Transcriptome Analysis of Escherichia coli O157:H7 Exposed to Lysates of Lettuce Leaves

Abstract

Harvesting and processing of leafy greens inherently cause plant tissue damage, creating niches on leaves that human pathogens can exploit. We previously demonstrated that Escherichia coli O157:H7 (EcO157) multiplies more rapidly on shredded leaves than on intact leaves (M. T. Brandl, Appl. Environ. Microbiol. 74:5285-5289, 2008). To investigate how EcO157 cells adapt to physicochemical conditions in injured lettuce tissue, we used microarray-based whole-genome transcriptional profiling to characterize gene expression patterns in EcO157 after 15- and 30-min exposures to romaine lettuce lysates. Multiple carbohydrate transport systems that have a role in the utilization of substrates known to be prevalent in plant cells were activated in EcO157. This indicates the availability to the human pathogen of a variety of carbohydrates released from injured plant cells that may promote its extensive growth in leaf lysates and, thus, in wounded leaf tissue. In addition, microarray analysis revealed the upregulation of numerous genes associated with EcO157 attachment and virulence, with oxidative stress and antimicrobial resistance (including the OxyR and Mar regulons), with detoxification of noxious compounds, and with DNA repair. Upregulation of oxidative stress and antimicrobial resistance genes in EcO157 was confirmed on shredded lettuce by quantitative reverse transcription-PCR. We further demonstrate that this adaptation to stress conditions imparts the pathogen with increased resistance to hydrogen peroxide and calcium hypochlorite. This enhanced resistance to chlorinated sanitizers combined with increased expression of virulence determinants and multiplication at sites of injury on the leaves may help explain the association of processed leafy greens with outbreaks of EcO157.

Leafy vegetables are the second most common vehicle associated with outbreak-related cases of food-borne illness in the United States. Indeed, Escherichia coli O157:H7 (EcO157) infections are increasingly linked to leafy produce, with 27 outbreaks of EcO157-associated disease between 1995 and 2008 (44; www.michigan.gov/documents/mda/ecoli_253034_7.pdf). Infection with EcO157 is of particular concern due to its potential to cause severe disease; 6 of 11 deaths related to food-borne illness in 2006 were linked to this bacterium, even though only 2% of all reported food-borne illnesses were attributed to EcO157 that year (13). Of the documented outbreaks associated with EcO157 between 1991 and 2002, more than half (61%) were associated with either lettuce, salad, or coleslaw (57). Since then, several large outbreaks of EcO157 infection have occurred related to contaminated bagged spinach, and bagged Iceberg and romaine lettuce (12, 13; www.michigan.gov/documents/mda/ecoli_253034_7.pdf; www.doh.wa.gov/Publicat/2008_news/08-092.htm). The relative importance of preharvest to postharvest contamination, as well as the factors that allow EcO157 to persist in a nonhost environment such as leafy vegetables, remain unclear. Although increased processing of produce to meet rising consumer demand has potentially contributed to the emergence of EcO157 outbreaks (64), this alone cannot account for the 38.6% increase in the proportion of outbreaks associated with leafy greens from 1996 to 2005, whereas the consumption of leafy greens increased only 9% (28).

We have previously reported that the growth rate of EcO157 at 28°C on lettuce leaves over short periods of time increased with the severity of tissue damage inflicted by various mechanical means (8). Notably, in only 4 h, cell populations of EcO157 increased 11-fold on shredded lettuce compared to 2-fold on intact lettuce leaves. This indicated that contamination with a single cell of the pathogen theoretically could lead to an infectious dose on shredded lettuce in this short period of time, if as few as 10 cells are required for this pathogen to cause illness in humans (http://www.cfsan.fda.gov/∼ebam). The survival and growth of EcO157 in the wounds of lettuce leaf tissue will greatly depend on its ability to adapt to physicochemical conditions, such as changes in matric and osmotic water potential and in substrate availability, as well as the presence of inhibitory compounds released passively or actively by the plant tissue. Although the pathogen may benefit from the release of nutrients from the lysed plant cells and may find physical protection within the broken tissue, other factors brought about by the plant response to mechanical injury and microbial colonization of the wound site may dictate the outcome of a contamination event at the wound site. Mechanical plant injury induces several biochemical and signaling pathways involved in the local and systemic wound response (31, 39). These include secondary metabolism pathways leading to the synthesis of antimicrobial compounds such as flavanoids, alkaloids, aliphatic acids, and phenylpropanoids that accumulate at and around the wound site and may retard the growth of microbial colonists (14, 21, 58). Upon wounding, plants produce an oxidative burst that generates reactive oxygen and nitrogen species, including superoxide (O₂⁻) and hydrogen peroxide (H₂O₂) (42, 49) and nitric oxide (NO) (26, 32).

Despite increasing evidence of the enhanced attachment, survival, and growth of EcO157 on cut surfaces of processed lettuce, even after treatment with traditional sanitizers, there is little understanding of the physiological response of the pathogen to conditions encountered in damaged plant tissue. Knowledge of the molecular events that drive the adaptation of EcO157 and its metabolism at these sites may provide insight not only into the factors that enhance its survival but also into potential biochemical pathways to target for inhibition of the pathogen with hurdle technologies. In the present study, we investigated the transcriptional profile of EcO157 in romaine lettuce leaf lysates with the aim of characterizing the response of the pathogen to the chemical conditions in wounded lettuce tissue and better understand its behavior in minimally processed lettuce. We demonstrate here that shortly after being exposed to lettuce lysate, EcO157 upregulates multiple virulence and motility genes and shifts a great part of its metabolism to enable it to survive oxidative stress, osmotic stress, and the effect of antimicrobial or toxic plant compounds. In addition, we show that this adaptation to the presence of reactive oxygen species resulting from damaged plant cells imparts the pathogen with increased resistance to two sanitizers, namely, hydrogen peroxide and calcium hypochlorite.

MATERIALS AND METHODS

Preparation and inoculation of lettuce juice lysates and shredded lettuce.

Romaine lettuce, Lactuca sativa Cos, obtained from commercial suppliers was used for all experiments described herein. Leaves that were positioned in the middle of the rosette and were external to the tightly closed head of the romaine lettuce were used. These represented approximately the eleventh to the fifteenth leaf in age and are named hereafter “middle leaves.” Although young romaine lettuce leaves support great population sizes of EcO157, as previously reported by Brandl and Amundson (9), middle leaves are more similar physiologically and thus were likely to generate less variability between leaves as well as between lettuce heads in the chemical composition of the lysates. The middle leaves were rapidly crushed by hand using a mortar and pestle and immediately filtered through an 8-μm-pore-size filter (grade 40, Ashless; Whatman). The filtrate was centrifuged at 12,000 × g for 5 min to pellet chloroplasts and plant debris. For microarray analysis, quantitative reverse transcription-PCR (QRT-PCR) and measurement of reactive oxygen species, the leaf lysate supernatant was used immediately after centrifugation. For growth curve experiments in lettuce leaf lysate, the supernatant was sterilized by passage through a 0.45-μm-pore-size syringe filter (Fisherbrand). To quantify the growth and gene expression in EcO157 cells exposed to wounded lettuce tissue per se, lettuce leaves were finely shredded into 2-mm wide strips with a serrated knife and then cut again crosswise, resulting in lettuce pieces of ∼4 mm². This fine shredding ensured that a great part of the tissue that the pathogen was exposed to was indeed damaged tissue.

