Introduction

Salmonella enterica subspecies enterica (designated as S. enterica) are some of the most commonly known bacterial pathogens which cause human illness. Often the disease is associated with the consumption of contaminated foods like pork or poultry meat and eggs or egg products. Since recently, many human pathogenic organisms have been recognized to exist on plant root or leaf surfaces (Lyytikainen et al., 2004; Brandl, 2006), and even inside plant tissues (Kutter et al., 2006; Rosenblueth and Martínez-Romero, 2006). For example, outbreaks of Salmonellosis have increasingly been traced back to contaminated fresh produce (Viswanathan and Kaur, 2001; Sivapalasingam et al., 2004). For greenhouse grown produce, enteric pathogens are mainly introduced as a result of bad hygiene (Beuchat and Ryu, 1997). In the field however, contamination of vegetable crops may occur via soil amended with manure from agricultural animals which are known reservoirs for Salmonellae (Viswanathan and Kaur, 2001; Natvig et al., 2002). Both manure and irrigation water contribute significantly to the spread of human pathogens onto fields and the crops growing there (Natvig et al., 2002; Solomon et al., 2002; Islam et al., 2004).

In recent years, it became evident that serovars of S. enterica are not only able to attach to and proliferate on the surface of plant tissues (Zenkteler et al., 1997; Solomon et al., 2002) but can also colonize plant tissues internally (Kutter et al., 2006). For example, gfp-tagged strains of S. enterica colonized the interior of tomato plants when grown hydroponically (Guo et al., 2001; Guo et al., 2002) and various S. enterica serovars were able to colonize Medicago sativa and other leguminous plants endophytically and epiphytically (Dong et al., 2003; Wang et al. 2006). Also, an avirulent strain of S. enterica serovar Typhimurium colonized carrots and radishes which were grown on a field treated with S. enterica serovar Typhimurium-contaminated composted manure or irrigation water (Islam et al., 2004). Just recently, S. enterica serovar Typhimurium LT2 and DT104h were found to endophytically colonize barley sprouts during growth in an axenic system (Kutter et al., 2006), and various enterobacteria, including S. enterica, were found to be natural endophytes of Conzattia multiflora (Wang et al., 2006).

Typically, plant defense responses upon colonization by these pathogens were observed, for example during colonization of Medicago truncatula by S. enterica serovar Typhimurium which resulted in the induction of salicylic acid—dependent and—independent plant defenses (Iniguez et al., 2005). The induction of salicylic acid plant defense pathway (salicylic acid resistance) appeared correlated to the expression of bacterial genes for TTSS-SPI effector proteins, whereas the presence of flagella only induced the SA-independent plant defense (induced systemic resistance). A recent study involved the molecular response of axenically grown lettuce to colonization by S. enterica serovar Dublin, from which a differential expression was indicated of various virulence and pathogenicity-related genes of lettuce, over time (Klerks et al., 2007).

From these studies it is evident that human pathogens like S. enterica are able to colonize fresh produce endophytically and epiphytically and interact at a molecular level with the host plant. Some previous research has been described investigating the conditions required for plant colonization by human pathogens (Rosenblueth and Martínez-Romero, 2006; Toth et al., 2006). Concerning the bacterial genes that are required for attachment to plant roots, some genes have been identified to be crucial for attachment (Barak et al., 2005). However, fundamental ecological questions concerning the route of infection of lettuce by S. enterica and ecological factors influencing the colonization efficiency have not yet been described. Such data are of main importance to understand the mechanism of transmission, which in its turn is required to define preventive actions to reduce or even eliminate the risk of food contamination originating from agricultural systems.

This research studied plant and microbial factors that influence the colonization efficiency of a set of epidemiologically important human pathogenic S. enterica serovars in association with commercially important lettuce. The effect of differences in cultivars and S. enterica serovars on colonization efficiency was investigated with respect to prevalence and degree of colonization. The role of the natural endophytic microflora in determining plant susceptibility was assessed by performing correlation analyses between the Shannon index (H) or the species richness and the number of S. enterica CFU per g lettuce. Finally, the contribution of root exudates to colonization efficiency of lettuce by S. enterica serovars was tested and bacterial genes that were induced over time in the presence of root exudates were identified by micro-array analysis.

Materials and methods

Plant material and bacterial strains

Liquid cultures of S. enterica serovar Dublin, S. enterica serovar Typhimurium, S. enterica serovar Enteritidis, S. enterica serovar Newport and S. enterica serovar Montevideo, were kindly provided by Dr H Aarts (RIKILT, The Netherlands) after overnight growth at 30 °C in tryptic soy broth. The cultures were maintained by plating on selective Hektoen enteric agar (Biotec Laboratories Ltd, Ipswich, Suffolk, UK) and were increased by overnight enrichment at 37 °C in buffered-peptone water (BPW).

