Application of MALDI-TOF mass spectrometry and specific PCR for tracking of E. coli O157:H− strain 431/97 in Batavia lettuce

In this study, lettuce roots and leaves were contaminated with enterohemorrhagic Escherichia coli O157:H− strain 431/97 under greenhouse conditions. Furthermore, the internalization of strain 431/97 in lettuce roots and leaves was examined. To track the inoculated bacteria during the experiments and to differentiate them from the autochthonous microbiota, a combined protocol including matrix-assisted laser desorption/ionization time of flight mass spectrometry (MALDI-TOF MS) and PCR was established. At different time points after inoculation of Batavia lettuce with 5.5 × 106 cfu/mL (high inoculation level) or 3.0 × 101 cfu/mL (low inoculation level) E. coli O157:H− strain 431/97 as well as sterile 0.9% (w/v) sodium chloride solution (negative control), samples from the root and the leaf were taken and surface disinfected with gentamicin. After homogenization, microorganisms were isolated from the samples and analyzed by MALDI-TOF MS. Analysis of the root samples resulted in bacterial counts of 1.0 × 102–1.0 × 106 cfu/0.25 g depending on the inoculated viable counts and the incubation period. In the leaf samples, strain 431/97 was not detected. The investigation of the viable cell counts of E. coli O157:H− 431/97 following irrigation of the leaves resulted in bacterial counts of 102 cfu/0.25 g for the disinfected leaf samples. Thus, the established protocol is suitable for detecting the investigated strain under greenhouse conditions in plant infection experiments. This strain may indeed survive in the soil, but did not enter the plant via the root in detectable numbers. Contrarily, viable counts exceeding the generally accepted infective dose of less than 100 cells for enterohemorrhagic E. coli were determined internalized after irrigation of the leaves. As this may pose a risk for the consumer, the present study provides a valuable set of tools for further research.


Background
Enterohemorrhagic Escherichia coli (EHEC) typically produces Shiga toxins (Stx), is able to form attaching and effacing lesions to epithelial cells and may carry large virulence plasmids [25,31]. They may cause watery diarrhea and serious complications such as bloody diarrhea, hemorrhagic colitis and the hemolytic uremic syndrome in humans [25]. These complications are caused by several toxins, such as Shiga toxin, cytolethal distending toxin V, enterohemolysin, and subtilase cytotoxin, as well as accessory virulence factors [24,34]. Most outbreaks occurred by strains of the serotypes O157:H7 and O157:H − , as well as the non-O157 serotypes O26, O91, O103, O111, O118, O145 and O166 [6,19,20]. Contrary to other pathogens such as Salmonella and Listeria, the infective dose of EHEC is with less than 100 cells very low [25]. The main reservoirs of EHEC are cattle and small ruminants, which remain asymptomatic [22]. In the past, the ingestion of raw or undercooked animal products such as milk, cheese and meat was considered as the main route of transmission [30]. More recently, plants which are consumed raw, such as lettuce and sprouts, were also identified to be of importance [10,12,16,36]. Some of the largest EHEC outbreaks were associated with plants, such as the O157:H7 outbreak in Japan with radish sprouts in 1996 [23], the O157:H7 outbreak in the United States of America with spinach in 2006 [37], and the O104:H4 outbreak in Germany with fenugreek sprouts in 2011 [1], which was caused by a hybrid strain of enteroaggregative and enterohemorrhagic E. coli. EHEC may get in contact with plants via water, soil, where they were shown to survive for several months [4], manure (either applied on purpose as fertilizer or accidental contamination during flooding), wild animals, and insects [5]. Additionally, E. coli is reported to be mobile in the plant [15]. According to current knowledge, the plant itself is not only a carrier, but also its role as a secondary host for EHEC is discussed [17]. EHEC strains get in contact with plant surfaces by adherence and adhesion, and can survive and multiply as an epiphyte on them [3]. EHEC O157:H7 strain Sakai was shown to attach to and internalize into lamb's lettuce roots under green-house conditions [13]. EHEC may also integrate into already existing biofilms [38] on leaf surfaces as well as invade into the leaf via stomata [18] resulting in an endophytic lifestyle.
Methods to trace specific microorganisms in the plant environment are basic to study the interaction of EHEC and plants in the presence of the autochthonous microbiota of the rhizo-and phyllosphere. Therefore, a fast protocol combining MALDI-TOF MS with specific PCR to trace the selected E. coli O157:H − strain 431/97 during plant experiments under real-life conditions was established and successfully applied in this study.

