Salmonella must be viable in order to attach to the surface of prepared vegetable tissues


Elizabeth J. Saggers, Institute of Food Research, Norwich Research Park, Colney, Norwich, NR4 7UA, UK.


Aims:  The aims of the current study were to explore the site of bacterial attachment to vegetable tissues and to investigate the hypothesis that Salmonella must be living in order to attach to this site(s).

Methods and Results:  Scanning electron micrographs of intact potato cells showed that Salm. serotype Typhimurium attached to cell-wall junctions; suggesting a high-level of site selectivity. Inactivation of Salm. Typhimurium using heat, ethanol, formalin or Kanamycin resulted in cells that could be no longer attached to these sites. Attachment of a Gfp+ strain of Salm. Typhimurium to cell-wall material (CWM) was examined via flow cytometric analysis. Only live Salm. Typhimurium attached to the CWM.

Conclusions: Salmonella serotype Typhimurium must be metabolically active to ensure attachment to vegetable tissues. Attachment preferentially occurs at the plant cell-wall junction and the cell-wall components found here, including pectate, may provide a receptor site for bacterial attachment.

Significance and Impact of the Study:  Further studies into individual plant cell-wall components may yield the specific bacterial receptor site in vegetable tissues. This information could in turn lead to the development of more targeted and effective decontamination protocols that block this site of attachment.


Minimally processed fruits and vegetables have recently undergone an increase in consumer demand because of their healthy image and convenience. The product range includes ready-to-eat prepared vegetable tissues, such as carrot sticks and shredded lettuce and prepared fruits. Microbiological safety is a key issue for the entire product range, as they are intended for consumption raw, without further preparation or cooking. Contamination of these products is predominantly by Gram-negative bacteria, in particular, members of the Pseudomonadaceae and Enterobacteriaceae (Brocklehurst et al. 1987; Brocklehurst 1994). In addition, it is well established that some products can also contain potential pathogens (Nguyen-the and Carlin 1994; Beuchat 1996; Francis et al. 1999), and some have been implicated in an increasing number of outbreaks of food-borne illness (Long et al. 2002). Between 1st January 1992 and 31st December 2000, 5·6% of outbreaks of food-borne illness in England and Wales were attributed to salad vegetables and fruit, with Salmonella being the most frequently implicated bacterial pathogen (Long et al. 2002). Lettuce, in particular, has been connected to outbreaks of Salmonella serotype Typhimurium (Horby et al. 2003), Escherichia coli (Ackers et al. 1998; Hilborn et al. 1999) and Shigella sonnei (Kapperud et al. 1995).

We regard bacterial colonization of prepared fruit and vegetable tissues to be the result of three phases: an initial attachment phase, a consolidation phase, which may involve the production of extracellular polymer, and subsequent growth to form microcolonies. Most studies of colonized vegetable tissue surfaces concentrate on the growth of bacteria on the product rather than on the initial attachment and consolidation phase. However, commercial decontamination processes are usually applied to prepared vegetable tissues within a few minutes of dicing, chopping or shredding, i.e. during the initial attachment and colonization phases. It is recognized that the use of biocides as decontaminants is not always successful (Adams et al. 1989; Garg et al. 1990; Brocklehurst 1994; Zhang and Farber 1996). Accordingly this work is part of a wider ongoing study to understand the mechanisms used by bacteria for attachment to tissues. In particular, this paper focuses on the competencies of the organism in the attachment process with a view to exploiting this knowledge to improve postharvest technologies and yield more effective decontamination protocols.