Seeds of commercially available lettuce cultivar Tamburo, Nelly and Cancan were kindly provided by Mr Raats (Nickerson-Zwaan, The Netherlands). The seeds were sterilized in a solution of 1% sodium hypochloride and 0.01% Tween-20, and then rinsed in water (twice) for 1 min each. Subsequently, the seeds were air-dried for 1 h and stored.

Association of Salmonella serovars with lettuce Tamburo grown on Salmonella-contaminated soil

To determine if S. enterica serovars are able to colonize lettuce seedlings via the roots under seminatural conditions, lettuce seeds (cultivar Tamburo) were planted on S. enterica-contaminated manure-amended soil in a greenhouse. Fresh manure was collected from a Dutch organic dairy farm. Soil was collected from a field (60 kg of top layer of 20 cm) from the organic experimental farm the Droevendaal (Wageningen, the Netherlands). The soil consisted of 89% sand, 8% silt, 3% clay, a total nitrate (N) and carbon (C) of 2135 and 22400 mg per kg, 11% moisture and had a pH of 7.14. The manure contained 28.7% acid detergent fiber, 40.3% neutral detergent fiber, a total dissolved organic N and C of 740 mg per kg and 8167, 220 mg per kg ammonium, 8.14 mg per kg nitrate and had a pH of 6.8. Both substrates tested negative for presence of S. enterica, which was determined by plating directly on selective Hektoen enteric agar and by testing the total DNA extracts from 10 ml BPW enrichments of three random samples of 1 g of each substrate using real-time PCR analysis (Klerks et al., 2004). The manure was first inoculated with S. enterica serovar Typhimurium, S. enterica serovar Enteritidis or enterica serovar S. Dublin (108 CFUs per g fresh weight) and mixed thoroughly before addition to soil at a ratio of 1:10 fresh weight. The final S. enterica cell density was 107 CFU per g mixture. In total, 74 pots of 50 ml of each contaminated soil–manure mixture were prepared per S. enterica serovar. The negative control pots (74) consisted of non-S. enterica-inoculated manure–soil mixture. One lettuce seed was added to each pot. The seeds were covered with soil after planting. All 296 pots were placed on saucers in a greenhouse with 16 h of artificial light at 18 °C and 80% humidity, and twice a day plants were watered carefully on the saucers to avoid splashing. After 6 weeks, each plant was harvested by cutting the plant at the stem just above the soil, weighed and thoroughly rinsed once in 30 ml of sterilized water prior to analysis. Each wash fraction was centrifuged (5 min at 6000 g), the pellet was re-suspended in 100 μl of BPW and 40 μl was plated on Hektoen enteric agar, in duplicate. Next, the plants of each treatment were randomly divided in two sets. Each plant of the first set of plants was ground in 1 ml of BPW. From the second set of plants each shoot was surface disinfected in 70% ethanol and washed twice in sterile water prior to grinding in 1 ml of BPW (Klerks et al., 2007). Of each suspension with ground plant material 40 μl was plated on Hektoen enteric agar, in duplicate.

Differential colonization of lettuce cultivars with S. enterica serovars

To evaluate whether the endophytic colonization efficiency of lettuce by S. enterica is dependent on the lettuce cultivar or the S. enterica serovar, three lettuce cultivars were tested with five S. enterica serovars for the degree of endophytic colonization. First, sterile seeds of three cultivars Cancan, Tamburo and Nelly, were sprouted in sterile 0.5% Hoagland's water agar (Sigma Aldrich, Irvine, UK) in 15 ml glass tubes and incubated for 2 weeks at 21 °C with light/dark intervals of 12 h in a closed container with glass cover and placed in a growth chamber. Next, 45 seedlings of each lettuce cultivar were inoculated with five S. enterica serovars (S. enterica serovar Typhimurium, S. enterica serovar Enteritidis, S. enterica serovar Dublin, S. enterica serovar Montevideo and S. enterica serovar Newport), resulting in nine replicates per combination in three blocks. A Fresh BPW culture of each S. enterica serovar was 100-fold diluted in sterile Hoagland's solution and 10 μl (107 CFU ml−1) was carefully pipetted into the agar close to the roots of each lettuce seedling. Contact with leaves was avoided. After 7 days of incubation the shoots were harvested by removing the roots at the transition region. The shoot tissue of each plant was weighed and subsequently surface disinfected by washing for 10 s in 70% ethanol and rinsing twice with sterile water (Klerks et al., 2007). Then, leaf tissue was thoroughly ground in cold BPW. For molecular analyses 100 μl of suspension was used to extract DNA using the Plant DNeasy DNA extraction kit (Qiagen, Westburg, Leusden, The Netherlands). A dilution series was prepared (non-diluted, 20 × and 200 × diluted) from the ground tissue suspension and 40 μl of each dilution series was plated in duplicate on S. enterica-selective Hektoen enteric agar and incubated overnight at 37 °C. The colonies grown from the dilution series were counted and a random selection of S. enterica-like colonies and other colonies (68 in total) from each of the cultivar–serovar combinations were enriched by growing overnight in BPW at 37 °C. The bacteria were pelleted from these enrichment cultures by centrifuging 500 μl at 6000 g for 5 min. Next, the DNA was extracted from the pellet according to the protocol of the microbiological DNA extraction kit (MoBio Laboratories, Solana Beach, CA, USA). The purified DNA was eluted in 100 μl of elution buffer and stored at −20 °C until further use.