Bacterial strain and cultivation conditions
Shiga toxin-negative sorbitol-fermenting E. coli O157:H − strain 431/97 was used in this study. This strain was isolated from a 3-year-old child with watery diarrhea [32]. It harbors the 121-kb-plasmid pSFO157 which encodes for several putative virulence loci, such as EHEC hemolysin (hly), type II secretion system proteins (etp), and Sfp fimbriae (sfp) [2]. Three genes were used to differentiate this strain from other Enterobacteriaceae which might be present in the autochthonous microbiota of the soil or plant: the O157-antigen specific gene rfbE, which is positive in this strain, the sfpA gene, which is indicative for the presence of pSFO157, and the Shiga toxin 2a gene, which is absent in the strain 431/97. Thus, this strain has the genetic characteristics rfbE + , sfpA +

Lettuce cultivar and cultivation
For establishing of the disinfection procedures, commercial oak leaf lettuce purchased at a local supermarket was used.
For all other experiments, Batavia lettuce of the cultivar Alanet (Enza Zaden Beheer B. V., Enkhuizen, The Netherlands) was grown in a biosafety level 2 greenhouse (Agroscope, Waedenswil, Switzerland) in natural field soil (sandy loam) under 16-h light and 8-h darkness conditions. The humidity in the greenhouse was 50%, the daytime (light) temperature was 22 °C and the nighttime (darkness) temperature 18 °C.

Disinfection of lettuce surfaces with gentamicin
Six 0.25 g samples of oak leaf lettuce were prepared by cutting the leaves with a sterile scalpel and weighing the pieces on an analytical scale. One sample remained untreated as a control. The other five samples were inoculated with 10 µL of a bacterial suspension containing 1.0 × 10 9 cfu/mL of strain 431/97 prepared as described above. The leaves were dried under a biosafety cabinet. Four of these samples were incubated in an aqueous solution with a final concentration of 50 µg/mL gentamicin (Carl Roth GmbH & Co. KG) for 1, 5, 10 or 20 min and then rinsed in 50-mL sterile demineralized water. The fifth sample was not treated. All samples were transferred to 2.25 mL of sterile 0.9% (w/v) sodium chloride solution and mixed on a vortex shaker at 100×g for 1 min. One mL of each of these suspensions was transferred into 20-mL modified trypticase soy broth (30 g/L trypticase soy broth (Becton, Dickinson and Company, Heidelberg, Germany), 1.5 g/L bile salts (Becton, Dickinson and Company), 1.5 g/L di-potassium hydrogen phosphate trihydrate (Merck KGaA); pH 7.4 ± 0.2) and incubated on a rotary shaker at 160 rpm under aerobic conditions at 37 °C over night. Bacterial growth was determined in a spectrophotometer as an increase in OD 600 in comparison to the sterile broth.

Impact of gentamicin residues on bacterial growth
To investigate whether the gentamicin solution could be completely washed off from the samples, they were treated with gentamicin as described above for 5, 10 and 20 min. The treated samples were again transferred into physiological sodium chloride solution, blended, and a 1-mL aliquot was transferred into modified trypticase soy broth. To the remaining sodium chloride solution still containing the leaf, an E. coli O157:H − strain 431/97 solution with a viable count of 2.5 × 10 3 cfu/mL was added. The samples were mixed at 100×g with a shaker and incubated for 35 min at room temperature. They were treated further as described above, and additionally spread-plated on Fluorocult violet red bile agar (VRBagar; Merck KGaA) and on plate count agar (PC-agar; 5 g/L tryptone (Becton, Dickinson and Company), 2.5 g/L yeast extract (Becton, Dickinson and Company), 1 g/L glucose (Merck KGaA), 12 g/L agar (Becton, Dickinson and Company); pH 7.0 ± 0.2). The plates were incubated at 37 °C for 24 h, and the viable counts were determined.