Bacterial strains and growth conditions.

A spontaneous rifampin-resistant mutant of Escherichia coli O157:H7 strain EDL933 (ATCC 43895) was selected from Luria-Bertani (LB) agar plates containing 100 μg of rifampin (MP Biomedicals)/ml, incubated at 37°C, and designated MB211. Strain EDL933 grew and survived in shredded lettuce very similarly to EcO157 strain H1827, the causal agent of a lettuce-linked outbreak (8; data not shown). Gene knockout mutants of E. coli O157:H7 strain MB211 were generated by using the λ red recombinase system (20) by replacing the target gene in strain MB211 with a kanamycin resistance cassette (Table 1) . Bacterial strains expressing the green fluorescent protein (GFP) were constructed by transformation with the plasmid pGT-KAN, as described previously (10). For growth curves and inoculation onto lettuce or into lettuce leaf lysates, all strains were grown overnight in M9 minimal medium supplemented with 0.2% glucose (M9-glucose) and 100 μg of rifampin/ml on a rotary shaker at 28°C.

TABLE 1.

Bacterial strains examined in this study

Strain	Genotypea	Source
MB211	E. coli O157:H7 EDL933, spontaneous Rifr	This study
MB630	MB211 marA::kan, Rifr Kanr	This study
MB634	MB211 marR::kan, Rifr Kanr	This study
MB648	MB211 nemA::kan, Rifr Kanr	This study
MB650	MB211 nemR::kan, Rifr Kanr	This study
MB652	MB211 ycfR::kan, Rifr Kanr	This study
MB654	MB211 yqhD::kan, Rifr Kanr	This study
MB657	MB211 oxyR::kan, Rifr Kanr	This study

^aRif^r, rifampin resistance; Kan^r, kanamycin resistance.

For growth curve experiments conducted in lettuce leaf lysates, overnight cultures were first washed twice with potassium phosphate buffer (10 mM, pH 7.0; KP buffer) before transfer to the freshly prepared lysate. For experiments measuring early gene expression in lettuce leaf lysate (by microarray and QRT-PCR) or on shredded lettuce (by QRT-PCR), overnight cultures were transferred to fresh M9-glucose and grown for several hours into the mid-log phase of growth at 28°C and 150 rpm and then washed twice with KP buffer before inoculation. The lysates were inoculated with EcO157 cells in the mid-log phase of growth in minimal medium in order to isolate the bacterial responses to romaine lettuce lysates from changes in gene expression solely caused by the transition out of stationary phase (6). Lysates were inoculated at 5 × 10⁶ CFU/ml for growth experiments and at 10⁸ CFU/ml for microarray analysis and QRT-PCR and then incubated at 28°C with shaking at 150 rpm. In order to evaluate gene expression in EcO157 in lettuce lysates, samples for RNA extraction and subsequent microarray or QRT-PCR analysis were taken at 15 or 30 min after exposure of mid-log-phase EcO157 cells to freshly prepared lysates. Short incubation periods in the lysates at 28°C were used in order to characterize the early response of the pathogen to fluids leaking out of leaf cells after injury occurred, at an ambient daytime temperature that would be present in the field during growth and harvesting, or during processing under conditions that would fail to maintain cool temperatures. Two and three replicate suspensions in lysates were prepared for growth curves and for microarray analysis and QRT-PCR, respectively. All experiments were performed in duplicate with different lettuce heads and inocula on separate days.

For comparative growth studies of the parental and mutant strains on shredded lettuce and for collection of bacterial RNA to perform QRT-PCR, samples consisted of 10 and 100 g, respectively. All samples were inoculated at 5 × 10⁶ CFU/g and incubated at 28°C for 30 min. The bacteria were then recovered from the shredded lettuce by stomaching as described below and either dilution plated to measure population sizes or handled for subsequent RNA extraction and QRT-PCR. Two replicate bags of shredded lettuce were prepared for the above experiments and three replicate wells per RNA sample (i.e., per replicate bag) were used to perform QRT-PCR and compute mean gene expression ratios.

Recovery and measurement of bacterial populations.

Samples of EcO157 suspensions in lettuce lysates were diluted in 10 mM KP buffer, dilution-plated immediately with an automated plater (Autoplate 4000; Spiral Biotech) onto LB agar containing 100 μg of rifampin/ml, and the plates were incubated overnight at 37°C. For shredded lettuce, 100 ml of KP buffer was added to each bag, which then was placed in a stomacher (Seward) on high for 2 min. The resulting leaf washings were dilution plated as described above to estimate population sizes of the pathogen and mutants or were immediately filtered with a 20-μm-pore-size nylon filter (Millipore) to remove plant debris and collect bacterial cells for RNA extraction in the filtrate.

RNA extraction procedures and QRT-PCR.

Ice-cold phenol-ethanol (5%:95%) solution was immediately added to suspensions in leaf lysates or to filtered leaf washings, and the mixture was incubated on ice for 30 to 60 min. The cells were then centrifuged, and the pellet was stored at −80°C. RNA extraction was performed with the Promega SV Total RNA kit according to the manufacturer's instructions, except that bacterial pellets were first treated with 50 mg of lysozyme (Fisherbrand)/ml and 1 U of anti-RNase (Ambion)/μl. Total RNA was quantified with a Nanodrop ND 1000 spectrophotometer (Thermo Scientific), examined for quality on an Agilent Bioanalyzer, and stored at −80°C until used for microarray analysis or QRT-PCR. Before QRT-PCR, total bacterial RNA was treated with Turbo DNase I (Ambion), and the absence of DNA was confirmed by QRT-PCR in the absence of reverse transcriptase using primers for gyrB (Table 2). All QRT-PCR was performed with a Brilliant II SYBR green QRT-PCR 1-Step kit (Stratagene) on an MxPro 3000P Cycler (Stratagene). For each gene, the ratio of expression in EcO157 in the lysates or on shredded lettuce compared to that in M9 culture was normalized to the expression of gyrB based on the equation by Pfaffl in 2001 (52a). Lysates were tested for the presence of lettuce RNA by using the forward primer 5′-ATCTGCGGACAACCAATGAG-3′ and the reverse primer 5′-CACTAAAACGGGGAGGAATG -3′ directed to the L. sativa chloroplast RNA polymerase beta subunit RNA (GenBank Accession YP_398317). Only insignificant amounts of lettuce RNA were detected; these were estimated to have minimal effect on calculated expression ratios of EcO157 genes.

TABLE 2.

