Six VTEC strains were tested in this study: E. coli O157:H7 (EDL933) (ATCC 700972), O26:H11 (EC20070549), O103:H2 (EC19970811), O104:H4 (NML#11-3088), O111:NM (EC20070546), and O145:NM (EC19970355). E. coli EDL933 was included in the present study because it is prototype O157:H7 strain. Serotypes O26, O103, O111, and O145 are predominant non-O157 serotypes worldwide, therefore strains representing these serotypes have been investigated in this study. In addition VTEC O104:H4 isolated from German outbreak was selected, because this strain is highly virulent and is linked to fresh produce outbreak. All the strains except EDL933 were of human origin. E. coli O157:H7 (EDL933) was isolated from raw hamburger meat linked to an outbreak of HC. Further details regarding these strains have been published previously (Mei et al., 2015). E. coli O157:H7 (EDL933) possesses genes encoding Stx1 and Stx2, intimin gene (eae) and flagellin gene (fliC). VTEC O26:H11 (EC20070549) and O103:H2 (EC19970811) do not have stx2 genes. VTEC strains O111:NM (EC20070546) and O145:NM (EC19970355) do not have genes encoding Stx2 and flagellin. Enteroaggregative E. coli strain O104:H4 (NML#113088) does not possess genes encoding Stx1 and intimin. For simplicity, VTEC strains will be referred to by their serotype as each strain has its distinct serotype. All the strains were grown in Tryptic soy broth (TSB) at 37°C with shaking (180 rpm) for 18 h. Subsequently 0.5 mL of 18 h culture of each VTEC strain was inoculated into 50 mL pre-warmed TSB and incubated for 3 h at 37°C with shaking (180 rpm). The bacterial cells in logarithmic phase were collected, washed twice with sterile distilled water and used to inoculate lettuce samples.

Romaine lettuce was purchased from a local grocery store in Guelph. The outer leaves were removed and the remaining leaves were cut into 4 × 10 cm slices. Twenty-five grams of the leaves were weighed, placed into sterile polyethylene stomacher bags, and inoculated with 5 mL of bacterial suspension (10⁸ CFU/g). The leaves were massaged for 2 min to distribute the bacterial suspension evenly on the leaves. Subsequently, 195 mL of H₂O₂ was added to the bag to achieve a final concentration of 50 mM. The negative control sample contained lettuce leaves with 5 mL of autoclaved distilled water and 195 mL H₂O₂. The non-treated control sample contained 5 mL of bacterial suspension and 195 mL of autoclaved distilled water. The bags were incubated at room temperature for 40 min. After incubation, 1 mL of culture was taken from each stomacher bag and serially diluted with 0.1% (w/v) buffered peptone water. Subsequently, 100 μL of each dilution was spread on cefixime/tellurite—Sorbitol MacConkey agar (CT-SMAC) plates in duplicate. The plates were incubated for 18 h at 37°C, and the CFU/mL was calculated. Additionally, 4 mL samples from controls and each stomacher bag treated with 50 mM H₂O₂ were collected for RNA extraction.

Bacterial cells were concentrated by centrifugation before RNA extraction. After taking samples from controls and each stomacher bag treated with 50 mM H₂O₂, each 4 mL sample was added to 8 mL (2 volume) of RNAprotect Bacteria Reagent (Qiagen). The samples were centrifuged for 10 min at 5000×g and supernatants were decanted prior to RNA extraction. RNA was isolated using the RNeasy Mini Kit (Qiagen, Mississauga, ON, Canada), following the manufacturer’s instructions for Gram-negative bacteria. Contaminating genomic DNA was removed from each RNA preparation using the Turbo DNA-free^TM kit (Ambion, Cambridge, UK), according to the manufacturer’s instructions for DNase treatments. Total RNA concentration was determined using a Thermo Scientific Nanodrop 2000 (ON, Canada). Subsequently, 0.3 μg of RNA was reverse-transcribed using SuperScript III (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. A no reverse transcriptase control (No RT) was included for each DNase-treated RNA sample.