Detection and identification of S. enterica

S. enterica serovars were isolated from surface disinfected lettuce seedlings by dilution plating on selective Hektoen enteric agar. Surface disinfection of lettuce plants was performed as mentioned above. Total DNA was extracted from surface-disinfected lettuce tissue that was ground in a BPW suspension. For DNA extraction the Plant DNeasy DNA extraction kit was applied on 100 μl of suspension that was added to 400 μl of lysis buffer supplied with the kit. Further treatment was according to the supplied protocol of the DNA extraction kit (Qiagen). The purified DNA was eluted in 200 μl of elution buffer and stored at −20 °C. DNA was isolated as mentioned above, followed by Taqman PCR amplification using primers and probe sequences that were described in Klerks et al., 2004.

Relationship between Salmonella colonization and the natural endophytic microbial community

To investigate a possible relationship between endophytic colonization by S. enterica and natural endophytic microbial communities, the diversity index, Shannon index (H) (Shannon and Weaver, 1963) and the species richness of natural endophytic bacteria were determined for different lettuce cultivars. Denaturing gradient gel electrophoresis (DGGE) (Muyzer et al., 1993) was performed using the DNA extracts from three lettuce plants with high endophytic S. enterica populations and one S. enterica-negative plant of each serovar–cultivar combination from the previous experiment. These DNA samples (60 in total), and additional control DNA samples (DNA of each S. enterica serovar) were first subjected to 16S rDNA PCR using primers directed to bacterial ribosomal DNA that excluded amplification mitochondria and chloroplasts 16S ribosomal DNA sequences (Chelius and Triplett, 2001). The PCR mix (final volume 24 μl) consisted of 200 μM dNTP, 3.75 mM MgCl2, 1 × Stoffel buffer (Applied Biosystems, Foster City, CA, USA), 0.4 μM primer799F (Chelius and Triplett, 2001), 0.4 μM primer1492R (Lane, 1991), 0.05% bovine serum albumin and 0.1 U Amplitaq Stoffel polymerase. To each reaction tube 1 μl of the total DNA extract was added before PCR amplification was started. The PCR program was set at an initial incubation of 3 min at 95 °C, followed by 30 cycles of 20 s at 94 °C, 40 s at 53 °C and 40 s at 72 °C. The reaction was stopped after 7 min incubation at 72 °C.

The primary PCR was followed by a second PCR using the primers U968 (Engelen et al., 1995) and R1378 (Heuer and Smalla, 1997); the latter contained a strong 5′-GC-clamp required for DGGE analysis. The second PCR mix (total volume of 49 μl) consisted of 200 μM dNTP, 1 × SuperTaq buffer (Applied Biosystems), 0.4 μM primerU968, 0.4 μM primer R1378 and 1 U SuperTaq polymerase. To each reaction mix 1 μl of primary PCR product was added and the second PCR was started with an initial incubation of 4 min at 94 °C. This was followed by 30 cycles of 1 min at 94 °C, 1 min at 55 °C and 2 min at 72 °C. The PCR was stopped after 10 min incubation at 72 °C followed by 5 min incubation at 10 °C.

Of each second PCR product 20 μl was mixed with 10 μl of loading buffer and applied to the gradient gel. The 6% (w/v) polyacrylamide gels in 0.53 TAE buffer (20 mM Tris-acetate (pH 7.4), 10 mM sodium acetate, 0.5 mM disodium ethylenediamine tetraacetic acid) contained a linear denaturing gradient of 45–65% of urea and formamide. The gels were run for 15 h at 60 °C and 100 V. After electrophoresis, the gels were stained for 30 min with SYBR Gold I nucleic acid gel stain (Molecular Probes Europe, Leiden, the Netherlands) and bands were visualized using a Docugel V system apparatus with ultraviolet light (Biozym, Landgraaf, The Netherlands).

Response of Salmonella serovars to root exudates of lettuce Tamburo

To determine if root exudates affect the movement of S. enterica serovars, the occurrence of chemotaxis was studied. First, lettuce Tamburo plants were grown in Hoagland's agar (0.5%) for 4 weeks under axenic conditions. From each plant 1 g of agar was collected that contained root exudates. To remove the organic-soluble compounds from water-soluble compounds, 1 ml of water and 1 ml of ethyl acetate was added to the agar and ground using mortar and pestle. The suspension was centrifuged at max speed (8000 g) for 5 min. The upper layer (water-phase (WP)) was transferred to a new tube, 1 ml of fresh ethyl acetate was added and mixed thoroughly. After centrifugation the WP was transferred to a new collection tube and stored at −20 °C until further use. The final water-soluble exudate concentration was half the initial concentration in agar, as the final volume was similar to the volume of agar used for extraction of exudates.

