Assessing the microbial oxidative stress mechanism of ozone treatment through the responses of Escherichia coli mutants


Paula Bourke, School of Food Science and Environmental Health, Dublin Institute of Technology, Cathal Brugha Street, Dublin 1, Ireland. E-mail:


Aims:  To investigate the effect of the oxidative stress of ozone on the microbial inactivation, cell membrane integrity and permeability and morphology changes of Escherichia coli.

Methods and Results: Escherichia coli BW 25113 and its isogenic mutants in soxR, soxS, oxyR, rpoS and dnaK genes were treated with ozone at a concentration of 6 μg ml−1 for a period up to 240 s. A significant effect of ozone exposure on microbial inactivation was observed. After ozonation, minor effects on the cell membrane integrity and permeability were observed, while scanning electron microscopy analysis showed slightly altered cell surface structure.

Conclusions:  The results of this study suggest that cell lysis was not the major mechanism of microbial inactivation. The deletion of oxidative stress–related genes resulted in increased susceptibility of E. coli cells to ozone treatment, implying that they play an important role for protection against the radicals produced by ozone. However, DnaK that has previously been shown to protect against oxidative stress did not protect against ozone treatment in this study. Furthermore, RpoS was important for the survival against ozone.

Significance and Impact of the Study:  This study provides important information about the role of oxidative stress in the responses of E. coli during ozonation.


Ozone is a powerful antimicrobial agent owing to its potential oxidizing capacity. Previous studies have reported that ozone effectively inactivates bacteria in water (Levadnaya et al. 2009), vegetables, fruits (Selma et al. 2008) and apple juice (Patil et al. 2010). It is also known that ozone at high concentrations may be fatal to humans, and it has been reported that prolonged exposure at concentrations >1000 mg l−1 can cause death (Scott and Lesher 1963). However, it is a highly unstable compound with a relatively short life; shorter half-life in water (20 min) than in air (3 days) at 20°C (; accessed on 6 May 2011). Therefore, as a result of its rapid decomposition to molecular oxygen, ozone treatment leaves no residues, making it an environmentally friendly and safe antimicrobial agent for use in the food industry (Kim et al. 1999). Different factors such as temperature, pH and organic matter influence ozone decomposition, while the antimicrobial properties of ozone are attributed to the progressive oxidation of vital cellular components. The bacterial cell surface has been suggested as the primary target of ozonation. Ozone oxidizes sulfhydryl groups, amino acids of enzymes, peptides and proteins to shorter peptides, while it also oxidizes polyunsaturated fatty acids to acid peroxides (Victorin 1992).

The decomposition of ozone results in the generation of superoxide radicals (inline image), hydroperoxyl radicals (inline image) and hydroxyl radicals (˙OH) (Adler and Hill 1950; Hoigné and Bader 1975). However, micro-organisms develop mechanisms such as the superoxide dismutases, reductases, peroxidises and catalases to counteract the lethal effects of the reactive oxygen species (Imlay 2008).

In Escherichia coli, two such mechanisms are SoxR and OxyR which are redox responsive transcription regulators that have been well described (Pomposiello and Demple 2001). Both regulators are induced in the presence of inline image radicals (Greenberg et al. 1990) and activate various genes like soxS and sod which in turn confer protection against these radicals through DNA repair or removal of the radicals (Pomposiello and Demple 2001). The DnaK and RpoS are two regulators of general stress genes, which, although are not dedicated mechanisms of protection against oxidative radicals, have been previously shown to confer protection against them (Delaney 1990; Rockabrand et al. 1995; Loewen et al.1998). Similar radicals are produced during ozone treatments, and therefore, these genes are expected to play an important role in protection of cells against this technology.

Up to now, there is really no extensive information on the main cellular target of ozone treatment although damage to cell membranes and cytoplasmic contents has previously been proposed (Scott and Lesher 1963; Mudd et al. 1969; Pryor et al. 1991). In this study, with the use of deletion E. coli mutants in soxR, soxS, oxyR, rpoS and dnaK which have been shown to play an important role in the protection against reactive oxygen radicals, we attempted to investigate for first time the nature of the ozone treatment and its cellular targets. This work will enable to further enhance understanding of the mechanism of action of ozone treatments to bactericidal effects.

