol l )1 vanillin 1Æ Æ Æ3 mmol l )1 carvacrol 9Æ collapse of the K + gradient within 10 min. Viable cell counts showed that the addition of vanillin resulted in a reduction in cell numbers from 6Æ to 5Æ CFU ml )1, while the addition of carvacrol resulted in the complete loss of viability (Table 2). The addition of either vanillin or carvacrol had a similar effect on K + ion gradients in Lact. plantarum cell suspensions (Fig. 3b). At 50 mmol l )1 vanillin a continuous decrease in K þ in to near 0 after 40 min was measured compared with the initial decrease of K þ in in control cells which eventually stabilized, again a corresponding increase of K þ ext was observed. No loss of cell viability was observed with vanillin, while carvacrol caused a complete loss of cell viability (Table 2). K + ion gradients were still present in L. innocua cultures after 60 min although K þ in levels were reduced from 12Æ3 to5æ5 ppm ml)1 or 6Æ6 ppm ml )1 after 40 min following the additions of 50 mmol l )1 vanillin or 3Æ3 mmol l )1 carvacrol, respectively (Fig. 3c). A corresponding increase of K þ ext was again observed. However, the viable cell counts only slightly decreased from 1Æ to 9Æ CFU ml )1 following exposure to vanillin, but the addition of carvacrol was totally bactericidal (Table 2). Effect of vanillin on ph homeostasis The ph homeostasis of Lact. plantarum in the presence of vanillin was investigated further by flow cytometry analysis utilizing the ph-sensitive stain CFDA-SE. Cells with an intact membrane would be able to maintain their internal ph via ion channels and pumps in the presence of a moderate change of ph. Therefore the mean ratio of green fluorescence (525 nm) that is ph sensitive and orange fluorescence (575 nm) that is ph insensitive would also remain constant. This ratio would change depending on the external ph if the Mean ratio ( 575 / 525 nm) ph (50 mmol l 1 potassium phosphate buffer) Fig. 4 Fluorescence intensity ratios (575/525 nm) of Lactobacillus plantarum cell suspensions stained with CFDA-SE (20 lg ml )1 ) after exposure to 2 lmol l )1 nigericin (m), 20 mmol l )1 (d) or 100 mmol l )1 (j) vanillin. Exposure to 4% (v/v) ethanol (r) acted as a control cell membrane integrity was disrupted and H + ion gradients were no longer controlled. The mean ratio of cells treated with either 4% (v/v) ethanol (control) or 20 mmol l )1 vanillin remained fairly stable irrespective of the external ph (Fig. 4). The slight change in mean ratio may be due to the amount of stain that accumulates in the cell wall and is dependent on external ph values. Addition of the H + ionophore nigericin (2 lmol l )1 ) resulted in a large shift in the mean ratio as the external ph increased. The mean ratio shifted from a maximum value of 333 at an external ph (ph ext ) 4Æ5 to a minimum value of 220 at a ph ext 7Æ5 indicating membrane damage and loss of ph homeostasis. This shift was more obvious with the addition of 100 mmol l )1 vanillin resulting in a maximum mean ratio value of 378 at a ph ext 4Æ5 and a minimum value of 243 at a ph ext 7Æ5. Effect of vanillin on cellular energy (ATP) Membrane perturbation, decrease in respiration and loss of K + and H + ion gradients would suggest that energy generation within the cells would be detrimentally affected. The effect of vanillin on ATP concentration was therefore investigated. Carvacrol was again used as a positive control. Intracellular ATP (ATP in ) levels continued to increase in E. coli cell suspensions even after the addition of 50 mmol l )1 vanillin although at a slightly reduced rate when compared with control cells (Fig. 5a). Small quantities of external ATP (ATP ext ) were observed but these levels did not increase significantly in either the control or vanillin treated cells. Addition of 3Æ3 mmoll )1 carvacrol resulted in the rapid decrease of ATP in to undetectable levels within 10 min. This coincided with an observed increase of ATP ext

