Monitoring of transfer and internalization of Escherichia coli from inoculated knives to fresh cut cucumbers (Cucumis sativus L.) using bioluminescence imaging

Slicing may cause the risk of cross-contamination in cucumber. In this study, knife inoculated with Escherichia coli (E. coli) was used to cut cucumbers, bioluminescence imaging (BLI) was used to visualize the possible distribution and internalization of E. coli during cutting and storage. Results showed that the initial two slices resulted in greater bacterial transfer. The bacterial transfer exhibited a fluctuating decay trend, E. coli was most distributed at the initial cutting site. The contaminated area on the surface of cucumber slices decreased during the storage period, which can be attributed to the death and internalization of E. coli. The maximum internalization distance of E. coli was about 2–3 mm, and did not further spread after 30 min from inoculation. Hence, our results provide useful information for risk management in both home and industrial environment.


Bacterial inoculum preparation. The engineered bioluminescent E. coli strain was kindly provided by
Beijing Academy of Agriculture and Forestry Sciences. The recombinant plasmid pXX3 that the lux operon from Photorhabdus luminescens bacteria was obtained by digesting pXen13 with Notl and Xhol then ligated to the similarly digested inverse-PCR product of the transposon vector pKGT452Cβ amplified using the primers pKGT2F/pKGT2R, and carries Cmx resistance gene::luxCDABE::Tn1409. Plasmid pXX3 was transformed into E. coli DH5α, then propagated in the dam-and dcm-deficient E. coli strain ER2925. The strain construction method was described in detail 19 . An aliquot (100 μL) of E. coli suspension was cultured in 100 mL of LB broth medium supplemented with chloramphenicol (20 μg/mL) at 37 °C for 48 h to activate the bacteria. The culture was then diluted with sterile phosphate buffered saline (PBS) to obtain the final E. coli concentrations of approximately 1, 3, 5, and 7 log CFU/mL.

Kitchen knife inoculation.
A stainless steel kitchen knife (18 cm × 8.2 cm, 1.48 cm tick) was sterilized in 75% ethanol (vol/vol) for 5 min, its surface rinsed with SDW and then dried in cabinet for 30 min. The method refers to Kusumaningrum et al. 20 , with a slight modification. The kitchen knife was respectively dipped in various concentrations of the bacterial suspension for 5 min, following which, the inoculated kitchen knife was placed in a sterile biosafety cabinet to dry for 10 min. A sterile cotton swab was used to scrape the adhered E. coli on the knife surface to the 1 mL sterile PBS and then serially diluted (tenfold) in PBS. The diluted droplets (0.1 mL) were then plated on LB agar supplemented with chloramphenicol (20 μg/mL) and incubated at 37 °C for 48 h. The numbers of colonies on the surface of the kitchen knife were counted by the plate count method. The experiment was replicated three times.
Transfer of E. coli on the kitchen knife to cucumber during consecutive cuts. Simulating the force applied during actual cutting, three different concentrations of E. coli (3.34 ± 0.21, 5.20 ± 0.13 and 7.06 ± 0.25 log CFU/mL) were used to inoculate the kitchen knife, then cucumber was cut 9 times (1 cm thick per slice) longitudinally by the contaminated knife. Try to use the same force for each cut to reduce the errors. The cucumber slices of same cutting number were placed in a sterile plastic bag together with 100 mL sterile PBS (3 slices per bag). Placed the plastic bag containing cucumber slices in a homogenizer for 3 min to homogenize the sample to obtain the mixed solution, then drew 1 mL of mixed solution and tenfold serially diluted with PBS. The diluted solution (0.1 mL) was plated on LB agar supplemented with chloramphenicol (20 μg/mL) and incubated at 37 °C for 48 h. The number of E. coli on cucumber slices was counted by plate count method and logarithmic conversion was performed.
All treatments were replicated three times. The transfer rate from the kitchen knife to the cucumber slice was determined as 21 : Bioluminescence imaging of fresh-cut cucumber slices. Distribution

Statistical analysis.
Data analyses were performed with one-way analysis of variance using IBM SPSS statistics software 20.0 (IBM Corp., Armonk, NY, USA). Significant differences between groups were determined using Duncan's test. The level of significance was set at 5% for all analyses. Image processing software were IndiGo software (Berthold Technologies, Oak Ridge, TN, USA) and Image-Pro plus 6.0.0 (Media Cybernetics, Inc.). Data were plotted using OriginPro 8 software (OriginLab Corporation, USA). All methods were carried out according to relevant institutional, national, and international guidelines and legislation.