Scanning electron microscopy (SEM) of prepared leaf tissues has shown that bacteria can be found on the abaxial and adaxial surfaces (Carmichael et al. 1999). However, the natural flora of processed lettuce was found to be concentrated in the intercellular junctions of the leaf (Carmichael et al. 1999). Further colonization studies showed that E. coli preferentially attached to the cut surface of lettuce whereas Pseudomonas fluorescens preferentially attached to the intact surface (Takeuchi et al. 2000). Salmonella serotype Typhimurium attached to both cut and intact surfaces (Takeuchi et al. 2000). The cut surface, along with other leaf structures such as stomata, may allow bacteria to become internalized within tissues and, as a result, become protected from chlorine disinfection (Seo and Frank 1999; Takeuchi et al. 2000; Takeuchi and Frank 2001). The above studies indicate the preferential areas of attachment, but do not investigate the mechanisms of the attachment process. Recent studies by this laboratory have examined the attachment of the spoilage bacteria Ps. fluorescens and Pantoea agglomerans (Garrood et al. 2004) and the pathogenic bacteria Salm. Typhimurium (E.J. Saggers et al., unpublished) to vegetable tissue. Further studies attempted to prevent attachment of Salmonella to vegetable tissues by masking potential attachment sites using dead Salmonella. However, attachment of subsequently applied live Salmonella was unaffected. Following the apparent failure of dead cells to mask potential attachment sites, we hypothesized that attachment may be an active process requiring the bacteria to be viable.

In the study reported here, we explore the hypothesis that Salmonella must be living in order to attach to vegetable tissues and present evidence in support of this hypothesis. We also present evidence that implicates components of the plant cell wall as potential bacterial receptor sites important in the initial attachment of Salmonella to vegetable tissues.

Materials and methods


Salmonella serotype Typhimurium strain LT-2 (NCIMB 10248) was obtained from the National Collection of Industrial and Marine Bacteria, Aberdeen, UK. A Gfp+ strain of Salm. Typhimurium SL1344, (strain JH3016) was provided by Dr Isabelle Hautefort, Molecular Microbiology Group, Institute of Food Research, UK. This strain contains a single copy rpsM::gfp+ fusion inserted chromosomally (Hautefort et al. 2003). This fusion is expressed in all conditions tested and was used to fluorescently label the Salmonella cells.

Culture media

Stock cultures of Salm. Typhimurium LT-2 were stored on Heart Infusion agar (Oxoid) slopes at 1°C. At monthly intervals, it was subcultured to fresh Heart Infusion agar slopes which were incubated at 25°C for 24 h and subsequently stored at 1°C.

A stock culture of Salm. Typhimurium SL1344 strain JH3016 was stored at −80°C in Tryptone Soya Broth (Oxoid) plus glycerol (30% w/v).

Stock cultures were plated onto Trypticase Soy Agar (Oxoid, Basingstoke, UK) plates which were incubated at 25°C for 24 h and subsequently stored at 5°C for no more than 2 weeks.

Inocula were prepared in Trypticase Soy Broth (Oxoid), and all viable counts were made on Plate Count Agar (PCA; Oxoid CM325).

The potato variety used throughout the study was Maris Piper obtained from a local supermarket.

Preparation of live and inactivated inocula

Bacteria were grown successively at 25°C for 24 h, and then at 20°C for 24 h. The resultant population of approx. 109 viable cells ml−1 was diluted in peptone salt dilution fluid (PSDF) (I.C.M.S.F. 1978) to give a suspension that contained the desired number of viable bacteria.

To provide a heat-inactivated inoculum, the 20°C culture was diluted to 108 CFU ml−1 in preheated (50°C) PSDF and then heated at 50°C for 14 min. This resulted in inactivation of all cells within the inoculum (data not shown).

To provide an ethanol-inactivated inoculum, the 20°C culture was filtered using a 0·22-μm-pore-sized membrane filter (Millipore, Billerica, MA, USA) to remove cells from suspension. Cells were then re-suspended in 10 ml 70% (v/v) ethanol and incubated at 20°C for 5 min, the time taken (found experimentally) to kill the population (data not shown). Cells were then filtered again through a 0·22-μm filter and re-suspended in 10 ml PSDF to remove the ethanol.