RT-PCR oligonucleotide primers

Gene	Primer sequence (5′-3′)		Source or reference
	Forward	Reverse
ahpF	GCAGATTCGCCATATTGACG	GCCCTGACCAAACTCTTTCC	This study
gloA	ACTCACTGGCGTTTGTTGG	GACCGGCGTCTTTCTCTTC	This study
grxA	CGGGTTGCCCTTACTGTG	CGGGTTTACCTGCCTTTTG	This study
gyrB	GCAAGCCACGCAGTTTCTC	GGAAGCCGACCTCTCTGATG	This study
hslO	TGCAGGCGGTATGTTGTTG	CGTTGTTGCGGATTTCAGC	This study
katP	CGGGAAACTTCAGAAACCTC	GCCACAGTCTCCTCATCATC	This study
marA	CGAGGACAACCTGGAATCAC	TGCGGCGGAACATCAAAG	18
marR	CGCGGCGTGTATTACTCC	GGTTCGGCAACCTTTCTACC	This study
nemA	GATTTCCGTCAGGCTATTGC	CGCGTTTACCCAGTTGTTC	This study
nemR	CCATTACGCCACATATCACC	TATCACGGCCATTTTCCAG	This study
oxyS	GAGCGGCACCTCTTTTAACCCTTG	CCTGGAGATCCGCAAAAGTTCACG	56
soxS	GTCGTCGCAAAAAAATCAGG	TGGGAGTGCGATCAAACTG	This study
ycfR	TAAGCTCCATGTCATTTGCC	TTCCATGGAGGGTATTCGG	This study
yfiA	GACGCCACAATCAATACACC	CGTCTTTCACCGATGTTGC	This study
yqhC	CGCGTGTTTCGTTATGATGC	TTCGCGGATGATCTGTTTG	This study
yqhD	GGCAGCGTGAAAAAAACCG	AACCCCCAGCAGGAAAGTC	This study

Microarray procedures.

Microarray procedures for gene expression profiling were based on previously described methods (15, 25, 41, 72). Briefly, PCR products were spotted onto Ultra-GAPS glass slides (Corning) by using an OmniGrid Accent arrayer (GeneMachines), and DNA was cross-linked to the slides via UV radiation. Each array contained 4,262 open reading frames (ORFs) from E. coli K-12 strain MG1655 supplemented with the 1,125 ORFs from E. coli O157:H7 strain EDL933 (including 25 from plasmid pO157) that are not present on the K-12 genome, similar to a previously published array design (41). The coding portion of the EcO157 chromosome, including the 25 protein-coding genes found on the pO157 plasmid, was examined (4,745 ORFs). Microarray analysis was based on competitive hybridization between Cy5-DNA and Cy3-cDNA. Genomic DNA (2 μg) was labeled via dye incorporation of Cy5-dCTP (GE Healthcare) using Klenow fragment (New England Biolabs). Total RNA (20 μg) from each experimental condition and each biological replicate was labeled via incorporation of Cy3-dCTP (GE Healthcare) into a cDNA product with the Fairplay III microarray labeling kit (Stratagene), based on previously described methods (15, 25, 41). Hybridization of Cy5-DNA and Cy3-cDNA was carried out overnight at 42°C as previously described (15, 25, 41). After hybridization, the slides were scanned on a GenePix 4000B scanner (Axon Instruments; Molecular Devices), and individual spots were analyzed with GenePix Pro 6.0 software (Axon Instruments). Normalization was conducted as previously described (72). Briefly, spots with a reference signal lower than background plus two standard deviations or spots covered by an obvious blemish were excluded. Local background was subtracted from all spots, and a Cy3/Cy5 ratio was calculated. Further normalization to account for differences in dye incorporation included data centering by setting the median natural logarithm to zero for each group of spots in one sector (printed by one pin).

Three biological replicates were collected for each experimental condition on two separate days for a total of six replicates, and three separate arrays on different slides of were hybridized for each biological replicate (technical replicates). Technical replicates were analyzed for outliers with the Dixon “Q” test (23) using a critical value of 0.1, and values lying outside of the cutoff were discarded. All remaining technical replicate values were averaged to obtain the value for each biological replicate, and the data from all six biological replicates were further tested by using an unpaired t test with unequal variance using Genespring 10 (Agilent/Stratagene). Genes showing a >2-fold upregulation or downregulation and a Benjamini-Hochberg false discovery rate-adjusted P value of ≤0.05 were considered to be differentially regulated in leaf lysates compared to M9 cultures. The expression of a subset of genes of interest was confirmed by QRT-PCR.

Measurement of reactive oxygen species and challenge with H₂O₂ or calcium hypochlorite.

Lettuce leaf lysates were tested for the presence of reactive oxygen species. Amplex Red (N-acetyl-3,7-dihydroxyphenoxazine; Invitrogen) was added to the lysates at a final concentration of 50 μM in order to detect and quantify H₂O₂ by measuring the increase in fluorescence at 590 nm after excitation at 530 nm (45, 75), using a Synergy HT spectrofluorometer (BioTek) after a 30-min incubation of the assay reaction. All assays were carried out in duplicate by using clear-bottom, black-sided, 96-well assay plates (Corning). A standard curve was prepared with 3% certified H₂O₂ (Fisherbrand) in 50 mM KP buffer and 1 U of horseradish peroxidase (HRP; MP Biomedicals)/liter. Blank fluorescence measurements were taken from wells containing only buffer, Amplex Red, and HRP; this fluorescence value was set to zero for the standard curve. The concentration of H₂O₂ in lettuce leaf lysates was then estimated based on a best-fit linear regression line generated with the Gen5 Software, version 1.01.9 (BioTek). Catalase (Sigma) was added at 2 mg/ml to a subset of wells in order to confirm the specificity of H₂O₂ detection; an 83 to 89% reduction in H₂O₂ concentration was noted, compared to the 62 to 63% reported previously (45).

Studies of EcO157 survival to H₂O₂ were conducted in a solution of 8 mM H₂O₂ (in 0.5 mM KP buffer) at 28°C and 150 rpm. Challenge studies with calcium hypochlorite (Spectrum) were conducted in distilled deionized water, and the level of total and free chlorine was assessed as 4 and 2ppm, respectively, based on the Aquacheck test (Hach). After a 5-min exposure of the bacterial cells to chlorine, free chlorine was neutralized by the addition of a solution of sodium thiosulfate at a final concentration of 0.1% (wt/vol) (59), and the suspension was dilution plated onto LB agar containing 100 μg of rifampin/ml to measure the surviving bacterial population sizes.

RESULTS

Growth of EcO157 in lysates of romaine lettuce juice.

We previously reported that EcO157 multiplied faster on damaged lettuce leaves than on intact leaves, with greater growth corresponding to greater leaf damage (8). In order to create a uniform experimental condition that would reflect the chemical environment encountered in microsites at the cut surface of the leaves, the response of EcO157 to romaine lettuce lysates was tested. Over a period of 24 h, the growth pattern of EcO157 in the lysates was similar to that observed in M9-glucose (Fig. 1). After inoculation into the lettuce lysates, EcO157 cell concentrations remained stable for the first 4 to 5 h of incubation, after which the pathogen multiplied with a doubling time of 1.2 h to achieve a concentration of ca. 2 × 10⁹ cells/ml in stationary phase (Fig. 1). The 4- to 5-h lag phase and period of adaptation in the lysates was observed whether the inoculum cells were prepared from mid-log-phase or stationary-phase cultures in M9-glucose (Fig. 1). The final concentration of EcO157 in lettuce lysates after 24 h of incubation at 28°C was very similar to that in M9-glucose under the same conditions.

FIG. 1.