For the chemotaxis experiments (modification of Adler, 1966), micro-capillaries (volume of 50 μl, diameter of 1 mm) were filled with 0.2% of Hoagland's agar, including 0.5% of the metabolism marker 2,3,5-triphenyl tetrazolium chloride to measure bacterial movement. In two separate experiments the WP of three lettuce plants (50 μl of WP per plant) were each tested in duplicate for chemotaxis of S. enterica serovar Typhimurium, S. enterica serovar Enteritidis, S. enterica serovar Dublin and water (in total 12 tubes per serovar). Of each S. enterica serovar 50 μl of 107 CFU ml−1 of phosphate buffer was added to 0.5 ml tubes. Phosphate buffer was used since it lacks any substrate, but retains the viability of the bacterial cells. One end of a capillary was positioned horizontally inside a 0.5-ml tube (also horizontal and sealed with parafilm) to allow contact with the bacterial suspension or control solution. The other end of the capillary was placed in another 0.5 ml tube containing the WP sample or negative control (water, Hoagland's solution or phosphate buffer), carefully covered with parafilm to prevent evaporation. The capillaries were incubated horizontally overnight at 37°C prior to observation of chemotaxis by color transition inside the capillaries.

Gene expression of S. enterica serovar Typhimurium in the presence of root exudates

To investigate the molecular response of S. enterica serovars to the presence of root exudates, gene expression of S. enterica serovar Typhimurium exposed to root exudates was analyzed over time. First, fresh enriched bacterial cultures of S. enterica serovar Typhimurium were prepared, centrifuged, washed twice with phosphate buffer and finally re-suspended in phosphate buffer. Four tubes were prepared containing 900 μl bacterial suspension. One tube contained the WP of plant exudates of one chemotaxis-inducing plant (cultivar Tamburo) and one tube contained a non-chemotaxis-inducing WP of another Tamburo plant, whereas the control tubes contained either 0.1% sucrose or water. The suspensions were then incubated at 37 °C in a heating block. At different time intervals 100 μl of each bacterial suspension was collected and immediately transferred to 350 μl of lysis buffer of the RNeasy RNA extraction kit. The time series consisted of 0, 10, 20 min postinoculation. Next, total RNA was extracted from 100 μl of the bacterial suspensions using the RNeasy RNA extraction kit (Qiagen). RNA was eluted in 100 μl of RNase-free water and stored at −80 °C until further use.

From the time series cDNA was prepared to allow gene expression analysis using a thematic micro-array of S. enterica serovar Typhimurium designed to specifically detect virulence, growth and stress-related genes (Hermans, 2007). Each sample for cDNA synthesis was subjected to amino-allyl-dUTP labeling for subsequent Cy3 and Cy5 labeling, according to the protocol described by Hermans (2007). After cDNA synthesis and Cy3 or Cy5 labeling, the cDNA was precipitated according to general sodium acetate/ethanol precipitation. After drying of the pellet the cDNA was re-suspended (in duplicate) in filter sterilized hybridization buffer (0.2% sodium dodecyl sulfate, 5 × Denhardt's solution, 5 × sodium saline citrate, 0.5 × formamide and 0.25 μg salmon sperm). Next, the samples were incubated for 10 min at 65 °C. Finally, for each cDNA sample the Cy3 reference labeled fractions of each sample were combined and mixed separately with each Cy5-labeled cDNA fractions at a 1:1 ratio. The cDNA samples were boiled prior to application to the micro-array. Specific hybridization was analyzed using the ScanArray 3000 confocal laser scanner (GSI Lumonics, Kanata, ON, Canada), measuring the fluorescence of each spot for Cy3 and Cy5 and four background areas around each spot. After calculating the signal-to-noise ratio of each spot (Hermans, 2007), the data were corrected for inter-chip and intra-chip variations and a specific labeling as described by Hermans (2007). The corrected micro-array expression profile of each gene was compared between the treatments (two root exudates, 0.1% sucrose and water).

Statistical analysis

χ2 Analysis was performed to assess the difference between the number of emerged lettuce plants that were grown on S. enterica-contaminated manure-amended soil, and the plants grown on non-contaminated manure-amended soil. A similar analysis (χ2 test) was performed to determine the difference between the S. enterica serovars with respect to the number of colonized plants.

To determine the effect of lettuce cultivar or S. enterica serovar on the prevalence of S. enterica CFU on/in lettuce seedlings a nonparametric test (Kruskall–Wallis) was performed. The interaction was determined by a χ2 test on the basis of the number of positive plants (prevalence). To determine if there was an interaction between cultivar and serovar with respect to the degree of endophytic colonization of lettuce plants, univariate analysis of variance (including the interaction term cultivar*serovar) was performed on the number of CFU per g of fresh tissue.