Materials and methods

Bacterial strains and cultural conditions

The bacterial strains used in this study were ΔsoxR (E. coli JW 4024), ΔsoxS (E. coli JW 4023), ΔoxyR (E. coli JW3933), ΔrpoS (E. coli JW 5437), ΔdnaK (E. coli JW0013) mutants and their isogenic parent E. coli BW 25113 (Baba et al. 2006). All strains were obtained from the National BioResource Project, Japan (NIG, Japan). Strains were maintained as frozen stocks at −70°C in the form of protective beads, which were plated onto tryptic soy agar (TSA; Scharlau Chemie, Barcelona, Spain) and incubated overnight at 37°C to obtain single colonies before storage at 4°C. Working cultures were prepared by inoculating a single colony into tryptic soya broth without glucose (TSB-G) (TSB without dextrose; Scharlau Chemie) followed by overnight incubation at 37°C.

Preparation of cell suspensions

Cells grown in TSB-G were harvested by centrifugation at 8713 g for 10 min at 4°C. This temperature was chosen to create a microbial environment that mimics storage of food products at refrigeration temperatures. The cellpellet was washed twice with sterile saline, 0·85% (Sodium chloride; Scharlau Chemie). It was then re-suspended in saline, and the bacterial density was determined by measuring absorbance at 550 nm using McFarland standard (BioMérieux, Marcy-l’Etoile, France). The inoculum was then suspended in saline to obtain c. 108 CFU ml−1.

Ozone treatment

Ozone gas was generated using a corona discharge ozone generator (Model OL80; Ozone services, Burton, BC, Canada). Oxygen was supplied via air cylinder (Air Products Ltd, Dublin, Ireland), and the flow rate was controlled using a flow regulator. Escherichia coli strains that were suspended in saline were subjected to ozone treatment by passing ozone gas through the cell suspension in an ozone bubble column. A flow rate of 0·06 l min−1 with an ozone concentration of 6 μg ml−1 was applied for each treatment. The chosen conditions were appropriate for the collection of complete kinetic responses (up to 240 s) to characterize accurately the microbial resistance of the studied strains. The ozone concentration input was recorded using an ozone analyzer (built in ozone module OL80A/DLS; Ozone Services). Excess ozone was destroyed by an ozone destroyer unit (OzoneLab™ Catalytic ozone destructor, Ozone Services, Canada), which neutralizes ozone into oxygen using an organic catalyst prior to release. Samples were removed (at time intervals of 30 s) and kept on ice until further analysis. The plating was carried out after the end of the treatment (i.e. 240 s). Three independent ozone inactivation experiments for every strain using individual freshly prepared inoculum (108 CFU ml−1) for each experiment were performed.

Determination of microbial inactivation

The effect of the ozone treatment on the microbial inactivation of each chosen strain was determined in terms of reduction in viable (culturable) counts over time. Samples (1 ml) were withdrawn, serially diluted in maximum recovery diluent (MRD; Scharlau Chemie, Spain), and 0·1-ml aliquots of appropriate dilutions were surface plated onto TSA. To obtain a low microbial detection limit for samples that low microbial counts were expected, 1 ml was spread onto TSA plates as described in EN ISO 11290-2 method (ISO 11290-2 1998). The limit of detection was 1 log CFU ml−1. Plates were incubated at 37°C for 24 h, and colony forming units were counted. Results were reported as Log10 CFU ml−1. The possibility of recovery of injured cells was taken into account by further incubating the plates for 2–3 days to detect possible increase in the formation of visible colonies.

Inactivation kinetics

The inactivation kinetics of E. coli strains in saline showed a characteristic nonlinear behaviour (refer to Results). This behaviour was described by the biphasic model and is given by following equation (Cerf 1977). The GInaFiT tool was employed to perform the regression analysis of the microbial inactivation data (Geeraerd et al. 2005). The biphasic equation was selected based on preliminary statistical comparison of the different inactivation models described in GInaFiT:


where N (CFU ml−1) is the number of micro-organisms, N0 (CFU ml−1) is the initial number of micro-organisms, f is the fraction of initial population in a major subpopulation, 1 − f is the fraction of initial population in a minor subpopulation, kmax1 and kmax2 are the parameters that determine the inactivation rate.