Growth of EcO157 in M9-glucose medium and romaine lettuce lysates. All inoculum cells were grown in M9-glucose. Freshly prepared lettuce lysates were inoculated with stationary-phase cells grown overnight (▪) or with cells in the mid-log phase of growth (□). As a reference for growth rate, stationary-phase cells were inoculated also into M9-glucose (▴). The data represent the mean cell concentrations of two replicate cultures. Error bars represent one standard deviation of the mean.

Transcriptional response of EcO157 to lettuce lysates.

The global transcriptional response of EcO157 to romaine lettuce lysates was assessed at 15 and 30 min postinoculation in comparison to that of cells in the mid-log phase of growth in M9-glucose medium. At both sampling times, one-fifth of the genome showed differential expression of at least twofold (P < 0.05). At 15 min, 10.3% (487/4,745) and 11.3% (534/4,745) of the genes were upregulated and downregulated, respectively; similarly, 10.4% (494/4,745) and 9.4% (448/4,745) of the genes were upregulated and downregulated, respectively, at 30 min (Fig. 2). Differentially regulated genes were categorized by Clusters of Orthologous Groups (COG) designations, and the percentage of genes differentially regulated in each category was compared to the overall percentage at each time point (Fig. 2). Categories that showed a notable over-representation of upregulated genes at both time points included those related to cell motility, intracellular trafficking and secretion, and numerous genes lacking a COG designation (unclassified). Among the notable categories of genes that were downregulated in the lysates was a large number of genes associated with amino acid and nucleic acid transport and metabolism, as suggested by the disproportionate downregulation of their respective COG categories (Fig. 2).

FIG. 2.

Grouping into functional categories of differentially regulated genes in ECO157 in romaine lettuce lysates compared to M9-glucose medium, as determined by microarray analysis. The categories of orthologous genes (COG) were used for grouping. Bars represent the percentage of genes with decreased or increased expression in a given category after a 15-min or a 30-min exposure to lettuce lysates.

(i) Induction of flagellar machinery, TTS, and fimbrial genes.

Differentially expressed genes with a COG designation for cell motility or secretion included both flagellar and fimbrial genes, as well as structural genes for two separate type III secretion systems (Table 3). The locus of enterocyte effacement (LEE) pathogenicity island associated with attaching-and-effacing lesions in both enteropathogenic and enterohemorrhagic E. coli has been divided into five separate operons designated LEE1, LEE2, LEE3, LEE5, and LEE4 (36). A large number of the structural genes that encode the type III secretion (TTS) apparatus were upregulated in response to lettuce lysates (Table 3). The LEE effector proteins are located primarily in the LEE4 operon, and the genes for these proteins have been reported to be constitutively transcribed in EcO157 (61). Nevertheless, one of these genes, espF, was differentially expressed in our two test conditions (Table 3). A second TTS apparatus is encoded by EcO157 that has homology to the Salmonella pathogenicity island I (43); the structural genes in this second TTS system were also upregulated in response to lettuce lysates (Table 3). Recently, a survey of the EcO157 genome has identified 49 additional potential effector proteins (65). Of these putative effectors, nine were expressed at higher levels in lettuce lysates than M9-glucose conditions, although three of these nine may be pseudogenes (65). Finally, under this category, a number of genes involved in cell motility (che, fhi, flg, flh, and fli) and/or adhesion (fim and others) had increased transcription compared to that in M9-glucose (Table 3).

TABLE 3.

EcO157 regulons and virulence factors upregulated in lettuce lysates as determined by microarray analysis

Categorya	Gene	Time point		Function
		15 min	30 min
Pathogenicity island
LEE1 operon	Z5137	17.1	5.3	Putative TTS protein
	Z5136	4.4	4.3	Putative TTS protein
	escR	16.8	50.7	EscR; TTS apparatus protein
	escU	10.0	10.0	EscU; TTS apparatus protein
LEE2 operon	cesD	16.4	11.7	CesD; TTS system chaperone
	escC	18.1	22.1	EscC; TTS apparatus protein
	Z5125	4.2	3.9	SepD; TTS protein
	escJ		6.2	EscJ; TTS apparatus protein
	Z5123	3.7	4.1	Putative TTS protein
LEE3 operon	escN		2.2	EscN; TTS apparatus protein
	sepQ	3.4	7.4	SepQ; TTS apparatus protein
LEE5 operon	Z5111	13.9	10.6	Putative cesT (tir chaperone)
	sepL	25.8		SepL; TTS apparatus protein
LEE4 operon	espF	6.3		EspF; type III effector protein
Other TTS (SPI-1 homologs	Z4180		20.8	Putative lipoprotein TTS apparatus; prgK homolog
Salmonella)	Z4182	35.4	90.8	Hypothetical protein: prgH homolog
	Z4187		16.5	TTS apparatus protein; spaR homolog
	Z4189	6.4		Putative integral membrane protein TTS apparatus; spaP homolog
	Z4190	7.6	5.7	TTS system apparatus protein; spaO homolog
	Z4191		13.3	TTS apparatus protein; invJ homolog
	Z4195	10.8	14.5	TTS apparatus protein; invA homolog
	Z4196	4.7		Putative secreted protein; invE homolog
	Z4197		11.0	TTS apparatus protein; invG homolog
	Z4198	17.0	13.0	Putative regulatory protein for TTS apparatus; invF homolog
Putative effector proteins	Z0025	5.1		Hypothetical protein
	Z2240	8.6	22.5	Hypothetical protein
	Z2241	3.6		Hypothetical protein
	Z2242	8.7	8.9	Hypothetical protein
	Z2075		2.3	Unknown protein encoded by prophage CP-933O
	Z4329		2.2	Hypothetical protein
	Z5212	2.8	3.6	Hypothetical protein
	Z6024	6.5	5.1	Unknown protein encoded by cryptic prophage CP-933P
	Z3023		4.1	Putative secreted protein (distant YopM homolog)
Flagellar regulon (by operon)
Class 2	fliP		5.1	Flagellar biosynthesis protein (fliLMNOPQR)
	fliQ	10.4	17.6	Flagellar biosynthesis protein
	fliE	3.6		Flagellar basal body protein (fliE)
	fliJ	4.6		Flagellar protein (fliFGHIJK)
	flgB	3.1		Flagellar basal body rod protein (flgBCDEFGHIJKL)
	flgC	3.4	3.6	Flagellar basal body rod protein
	flgD		8.5	Flagellar biosynthesis, initiation of hook assembly
	flgE		2.9	Flagellar hook protein
	flgG	38.4		Flagellar biosynthesis, cell-distal portion of basal body rod
	flgH	5.6		Flagellar L-ring protein precursor
	flgI	2.9		Flagellar P-ring protein precursor
	fhiA	6.8	11.3	Flagellar biosynthesis (FlhA homolog)
	flhE	4.6	14.2	Flagellar protein (flhBAE)
	fliZ		6.5	Flagellar biosynthesis protein (fliAZY)
Class 3	cheB	4.3	5.4	Chemotaxis-specific methylesterase (cheRBYZ)
	cheW	2.8		Positive regulator of CheA protein activity (cheAW)
Fimbrial genes (by operon)b	yadM	3.4	5.7	loc2/putative fimbrial protein
	ybgP	12.8	7.4	loc4/putative chaperone
	ycbQ	5.2		loc5/putative fimbrialike protein
	Z1288 m	3.9	2.7	loc5/PapC-like porin protein involved in fimbrial biogenesis
	Z1536		3.4	loc6/putative usher protein
	Z3276	3.4		loc9/putative fimbrial protein
	yehB	5.5	7.9	loc9/putative outer membrane protein
	Z3596	4.4	3.3	loc10/putative minor fimbrial subunit
	Z3598	8.5	16.7	loc10/putative minor fimbrial subunit
	Z3600	3.9		loc10/putative fimbrial usher
	Z3601	6.0	9.0	loc10 /putative major fimbrial subunit
	yraH		2.2	loc11/putative fimbria-like protein
	Z4501	3.0		loc11/hypothetical protein
	Z4965	10.8	15.8	loc12/putative fimbrial subunit
	Z4968 m	3.7	3.3	loc12/PapC-like porin protein involved in fimbrial biogenesis
	Z4969	3.8		loc12/putative fimbrial chaperone
	Z5221		3.5	loc13/putative fimbrial protein
	Z5225	9.9	12.3	loc13/putative major fimbrial subuit
	fimA	4.2		loc14/major type 1 subunit fimbrin (pilin)
	fimI	2.7		loc14/fimbrial protein
	fimF	4.1		loc14/fimbrial morphology
	fimZ		3.8	loc3/fimbrial Z protein; probable signal transducer
	ppdD	2.2		ppdD/prelipin peptidase-dependent protein
OxyR regulon (oxidative	ahpC		8.0	Alkyl hydroperoxide reductase, C22 subunit
stress)	ahpF	2.9	7.6	Alkyl hydroperoxide reductase, F52a subunit
	grxA	8.7	29.5	Glutaredoxin 1 redox coenzyme
	katG		18.1	Catalase; hydroperoxidase HPI(I)
	trxC	3.2	9.4	Putative thioredoxin-like protein
	katP		31.3	EHEC-catalase/peroxidase
Other oxidative stress related	cydC	87.2		Cysteine/glutathione ABC transporter membrane/ATP-binding component
	fhuD	101.4		Hydroxamate-dependent iron uptake, cytoplasmic membrane component
	yeeD	13.0	22.1	Predicted redox protein, regulator of disulfide bond formation
	yeeE	21.4	16.5	Putative transport permease protein
	Z2347		35.4	Putative copper-zinc superoxide dismutase prophage CP-933R
	Z3312		5.6	Putative copper-zinc superoxide dismutase prophage CP-933V
	Z3974	6.5	9.2	Alkyl hydroperoxide reductase AhpD
Osmotic stress	betA	7.8	9.0	Choline dehydrogenase
	betB	5.4	6.0	Betaine aldehyde dehydrogenase
	caiA		5.1	Crotonobetainyl-coenzyme A dehydrogenase
	caiT		32.8	l-Carnitine/γ-butyrobetaine antiporter
	fixX		10.2	Putative ferredoxin
Mar regulon (antimicrobial	gatA	10.1	11.8	Galactitol-specific enzyme IIA of phosphotransferase system
resistance)	map		5.4	Methionine aminopeptidase, type I
	marA	9.4	19.2	Multiple antibiotic resistance; transcriptional activator
	marR	19.2	27.9	Multiple antibiotic resistance protein; repressor of mar operon
	ribA		3.6	GTP cyclohydrolase II
	srlA_2	3.2		PTS system IIC component
	tnaA	2.1		Tryptophanase
	yfaE		4.6	Putative 2Fe-2S ferrodoxin
Other antimicrobial	emrD	3.3	4.8	Multidrug resistance protein
	Z3494	5.8	9.0	Putative antibiotic efflux protein (norA homolog Salmonella)