The DGGE banding patterns were analyzed using Gelcompar II software (version 1.61; Applied Maths, Woluwe, Belgium) to allow comparison of the gels. Each gel contained four marker lanes for reference purposes and background corrections were performed prior to identification of bands with settings of 5% significance threshold. Correspondence of bands between different samples was performed with 1% dynamic range settings. The Shannon index of diversity (H) and species richness were calculated (Van Diepeningen et al., 2005) and correlated with the degree of endophytic colonization by S. enterica serovars (log CFU g−1), using Pearson's correlation analysis (SPSS). In addition, the log CFU g−1 was regressed on both the diversity parameters.

To determine the significance of the root exudate treatment relative to the controls, χ2 analysis was performed on 54 capillary tubes which were positive or negative for chemotaxis of S. enterica serovars to root exudates. In total, 18 of these 54 tubes were controls without exudates (water, Hoagland's solutions or phosphate buffer). Relative attractiveness of exudates to different S. enterica serovars was also compared with a χ2 test.

Results

Colonization by S. enterica serovars of lettuce Tamburo grown on manure-amended soil

Of the 74 seeds planted per S. enterica serovar, 70% lettuce plants emerged compared to 85% of the non-inoculated control plants. Of the emerged plants, 16 out of 56 were colonized with S. enterica serovar Dublin, 8 out of 48 with S. enterica serovar Enteritidis and 14 out of 53 with S. enterica serovar Typhimurium. The number of emerged plants was significantly different between the non-inoculated plants and the S. enterica-inoculated plants (χ2=9.231; P=0.026). The number of colonized plants did not differ significantly among serovars (χ2=2.056; P=0.358). The control plants were all negative for S. enterica. For S. enterica serovar Dublin 3 out of 28 plants were also positive for endophytic colonization after surface disinfection. No endophytic colonization was observed for surface-disinfected plants grown on soil contaminated with S. enterica serovar Enteritidis or S. enterica serovar Typhimurium. The endophytic colonization of lettuce plants grown on S. enterica serovar Dublin-inoculated manure-amended soil was only 164 CFUs per g lettuce for S. enterica serovar Dublin if averaged over all tested plants (Table 1).

Table 1 Mean of endophytic Salmonella CFU per serovar in association with lettuce Tamburo grown on manure-amended soil or Hoagland's agar, 6 weeks after planting of lettuce seeds
Full size table

Endophytic colonization of lettuce cultivars by S. enterica serovars on Hoagland's agar

The number of endophytically colonized plants (prevalence) was significantly affected by the S. enterica serovar (P=0.024) but not by the lettuce cultivar (P=0.727), as determined by the nonparametric Kruskall–Wallis test. The percentages of lettuce plants endophytically colonized were 59%, 85%, 93%, 85% and 89% for S. enterica serovars Dublin, Enteritidis, Montevideo, Newport and Typhimurium, respectively. There was no significant interaction between cultivar and serovar with respect to prevalence of S. enterica CFU in lettuce seedlings (χ2=3.11; P=0.215).

Univariate analysis of variance indicated that there was a significant interaction between S. enterica serovar and cultivar with respect to the degree of endophytic colonization (CFU per g leaf) (P=0.047). This suggested a difference in colonization pattern of a specific lettuce cultivar by the five S. enterica serovars, in particular S. enterica serovar Typhimurium colonized cultivar Nelly more easily than the other cultivars (Figure 1). The overall lettuce cultivar effect on internal colonization (CFU per g leaf) was not significant (P=0.116), while the S. enterica colonization was significantly affected by S. enterica serovar (P=0.004) (Figure 2). The endophytic colonization from inoculated Hoagland's agar was highest for S. enterica serovar Dublin and much lower for S. enterica serovar Typhimurium and S. enterica serovar Enteritidis. Although this trend between the S. enterica serovars tested on Hoagland's agar is similar to the trend obtained from the S. enterica serovars tested on manure-amended soil, the level of colonization was much higher in Hoagland's agar (Table 1).

Figure 1
figure 1

Box plot of the degree of colonization per serovar Salmonella Dublin, S. Enteritidis, S. Montevideo, S. Newport and S. Typhimurium for each cultivar Cancan (a), Nelly (b) and Tamburo (c) separately. CFU, colony forming unit.

Full size image
Figure 2
figure 2

Number of lettuce seedlings in different classes of degree of endophytic Salmonella colonization for each Salmonella serovar was tested.

Full size image

Natural endophytic microbial communities of S. enterica-colonized lettuce cultivars

Correlation analysis of the natural endophytic species richness based on number of DGGE bands versus the degree of endophytic colonization (log S. enterica CFU per g fresh weight) indicated a significant negative correlation (r=−0.31 and P=0.04). However, analyses of the cultivars separately did not result in significant correlations (Cancan, r=−0.508 with P=0.064; Nelly, r=−0.389 with P=0.151; Tamburo, r=0.039 with P=0.889) although a negative trend was observed with Cancan and Nelly, but not with Tamburo. Correlation analysis of the Shannon index of diversity (H), on the basis of the number and relative intensity of the bands on a sample lane, versus the degree of endophytic colonization (log S. enterica CFU per g) was not significant (r=−0.276; P=0.066).