The estimated numerical values of f, kmax1 and kmax2 were then used to calculate the time required to achieve a reduction by 5 log cycles (t5d) using the Solver in Microsoft Excel (Microsoft Corporation, Redmond, WA, USA) by equalizing log10 (N0) − log10 (N) = 5.

In 2001, the US Food and Drug Administration (FDA) published a final rule requiring fruit juice producers to achieve a 5-log reduction in critical pathogen levels (USFDA 2001). Therefore, a reduction of 5 log was considered in this study.

For statistical analysis, the t5d values were calculated for each strain, and average values and standard deviations of the replicated studies were determined. Means were compared using anova followed by LSD testing at P < 0·05 level (spss, ver. 15.0; SPSS Inc., Chicago, IL).

Determination of cell membrane integrity

Membrane integrity was examined by determination of the release of material absorbing at 260 and 280 nm (Virto et al. 2005). Ozonated samples extracted at preset time intervals were centrifuged at 19 603 g for 20 min at 4°C. Supernatant (200 μl) was added to a 96-well plate (UV–transparent flat-bottom microplates, Corning-Costar cat. no. 3635; Fisher Scientific, Dublin, Ireland), and absorbance values at 260 and 280 nm were recorded using a UV spectrophotometer (Synergy HT; Bio-Tek, Winnooski, VT, USA). Controls included (i) E. coli in saline and (ii) saline only. These controls were chosen to assess whether there is any release of components from the cell before ozone treatment. Three independent experiments were performed, and triplicate samples were analysed. The absorbance value of cell-free supernatant of saline sample was subtracted from the simultaneously recorded absorbance values (for cell-free supernatant of untreated sample at 0 s and ozone-treated samples at 30–240 s).

Determination of cell membrane permeability

Cell membrane permeability was determined using a hydrophobic probe, 1-N-phenylnaphthylamine (NPN). The quantum yield (emission efficiency) of NPN is greatly increased in a hydrophobic environment of glycerophospholipid compared with an aqueous environment (Träuble and Overath 1973; Liu et al. 2004). Normally, NPN is excluded from the intact bacterial cell membrane. When the outer membrane structure is damaged, NPN can partition into the hydrophobic interior of the damaged outer membrane resulting in increased fluorescence, while it is otherwise excluded by outer membrane lipopolysaccharide layer of E. coli. The increase in NPN fluorescence intensity used as an indicator for increased cell membrane permeability (Tang et al. 2010). A 100 mmol l−1 stock solution of NPN in acetone was diluted to a concentration of 100 μmol l−1 in saline. Ozone-treated E. coli cultures (160 μl) were pipetted into microtitre plate wells (black; Nunc, Thermo Fisher Scientific, Ireland) to which 40 μl of 100 μmol l−1 NPN (Sigma-Aldrich, Dublin, Ireland) was added, yielding an end concentration of 20 μmol l−1 NPN. Immediately after mixing, plates were read on a Bio Tek Synergy HT fluorescence plate reader (excitation wavelength, 360/40 nm and an emission wavelength, 460/40 nm). Controls included (i) E. coli in saline with NPN and (ii) saline with NPN. Three independent experiments were performed, and triplicate samples were analysed. The fluorescence value of the untreated cell sample (control sample) was subtracted from the simultaneously recorded fluorescence value of ozone-treated samples in the presence of 20 μmol l−1 NPN.

Scanning electron microscope (SEM) imaging

Samples for SEM were prepared according to the procedure employed by Thanomsub et al. (2002) with minor modification. In detail, control (untreated) and treated samples of the parent and ΔsoxR and ΔoxyR sensitive mutant strains were collected at intervals of 0 and 30 s, as other assays indicated rapid population reduction and release of intracellular components within a period of 30 s. Samples were concentrated by centrifugation at 8713 RCF for 10 min. The supernatant was discarded, and the cells were fixed in ice-cold 2·5% glutaraldehyde in 0·05 mol l−1 sodium cacodylate buffer (pH = 7·4) for 2 h. Cells were then washed with the same buffer three times and were then fixed in 1% osmium tetroxide in 0·05 mol l−1 sodium cacodylate for 2 h at 4°C. Cells were then washed once with the same buffer followed by three washes with distilled water. Samples were dehydrated in increasing concentrations of ethanol (50, 70, 80, 90, 95–99·5%). The dehydrated samples were freeze-dried (Labconco, FreeZone 6; Mason Technology, Dublin, Ireland), mounted on stubs using double-sided carbon tape and sputter coated with Au, using a Emitech K575X Sputter Coating Unit, to prevent surface charging by the electron beam. Samples were then sputter coated at a vacuum of 5 × 10−3 mbar around 30 s resulting in a coating of 10 nm. Further on the samples were examined using a FEI Quanta 3D FEG Dual Beam SEM (FEI Ltd, Hillsboro, OR, USA) at 5 kV. SEM analysis was employed to image the damage or alteration in cell surface structure of bacterial cell population, after ozone treatment.