^aLEE, locus enterocyte effacement; SPI-1, Salmonella pathogenicity island 1; Nle, non-LEE effector.

^bFimbrial loci designated as described previously (40).

(ii) Induction of stress-responsive genes.

Analysis of the EcO157 transcriptional profile also revealed that several genes involved in oxidative stress, osmotic stress, and antimicrobial resistance were upregulated significantly (Table 3 and Fig. 3). E. coli, including EcO157, senses hydrogen peroxide primarily through the activated (oxidized) form of the transcription factor OxyR (62) It is noteworthy that the large group of genes that are involved in resistance to oxidative stress and that had increased expression in the lysates included key members of the OxyR regulon (ahpF, ahpC, grxA, trxC, and katG) (Table 3). In addition, katP, a catalase/peroxidase gene that was recently identified as a member of the OxyR regulon (66), showed induction after the pathogen was incubated for 30 min in the lettuce lysates.

FIG. 3.

EcO157 operons related to oxidative stress (A), detoxification (B), and DNA repair (C). Genes are shown in order of location and direction of transcription in the operons, and their size is not drawn to scale. Genes with increased expression in lettuce lysates compared to M9-glucose medium are presented as solid black arrows. The upper and lower numbers within each arrow show fold upregulation at 15 and 30 min postexposure to lettuce lysates, respectively. Gene symbols above the arrows are the most recent gene names from E. coli genomes. Gene locus tags or gene symbols below the arrows, when present, indicate currently used, EDL933-specific gene designations. CRISPR, clustered regularly interspaced short palindromic repeats.

Oxidative stress damages enzymes with Fe-S clusters, releasing ferrous iron that reacts with H₂O₂ via the Fenton reaction, producing OH radicals that cause direct damage to DNA (37). Repair systems for iron-sulfur cluster synthesis and repair (isc), iron regulation (fit and fep), sulfur acquisition (cys, ssu, and tau) (Fig. 3A) and DNA repair-related genes (nrd and CRISPR-associated genes) (Fig. 3C) were all expressed at higher levels in the lettuce lysates than in M9-glucose culture. Additional genes (along with their putative corresponding transcriptional activators) that are associated with oxidative stress and were highly upregulated in response to lysates included ycfR (bhsA), a gene induced under multiple stress conditions and involved in biofilm formation (73), and yqhD, which has been linked to repair of lipid peroxidation (52) (Fig. 3B). Genes in the bet and cai/fix operons, which are associated with osmotic stress, also showed increased transcription (Table 3).

Increased expression of several genes involved in antibiotic resistance and detoxification of damaging compounds was observed (Table 3 and Fig. 3B). These included genes belonging to the multiple antibiotic resistance (Mar) regulon (including marR, marA, gatA, map, ribA, and yfaE). marRA also plays a role in increased tolerance to oxidative stress and is a member of the SoxR regulon (1, 5). Upregulated genes with a role in detoxification included the nemR (ydhM), nemA, and gloA operon, as well as the frmRAB (yaiN, adhC, and yaiM) operon, which contributes to formaldehyde degradation (Fig. 3B) (29).

(iii) Carbon utilization genes derepressed in lettuce lysates.

Carbohydrate transport in Gram-negative bacteria occurs through one of four systems: sugar-specific porins, ATP-binding cassette (ABC) superfamily proteins, phosphotransferase system (PTS) proteins, or one of several sugar-specific major facilitator superfamily (MFS) proteins (62). Among the sugar-specific ABC transporter systems upregulated in response to lettuce lysates are multiple operons in two further categories of carbohydrate uptake transporters (CUT); the CUT1 family, which transports oligosaccharides, glycerol-phosphates, and other sugars (ycjV, ycjP, and ycjO and malE), and the CUT2 family, which transports monosaccharides exclusively (Z5689, Z5691, and Z0415) (Fig. 4A) (63).