7 110 D.J. FITZGERALD ET AL. (a) ATP (nmol l 1 ml 1 ) ATP (nmol l 1 ml 1 ) ATP (nmol l 1 ml 1 ) (b) (c) Time (min) Fig. 5 Intracellular (closed symbols, solid lines) and extracellular (open symbols, dotted lines) ATP levels of glucose-energized Escherichia coli (a), Lactobacillus plantarum (b) and Listeria innocua (c) exposed to 50 mmol l )1 vanillin (j() or3æ3 mmol l )1 carvacrol (mn). Exposure to 2% (v/v) ethanol (r)) acted as a control. Compounds were added at 20 min (indicated by arrow). The values represent the mean of triplicate measurements level. Addition of 50 mmol l )1 vanillin to Lact. plantarum cell suspensions resulted in the stimulation of ATP in levels compared with levels recorded in the control cell suspensions (Fig. 5b). The addition of carvacrol decreased ATP in levels although this time only a small increase in ATP ext levels was observed. Reduced levels of ATP in were recorded in cell suspensions of L. innocua treated with vanillin when compared with control levels (Fig. 5c). However, significant levels of ATP in still remained and no ATP ext were detected. The addition of carvacrol resulted in a gradual decrease in ATP in with a corresponding increase of ATP ext levels. DISCUSSION Many natural compounds, including phenolic compounds derived from plants, exhibit strong antimicrobial properties and therefore have the potential to be applied to food products as novel preservatives (Beuchat and Golden 1989; Gould 1996; Friedman et al. 2002). However, there is relatively little knowledge about the specific mode of action of these compounds. In this study, we investigated the effect of vanillin on the cytoplasmic membrane of the food-related bacteria E. coli, Lact. plantarum, and L. innocua. The antimicrobial activity of vanillin was found to be dependent on the time of exposure, concentration and the target organism. Jay and Rivers (1984) reported that the inhibitory activity of vanillin was more effective against nonlactic Gram-positives than Gram-negative bacteria. The high MIC reported here for Lact. plantarum is in agreement with these findings, however the E. coli strain used here had the lowest MIC of the three organisms studied indicating that certain Gram-negative bacteria can be equally susceptible to the antimicrobial activity of vanillin. The inhibitory action of vanillin at MIC was found to be bacteriostatic in contrast to the more potent phenolic antimicrobials such as carvacrol and thymol that are bactericidal (Ultee et al. 1998; Friedman et al. 2002). Furthermore, it is recognized that if vanillin is to be used as a preservative at the MIC values established here it will impart the characteristic flavour and hence organoleptic quality in the individual food application needs to be examined. Initial experiments revealed that glucose-dependent respiration was inhibited with increasing levels of vanillin. These observations could indicate that vanillin affected membrane integrity; however the direct inhibition of respiratory enzymes or processes cannot be excluded using the method employed here. Cox et al. (2000) observed that tea tree oil, where the antimicrobial activity is attributed to terpinen-4-ol also inhibited respiration in E. coli. Phenolic compounds primarily target the cytoplasmic membrane due to their hydrophobic nature and will therefore preferentially partition into the lipid bilayer (Sikkema et al. 1994, 1995; Weber and de Bont 1996). Interactions between both lipids and membrane embedded proteins with the phenolic compound results in the destabilizing of the membrane and loss of integrity. The observed increase in the uptake of the nucleic acid stain PI upon exposure to vanillin suggested the integrity of the cell membrane was compromised; however a significant proportion of the cell populations still had functional membranes. Similar observations have been

8 MODE OF ANTIMICROBIAL ACTION OF VANILLIN 111 reported in cultures of E. coli and Staph. aureus when exposed to tea tree oil (Cox et al. 2000). This would indicate that the destabilizing effect of vanillin on the membrane is at a sublethal level for the majority of the microbial population and may provide an explanation for the bacteriostatic activity of vanillin observed here. One function of the cytoplasmic membrane is to act as a selectively permeable barrier for small ions such as H +, K +, Na + and Ca 2+. The gradients of these ions between the intra- and extracellular environments play an important role for the cell in the regulation of cytoplasmic ph, control of turgor pressure and generation of cellular energy (Kroll and Booth 1981; Booth 1985; Sikkema et al. 1995). This study has shown that vanillin caused a rapid leakage