Effect on microbial inactivation kinetics

There was a significant effect on microbial inactivation because of the ozone exposure and for all strains the microbial kinetics exhibited biphasic behaviour (Fig. 1). The inactivation parameters were estimated by the biphasic equation, and the t5d values (Table 1) were calculated as described in the Materials and methods. The parent strain and the ΔdnaK mutant were comparatively less susceptible to ozone treatment than ΔsoxR, ΔsoxS, ΔoxyR and ΔrpoS mutants as illustrated from the t5d values (P < 0·05) (Table 1). No significant difference between the four sensitive mutant strains was recorded. However, for most of the strains, a 5-log cycle reduction was achieved within 240 s. The incubation of the plates for additional 2–3 days at 37°C did not result in increase in the number of colony forming units.

Figure 1.

 Effect of ozone at a concentration of 6 μg ml−1 on microbial inactivation kinetics of different Escherichia coli strains in saline solution. (a) E. coli BW 25113 (parent strain); (b) Δrpos; (c) ΔsoxR; (d) ΔsoxS; (e) ΔoxyR and (f) ΔdnaK.

Table 1.   The t5d (time to achieve 5-log reduction) values and relative fluorescence values for Escherichia coli strains (different letters indicate a significant difference at the 0·05 level between each strain) obtained after ozone treatment at 6 μg ml−1 in saline solution
E. coli strainkmax1kmax2RMSEt5d (s)*Relative 1-N-phenylnaphthylamine (NPN) fluorescence (after 30 s of ozone treatment)
  1. RMSE, root mean square error; kmax1 and kmax2, parameters determining inactivation rate; t5d, time required to achieve 5-log reduction.

  2. *The 1-N-phenylnaphthylamine (NPN) uptake values represent fluorescence units (±SD) after subtraction of cell control before ozone treatment.

BW 25113 (parent strain)0·3403 ± 0·1500·0214 ± 0·0010·11204·96a44 ± 8·71
ΔrpoS0·2746 ± 0·0330·0263 ± 0·0040·23129·19b40 ± 3·60
ΔsoxR0·3312 ± 0·0570·0325 ± 0·0040·2392·06b53·33 ± 12·4
ΔsoxS0·2874 ± 0·0340·0255 ± 0·0040·22117·08b45 ± 7·5
ΔoxyR0·3176 ± 0·0480·0317 ± 0·0030·19107·06b40 ± 5·40
ΔdnaK0·2478 ± 0·0270·0175 ± 0·0020·15258·61a38·9 ± 9·8

Effect on cell membrane integrity and membrane permeability

When E. coli strains were treated with ozone, the absorbance at 260 nm increased immediately after 30 s of treatment irrespective of the parent strain or mutant strains studied (Fig. 2a). The maximum absorbance at 260 nm was noticeable for ΔoxyR and ΔrpoS mutants. ΔoxyR mutant showed significantly higher absorbance values at 260 nm (< 0·05) after 30 and 60 s of ozone treatment compared with parent and other mutant strains studied. Ozone treatment of 90 s showed higher absorbance values (< 0·05) at 260 nm for ΔoxyR and ΔrpoS mutants. Ozonation for the longest treatment time (240 s) resulted in lower absorbance values for parent, ΔsoxR, ΔsoxS and ΔdnaK mutants compared with ΔoxyR and ΔrpoS mutants. The absorbance at 280 nm after ozone treatment was not significant for all the strains studied. In all cases, the 260/280 ratio remained constantly at levels equal or above 2.

Figure 2.