FIG. 4.

EcO157 operons related to carbohydrate uptake and metabolism from three categories: ABC transporters (A), PTS proteins (B), and MFS proteins (C). Genes are shown in order of location and direction of transcription in the operons, and their sizes are not drawn to scale. Genes with increased expression in lettuce lysates compared to M9-glucose medium are presented as solid black arrows. The upper and lower numbers within each arrow show the fold upregulation at 15 and 30 min postexposure to lettuce lysates, respectively. Gene symbols above the arrows are the most recent gene names from E. coli and Shigella genomes. Gene locus tags or gene symbols below the arrows, when present, indicate currently used, EDL933-specific gene designations. ABC, ATP-binding cassette; CUT, carbohydrate uptake transporter; PTS, phosphotransferase system; MFS, major facilitator superfamily. *, Substrate-specific binding protein.

The transcription of malE, encoding a maltooligosaccharide-binding protein, was 348-fold higher after 15 min of EcO157 incubation in the lysates than in M9 culture (Fig. 4A). This gene was also expressed at very high levels later at 30 min of incubation, but the mean expression in lysates compared to that in the control treatment was not significantly different by the t test (P > 0.05) due to high variability in transcription among all replicates. malZ, which encodes a maltodextrin glucosidase that breaks down maltodextrins to glucose and maltose (7), was increased in expression 11.3- and 23.7-fold at 15 and 30 min, respectively.

PTS genes upregulated in response to lettuce lysates included gatA, a gene encoding a component of a galactitol-specific transporter and a member of the Mar regulon (Table 3), as well as the gatR repressor gene (48), which showed a 7.8-fold downregulation at 30 min of incubation. PTS genes specific for sorbose, N-acetylgalactosamine, sorbitol/glucitol, and another galactitol import system were also upregulated (Fig. 4B).

uhpC, an MFS member and regulator of the UhpT glucose 6-phosphate antiporter, increased in expression 66.2- and 73.3-fold in lysates at 15 and 30 min, respectively. Also in the MFS, multiple genes in a sucrose transport operon, and galP, specific for galactose transport, had increased transcription (Fig. 4C). The galactose utilization gene yafB, a putative aldose reductase that may reduce galactose to galactitol, showed a 12.3-fold induction at 15 min of incubation.

Measurement of reactive oxygen species in lettuce lysate.

Because upregulation of multiple genes in the OxyR regulon was observed, the concentration of H₂O₂ present in the lettuce lysates was quantified. The Amplex Red assay revealed that immediately after preparation, the lysates contained ∼25 μM H₂O₂ and showed a steady decline in concentration to ∼5 μM after 30 min (data not shown).

QRT-PCR on select genes of EcO157 exposed to lettuce lysates and shredded leaves.

QRT-PCR on EcO157 cells recovered from inoculated lysates and inoculated shredded leaves, both incubated for 30 min at 28°C, confirmed the increased transcription observed by microarray analysis of a subset of genes of interest (ahpF, grxA, ycfR, marR, marA, nemR/Z2666, nemA, gloA, yqhC, and yqhD) in comparison with expression in M9-grown cells (Fig. 5). QRT-PCR also allowed for the detection of the increased expression of oxyS, a small noncoding RNA that is activated by the oxidized form of OxyR and that was not spotted on the arrays. For all of the genes above, the magnitude of upregulation was greater in EcO157 cells incubated in lettuce lysates than in shredded lettuce.

FIG. 5.

QRT-PCR quantification of change in expression of EcO157 genes determined to be upregulated in lettuce lysates compared to M9-glucose medium by microarray analysis. Upregulation of genes was measured 30 min after inoculation of EcO157 in lettuce leaf lysates or shredded lettuce. Each bar represents the mean upregulation in three replicate lysate samples and in two replicate shredded lettuce samples. The error bars represent one standard error of the mean.

QRT-PCR showed that genes belonging to the OxyR regulon (ahpF, grxA, oxyS, and katP) were not upregulated in cells of the oxyR mutant, in contrast to the parental strain cells, after exposure to lettuce lysates (Table 4). This loss of upregulation of ahpF, grxA, oxyS, and katP genes in the oxyR mutant confirmed that OxyR was required for the activation of these genes in lettuce lysates. Several other oxidative stress-related genes that are not known to be under OxyR control were not differentially transcribed in the oxyR mutant compared to the parental strain under the conditions described above. Therefore, identification of a redundant oxidative stress-responsive pathway that enabled EcO157 to survive in the lysates despite a mutation in OxyR was not achieved.

TABLE 4.

Change in gene expression in E. coli O157:H7 strain EDL933 and its oxyR deletion mutant in response to romaine lettuce lysate as determined by QRT-PCR

Gene	Fold increase (SD)a
	Wild type	oxyR mutant
OxyR regulon
ahpF	2.3 (0.2)	1.0 (0.0)
grxA	18.8 (5.8)	0.6 (0.2)
oxyS	10.8 (5.6)	0.9 (0.2)
katP	5.9 (0.1)	0.6 (0.0)
OxyR independent, H2O2 responsiveb
ibpA	0.6 (0.2)	0.5 (0.2)
ibpB	0.4 (0.0)	0.3 (0.0)
soxS	0.4 (0.0)	0.4 (0.1)
yfiA	0.1 (0.0)	<0.1 (0.0)
Other genes
hslO	0.5 (0.0)	0.4 (0.1)
marA	5.2 (2.6)	5.9 (1.3)
marR	5.9 (0.5)	3.9 (0.9)
nemA	52.3 (28.6)	20.5 (5.3)
nemR	27.0 (1.7)	17.6 (0.9)
norV	1.2 (0.9)	2.1 (0.8)
ycfR	62.7 (18.2)	107.6 (2.1)
yqhD	14.5 (0.1)	12.2 (1.4)

^aCalculated in reference to expression in EcO157 cells cultured in M9 medium with 0.2% glucose to the mid-log phase of growth. n = 2 biological replicates.

^bAs described elsewhere (66, 74).

Comparative behavior of EcO157 mutants in lettuce lysates and shredded leaves.

We investigated the role of EcO157 genes that showed increased expression, in the survival of the pathogen in lettuce lysates. Several EcO157 mutants were constructed, and none of the mutants in oxyR, nemA, nemR/Z2666, marA, marR, yqhD, or ycfR exhibited reduced survival or growth over 5 h in lettuce lysates compared to the parental strain. In addition, neither the marA nor the marR mutant showed a difference in behavior on shredded lettuce compared to the parental strain (data not shown).

Survival of EcO157 challenged with H₂O₂ or Ca(OCl)₂ after exposure to lettuce lysates.

In light of the numerous oxidative stress-related genes that were upregulated in EcO157 under conditions resulting from lettuce leaf damage, the effect of exposure to lettuce lysates on the subsequent resistance of the EcO157 cells to oxidative compounds was investigated. When previously grown in M9-glucose, the EcO157 parental strain and oxyR mutant decreased in concentration 2.6- and 3.7-fold, respectively, after exposure to 8 mM H₂O₂ for 70 min (Fig. 6). The mutant survived at a slightly lower rate than the parental strain in the first 40 min of incubation in H₂O₂ (Fig. 6). Most importantly, after exposure to romaine lettuce lysates for 30 min, both strains exhibited enhanced survival in response to H₂O₂ challenge compared to M9-grown cells (Fig. 6).