Internalization of E. coli O157:H − 431/97 into the roots
The graphical abstract shows a flow diagram of the inoculation of plants as well as the subsequent analyses. Six trays of 7-day-old Batavia lettuce plants were used for the experiments. Two trays were watered with 200 mL of a suspension containing 5.5 × 10 6 cfu/mL (high inoculation level, H) E. coli O157:H − 431/97, two trays with 200 mL of a suspension containing 3.0 × 10 1 cfu/mL (low inoculation level, L) E. coli O157:H − 431/97, and two trays with 200 mL sterile 0.9% (w/v) sodium chloride solution (negative control, N). The suspensions were applied to the soil surface using a sterile 20-ml disposable plastic pipette, without contaminating the lettuce leaves.
Samples were drawn 1, 7 and 14 days after the inoculation. After 14 days, lettuce plants reached a height of approximately 80 mm. The plants were carefully extracted from the soil, and cut with a sterile scalpel. Roots and leaves weigh less than 0.25 g per plant at this stage; thus, roots and leaves of several plants per tray were pooled to achive this mass. Therefore, they were transferred into a sterile petri dish. The leaves of one tray per treatment were disinfected with gentamicin for 10 min as described above, while the leaves of the second tray remained untreated. The disinfected leaves were rinsed twice with 40-mL demineralized sterile water. All plant samples were transferred into sterile mortars, and 2.25-mL sterile physiological sodium chloride solution was added to each sample. The plant tissue was ground, and the resulting solution was diluted decimally in physiological sodium chloride solution. Appropriate dilutions were spread-plated in duplicate on sorbitol Mac Conkey agar (SMAC, Merck KGaA), and the plates were incubated at 37 °C for 24 h. Sorbitol-fermenting bacteria, such as the investigated strain 431/97, metabolize this carbohydrate into acid, which in turn is used in combination with the indicator neutral red to turn the colonies red. Sorbitol-negative bacteria on the other hand, such as taxonomically closely related Shigella and Salmonella, show colorless colonies. The viable counts were determined and the colonies were assessed for their morphology. Selected colonies displaying different morphologies were sub-cultured on sorbitol Mac Conkey agar for purification, and assigned consecutive numbers.

Internalization of E. coli O157:H − 431/97 into the leaves
Six trays of 12-day-old Batavia lettuce plants were used for the experiments. The experimental setup was as described above, but this time the bacterial suspension or the sodium chloride solution was dispersed over the plant leaves and left to dry until the first sampling on the next day. Plant leaf samples were drawn and analyzed as described above.

Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) analysis of selected isolates
As a basis for further analysis, a main spectrum (MSP) of E. coli O157:H − strain 431/97 was generated using the MALDI-TOF mass spectrometer ultrafleXtreme (Bruker Daltonics GmbH, Bremen, Germany) from 12 spectra of this strain with the MALDI Biotyper 3.0 software, using the Biotyper MSP Creation Standard Method. The maximum quantity error of each spectrum was set to 2000 ppm, and of the main spectrum to 200 ppm. For the main spectrum, the minimum peak frequency was set to 25%, and the maximum number of peaks to 70. This main spectrum was included into the MALDI Biotyper 3.0-Database as Escherichia coli 431_97_drd (Bruker Daltonics GmbH).
Colonies were chosen from the agar plates of the internalization experiments for subsequent analysis. Sixteen isolates were chosen which displayed the same morphology on plate as the investigated E. coli O157:H − 431/97 strain, and 15 isolates which displayed a different morphology and were, therefore, considered as members of the autochthonous microbiota. The putative E. coli isolates originated from the experiments described in Table 1 (see "Results" section), whereas the origin of the putative autochthonous microbiota isolates is shown in Table 2 (see "Results" section).
Colony material of the pure cultures prepared as described above was transferred in duplicate onto spots of the MALDI-TOF MS target (MTP ground steel, Bruker Daltonics GmbH) with sterile tooth-picks. Eight spots per target were loaded with E. coli O157:H − 431/97 as a reference. The bacteria were inactivated with 0.8 µL 70% formic acid per spot. To the dried spots, 0.8 µL matrix solution (10 mg α-cyano-4-hydroxycinnamic acid (Bruker Daltonics GmbH) dissolved in 1 mL 50% acetonitrile/2.5% trifluoroacetic acid solution) was added, and this solution was air-dried. The plates were stored in the dark in sterile containers under argon atmosphere at room temperature. Prior to analysis, a bacterial test standard (BTS, Bruker Daltonics GmbH) covering a mass range from 3.6 to 17 kDa was added to selected spots (for detailed protein masses see [14]). Sum spectra were generated from samples after 1200 laser shots per sample at 1000-Hz frequency in the positive linear mode in mass-charge (m/z) ratios ranging from 2000 to 20,000 Da. Baseline subtraction and smoothing of the spectra was performed with the MALDI flexAnalysis 3.  For the negative control without gentamicin treatment as well as for the treatments for 1 and 5 min, bacterial growth was observed. Longer treatment periods of 10 and 20 min inhibited bacterial growth, as the optical densities were equivalent to the un-inoculated medium control (10 min: increase in OD 600 0.0009 ± 0.0001; 20 min: increase in OD 600 0.0050 ± 0.0002).