of intracellular K + ions and a resulting increase in extracellular K +. Ion gradients completely collapsed in cell suspensions of Lact. plantarum, however partial gradients still remained in cell suspensions of E. coli and L. innocua. Carvacrol and tea tree oil have also been shown to cause leakage of K + ions (Ultee et al. 1999; Cox et al. 2000). In this study carvacrol was used as a positive control and the results obtained here were consistent with those of Ultee et al. (1999). However, unlike vanillin, the addition of carvacrol resulted in the complete loss of viability. If membrane permeability was increased for K + ions then it would be assumed that H + ion gradients would also be affected. The addition of vanillin (100 mmol l )1 ) or the proton ionophore nigericin (2 lmol l )1 )tolact. plantarum cultures resulted in the loss of ph homeostasis. However, the addition of 20 mmol l )1 vanillin did not affect ph homeostasis and further demonstrates the concentration dependence of vanillin activity. Proton ion gradients generate the two components of the proton motive force, the ph gradient (DpH) and the electrical potential (Dw) that are used to generate ATP via the membrane-located ATPase (Sikkema et al. 1995; Davidson 1997). The loss of ion gradients, particularly the H +, could be expected to be detrimental for the generation of cellular energy (ATP), while it may affect the cell indirectly through decreased enzyme activity, although this inhibition would be dependent on both the external ph and the optimal ph at which any particular enzyme functions. In this study addition of vanillin resulted in either the stimulation of ATP (Lact. plantarum) or slightly reduced levels of ATP production (E. coli and L. innocua) when compared with levels produced in control cell suspensions. These results indicate an apparent contradiction. Although the generation of ATP, at least for a short period of time could be possible through the passive influx of H + via the ATPase enzyme, a partial K + gradient still existed in E. coli and L. innocua cell suspensions that may have been sufficient to drive ATP generation. Lactobacillus plantarum are oxygen-tolerant organisms that do not produce ATP via an oxidative phosphorylation system, instead producing ATP via substrate level phosphorylation. This energy-generating system could potentially remain unaffected by vanillin in the short term i.e. glucose as a substrate was present initially, while the same hypothesis may also hold true for the generation of ATP via glycolysis only in the other two strains. The addition of carvacrol resulted in the rapid depletion of intracellular ATP levels and a corresponding increase in extracellular ATP levels indicating leakage of ATP in E. coli and L. innocua. Helander et al. (1998) also observed a similar change in ATP levels when E. coli cells were exposed to either carvacrol or thymol. Ultee et al. (2002) reported a novel mode of action for carvacrol inhibition of B. cereus. It was proposed that carvacrol acts as a trans-membrane carrier of monovalent cations by exchanging its hydroxyl H + for another ion such as K +. The observations made in our study showing K + leakage and also the loss of ph homeostasis (Lact. plantarum only) would suggest that vanillin does not work in the same manner. This is further substantiated by earlier work showing the importance of the aldehyde group rather than the hydroxyl group of the vanillin structure in the inhibitory activity of vanillin against Saccharomyces cerevisiae (Fitzgerald et al. 2003). The extent of membrane damage induced by any given compound can be related to its intrinsic hydrophobicity that can be determined experimentally by its partition coefficient in octanol-water (P o/w ), compounds with a higher P o/w will partition further into the cell membrane (Weber and de Bont 1996). Vanillin has a log P o/w of 1Æ09 while carvacrol has a log P o/w of 3Æ64 (Ultee et al. 2002) which could account for the relatively weak membrane perturbation observed upon exposure to vanillin and the severe membrane damage caused by exposure to carvacrol reported here and elsewhere (Kim et al. 1995; Helander et al. 1998; Ultee et al. 1998). Phenolic compounds have been shown to inhibit DNA, RNA and protein synthesis (Nes and Eklund 1983), glucose uptake (Evans and Martin 2000) and enzyme activities (Rico-Munoz et al. 1987; Wendakoon and Sakaguchi 1995; Kreydiyyeh et al. 2000). Furthermore, several reports have detailed the binding interactions between vanillin and proteins including bovine serum albumin, soy, fababean and milk proteins (Ng et al. 1989; Li et al. 2000; Chobpattana et al. 2002). Therefore we cannot rule out the possibility that vanillin could also inhibit key membrane proteins or other cellular functions. In summary our observations using the food-related bacteria E. coli, Lact. plantarum, andl. innocua have shown that the inhibitory activity of vanillin resides primarily in its ability to detrimentally affect the integrity of the cytoplasmic membrane, with the resultant loss of ion gradients, ph homeostasis and inhibition of respiratory activity. Energy generation remains largely unaffected or can indeed be stimulated for at least a short period of time (1 h). The

9 112 D.J. FITZGERALD ET AL. extent of the membrane damage appears to be sublethal in the majority of cells within an inhibited microbial population, exhibited as a bacteriostatic action of inhibition at MIC. ACKNOWLEDGEMENTS We thank Dave Hart (IFR) for his assistance with the potassium measurements. This work was supported at the Institute of Food Research by a Unilever studentship awarded to D. Fitzgerald. REFERENCES Aziz, N.H., Farag, S.E., Mousa, L.A.A. and Abo-Zaid, M.A. (1998) Comparative antibacterial and antifungal effects of some phenolic compounds. Microbios 96, Beuchat, L.R. and Golden, D.A. (1989) Antimicrobials occurring naturally in foods. Food Technology 43, Booth, I.R. (1985) Regulation of cytoplasmic ph in bacteria. Microbiological Reviews 49, Breeuwer, P., Drocourt, J.-L., Rombouts, F.M. and Abee, T. (1996) A novel method for continuous determination of the intracellular ph in bacteria with the internally conjugated fluorescent probe 5 (and 6-)- carboxyfluorescein succinimidyl ester. Applied and Environmental Microbiology 62, Burri, J., Graf, M., Lambelet, P. and Löliger, J. (1989) Vanillin: more than a flavouring agent a potent antioxidant. Journal of the Science of Food and Agriculture 48, Burt, S.A. and Reinders, R.D. (2003) Antibacterial activity of selected plant essential oils against Escherichia coli O157:H7. Letters in Applied Microbiology 36, Carson, C.F., Mee, B.J. and Riley, T.V. (2002) Mechanism of action of Melaleuca alternifolia (tea tree) oil on Staphylococcus aureus determined by time to kill, leakage, and salt tolerance assays and electron microscopy. Antimicrobial Agents and Chemotherapy 46, Cerrutti, P. and Alzamora, S.M. (1996) Inhibitory effects of vanillin on some food spoilage yeasts in laboratory media and fruit purees. International Journal of Food Microbiology 29, Chobpattana, W., Jeon, I.J., Smith, J.S. and Loughin, T.M. (2002) Mechanisms of interaction between vanillin and milk proteins in model systems. Journal of Food Science 67, Cox, S.D., Mann, C.M., Markham, J.L., Bell, H.C., Gustafson, J.E., Warmington, J.R. and Wyllie, S.G. (2000) The mode of antimicrobial action of the essential oil of Melaleuca alternifolia (tea tree oil). Journal of Applied Microbiology 88, Davidson, P.M. (1993) Parabens and phenolic compounds. In Antimicrobials in Foods, 2nd edn. ed. Davidson, P.M. and Branen, A.L. pp New York: Marcel Dekker Inc. Davidson, P.M. (1997) Chemical preservatives and natural antimicrobial compounds. In Food Microbiology Fundamentals and Frontiers ed. Doyle, M.P., Beuchat, L.R. and Montville, T.J. pp Washington DC: American Society for Microbiology. Davidson, P.M. and Naidu, A.S. (2000) Phyto-phenols. In Natural Food Antimicrobial Systems ed. Naidu, A.S. pp London: CRC Press. Evans, J.D. and Martin, S.A. (2000) Effects of thymol on ruminal microorganisms. Current Microbiology 41, Fitzgerald, D.J., Stratford, M. and Narbad, A. (2003) Analysis of the inhibition of food spoilage yeasts by vanillin. International Journal of Food Microbiology 86, Fitzgerald, D.J., Stratford, M., Gasson, M.J. and Narbad A. (2004) The potential application of vanillin in preventing yeast spoilage of soft drinks and fruit juices. Journal of Food Protection 67, Friedman, M., Henika, P.R. and Mandrell, R.E. (2002) Bactericidal activities of plant essential oils and some of their isolated constituents