Response of S. enterica to lettuce root exudates

The root exudates of three Tamburo plants resulted in metabolic activity (red coloring) inside the micro-capillaries of S. enterica serovar Typhimurium (8 out of 12 positive), S. enterica serovar Dublin (5 out of 12) and S. enterica serovar Enteritidis (0 out of 12, that is no activity observed) (Figure 3). Each control tube (treatments without exudates was negative for color transition (18 out of 18), which indicated that no random, passive movement due to diffusion occurred in these capillaries. A χ2 test on the total number of positive samples (movement inside tube) between the control tubes and the tubes containing root exudates showed a significant difference (χ2=8.56; P=0.05). There were also significant differences among the S. enterica serovars with respect to number of positive reactions in response to the root exudates (χ2=11.8; P=0.01). There was no differences between S. enterica serovar Typhimurium and S. enterica serovar Dublin (χ2=1.5; P=0.53), but there were differences between S. enterica serovar Dublin and S. enterica serovar Enteritidis: (χ2=6.3; P=0.05) and between S. enterica serovar Typhimurium and S. enterica serovar Enteritidis (χ2=12; P=0.01).

Figure 3
figure 3

Chemotaxis of Salmonella serovars in microcapillary tubes with root exudates or control solutions traced by reaction with TTC. The left side of a microcapillary tube was placed in a suspension of Salmonella (or water as control) present in a 0.5 ml eppendorf tube (A). The right end of the micro-capillary tube (B) was inserted into another 0.5 ml tube containing either the water fraction of root exudates or control solution (phosphate buffer, Hoagland's solution or water). Movement of Salmonella serovars was visualized by tetrazolium (red color). The upper micro-capillary tube was positive for movement of S. Typhimurium toward lettuce root exudates (C). The middle tube indicated movement of S. Dublin to root exudates. The bottom micro-capillary was negative for chemotaxis, having Salmonella inoculated on the left end and a control solution on the right end of the capillary. TTC, 2,3,5-triphenyl tetrazolium chloride.

Full size image

Gene-expression of S. enterica serovar Typhimurium in response to lettuce root exudates

To test the response of S. enterica serovars upon exposure to root exudates, micro-array analysis was performed on the time series of S. enterica serovar Typhimurium (0, 10 and 20 min postinoculation) of each treatment. After normalization (according to Hermans, 2007), most of the genes were equally expressed between the different treatments. However, some genes did show differential expression levels in time between the treatments (chemotaxis-inducing root exudate, a non-inducing root exudate, 0.1% sucrose and water) (Table 2). The differentially expressed genes (due to chemotaxis-inducing root exudates) appeared either associated with pathogenicity or pointed toward a relationship with a sugar-like carbon source.

Table 2 Corrected gene expression levels from micro-array analysis of S. Typhimurium incubated with water, 0.1% sucrose, a chemotaxis-inducing root exudate and a non-chemotaxis-inducing root exudate, in time
Full size table

The genes related to a sugar-like carbon source were OtsA (trehalose-6-phosphate synthase), which utilizes glucose-6-phosphate as substrate (Giζver et al., 1988), UhpC (hexose phosphate utilization protein), which is a sensor for external glucose-6-phosphate) (Schwϕppe et al., 2003), MetE (methyltetrahydropteroyltriglutamate-homocysteine S-methyltransferase), a vitamin B12-independent enzyme (Urbanowski and Stauffer, 1989), DsrA (putative anti-silencer RNA), a regulator of transcription to express RcsA promoter, which in its turn is responsible for capsular polysaccharide synthesis (Sledjeski and Gottesman, 1995), RseA (sigma-E factor regulatory protein), which is involved in the storage of sigma which is released during stress (Ades et al., 1999), SsaH and SsaM (putative effector proteins), both regulators of secretion of the type III secretion system (Lee et al., 2000) and SpaO, involved in surface presentation of antigens, secretory proteins. Other genes that were differentially expressed were related to specific limiting factors like iron (gene SitD) or anaerobic respiration (TtrA).

Discussion

Several human pathogenic organisms have been recognized to exist on plant root or leaf surfaces (Lyytikainen et al., 2004; Brandl, 2006), and even inside plant roots (Kutter et al., 2006; Rosenblueth and Martínez-Romero, 2006). However, only few studies have investigated the physiological or molecular interaction between human pathogenic bacteria and plants (Plotnikova et al., 2000; Iniguez et al., 2005; Prithiviraj et al., 2005; Kutter et al., 2006; Klerks et al., 2007). Even fewer studies have been described concerning the conditions required for plant colonization by these pathogens (Toth et al., 2006, Brandl, 2006). To develop preventive strategies it is important to elucidate the critical points of plant colonization by the pathogen. This research presents plant and microbial factors that influence the (endophytic) colonization efficiency of human pathogenic S. enterica serovars in association with lettuce.