Kanamycin inactivation of the inoculum was achieved by dilution of the 24 h culture to the desired concentration. Kanamycin (200 μg ml−1) was added and the culture incubated for 30 min at 20°C after which time all cells were unable to form colonies on PCA (data not shown). The cells were then removed from the suspension by centrifugation and re-suspended in 10 ml phosphate-buffered saline (PBS) to provide the Kanamycin-inactivated inoculum.

Formalin-inactivation of the inoculum was achieved by fixation in 4% (v/v) formalin for 2 min. The cells were removed from the suspension by centrifugation and re-suspended in 10 ml PBS to provide the formalin-inactivated inoculum.

Preparation of inoculated potato tissue samples for scanning electron microscopy

The surface of potatoes was sterilized by spraying with 70% (v/v) ethanol. A sterile knife was used to remove two opposing surfaces of the potato revealing an area of sterile inner potato tissue. A sterile cork borer (8 mm diameter) was then used to produce cores of sterile potato tissue 4 cm long. The potato cores were blanched by immersion in boiling water for 30 s. Blanching allowed the potato cells to separate from each other when snapped but still remain as intact cells (Parker et al. 2001). Blanching for 30 s resulted in a gradient of cell separation across the potato tissue core with raw (un-separated) cells in the centre of the core and separated cells at the outer surface. Post-blanching, cores were rinsed in 100 ml cold sterile glass distilled water for 5 min to remove excess starch. After rinsing, each core was aseptically snapped in half and a 3-mm section aseptically cut from each snapped end and placed, snapped face uppermost, in a beaker containing the inoculum to be investigated.

The potato tissue was exposed to the inoculum for 10 min at 20°C to allow attachment to the tissue. Tissue samples were then rinsed twice to remove unattached cells. Rinsing was by placing tissue samples in a beaker of 100 ml sterile PSDF which was stirred for 1 min at 150 rev min−1 using a magnetic stirrer.

After rinsing, the tissue samples were prepared for SEM according to a previously described method (Parker and Waldron 1995). In summary, the potato tissue was fixed in 3% (w/v) glutaraldehyde in 0·05 mol l−1 cacodylate buffer (pH 7·2) for 2 h, dehydrated in an ethanol series and transferred to acetone. They were then dried by critical point method using liquid CO2 as the transition fluid and mounted, snapped surface uppermost, onto aluminium stubs using silver conducting paint. All samples were then sputter coated with a layer of gold and imaged in a Leica Cambridge Stereoscan 360 SEM.

Preparation of potato cell-wall material

Cell-wall material (CWM) was prepared based on the method of Parker and Waldron (1995). In summary, the surface of the potato was sterilized as described earlier and the potato sliced transversely (6 mm thick). Parts within the vascular ring were cut out from slices from the middle third of the axis. The tissue was frozen in liquid nitrogen and stored. Batches of potato tissue (500 g) were blended for 3 min in 1·5% sodium dodecyl sulfate (1000 ml) + 5 mmol l−1 Na2S2O5 + 5 ml octanol in a Waring Blender (Christison Particle Technologies Ltd, Gateshead, UK). The mixture was filtered on a 2-mm mesh to remove unblended tissue and the filtrate was homogenized with an Ystral homogenizer (Ystral GmbH, Dottingen, Germany) at 16 000 rev min−1 for 1 min, then filtered using a 200-μm mesh nylon cloth (BioDesign Inc. of New York, Carmel, NY, USA). The material retained was washed with water and then suspended in 0·5% (w/v) sodium dodecyl sulfate (500 ml) + 5 mmol l−1 Na2S2O5 + 2·5 ml octanol and then ball-milled for 4 h at 60 rev min−1 at 4°C in a 2·5-l pot (Capco Test Equipment Ltd, Wickham Market, Suffolk, UK). The liquor was poured out and each ball was washed with water. Most of the starch was removed at this stage by filtering on a 100-μm mesh nylon cloth. The material was washed with 10 l of water, re-suspended in water and homogenized (16 000 rev min−1 for 1 min), and then filtered and washed again on a 100-μm mesh nylon cloth. Absence of starch was assessed by staining with a solution of iodine in potassium iodide. The CWM was re-suspended in water and frozen at −20°C.