FIG. 6.

Time course of survival of EcO157 parental (squares) and OxyR-minus mutant (triangles) in hydrogen peroxide after preincubation in M9-glucose or in lettuce lysates. Cells were grown to exponential phase in M9-glucose (open symbols) or were incubated in lettuce lysates for 30 min (closed symbols) before exposure to 8 mM H₂O₂. The data represent the mean cell concentration of EcO157 in two replicate suspensions in H₂O₂, which were each prepared from different inoculum cultures in M9-glucose or lysates. The error bars represent one standard deviation of the mean.

In order to test whether this protective response generated by exposure to lettuce lysates was also effective against a chlorine challenge, EcO157 cells were treated with a calcium hypochlorite solution containing 2 ppm of free chlorine, both before and after exposure to romaine lettuce lysates (Fig. 7). Cells that had been primed by a 30-min incubation in lettuce lysates survived at greater population sizes after a 5-min exposure to chlorine than M9-grown cells.

FIG. 7.

Survival of EcO157 cells to calcium hypochlorite after preincubation in M9-glucose or in lettuce lysates. Cells were grown to exponential phase in M9-glucose or were incubated in lettuce lysates for 30 min, before transfer to a calcium hypochlorite solution containing 2 ppm of free chlorine. Sodium thiosulfate was added to neutralize the free chlorine after 5 min of incubation of the suspension. The data represent the mean cell concentration of EcO157 in four replicate suspensions before (▪) and after (□) exposure to calcium hypochlorite. The error bars represent one standard deviation of the mean.

DISCUSSION

The global response of EcO157 to physicochemical conditions present at wound sites in processed lettuce was modeled using leaf lysates obtained by rapidly crushing romaine lettuce leaves. Clearly, EcO157 experienced a period of adaptation after inoculation into the lettuce lysates, as revealed by its lag phase within the first 4 to 5 h of incubation. The population dynamics of the pathogen in the lysates were very similar whether the cultures in minimal medium that were used as inoculum were in the stationary or the mid-log phase of growth. Thus, the period of adaptation that we observed was probably not caused by a change in growth phase but rather by a physiological change in response to conditions prevailing in the lysates. Minimal medium was assessed as more relevant than rich medium to grow the inoculum cells because, first, its low complexity was likely closer to that of the lettuce lysates. Preliminary microarray analysis with LB-cultured cells as the inoculum indicated that a disproportionately high number of housekeeping genes were upregulated in EcO157 in the lettuce lysates in response to the transition from a rich and complex environment to a low-nutrient milieu (data not shown). Second, the choice of minimal medium as a control environment in the gene expression studies was based on the hypothesis that EcO157 cells landing in lettuce lesions most likely originate from an environment of low nutritional complexity. As discussed below, the upregulation in the lysates and shredded lettuce of a large number of EcO157 genes involved in adaptation to stresses caused by compounds that are prevalent in wounded plant tissue demonstrate that our control environment allowed us to obtain transcriptional data relevant to the exposure of the pathogen to injured leaf tissue.

After a period of acclimation, EcO157 displayed a remarkable ability to multiply in the lettuce lysates and had a growth rate equivalent to that in minimal medium with glucose, suggesting that the nutritional content of lysed romaine lettuce cells was sufficient to support growth of the pathogen under warm temperature conditions. This growth represented 10 cell generations of the pathogen and therefore an infectious dose sufficient to cause disease (http://www.cfsan.fda.gov/∼ebam). Microarray analysis of EcO157 gene expression during the adaptation phase in the lysates revealed that the cells were adjusting to alternate nutrient sources, as well as to a variety of stresses such as the presence of reactive oxygen species and of antimicrobial or toxic compounds, and to osmotic stress. A subset of upregulated genes that are involved in oxidative stress and antimicrobial resistance were tested by QRT-PCR in both lettuce lysates and shredded lettuce. The similar pattern of expression of these genes in both environments validated the use of lysates to approximate some of the early chemical conditions in the lesions of processed lettuce leaves. The greater increase in expression of these genes in the lysates compared to shredded lettuce may be due to the greater homogeneity of conditions in the lysates, since EcO157 inoculum cells may have landed on the shredded lettuce not only on the cut tissue but also on undamaged surfaces where these chemical stresses may have been absent.

Carbohydrate utilization.

Approximately 2% of the fresh weight of romaine lettuce leaves are made up of carbohydrates, compared to 94% water (19). As photosynthetic organisms, lettuce plants are rich in a variety of sugars including, but not limited to, sucrose, glucose, fructose, galactose, and mannose (63, 68). Glucose and maltose are present in leaves as products of starch degradation (70), and sucrose is the primary sugar transported in the phloem (63). Multiple carbohydrate transport genes in EcO157 were upregulated in response to lysates, including those specific for the disaccharides maltose and sucrose, and at least two different monosaccharide transporters belonging to the carbohydrate uptake transporters 1 (CUT1) family. The most impressive induction of a carbohydrate transport gene was that of malE, showing a 348-fold increase in transcription only 15 min after exposure of EcO157 to lysate. This rapid and large induction indicates that maltose may be a critical plant metabolite in the survival and/or growth of EcO157 in damaged lettuce leaf tissue. Overall, it is clear that injured plant cells release a wide range of carbohydrates that EcO157 can utilize as substrates for growth once its arsenal to a variety of stresses in the lysates has been deployed.

Oxidative stress and DNA repair.

A large number of EcO157 genes with increased expression in the lysates belong to operons that are responsive to oxidative stress. The two best-characterized oxidative stress-responsive transcription factors in E. coli are SoxR and OxyR (33, 55). SoxR has been shown to detect superoxide radicals via oxidation of its iron-sulfur [2Fe-2S] cluster, whereas OxyR detects H₂O₂ via oxidation of specific cysteine residues (30, 62). The oxidation of these regulatory proteins per se initiates transcription of their respective regulons. This explains the lack of increased transcription of oxyR and soxR observed in the microarray analysis, while genes that are members of their regulons were upregulated. The oxidative stress apparently experienced by EcO157 in the lysates and on shredded lettuce likely was caused by the presence of reactive oxygen species such as O₂⁻ and H₂O₂ that are generated during the oxidative burst in plant cells as a result of mechanical injury (14, 49).