Results and discussion
Evaluation of kitchen knife inoculation. Concentration of E. coli inoculated on a kitchen knife was shown in Table 1. When the inoculation concentrations were 1.09 ± 0.08, 3.34 ± 0.21, 5.20 ± 0.13 and 7.06 ± 0.25 log CFU/mL, the amount of E. coli on knife were 0.36 ± 0.10, 2.85 ± 0.20, 4.70 ± 0.28 and 6.49 ± 0.01 log CFU/knife. The number of E. coli inoculated on the knife increased with the increase of the inoculation level, the number was about 0.5-log less than the inoculation solution. Kusumaningrum et al. 20 reported that the amount of Salmonella Enteritidis and Staphylococcus aureus on stainless steel surfaces declined with the decrease of the inoculation concentration, the change pattern was similar to this study. Furthermore, previous reports pointed out that the decay model describing the transfer behavior of pathogenic bacteria during slicing of vegetable was less accurate when using the low inoculum 15 . Thus, the transfer number of E. coli on the knife was small under low inoculation concentration, which was not conducive to researching and monitoring the transfer of E. coli. In order to reveal E. coli transfer, higher inoculation concentrations (3.34 ± 0.21, 5.20 ± 0.13 and 7.06 ± 0.25 log CFU/mL) were selected for subsequent experiments.
E. coli transferred from kitchen knife to cucumber. Experimental data showed that the transfer amount and rate of E. coli exhibited a fluctuating decay pattern during cutting process (Table 2). It was the highest at the first cut in all treatments, after that data showed a sharp decrease, then slightly increased and leveled off, finally decreased again as the number of slices increased. Moreover, the number of E. coli remaining on cucumber slices after 9 cuts was below 100. At low inoculum level (3.34 ± 0.21 log CFU/mL), the recovery amount (slice 1: 0.95 ± 0.10; slice 2: 0.68 ± 0.14; slice 3: 0.31 ± 0.14 log CFU/cucumber slice) and transfer rate (slice 1: 1.259; slice 2: 0.676; slice 3: 0.288%) of E. coli on the first 3 slices decreased significantly as the cutting number increased. Then they increased significantly at the fourth slice (recovery amount: 0.65 ± 0.11 log CFU/ cucumber slice; transfer rate: 0.631%) and the changes on the following data points became stable. When the ninth slice was reached, the recovery amount (0.26 ± 0.16 log CFU/cucumber slice) and transfer rate (0.257%) of E. coli were decreased rapidly. This was similar to the previous study, the bacterial transfer that occurs after cutting food with a blade inoculated with 10 4 CFU/mL bacterial solution showed a fluctuating decrease trend 23 . At moderate and high inoculation levels (5.20 ± 0.13 log CFU/mL and 7.06 ± 0.25 log CFU/mL), higher amount and transfer rate of E. coli appeared (3.36 ± 0.03 and 3.75 ± 0.08 log CFU/cucumber slice; 4.571% and 0.182%).  24,25 . Stainless steel is a hydrophilic surface, which can be used as a medium for bacteria to attach, may promote the release of pathogens during the food preparation process and relocate them to the surface of high-moisture food 26 . The cutting speed and action were kept as consistent as possible in this study, all cutting action were the longitudinal cutting that perpendicular to the cutting board, therefore cutting speed and action might have little effect on the transfer of E. coli. However, the force was variable during the cutting process, the fluctuating transfer of E. coli might be affected by force 7 . The similar trends were observed by other researchers with regard to other pathogenic bacteria that transfer through surface of stainless steel to cucumbers, lettuce and celery 20,27,28 . Furthermore, in previous researches, the transfer behavior of pathogenic bacteria during the slicing process of fish, meat and vegetables were similar which could use the decay model to describe [29][30][31] . The transfer of E. coli was probably because the shearing force on the knife surface, which was not determined in our study, might have affected the removal of loosely attached cells, resulting in the transfer of bacterial cells on the knife to the cucumber slices 32 . As the slice continued, the number of cells transferred back to the knife gradually decreased and cells transferred to cucumer slice with the next cut gradually decreased, this might be because the knife was simultaneously affected by adhesion and hydrophobicity. When the attachment between E. coli and knife was not as strong as E. coli and cucumber, the cells might detachment from the knife and movement to cucumber slice so that transfer becomes more feasible. The remain of exudate released from the cucumber slices altered the hydrophobicity of the knife surface and affected the transfer of E. coli. This caused some transferred cells to move back onto the knife 33 . Furthermore, the dual-or multiple-species of endophytes in cucumber exudate might affect the adhesion of E. coli and stainless steel 34 . Changes in the transfer rate during the continuous cutting process, which might be because consecutive cutting was not a static, neither an easy-to-control process 21 .
BL images of E. coli transfer and distribution on cucumber slice after cutting. The high inoculation (7.06 ± 0.25 log CFU/mL) was used to reveal the transfer of E. coli during the cross-contamination process  Fig. 1a. Our results suggested that the adhesion of the initial cleavage site was strongest due to mechanical action, most E. coli transferred to the upper section. The cells transferred back to the knife due to the hydration of the exudate, E. coli would be randomly distributed on the tissues in the middle and lower parts of the cucumber slices. The E. coli luminescence signal detected on the vascular bundles, xylem vessel and placental tissues was the strongest at the site of inoculation, representing the most transfer of strains. This might be because these tissues have xylem tissues that could transport nutrients and water to cucumbers 12 , which provided nutrients for growth of E. coli. These tissues would accumulate and adhesion of E. coli more easily than other tissues and the detected luminescent signal will be stronger than other tissues. The 3D surface chart further showed that more luminescence signals were detected at the initial cleavage site and stronger luminescence signals were detected in the xylem tissues (Fig. 1b).