against Campylobacter jejuni, Escherichia coli, Listeria monocytogenes, and Salmonella enterica. Journal of Food Protection 65, Gould, G.W. (1996) Industry perspectives on the use of natural antimicrobials and inhibitors for food applications. Journal of Food Protection (Suppl.), Helander, I.M., Alakomi, H., Latva-Kala, K., Mattila-Sandholm, T., Pol, I., Smid, E.J., Gorris, L.G.M. and von Wright, A. (1998) Characterization of the action of selected essential oil components on Gram-negative bacteria. Journal of Agricultural and Food Chemistry 46, Hocking, M.B. (1997) Vanillin: synthetic flavoring from spent sulfite liquor. Journal of Chemical Education 74, Jay, J.M. and Rivers, G.M. (1984) Antimicrobial activity of some food flavoring compounds. Journal of Food Safety 6, Kamat, J.P., Ghosh, A. and Devasagayam, T.P.A. (2000) Vanillin as an antioxidant in rat liver mitochondria: inhibition of protein oxidation and lipid peroxidation induced by photosensitization. Molecular and Cellular Biochemistry 209, Kim, J., Marshall, M.R. and Wei, C. (1995) Antibacterial activity of some essential oil components against five foodborne pathogens. Journal of Agricultural Food Chemistry 43, Kreydiyyeh, S.I., Usta, J. and Copti, R. (2000) Effect of cinnamon, clove and some of their constituents on the Na + -K + -ATPase activity and alanine absorption in the rat jejunum. Food and Chemical Toxicology 38, Kroll, R.G. and Booth, I.R. (1981) The role of potassium transport in the generation of a ph gradient in Escherichia coli. Biochemical Journal 198, Lambert, R.J.W., Skandamis, P.N., Coote, P.J. and Nychas, G.-J.E. (2001) A study of the minimum inhibitory concentration and mode of action of oregano essential oil, thymol and carvacrol. Journal of Applied Microbiology 91, Li, Z., Grün, I.U. and Fernando, L.N. (2000) Interaction of vanillin with soy and dairy proteins in aqueous model systems: a thermodynamic study. Journal of Food Science 65, López-Malo, A., Alzamora, S.M. and Argaiz, A. (1995) Effect of natural vanillin on germination time and radial growth of moulds in fruit-based agar systems. Food Microbiology 12, Nes, I.F. and Eklund, T. (1983) The effect of parabens on DNA, RNA and protein synthesis in Escherichia coli and Bacillus subtilis. Journal of Applied Bacteriology 54, Ng, P.K.W., Hoehn, E. and Bushuk, W. (1989) Binding of vanillin by fababean proteins. Journal of Food Science 54, Ramachandra Roa, S. and Ravishankar, G.A. (2000) Vanilla flavor: production by conventional and biotechnological routes. Journal of the Science of Food and Agriculture 80,

10 MODE OF ANTIMICROBIAL ACTION OF VANILLIN 113 Rico-Munoz, E., Bargiota, E.E. and Davidson, P.M. (1987) Effect of selected phenolic compounds on the membrane-bound adenosine triphosphatase of Staphylococcus aureus. Food Microbiology 4, Sikkema, J., de Bont, J.A.M. and Poolman, B. (1994) Interactions of cyclic hydrocarbons with biological membranes. Journal of Biological Chemistry 269, Sikkema, J., de Bont, J.A.M. and Poolman, B. (1995) Mechanisms of membrane toxicity of hydrocarbons. Microbiological Reviews 59, Ultee, A., Gorris, L.G.M. and Smid, E.J. (1998) Bactericidal activity of carvacrol towards the food-borne pathogen Bacillus cereus. Journal of Applied Microbiology 85, Ultee, A., Kets, E.P.W. and Smid, E.J. (1999) Mechanisms of action of carvacrol on the food-borne pathogen Bacillus cereus. Applied and Environmental Microbiology 65, Ultee, A., Bennik, M.H.J. and Moezelaar, R. (2002) The phenolic hydroxyl group of carvacrol is essential for action against the foodborne pathogen Bacillus cereus. Applied and Environmental Microbiology 68, Walsh, S.E., Maillard, J.-Y., Russell, A.D., Catrenich, C.E., Charbonneau, D.L. and Bartolo, R.G. (2003) Activity and mechanisms of action of selected biocidal agents on Gram-positive and -negative bacteria. Journal of Applied Microbiology 94, Walton, N.J., Mayer, M.J. and Narbad, A. (2003) Molecules of interest: vanillin. Phytochemistry 63, Weber, F.J. and de Bont, J.A.M. (1996) Adaptation mechanisms of microorganisms to the toxic effects of organic solvents on membranes. Biochimica et Biophysica Acta 1286, Wendakoon, C.N. and Sakaguchi, M. (1995) Inhibition of amino acid decarboxylase activity of Enterobacter aerogenes by active components in spices. Journal of Food Protection 58,