All stress-associated and virulence genes tested in this study are described in Table 1. Transcription of key stress and virulence genes (oxyR, sodA, soxR, uspA, rpoS, phoA, dps, cspA, cspC, cspE, gadA, gadB, gadW, mutS as well as eae, stx1A, stx2A, and fliC) was evaluated using a comparative real-time PCR. Each 25-μL reaction contained 1 μL of reverse-transcribed cDNA, 12.5 μL of Power SYBR^® Green PCR Master Mix, 0.25 U AmpErase^® Uracil N-Glycosylase (UNG; Applied Biosystems), 2.0 μM of each primer and 6.25 μL nuclease-free water. All primers used in the study were previously described (Mei et al., 2015). Amplification and detection were carried out on a MX3500^® Multiplex Quantitative PCR System (Stratagene, La Jolla, CA, USA) with an initial temperature of 50°C for 2 min. Following denaturation at 95°C for 10 min, reactions were cycled 40 times as follows: amplification at 95°C for 15 s, annealing at 60°C for 30 s, extension at 72°C for 30 s. Subsequent melt curve analysis involved heating the products to 95°C for 1 min, followed by cooling to 55°C for 30 s, and heating to 95°C while monitoring fluorescence. No template control and No RT were included for each assay and no Ct values were obtained for all negative controls after 40 cycles of PCR (data not shown). Several well-known candidate reference genes including 16S rRNA, tufA/B, mdh, pyrC, gatB, recA, serC, frr, rpsU, udp, mdoG, rpoA, and arcA were tested for expression stability as previously described (Mei et al., 2015). Only 16S rRNA gene expression was stable in all VTEC strains on lettuce under experimental conditions. Therefore, 16S rRNA was used as reference gene in this study to normalize the data. Serial dilutions of the cDNA template were examined with 16S rRNA primers. Using 50-fold dilutions the Ct value was around 12, and with other primers for different target genes all the Ct values were between 20 and 35. Therefore, 50-fold dilutions of cDNA samples were used in the experiment.

The effect of H₂O₂ treatment on survival and gene transcription of VTEC strains present on lettuce was investigated by at least three independent experiments. Each biological sample was run in duplicate on real-time RT PCR. Relative mRNA levels were determined according to the method described by Hellemans et al. (2007). Gene transcription data were analyzed using Student’s t-test. Data on reduction of VTEC populations on lettuce following exposure to H₂O₂ were analyzed using one-way ANOVA.

Superoxide dismutases (SODs) and catalases are employed by E. coli to respond to superoxide and peroxide stress. There are three SODs, including MnSOD (sodA), FeSOD (sodB), and CuZnSOD (sodC), and two catalases (HPI and HPII, encoded by katG and katE, respectively) in E. coli. These SODs and catalase genes are regulated by two major oxidative stress regulons, OxyR and SoxRS (Chiang and Schellhorn, 2012). In the presence of oxidative stress, OxyR can sense H₂O₂ and be converted to the oxidized form, subsequently activating transcription of the OxyR regulon genes, such as katG, dps, and oxyR, which protect the cell from H₂O₂ toxicity (Storz et al., 1990a; Farr and Kogoma, 1991; Chiang and Schellhorn, 2012). In this study, the transcription of oxyR was downregulated in all VTEC strains (Figures 1–6), this may cause a reduction in catalase expression, increasing peroxide sensitivity.

SoxRS is another key regulon triggered under oxidative stress in E. coli. It is positively regulated by superoxide-generating agents such as paraquat and constitutes a two-stage regulatory system, in which SoxR, activated as a transcriptional activator, induces the expression of soxS and the resulting increased levels of SoxS protein regulate the transcription of the various genes important for responding to oxidative stress (Demple and Amabile-Cuevas, 1991; Nunoshiba et al., 1992; Chiang and Schellhorn, 2012). The genes controlled by soxRS include sodA (Mn-containing SOD), nfo (DNA repair endonuclease IV), micF (antisense regulator of ompF), and zwf (glucose-6-phosphate dehydrogenase) (Christman et al., 1985; Greenberg et al., 1990; Tsaneva and Weiss, 1990). In the present study, following exposure of VTEC strains on lettuce to 50 mM H₂O₂, the transcription of soxR in E. coli O103:H2 slightly increased (1.5-fold), whereas, the transcription of soxR decreased to below the detectable level in O157:H7. Therefore, SodA reduction may be attributed to the reduction or only slight induction of soxR. Manchado et al. (2000) reported that the expression of OxyR-regulated genes (katG and dps) were induced when the concentration of H₂O₂ was from 1 to 100 μM, while the higher concentration (≥500 μM H₂O₂) resulted in the upregulation of soxS and sodA. That study suggests that the expression of oxyR or soxRS is dose-dependent. In addition, downregulation of soxR was also observed in VTEC broth cultures exposed to oxidative stress (Mei et al., 2015), however, the transcription of sodA was significantly (P ≤ 0.05) increased in pure cultures of all VTEC strains. The inconsistent results of expression of soxR and sodA may indicate that (1) sodA can react to superoxide stress independently; (2) SoxRS is not the sole regulator of sodA gene.