The genes that were upregulated in the lysates and that belong to the Sox regulon also are members of the Mar regulon, namely, ribA, marR, and marA (Table 3). Although sodA and not sodC is part of the Sox regulon (54), two sodC-like Cu/Zn periplasmic superoxide dismutases encoded in EcO157 (Z2347 and Z3312) had increased transcription in response to lettuce lysates. D'Orazio et al. (24) reported that single EcO157 mutants in each of these genes displayed reduced survival to H₂O_2. Therefore, these genes were induced in EcO157 possibly due to the presence of H₂O₂ in lettuce lysates. Contrarily to the SoxR regulon, key genes of the OxyR regulon (ahpCF, grxA, trxC, katG, and katP) had increased expression in the lysates. Zheng et al. (74) demonstrated by microarray analysis that a core set of OxyR-regulated genes, namely, dps, katG, grxA, ahpCF, and trxC, are induced considerably in EcO157 cells exposed to 1 mM H₂O₂; all but one of these genes (dps) were upregulated more than eightfold in response to romaine lettuce lysates in the present study. Glutathione reductase (gor), a member of the OxyR regulon, was not differentially regulated; however, cydC, which is involved in transport of glutathione into the periplasm in order to regulate redox homeostasis (53), was highly upregulated after 15 min of incubation of EcO157 in the lysates. In its oxidized form, the OxyR transcription factor also upregulates the small noncoding RNA, oxyS (3). The increased expression of oxyS in lettuce lysate-exposed cells provided further support for the broad induction of the OxyR regulon under our investigated conditions.

Both H₂O₂ and superoxide can damage enzymes with Fe-S clusters via release of free iron, which can then lead to direct DNA damage by reactive iron molecules, as well as by ·OH radicals released when H₂O₂ reacts with ferrous iron (37). The IscR-regulated gene cluster iscRSUA-hscBAfdx-iscX, which is associated with Fe-S cluster assembly and repair (34), was highly upregulated at 30 min postexposure to lettuce lysates (including upregulation of iscR itself as soon as 15 min postinoculation). Perhaps in order to recruit Fe and S required for Fe-S cluster repair, genes with a role in sulfur acquisition in the cys, ssu, and tau operons (67), and others related to iron transport in the fit (50) and fep (51) operons, were upregulated. Also upregulated in the lysates were genes involved in DNA repair, such as two clusters of ribonucleotide reductase genes (nrdAB and nrdHIEF) (46, 48).

Several other genes involved in oxidative stress were part of the transcriptional signature of EcO157 cells in lettuce lysates. These included the “biofilm through hydrophobicity and stress response” (bhsA) gene (formerly ycfR) and yqhD. bhsA was observed previously to be expressed in E. coli in response to the presence of H₂O₂ (69, 74) and carbon monoxide (47). yqhD encodes an aldehyde reductase that recently was described as protecting against lipid peroxidation stress resulting from various reactive oxygen species-generating compounds (52).

The Amplex Red fluorescence assay revealed that H₂O₂ indeed was present in the lettuce lysates and that its concentration decreased over time. The failure of catalase to completely suppress the Amplex Red signal in the lysates suggests that reactive oxygen species other than H₂O₂ may have been present. The allocation of an extensive part of the transcriptional machinery to the increased expression of oxidative stress genes, as shown by microarray analysis and confirmed by QRT-PCR also on shredded lettuce, indicates that EcO157 cells are exposed early to significant amounts of oxygen radicals resulting from the plant response to injury. The large number of genes/operons recruited to adapt to these stressful conditions may explain why the oxyR mutant and other mutants in oxidative stress genes tested here were unaffected in survival and growth in the lysates. Likely, redundancy in this vast array of responses to oxidative stress may have ensured that sufficient cell protection is achieved, even in the event of some nonfunctional pathway. Besides the obvious adaptation of EcO157 to oxidative stress in the lysates, the degradation of reactive oxygen species over time, as assessed with the Amplex Red assay, may have contributed to a more favorable milieu and thereby enabled the subsequent growth of the pathogen in the lysates 4 h after incubation.

Despite the obvious importance of oxidative stress, the increased activity of genes such as bet and cai-fix, which are part of two osmotic stress-responsive operons, suggests that additional stresses are imposed on EcO157 under conditions brought about by plant cell damage and lysis. Most likely, lysed plant cells release numerous compounds that can act as osmolytes and may have induced an osmotic stress response in the pathogen. It is possible that the downregulation of proVWX in EcO157 in the lysates (data not shown) is a consequence of the absence or insufficient amounts of glycine betaine and proline to import into the cells in order to remediate the effects of osmotic stress on the bacterial cells.

Antimicrobial stress.

A number of genes involved in antimicrobial resistance and detoxification of noxious compounds showed increased transcription in lysates, presumably in response to compounds freed from the ruptured plant cells; these included genes that overlap with the cellular response to oxidative stress. These genes consisted of multiple members of the Mar regulon, including the mar operon itself, the nem operon, and the frm operon, which is involved in formaldehyde degradation (29). It is noteworthy that the mar operon is activated by salicylic acid (16), a derivative of the polypropanoid pathway (22) and a key signaling molecule in the plant global response to stress, including injury (38). In addition, minimally processed lettuce contains numerous phenolic compounds (11, 35), as well as intermediate compounds of the phenylpropanoid pathway that possess antimicrobial properties (4). Thus, the reactive oxygen species found at wound sites, and the various aromatic compounds present in leaf tissue, may explain the broad transcriptional response in EcO157 that relates to oxidative and antimicrobial stress.

Flagellar machinery, TTS system, and fimbriae.

Microarray analysis revealed that a large number of EcO157 genes encoding proteins in both TTS systems, in flagellar systems, and in fimbriae were upregulated in response to lettuce lysates, compared to growth in M9 medium. The role of these bacterial virulence determinants and appendages in the adaptation to conditions resulting from damaged leaf tissue remains unclear. The increased expression of various virulence genes, including TTS genes, in EcO157 cells shortly after nutrient replenishment in fresh culture broth has been reported (2). It is also possible that signals in the lettuce lysates primed the pathogen for attachment to lettuce tissue and that the TTS, the flagella, and the fimbriae serve this function. Flagellum genes that had increased expression in the lysates belonged primarily to the class 2 or “middle” genes, which are involved in formation of the basal flagellar structure. Flagellar and TTS systems have significant functional and structural homology (17, 27). EcO157 mutants deficient in the flagellar subunit FliC or the TTS protein EscN had reduced attachment to both spinach and lettuce leaves (71), whereas the LEE protein EspA mediated attachment of EcO157 to arugula, spinach, and lettuce, albeit after growth of the inoculum under culture and temperature conditions inducing TTS genes (60).

Our findings reveal that EcO157 has the ability to respond and adapt to stressful conditions in injured lettuce tissue and to use an array of substrates to multiply in that environment. The resulting change in transcriptional activity and metabolism leads to a particular physiology that may greatly affect its ability to colonize wounded leaf tissue, its response to decontamination treatment and perhaps also its ability to infect a host. The produce industry has relied primarily on oxidants, such as chlorinated compounds, to sanitize lettuce and wash water during processing. The significant upregulation of several oxidative stress-responsive regulons in EcO157 cells in lettuce lysates and shredded leaves observed in our study imply that the pathogen is well adapted to cope with oxidative assault after its initial exposure to injured leaf tissue. The enhanced survival to H₂O₂ of EcO157 cells primed in lettuce lysates compared to that of cultured cells, as well as the similar trend observed with Ca hypochlorite challenge, supports this hypothesis. This increased resistance to chlorine sanitizers, combined with the growth of the pathogen in wounds on the leaves due the availability of a variety of carbohydrates, may help explain the association of processed leafy greens with outbreaks of EcO157 in the recent years. On the other hand, knowledge of the numerous stresses that EcO157 faces at the wound sites of cut lettuce may provide clues for new targets to use in hurdle technologies for the reduction of microbial contamination of processed produce.