 Release of cellular material absorbing at (a) 260 nm and (b) 280 nm from Escherichia coli cells treated with ozone at a concentration of 6 μg ml−1 in saline solution: ◆- parent strain BW 25113; bsl00001- Δrpos; bsl00066soxR; △- ΔsoxS; ○-ΔoxyR; •-ΔdnaK.

The uptake of NPN by E. coli strains after 30 s of ozone treatment is shown in Table 1. For all E. coli strains, ozone treatment of 30 s resulted in increased NPN uptake. Further exposure to ozone did not give a significant increase in fluorescence. Additionally, there were no significant differences observed between NPN uptake values (relative fluorescence values) for different E. coli strains studied (Table 1).

SEM examination of ozone-treated Escherichia coli

All E. coli strains showed a rapid reduction in population and a release of intracellular components following 30 s of ozone treatment. Therefore, for SEM analysis, time 0 and 30 s samples were chosen, and the analysis was performed for the parent strain and two of the most sensitive mutant strains; ΔoxyR and ΔsoxR. Detailed observation of E. coli cells after SEM analysis showed slightly altered cell surface structure and damage to the cell surface but to a less extent compared with the untreated cells (Fig. 3). The surface of ozone-treated E. coli appeared slightly rough compared with the nonozonated cells.

Figure 3.

 Scanning electron micrograph of untreated and ozone-treated (at a concentration of 6 μg ml−1 in saline solution) Escherichia coli. Damaged or altered cells after ozone treatment are indicated by arrows. (a) E. coli BW 25113- untreated; (b) E. coli BW 25113- ozone treated (30 s); (c) ΔoxyR- untreated; (d) ΔoxyR- ozone treated (30 s); (e) ΔsoxR- untreated and (f) ΔsoxR- ozone treated (30 s).


In this work, for the first time, we attempted to collect information regarding the nature and the main cellular targets of ozone treatment by the use of carefully selected mutant strains. Interestingly, the same intensity of ozone treatment had different effects on mutants in genes, which have previously shown to play an important role in oxidative stress. These lead to some interesting conclusions regarding ozone treatment.

The lowest t5d values for ΔsoxR, ΔsoxS, ΔoxyR and ΔrpoS mutants highlight the importance of oxidative stress–related genes for the protection of E. coli against ozone treatment. The SoxRS regulon (superoxide response regulon) has previously been shown to play an important role in the protection against ozone treatment in E. coli (Jimenez-Arribas et al. 2001). The SoxR is the activator, which in its oxidized form enhances transcription of soxS that encodes for a transcriptional activator of several genes of the SoxRS regulon (e.g. sodA and nfo; Pomposiello and Demple 2001). These genes are directly responsible for the removal of superoxide anions or repair of superoxide-damaged macromolecules such as DNA. These could explain why both ΔsoxR and ΔsoxS mutant were sensitive to ozone treatment. In addition, this provides strong evidence that one of the main cellular targets of ozone treatment is the DNA. Previous studies also reported that ozone causes damage to DNA which if unrepaired results on extensive breakdown of DNA in E. coli and consequently loss of cell viability (Hamelin and Chung 1974; Hamelin et al. 1977, 1978). The damage of the chromosomal DNA could evidently be one of the reasons for the inactivation of E. coli by ozone (Ishizaki et al. 1987).

OxyR is another transcriptional regulator required for the induction of hydrogen peroxide (H2O2) inducible genes like katG, ahpCF grxA (Zheng et al. 1998, 2001). The interaction of H2O2 with iron localized along the phosphodiester backbone of nucleic acids leads to cell death upon exposure to H2O2 (Storz and Imlay 1999). Strains with oxyR deletions are unable to induce this regulon and are hypersensitive to H2O2 (Christman et al. 1989). Hence, the absence of the H2O2 inducible gene activator in ΔοxyR mutant resulted in increased sensitivity to ozone (refer to Table 1 and Fig. 1).

One of the most important regulators of stress genes involved in general stress resistance is RpoS. The RpoS subunit of RNA polymerase is the master regulator of general stress response in E. coli, positively regulating more than 500 (10%) genes (Hengge 2009). RpoS is known to regulate the expression of genes that are important against oxidative stress (katP, ahpCF) (Loewen et al. 1998). However, until now, it has not been shown whether it plays any role in protecting the cells against ozone treatment. Here, we demonstrate that RpoS has an important role in the resistance of E. coli against ozone. This finding can lead to further research focusing on the identification of the specific genes of the RpoS regulon, which are important for protection against ozone treatment.