Escherichia coli contains a large CspA family, consisting of nine homologous cold inducible proteins, CspA to CspI, among which, CspA is the major one produced at 10–24°C, and is negatively downregulated by cspC. Whereas, CspC and CspE are constitutively produced at 37°C and are not temperature regulated (Yamanaka et al., 1994; Etchegaray et al., 1996; Phadtare and Inouye, 2001). It has been reported that CspC and CspE are important regulators of the expression of RpoS, a global stress response regulator, and UspA, universal stress protein A responding to various general stresses (Nystrom and Neidhardt, 1992, 1994). In addition, RpoS-controlled genes such as dps and katG are upregulated or downregulated by the overexpression or deletion of cspC and cspE (Phadtare and Inouye, 2001; Phadtare et al., 2006). Therefore, CspC and CspE play important roles in the stress response of E. coli. Upregulation of cspC and cspE in all six VTEC strains on lettuce was observed in the present study (Figures 1–6). However, uspA and rpoS were downregulated, and a minor change in the transcription of dps was observed. Interestingly, the transcription of cspC was downregulated in broth cultures of all VTEC strains exposed to 2.5 mM H₂O₂ at 37°C (Mei et al., 2015). On the other hand, in the present study cspC was upregulated in VTEC strains on lettuce treated with 50 mM H₂O₂ at room temperature. Previous investigation showed that CspC and CspE cannot be induced under some stress conditions, like 0.5 M NaCl, 0.5 M KCl, 5% ethanol, pH 10, pH 4, temperature of 15 or 50°C (Phadtare and Inouye, 2001). It is possible that a different concentration of H₂O₂ used for lettuce decontamination as well as environmental factors such as H₂O₂ released from fresh-cut lettuce leaves as well as nutrients present in lettuce samples and natural microbiota contributed to the induction of cspC and cspE. More interestingly, contrary results of transcription of cspC and cspE observed in response to H₂O₂ treatment of VTEC pure cultures and in the same strains present on lettuce, suggest that VTEC may react to the same stress differently in different environments. However, it is not possible to rule out that induction of cold stress genes in present study was more of the effect of temperature change combined with peroxide treatment. Centrifugation of bacterial cells using refrigerated centrifuge, following VTEC exposure on lettuce to H₂O₂ could influence transcription of cold shock genes. Further studies are required to gain a better understanding of the stress conditions that affect the cspC and cspE gene transcription in VTEC present on lettuce.

The gene phoA, encoding alkaline phosphatase, was induced under a phosphate-limited condition, but was not synthesized in normal growth medium (Han and Lee, 2006). The phoA is a member of pho regulon, regulated by a two-protein system PhoR–PhoP (Sola-Landa et al., 2003). The phoA was significantly downregulated in all the VTEC strains on lettuce tested in this study (Figures 1–6). Furthermore, downregulation of phoA was observed in most VTEC strains when pure cultures were treated with H₂O₂ (Mei et al., 2015). Therefore, further studies are needed to test, if H₂O₂ treatment changes the sensitivity of VTEC strains to phosphate-limited condition.

The H₂O₂ can cause lethality of the bacterial cells through several mechanisms. It has been proposed that the primary cause of cell inactivation by H₂O₂ or other oxidative agents is DNA damage (Storz et al., 1990b). Thus, it is important for the cells to cope with stresses by inducing the production of a variety of DNA repair enzymes as well as catalases. Glutathione also acts to protect cells from oxidative stress. In addition, Dps—the DNA binding protein from starved cells, protects cells during environmental stresses, including oxidative stress and nutritional deprivation (Almiron et al., 1992; Calhoun and Kwon, 2011). Dps protects cells from harmful oxidative radicals by DNA binding, iron storage, and by binding and oxidizing Fe ions at ferroxidase centers. Furthermore, Dps may regulate the expression of DNA repair enzymes and catalases necessary for stress resistance. Zheng et al. (2001) reported that the expression of dps considerably increased when the cells were exposed to 1 mM H₂O₂ for 10 min. However, only minor upregulation of dps was observed in this study for most of VTEC strains on lettuce (Figures 1–6).

MutS is a member of methylation-dependent MMR system which helps to maintain chromosome stability in E. coli. MutS binds to the mismatches and initiates the long-patch MMR on daughter DNA strands (Modrich and Lahue, 1996; Schofield and Hsieh, 2003). MMR has also been shown to be involved in the repair of oxidative DNA damage which causes spontaneous lesion 7,8-dihydro-8-oxo-guanine (8-oxoG or GO; Michaels and Miller, 1992; Tchou and Grollman, 1993; Lu et al., 2001). Studies have shown that the amount of MutS in E. coli remarkably decreased when cells were in the stationary phase and under starvation stress and the expression of MutS repair protein was negatively regulated by the RpoS and Hfq global regulators (Feng et al., 1996; Tsui et al., 1997; Li et al., 2003). In the present study, mutS was upregulated following exposure of VTEC on lettuce to H₂O₂ (Figures 1–6). This upregulation may be caused by significantly downregulated expression of RpoS, which is a negative regulator of mutS.