Survival of E. coli on cucumber slices during storage. BL images showed that the contaminated area of E. coli on all cucumber slice samples gradually decreased with the decrease of storage temperature and the extension of storage time. More specifically, the luminescent signal of the E. coli under 37 °C storage condition was strongest (Fig. 2). This might be because 37 °C was the suitable temperature for the growth of E. coli. E. coli was irregularly distributed on vascular bundles, xylem vessels, placenta and mesophyll tissues where the signal color were the reddest and the signal were the strongest. The survival of E. coli on cucumber slice after cutting 5   Fig. 3. It was found that more number of living E. coli were observed on the cucumber slices in the early stage of inoculation 30 min after cross-contamination (Fig. 3a), while more E. coli deaths were observed after 2 h of inoculation (Fig. 3b). The death of E. coli might be due to the gradual consumption of water and nutrients on the surface of cucumber slices with the extension of storage time. This might also be caused by the release of hydrogen peroxide from damaged plant tissue that causes oxidative stress in the bacterial cells and the presence of competing microorganisms 20 , thereby harming or inactivating E. coli 35 . However, the antibacterial hydrogen peroxide produced by the wound will temporarily affect the attachment of E. coli and tends to decrease after injury 36 , so some bacteria will survive 37 . Under other temperature conditions (25 °C, 10 °C, 4 °C), E. coli mainly colonized and distributed on the margins, placenta and vascular system tissues. However, the attenuation degree of E. coli in the same contaminated area was much lower than 37 °C with the storage time. These temperatures (25 °C, 10 °C, 4 °C) are not suitable for the growth of E. coli, leading to differences in the growth and attenuation metabolism of E. coli on cucumbers. However, previous studies have shown that pathogenic bacteria could survive at 4 °C and grow at 10 °C, indicating that unintentional abuse of temperature could cause the growth of pathogenic bacteria, enough to reach potentially hazardous levels 38 . The survival ability of E. coli was stable due to the decreased ability of self-protection and regulation of plant cells during low temperature storage of cucumber 39 . The increased activity of POD and SOD could kill the superoxide free radical produced in adversity 40 , reduce the damage to E. coli. Meanwhile, the exudate released from the cucumber slices provided sufficient nutrition for the survival of the bacteria and maintained their activity. These results showed that there were still potential safety hazards in cutting cucumbers with contaminated knives even under low temperature conditions, the distribution of E. coli tended to be on the initial cutting site and nutrient-rich tissues. In addition, it was observed that the transfer area and change trend of E. coli after cutting cucumber using a kitchen knife inoculated with 10 5 and 10 3 CFU/mL of E. coli during storage was similar to the concentration of 10 7 CFU/mL (data not shown).   (Fig. 4a). The maximum internalization distance was 2-3 mm after 30 min from inoculation. The 3D surface chart further clarified that the bioluminescent signals were the strongest on the inoculated surface, and became weaker with an increase in internal distance (Fig. 4b). Regarding the internalization distance, the result obtained in the present study was in agreement with previous research findings 41 indicating that microorganisms could penetrate into the medium, the penetration distance was in the first 2-3 mm of the medium surface, rendering the bacteria more difficult to detect. The internalization of E. coli at different storage times and temperatures was shown in Fig. 5. The internalized area of all samples decreased rapidly with storage time, being constant stable after 60 min from inoculation. This might be due to the temporarily affect the growth of E. coli by cucumber exudate, which led to a rapid decrease in internalized E. coli. After a period of storage, the bacterial cells had adapted to the growth environment and could survive then stabilized. E. coli reached the maximum internalization distance after 30 min from inoculation at 37 °C without change with the extension of storage time. The internalization distance of E. coli declined with temperature decreased, and remained unchanged after 30 min. Less E. coli internalization distance and area was observed at refrigeration temperatures (4 and 10 °C), but the viability of E. coli was weaker at 4 °C. The internalization is probably due to the fact that vascular bundles, xylem vessels function in transporting nutrients and water, thus providing access to E. coli to the internal tissues 42,43 . But the refrigeration temperatures were not suitable for bacterial growth, they could reduce the activity of E. coli. Results further proved that the decreased of the contaminated area on the cucumber slice and the luminescence signal were caused by the internalization of E. coli as well as the death of E. coli.

Conclusion
This study determined that the concentration of E. coli contaminated on the surface of the kitchen knife was around 0.5 log lower than the inoculation concentration. Consecutive cutting of cucumber slices with a contaminated knife would cause cross-contamination, which was a dynamic process. The maximum transfer area of E. coli was at the initial cutting site of cucumber slices. Once E. coli adhered to the vascular bundles and xylem tissues, it would spread along these tissues into the internal tissues of cucumber and the maximum internalization distance was 2-3 mm. The contaminated area of E. coli on cucumber slices gradually decreased with storage time, which might be due to the death and internalization of E. coli. Quantifying and visualizing the relevant cross-contamination during cucumber cutting validated the transfer and internalization of bacterial and could provide a scientific basis for risk management strategies to reduce, prevent or eliminate cross-contamination in the kitchen as well as in the industrial environment.