Measurement of the attachment of Salmonella serotype Typhimurium to potato CWM by flow cytometry

Flow cytometry was used to enumerate the number of live or dead Salm. Typhimurium JH3016 attached to potato CWM. Frozen stock CWM was defrosted and washed with PBS through a 30-μm nylon mesh filter using a Nalgene filter unit. All PBS used in this part of the study was filtered through a 0·22-μm-pore-sized filter (Millipore) to reduce the background noise during flow cytometry analysis. The CWM was divided equally into aliquots, and each was re-hydrated for 2 h in 10 ml PBS at 20°C rotating at 120 rev min−1 on an orbital shaker.

Hundred microlitres of viable, formalin- or Kanamycin-inactivated Salm. Typhimurium JH3016 were added to an aliquot (2·5 g wet weight) of the re-hydrated CWM and incubated at 20°C, 120 rev min−1 for 30 min to allow attachment. After this time, samples were washed using 100 ml PBS through a 30-μm filter to retain bacteria attached to the CWM. The CWM was then collected into a stomacher bag and stomached (Seward 80 Biomaster; Fisher Scientific, Loughborough, UK) for 60 s to remove the attached bacteria. The sample was washed through a 30-μm filter and the filtrate washed through a 5-μm filter to remove any residual CWM. The number of bacteria in the filtrate, and therefore previously attached to the CWM, was then enumerated by flow cytometry. For this, the sample containing viable cells was fixed in 4% (v/v) formalin for 2 min. After fixation of the ‘viable’ sample, and in the case of all samples of inactivated cells, the filtrate was centrifuged at 15 000 g for 5 min and washed twice by centrifugation with PBS. Samples were then immediately analysed by flow cytometry. This used a FACScalibur flow cytometer (Becton Dickinson, Franklin Lakes, NJ, USA) equipped with a 15-mW air-cooled argon ion laser as the excitation light source (488 nm). In order to perform absolute counts, samples were mixed with a known number of FluoSpheres® carboxylate-modified crimson fluorescent microspheres (Molecular probes F-8816; Invitrogen, Paisley, UK). The beads were detected on a separate channel from the constitutively GFP+-expressing Salmonella strain JH3016. All parameters were collected by using amplification gains set on LOG mode. Acquisition was stopped when 10 000 beads were counted, allowing the volume of sample used to be determined. The number of GFP+-expressing Salmonella cells detected in that same volume was subsequently calculated after analysis with CellQuest 3·3 software (Becton Dickinson) and converted into number of Salmonella cells per millilitre. Control samples of filtered PBS and uninoculated CWM were also analysed to establish background noise within the samples.

To validate the absolute counts obtained from calculations of bead and GFP+ fluorescence, a viable count was performed on each sample before fixation for flow cytometry. Fifty microlitres of the sample were inoculated onto the surface of duplicate plates of PCA using a Spiral Plate Maker (Don Whitley Scientific, Shipley, UK). PCA plates were incubated at 30°C for 24 h before enumeration.