Food-borne pathogenic E. coli must be able to survive in the extremely acidic environment of the stomach and resist very low pH (1.5–3.0) for several hours (Peterson et al., 1989). Previous studies have shown that a total of 12 genes comprise an acid fitness island, including a glutamate decarboxylase (GAD) system and three transcriptional regulators (GadE, GadX, and GadW) of the GAD enzymes, which renders E. coli the ability to survive strong acidic stress (Hommais et al., 2001, 2004; Ma et al., 2002; Bergholz et al., 2007; Mates et al., 2007). E. coli produce two isozymes of GAD encoded by the gadA and gadB genes, which are induced by GadX at any pH, while GadW represses expression of gadX. GadW activates the expression of gadA and gadB only in the absence of GadX (Ma et al., 2002; Tucker et al., 2003). In addition, it has been reported that the expression of gadA and gadBC were induced when bacteria were cultured in acidified medium, treated with acetate, and during entry into stationary phase (Selinger et al., 2000; Arnold et al., 2001; Tucker et al., 2002). However, the expression of regulators GadX and GadW, under the same conditions, were unknown. In the present study, gadA, gadB, and gadW were significantly downregulated in all six VTEC strains on lettuce (Figures 1–6). We suspect that the expression of gadX was also downregulated in response to H₂O₂, which subsequently resulted in the downregulation of gadA and gadB. Furthermore, the results indicate that following H₂O₂ treatment, VTEC on lettuce may become more sensitive to acid stress, which decreases the viability of pathogens under low pH.

Production of Stx is the definitive virulence factor of VTEC O157:H7 and non-O157 serotypes (Croxen et al., 2013; Scheutz, 2014; Smith et al., 2014; Franz et al., 2015). Stx produced by VTEC can be classified into two types Stx1 and Stx2. There are three subtypes of Stx1 (a, c, d) and seven subtypes of Stx2 (a, b, c, d, e, f, g). Although both toxins could cause bloody diarrhea and HUS, a specific subset of stx2 subtypes (stx2a, stx2c, and stx2d) have a higher association with HC and HUS than stx1 subtypes or other stx2 subtypes (Scheutz, 2014). Stx are encoded on bacteriophage genomes that are integrated into bacterial chromosome. As such, the biology of the Stx-encoding phages influences the expression of stx1 and stx2. Production and release of toxins depend on induction of these bacteriophages (Croxen et al., 2013; Scheutz, 2014). Amongst the VTEC strains tested in this study, E. coli O157:H7 produced both Stx1 and Stx2, E. coli O104:H4 produced only Stx2, and the remaining four strains produced only Stx1. In this study, both stx1A and stx2A genes were dramatically downregulated in all VTEC strains on lettuce following exposure to H₂O₂ (Figures 1–6).

Many VTEC strains produce attaching and effacing (AE) lesions, which are controlled by the locus of enterocyte effacement (LEE; Croxen et al., 2013; Scheutz, 2014). Although AE lesions are not essential for bloody diarrhea and HUS, the majority of strains implicated in these syndromes are LEE-positive. Intimin, encoded by eae (E. coli attaching and effacing) gene, is responsible for intimate adhesion of VTEC to the intestinal epithelium and formation of AE lesions. Intimin binds to the cell receptor Tir, which is translocated by the pathogen to the enterocyte via a type III secretion system (Croxen et al., 2013). The presence of eae gene is strongly correlated to the presence of genes encoding many other virulence factors (Franz et al., 2015). In addition, VTEC strains with eae and stx2 genes have been associated with HUS and bloody diarrhea (Scheutz, 2014). In the present study, the transcription of eae (except for E. coli O104:H4 which is LEE-negative) changed only slightly following H₂O₂ treatment (Figures 1–6). Thus, further studies about eae regulation and the intimin-related virulence of VTEC strains under different stress conditions are needed.

Production of flagella is another contributor to the pathogenicity of VTEC (Gyles, 2007). Flagella are mainly responsible for motility, chemotaxis, and secretion of virulence factors. The key structural component of the flagellum filament is encoded by the fliC gene. In our studies, the transcription of fliC was dramatically downregulated in all four motile VTEC strains on lettuce treated with H₂O₂ (Figures 1–6). In addition, fliC was downregulated in pure broth cultures exposed to H₂O₂ (Mei et al., 2015). These results suggest that H₂O₂ treatment affects motility of VTEC strains.