In this study S. enterica serovars colonized lettuce endophytically as was shown before (Cooley et al., 2003; Klerks et al., 2007). However, we showed for the first time a differential interaction between S. enterica serovars and plant cultivars, besides serovar-dependent host susceptibility with respect to the degree of colonization. This result points to differences in susceptibility of the cultivars, but also differences between the S. enterica serovars with respect to colonization of lettuce seedlings. However, the difference between the serovars in apparent lettuce colonization might be partially determined by the survival of the different S. enterica serovars in manure-amended soil. S. enterica serovars were found to remain viable for at least several months even though a decline was observed (as expected) (Holley et al., 2006). Especially, in the first weeks after inoculation of S. enterica into the manure prior mixing with soil, no significant difference in survival rate (that is decline in viable cell numbers) between S. enterica serovars is expected. Of course an effect of the survival rate cannot be completely excluded, but it will only have a minor influence on the colonization efficiency compared to other factors like chemotaxis.

The observed differential S. enterica serovar-dependent host susceptibility suggested the presence of host-adapted serovars but also more resistant cultivars. Also, with respect to prevalence and degree of colonization, a large difference was present between soil-grown plants and axenically grown plants. This effect is mainly attributed to the absence/presence of rhizosphere bacteria. Since no bacteria are present that colonize the roots in an axenic system, the roots are easily accessible for the inoculated Salmonellae. The rhizosphere of roots grown in soil however are known to contain many different soil bacteria that colonize the roots already at the sprouting stage (Yang and Crowley, 2000), herewith protecting the roots with a shield of indigenous soil bacteria (Cooley et al., 2003; Berg et al., 2005). Moreover, the steep gradient of root-exuded compounds in soil is continually modulated by the indigenous soil microflora. The Salmonellae have to compete with these environmental bacteria to establish in the rhizosphere (Gagliardi and Karns, 2000; Cooley et al., 2003; Ibekwe et al., 2006), which leads to lower S. enterica cell densities close to the roots. This suggests that the colonization efficiency is strongly dependent on accessibility of the plant roots to the S. enterica serovars.

Next to being exposed to plant defenses, S. enterica serovars also have to compete with the plant natural endophytic microflora for a certain niche with carbon sources (Leveau and Lindow, 2000). Natural endophytes mainly reside at the intercellular spaces between the plant cells and in the vascular systems of the plant. However, also S. enterica has a preference for these regions, implying a direct interaction between the endophytical Salmonellae and the natural endophytic microflora. Indeed, the species richness of the natural endophytic microbial community of lettuce cultivars was negatively correlated to the number of endophytic S. enterica CFUs per g shoot tissue. When tested for each cultivar separately cultivar Cancan and Nelly showed a negative trend, but not cultivar Tamburo. This suggests that the microflora of Cancan and Nelly were more antagonistic to S. enterica, or at least limiting the endophytic colonization by S. enterica serovars compared to cultivar Tamburo. Thus, the natural endophytic microflora appeared to contribute significantly to the level of susceptibility of the lettuce cultivar for S. enterica serovars. Of course, measuring the complexity of the microbial community is inherently biased since only the most abundant species (maximum of approximately 100) are visualized by DGGE. Therefore, the use of the Shannon index with DGGE data can only be valid for quantitative analysis if the relative difference between treatments is tested (Shannon and Weaver, 1949). Thus, a less abundant species might have a higher impact on the colonization efficiency of S. enterica serovars than the visualized species from DGGE analysis.

In this study, a thematic micro-array was used (Hermans, 2007), thus not all expressed genes of S. enterica serovar Typhimurium were detected. Even though this micro-array covered the most important functional groups (that is, virulence, stress, metabolism and quorum sensing) that are expected to play a role in chemotaxis. Indeed, it appeared that genes for general metabolism are of importance for S. enterica serovars to actively move to the plant roots. The active movement of S. enterica serovars to the roots points to a tentative route of infection that is different from previously published passive contamination due to soil splashing, irrigation, insect transmission or even passive uptake through roots (Solomon et al., 2002). The chemotaxis experiment and micro-array data both confirmed the presence of an organic compound in the lettuce root exudates that is used as carbon source by the S. enterica serovars. Chemotaxis of S. enterica serovars to sugar suspensions was already described in the late 1970s (Melton et al., 1978) and it is also well known that root exudates contain various (mono)-saccharides like fructose, glucose and so on (Neumann and RÏ•mheld, 2001). Many bacteria colonize the roots of plants or persist in the rhizosphere by using these plant root exudates as carbon source (Curl and Truelove, 1986; Rediers et al., 2003). Thus, it is highly plausible that sugar-like compounds in the root exudates are responsible for the active movement of S. enterica serovars to the roots.