The parent strain as well as the ΔdnaK mutant was less susceptible to ozone treatment than the other mutants. This observation is highly interesting as DnaK has previously been shown to play an important role in the resistance of E. coli cells against H2O2 (Delaney 1990; Rockabrand et al. 1995) mainly through the protection of proteins from oxidative damage (Echave et al. 2002). The current results suggest that proteins do not seem to be the main cellular target of ozone treatment. Additional studies could be focusing on identifying the role of DnaK, if any, against specific radicals.

Macromolecules such as proteins and lipids have different susceptibilities to oxidative damage within a range of timescales, which makes the characterization of the oxidative stress mechanism a complicated task (Semchyshyn et al. 2005). Release of intracellular components after ozonation was not observed for all E. coli strains in the current study. This is obvious from the fact that the absorbance observed at 260 nm was significantly lower than that reported previously (Komanapalli and Lau 1996; Curtiellas et al. 2005). Furthermore, the 260/280 ratio remains constantly at levels equal or above 2 which suggests that no proteins were released during ozone treatment. The above is also confirmed by the SEM results, which do not show any significant damage of the cells. Similar conclusions could be made from the results obtained with NPN assessing the cell membrane permeability. More specifically, the parent and mutant E. coli strains showed increased uptake of NPN followed by ozonation, which is normally excluded by Gram-negative bacteria. Disruption of the outer membrane weakens the bacteria and allows the permeability of large, hydrophobic molecules (Murray et al. 2009) like NPN indicating damage to the outer membrane (Helander and Mattila-Sandholm 2000). It is evident from the current results that the membrane damage by ozonation is less when compared with that reported by using chitosan (Liu et al. 2004) and lactic acid (Alakomi et al. 2000) as antimicrobial agents. Liu et al. (2004) observed the relative fluorescence value of around 400 after the addition of chitosan solution (0·5%) to E. coli suspensions. However, Alakomi et al. (2000) reported relative fluorescence values of 288 or 357 after the addition of 5 mmol l−1 or 10 mmol l−1 lactic acid and fluorescence value of only 40 after the addition of 1 mmol l−1 EDTA to E. coli suspensions. Similarly, in the current study, lower relative fluorescence values were recorded (refer Table 1) indicating less membrane damage after ozone treatment.

The inner membrane permeability can also be determined by measuring the release of β-galactosidase activity from E. coli. Previously, it has been reported that β-galactosidase activity was decreased rapidly following the exposure of ozone (Takamoto et al. 1992).

Two main theories appear in the literature about the microbial inactivation mechanism during ozonation: (i) cell lysis (Scott and Lesher 1963; Mudd et al. 1969; Hamelin et al. 1978; Hunt and Marinas 1999; Thanomsub et al. 2002), (ii) interaction with cell constituents (Ingram and Haines 1949; Hamelin et al. 1977, 1978; Ishizaki et al. 1987). According to the latter, ozone diffuses through the membrane and react with cell (vital) component. Similarly, Perrich et al. (1975) concluded that cell lysis was not the main mechanism for E. coli inactivation and that the cells remained morphologically intact after inactivation.

In the current study, and based on all the applied analysis, it was observed that cell lysis was not the main mechanism of inactivation. The ability of ozone to diffuse through the membrane appears to damage the cell constituents, thereby negatively impacting on their metabolic activity and consequently leading to the final inactivation of the cells.


From the present results, it was evident that cell lysis was not the major mechanism of inactivation shown in this study. Experiments performed with mutants in genes conferring protection against oxidative stress demonstrated the important role of the soxRS and the oxyR regulon in protection against ozone treatment. DnaK that was shown previously to play a role in H2O2 resistance did not appear to protect against ozone. The role of specific cellular targets as well as the identification of genes from the RpoS regulon playing a role in protection against ozone treatment is of further investigation. In the present work, a primary step was made to elucidate the nature of the ozone treatment and the cellular targets generally involved in the inactivation of cells by ozone.


Funding for this research was provided under the National Development Plan 2000–2006, through the Food Institutional Research Measure, administered by the Department of Agriculture, Fisheries & Food, Ireland.