Salmonella serotype Typhimurium must be viable in order to attach to potato tissue and cell-wall material

The mechanism of Salmonella attachment to vegetable tissue is unknown. Previous unpublished work by this laboratory led to the hypothesis that cells must be viable in order to attach to vegetable tissue. To explore this hypothesis, viable and inactivated cells of Salm. Typhimurium were incubated with potato tissue to allow attachment to occur and the tissues were visualized using SEM. Figure 1 shows that only viable cells of Salm. Typhimurium attached to the potato tissue. The method of inactivation appeared to be irrelevant; both heat-inactivated (Fig. 1b,d) and ethanol-inactivated (data not shown) organisms failed to attach. Figure 1(a) also identified the plant cell wall as the preferred site of attachment for Salm. Typhimurium. In particular, they appeared to attach to the pectin layer at the cell-wall junction. Cooking (via blanching) separates cells by partially breaking down this pectin layer resulting in a larger exposure of pectin-rich material. The image of cooked tissue (Fig. 1c) presented here shows less pectin present at the cell-wall junction and also less Salmonella attachment. It is, therefore, possible that pectin is one component of the cell wall that is a bacterial attachment site.

Figure 1.

 Scanning electron micrographs of Salmonella serotype Typhimurium LT-2 attached to raw and blanched (cooked) potato tissue. (a) Live Salm. Typhimurium LT-2 incubated with raw potato tissue showing attachment to cell-wall junction. (b) Heat-inactivated Salm. Typhimurium LT-2 incubated with raw potato tissue; no attachment occurred. (c) Live Salm. Typhimurium LT-2 incubated with blanched potato tissue; blanched potato tissue showing less pectin present and less attachment. (d) Heat-inactivated Salm. Typhimurium LT-2 incubated with blanched potato tissue; no attachment occurred. Scale bar is 10 μm.

Plant cell-wall components are a site of bacterial attachment to potato tissue

The SEM images of intact potato tissue indicated the plant cell-wall junction was the preferred site of attachment for live Salm. Typhimurium (Fig. 1a,c). Purified CWM contains only cell-wall components and no other cellular material (Parker and Waldron 1995) and so can be used to confirm that (i) a component of the cell wall was the site of Salmonella attachment to potato tissue and (ii) that this attachment required Salmonella to be viable.

Attachment of Salmonella to potato CWM was determined via flow cytometric analysis using a GFP+ strain of Salm. Typhimurium, strain JH3016. Samples derived from CWM incubated with live JH3016 fluoresced in the GFP+ range indicating organisms were present in this sample and therefore had attached to the potato CWM (Fig. 2a). The count of fluorescent particles was used to enumerate the beads and GFP+-expressing Salmonella. The number of viable cells derived from this count was 4·1 × 106 CFU g−1. This count was validated by the number derived from the conventional viable counts carried out simultaneously (3·3 × 106 CFU g−1).

Figure 2.

 Flow cytometric analysis of attachment of Salmonella serotype Typhimurium JH3016 to potato cell-wall material (CWM). (a) Fluorescence readings of samples of cell-wall material incubated with live JH3016. (b) Fluorescence readings of samples of CWM incubated with formalin-inactivated JH3016. (c) Fluorescence readings of samples of CWM in phosphate buffered saline (PBS). (d) Fluorescence readings of samples derived from PBS only. Readings are means of duplicate experiments.

In contrast, in the samples derived from CWM incubated with formalin-inactivated JH3016, only small counts of fluorescent particles were seen (Fig. 2b), equating to 2·4 × 104 CFU g−1. Parallel traditional viable counts did not recover any viable Salmonella; therefore, these counts may have been other particles that fluoresced in the GFP+ range. The same was observed in samples incubated with Kanamycin-inactivated JH3016 (data not shown) and in the control samples; CWM only and PBS (Fig. 2c,d). These results confirm that the plant cell wall contains a receptor site for bacterial attachment to potato tissue and that for this attachment to occur the bacteria must be viable.


Bacterial contamination of ready-to-eat vegetable tissues is an emerging issue with regards to the safety of these products. Many studies have concentrated on the growth of bacteria in the products rather than on the initial attachment phase and the subsequent consolidation event. With decontamination protocols occurring within minutes of processing, the understanding of the initial phases of attachment and consolidation are key to the development of new, more effective decontamination protocols.