Interestingly, in the presence of root exudates and observed chemotaxis, also pathogenicity genes of S. enterica serovar Typhimurium were induced that are crucial for root attachment and subsequent colonization (Barak et al., 2005). For example, SsaH and SsaM are regulators of the type III secretion system and SpaO is involved in the surface presentation of antigens that is secretory proteins. The observed induction of DsrA implies the activation of processes that allow attachment, since it is involved in the synthesis of capsular polysaccharides, an important factor in the attachment to host cell surfaces (Eriksson de Rezende et al., 2005). Up to date it was not known that pathogenicity-related genes are already induced during chemotaxis toward the roots of plants. The findings of this study suggest that S. enterica serovars are conditioned for attachment to the plant root surface after being triggered by the root exudates to actively move toward the roots. Root exudates therefore have a dualistic result, which is the induction of chemotaxis of S. enterica to the roots and the simultaneous conditioning of S. enterica cells for host cell attachment. It is thus hypothesized that the capability of bacteria to condition for (plant) host cell attachment during chemotaxis is one of the most important factors for pathogenicity or colonization efficiency. This is in line with the observation that from the same lettuce cultivar both chemotaxis-inducing root exudates and non-inducing root exudates were extracted. This can be explained by the simple fact that most probably a difference in available root exudate concentration was obtained. Indeed plant colonization efficiency is dependent on the concentration or availability of the root exudates in the rhizosphere. This observed concentration-dependent chemotaxis should be investigated further to determine the relationship between exudate concentration and chemotaxis-dependent colonization efficiency.

Typically, amino acids could also play an important role during chemotaxis or survival close to roots, as the MetE gene, induced in the presence of mono-cysteine, was also upregulated (Urbanowski and Stauffer, 1989). Further analysis and identification of these organic compounds by high performance liquid chromatography separation and subsequent gas-chromatography mass spectrometry or liquid chromatography mass spectrometry might be of great interest to determine which compounds are responsible for attracting S. enterica serovars to the roots. In combination with gene-expression profile determination by reverse transcriptase (RT)-PCR this eventually might lead to identify marker genes for chemotaxis.

Tentative route of infection

To develop preventive strategies for infection of lettuce plants by S. enterica it is important to elucidate the critical points of plant colonization by the pathogen and to define a tentative route of infection. Basically, S. enterica serovars that are applied with manure onto or into the soil need to overcome several barriers to finally colonize the lettuce plant systemically. The impact of these barriers on the colonization efficiency is dependent on the serovar, the cultivar and the plant environment. Initially, the S. enterica serovars are triggered by sugar-like root exudates and due to chemotaxis the bacteria move actively toward the roots, meanwhile being conditioned for host cell attachment (this paper). The ability of S. enterica serovars to produce flagella and highly sensitive sensors for chemotaxis or quorum sensing compounds contributes largely to the rate of their movement (Melton et al., 1978). In close proximity to the roots, the attachment-conditioned Salmonellae compete with the rhizosphere bacteria to gain intimate access to the roots. The rhizosphere is recognized to serve as reservoir for human pathogenic bacteria (Berg et al., 2005), in which the availability of anaerobic and aerobic respiration pathways might enable S. enterica cells to compete with rhizosphere bacteria for nutrients. The Salmonellae efficiently attach to the roots, which is primarily dependent on the presence of using curli and lipopolysaccharides (Barak et al., 2005). In general the colonization of the plant roots by bacteria depends on the presence of flagella, the O-antigen of lipopolysaccharides, the growth rate and the ability to grow on root exudates (Lugtenberg and Dekkers, 1999). Upon root colonization the Salmonellae form a biofilm on the roots at natural openings or wounds, but also at the intercellular spaces between epidermal cells (Klerks et al., 2007). This suggests a preference for the intercellular spaces and not the plant cell itself.

Upon cell disruption or presentation of membrane-bound pathogenicity factors like flagella (Iniguez et al., 2005), the plant responds in a hypersensitive manner with subsequent triggering of induced systemic resistance. As soon as the type III secretion system translocates effector proteins into the host cells, the plant may respond by inducing the salicylic acid defense pathway (salicylic acid resistance). To what extent the plant responds by induced systemic resistance; and salicylic acid resistance depends on its ability and efficiency to recognize bacterial pathogenicity factors, degrading action and/or host cell penetration, which for each plant cultivar and species is different. Finally, endophytic spread follows that of plant pathogens by first infecting the vascular parenchyma and then the invasion of the xylem to allow systemic infection (Klerks et al., 2007).

Frequent intimate interaction between human pathogens and plants may make different plant species susceptible to adjusted strains and become secondary hosts. Eventually, this could lead to a higher prevalence of S. enterica serovars inside produce so that these cannot be removed by sanitation procedures. Even though, it should be stressed that, although the number of Salmonellosis outbreaks associated with consumption of produce is increasing, the risk is still relatively low compared to that of Salmonellosis associated with poultry or eggs. Even so, the potential presence of human pathogens inside lettuce should be taken into consideration as an emerging risk for human health.