Gentamicin is effectively removed from the leaves by rinsing
One aim of this study was to analyze bacteria that were not freely available on the leaf surface, because they, for example, entered stomata or were internalized into the plant tissue. Gentamicin residues on the leaf surface might inhibit the growth of these bacteria when they were set free in sample preparation, and, thus, affect the subsequent viable count results. To investigate this possible effect, the leaves were contaminated with E. coli O157:H − 431/97 as described above. After gentamicin treatment, the leaves were rinsed in physiological sodium chloride solution. These solutions of both controls, the native non-disinfected leaves and the negative control, showed the respective expected positive and negative results in modified trypticase soy broth (data not shown). For the gentamicin-treated leaves, no growth was determined, confirming the above results. Consecutively, the sodium chloride solutions obtained from disinfected leaves were inoculated with low numbers of E. coli O157:H − 431/97 to determine a possible antimicrobial effect. The counts on VRB-agar remained constant at 2.8 × 10 2 -2.9 × 10 2 cfu/mL (Fig. 1), while as expected slightly higher counts of 2.8 × 10 2 -1.0 × 10 3 cfu/mL were detected on PC-agar (Fig. 1).

Internalization of E. coli O157:H − 431/97 via the rhizosphere
After inoculation of the soil of lettuce plants with high (H; 5.5 × 10 6 cfu/mL) or low (L; 3.0 × 10 1 cfu/mL) amounts of E. coli O157:H − 431/97, the plants were grown for up to 2 weeks. Samples of roots with adhering soil as well as leaves were prepared as described above after 1, 7 and 14 days and streaked on SMAC agar. The colonies on SMAC agar were visually distinguished in potential EHEC colonies (red color) and autochthonous microbiota (colorless), and were evaluated separately. The rhizosphere samples were analyzed without prior treatment. One day after the inoculation, the viable counts of the autochthonous microbiota on SMAC agar ranged at about 10 8 cfu/g root (Fig. 2). The counts of samples N and L at day 1 could not be evaluated due to too high viable counts. These counts declined by approximately two log units within the first week, and remained stable thereafter for the experimental period. The inoculated E. coli O157:H − strain 431/97 was only detected in rhizosphere samples which were inoculated with high doses of this strain. One day after inoculation, the EHEC counts were 9.2 × 10 6 cfu/g root with adhering rhizosphere. These counts dropped significantly to 5.0 × 10 1 cfu/g during the first week, and remained stable thereafter (Fig. 2). The leaf samples were evaluated after treatment with gentamicin. While an autochthonous microbiota was detected in the range of 10 4 -10 5 cfu/g leaf, no EHEC cells were found (data not shown).

Internalization of E. coli O157:H − 431/97 via the phyllosphere
Lettuce plants were inoculated on their leaves with high (H; 2.3 × 10 6 cfu/mL) or low (L; 3.0 × 10 1 cfu/ mL) doses of E. coli O157:H − strain 431/97, and grown for up to 2 weeks. The leaves were either analyzed as such or treated with gentamicin prior to viable count analysis. Without disinfection, the viable counts of the leaves ranged as expected between 1.4 × 10 3 and 4.3 × 10 5 cfu/0.25 g leaf 1 day after contamination. Subsequently, the viable counts ranged from below the detection limit (1.0 × 10 2 cfu/0.25 g leaf ) to 2.0 × 10 3 cfu/0.25 g leaf (Fig. 3). In the disinfected leaf samples, the viable counts on SMAC agar were between 1.0 × 10 2 and 3.4 × 10 5 cfu/0.25 g leaf 1 day after contamination, and decreased over the incubation period (Fig. 3). One day post contamination, the EHEC strain was detected both in the non-disinfected and in the disinfected leaf sample initially inoculated with the high EHEC dose. The EHEC counts of the non-disinfected sample were approximately six times higher than those of the disinfected sample. After 7 days of incubation, one non-disinfected leaf sample was positive for EHEC with 7.5 × 10 1 cfu/0.25 g leaf, but below the quantification limit.