Previously, Salmonella Typhimurium has been shown to attach equally to both cut and intact surfaces of lettuce tissue (Takeuchi et al. 2000). This study demonstrates Salm. Typhimurium preferentially binds to material at the cell-wall junctions of intact potato tissue. It was also shown to attach to isolated CWM. These results indicate that a plant cell-wall component is one potential bacterial receptor site for Salmonella. Studies to identify the particular component of the plant cell wall are ongoing.

Although a potential site of attachment has been identified, the mechanism by which Salmonella attach to vegetable tissues is unknown. Attempts to mask receptor sites on potato tissue with dead Salmonella lead to the hypothesis that attachment requires cells to be metabolically active. Using four different methods of inactivation, we showed that Salm. Typhimurium must be viable in order to attach to vegetable tissues. This was in contrast to studies with E. coli where both live and inactivated bacteria attached to lettuce tissue (Soloman and Matthews 2006). In their study, Soloman and Matthews (2006) used glutaraldehyde to inactivate the cells. Glutaraldehyde is known to alter the adhesive properties of the bacterial membrane rendering the membrane more ‘sticky’. This alteration may account for why inactivated E. coli cells were able to attach to the tissue. In our study, the inability to attach was found to be a direct result of viability rather than modification of proteins on the membrane surface. Attachment by bacteria to plant tissues is believed to occur via surface structures on the bacterial membrane such as fimbrae, pili and lipopolysaccharide interacting with proteins on the plant cell wall (Mandrell et al. 2006). Formalin, ethanol and heat inactivation alters the protein structures on the bacterial membrane and so may prevent this recognition. Therefore, lack of attachment in these inactivated cells may be as a result of altered membrane structures rather than lack of metabolic activity. Kanamycin inactivates bacteria by inhibiting protein synthesis and does not affect the structure of proteins already present on the bacterial membrane. The inability of Kanamycin-inactivated Salm. Typhimurium to attach indicates that the attachment process requires protein synthesis and thus bacteria need to be viable. The plant pathogen Agrobacterium initially attaches loosely to the plant cell wall and then synthesizes cellulose fibrils that bind the organism tightly to the cell surface (Matthysse 1986). Salmonella serotype Typhimurium has recently been shown to possess the ability to synthesize cellulose (Zogaj et al. 2001). Therefore, it is possible that Salmonella uses a similar mechanism to attach to plant cell walls, i.e. an initial weak attachment by bacterial surface proteins interacting with the cell wall followed by the synthesis of cellulose to ensure strong attachment. This would explain the need for protein synthesis in the attachment process. The genes and mechanism involved in cellulose production in Salmonella are being elucidated (Zogaj et al. 2001; Barak et al. 2005).

It is recognized that the use of biocides as decontaminants is not always successful (Adams et al. 1989; Garg et al. 1990; Brocklehurst 1994; Zhang and Farber 1996); accordingly this work forms part of a wider ongoing study to understand the mechanisms used by bacteria for attachment to tissues with a view to developing new and more effective decontamination techniques. In particular, this paper focuses on the competencies of the organism and particularly the discovery that organisms must be viable for attachment to occur. This has implications for novel decontamination interventions, such as the potential use of a vaccine-like approach where dead bacteria could be used to prevent attachment of living bacteria. It is quite clear that if cells cannot attach when they are not viable, then the use of such a vaccine approach would be irrelevant. We intend to study further the initial attachment phase of Salm. Typhimurium to vegetable tissue using the genomic and proteomic techniques available to us with a view to elucidating further mechanisms that may be exploited to create novel and effective decontamination techniques that can be employed by the food industry.

In summary, this study indicates the pectin-rich area of the plant cell wall may act as a receptor site for bacterial attachment to vegetable tissue. Results also illustrate that in order for attachment to occur bacteria must be metabolically active and capable of protein synthesis.


The authors would like to acknowledge the financial support of the Biotechnology and Biological Sciences Research Council for this work. We also thank Dr Isabelle Hautefort for the flow cytometry analyses.