EHEC isolates can be correctly identified by MALDI-TOF biotyping and PCR
Altogether, 31 isolates were selected on account of their colony morphology as putative E. coli for subsequent PCR-and MALDI-TOF analysis. Sometimes, more than one colony per plate was chosen and denominated with a suffix. The expected results for the PCR for E. coli O157:H − 431/97 were rfbE + , sfpA + , and stx 2a − (for gel pictures see Additional file 3). 15 of the 31 isolates yielded no amplicons in all three PCR assays, thus being not the investigated E. coli strain, and were, therefore, discarded from further analysis. Of the 16 remaining potential EHEC colonies, 14 were identified as E. coli by MALDI-TOF analysis with score values between 2.3 and 3.0, indicating a definite identification at species level (see Table 1). The two remaining isolates were identified with score values of 2.226 (isolate 294 1) and 2.234 (isolate 289), respectively, which corresponds to a possible identification at species level (Table 1). For all strains, the correct profile of E. coli O157:H − strain 431/97 was the top hit of the results list (see tables S1 to S16 in Additional file 1). The peak lists and the mass spectra of all strains are given in Additional file 2. These results were complemented by PCR targeting the rfbE, stx 2a and sfpA genes (see figures S17 to S22 in Additional file 3) confirming the strain identity.
From the isolates of the autochthonous microbiota of the rhizo-and phyllosphere, 15 colonies were chosen for MALDI biotyping. The results of the top hits for each strain are displayed in Table 2. The peak lists and the mass spectra of all strains are given in Additional file 2. For 13 strains, MALDI biotyping provided a possible identification at species level, while two strains were identified at the genus level only. The dominant genus was Pseudomonas with six isolates, followed by Achromobacter with three. The other genera were identified only once each.

Discussion
Some of the largest EHEC outbreaks have been associated with plant foods such as leafy greens and sprouts.  [27]): quantitative real-time PCR (qPCR) targeting diverse genes (e.g., [11]), fluorescence in situ hybridization (FISH; e.g., [33]), the application of reporter genes such as lux (e.g., [29]), immunofluorescence (e.g., [40]), as well as the microscopic detection of bacteria expressing fluorescent proteins, such as green fluorescent protein (e.g., [39] The results of the current study show that the established protocol reliably traces E. coli O157:H − 431/97 in in vivo plant experiments. The investigated strain was already differentiated from the autochthonous microbiota as well as closely related pathogens such as Salmonella and Shigella on SMAC agar by its ability to metabolize sorbitol. This is not the case for other EHEC strains, which are incapable of sorbitol degradation and might, therefore, be misidentified. Therefore, a subsequent identification procedure as presented in this study is essential. Thus, the unique protein profile of the strain in question was integrated into the MALDI biotyping database. The mass peaks at mass/charge ratios of 3791.7, 3814.0, 3830.4, 6257.1, and 9742.8 can be considered as marker mass peaks for this strain. The putative E. coli O157:H − 431/97 colonies matched the deposited profile in the MALDI biotyping database and each time this was the top hit. This could be due to the fact that pathogenic E. coli contains an extensive accessory gene pool and that the E. coli core genome was found to consist of only 3100 genes, while the accessory genome was estimated to include up to 89,000 genes [21]. As already proposed by Christner et al. [9], this method is indeed applicable for strain-typing in environments containing different strains and species, such as, in the current study, the microbiota of the phyllo-and rhizosphere. The MALDI-TOF MS results were furthermore confirmed by complementary PCR in this study.
From these experiments, it was concluded that the investigated strain E. coli O157:H − 431/97 may survive in