Differential Attachment to and Subsequent Contamination of Agricultural Crops by Salmonella enterica

Abstract

U.S. salmonellosis outbreaks have occurred following consumption of tomato and cantaloupe but not lettuce. We report differential contamination among agricultural seedlings by Salmonella enterica via soil. Members of the family Brassicaceae had a higher incidence of outbreak than carrot, lettuce, and tomato. Once they were contaminated, phyllosphere populations were similar, except for tomato. Contamination differences exist among tomato cultivars.

Other than disease outbreaks caused by sprouted seeds, salmonellosis outbreaks associated with fresh produce have been attributed most frequently to consumption of tomatoes (6, 7), followed by consumption of cantaloupes (4). Although contamination of leafy greens by Shiga toxin-producing Escherichia coli has caused food-borne illness (8), outbreaks attributed to Salmonella enterica contamination of leafy greens have not been reported in the United States and are rare worldwide, except for those occurring in the United Kingdom (1, 13, 21, 22). Research to date has not addressed this discrepancy between those crops associated with salmonellosis; thus, we examined the ability of S. enterica to colonize agricultural crops seeded in contaminated soil, a probable route of preharvest contamination (5).

Enriched potting soil (total nitrogen, 0.14%; available P₂O₅, 0.09%; soluble potash, K₂O 0.02%; total iron, 0.25%; Canadian sphagnum peat moss, ground fir bark, compost, and sand in a proprietary blend) at a pH of 5.5 to 6.5 was irrigated once with an S. enterica suspension containing a mixture of eight strains from fresh produce salmonellosis outbreaks. Bacteria were struck from frozen stock and grown overnight on Luria-Bertani-kanamycin medium (40 mg/liter) plates at 37°C. Equal numbers (10⁴ CFU/ml) of the S. enterica serovars Baildon 05x-02123 (9), Cubana 98A9878 (20), Enteritidis 99A-23 (California Health Department [CHD], July 2005 tomato outbreak), Havana 98A4399 (20), Mbandaka 99A1670 (CHD, alfalfa seed isolate), Newport 96E01152C-TX (14), Poona 00A3563 (CHD, cantaloupe outbreak), and Schwarzengrund 96E01152C (14) comprised the bacterial suspension in sterile water. To verify that the black colonies recovered from Salmonella-Shigella agar with kanamycin (SS-Kan, 50 μg/ml) were inoculated strains, each strain was transformed with pKT-Kan, a broad-host-range vector that confers kanamycin resistance and green fluorescent protein expression (19). This plasmid has been shown to have no effect on the survival and growth of S. enterica on plants (2). Twenty-four hours later, surface-sanitized seeds (2) were sown in the contaminated soil, and pots were kept in a controlled environment growth chamber under a day-and-night cycle of 12 h, during which the day temperature was 26°C and the night temperature was 18°C, similar to that of an average spring season in northern California, and humidity was constant at 75%. Pots were irrigated every other day with 25 ml of sterile water, which resulted in no drainage.

Preliminary experiments revealed no differences among rhizoplane S. enterica populations on different plants (data not shown). Thus, only the phyllosphere was sampled at the one true leaf stage, as previously described (3). Seed germination was similar among each crop, and thus, seedlings of each crop were sampled on the same day. Briefly, phyllosphere samples were cut at the soil line, placed in sterile water, and vortexed for 30 s; sterile water was diluted, and aliquots were spread on SS-Kan agar for population enumeration. To the remainder of each sample, 1 ml of LB broth (with kanamycin) was added, and these enriched samples were incubated overnight (∼10 h) with 250 rpm shaking. If no colonies grew on the original SS-Kan enumeration plate, 10 μl of the enrichment was streaked on SS-Kan agar to confirm the presence of S. enterica in the plant samples. Plants were scored as positive if S. enterica populations could be enumerated, with or without enrichment. All media were obtained from Becton Dickinson (Franklin Lakes, NJ).

When seeds from six agricultural crops representing four plant families were sown in the same pots, radish, turnip, and broccoli (family Brassicaceae) seedlings had a significantly higher contamination incidence (CI) than those of carrot, lettuce, and tomato (P < 0.0001; Table 1). When seeds of leafy greens representing members of the families Apiaceae (cilantro and parsley), Asteraceae (radicchio, endive, and lettuce), and Chenopodiacae (spinach) were sown in the same pots of contaminated soil, radicchio and endive had a significantly higher CI than lettuce (P = 0.0064); however, there were no significant differences among lettuce cultivars (Table 1). Results from other studies concur with our findings of low CI for lettuce from seeds sown in S. enterica-contaminated soil (11) and no differences in CI among phyllosphere populations of different lettuce cultivars (18). However, this is the first report of differential S. enterica CI among Asteraceae members, radicchio ≥ endive > lettuce. It is also evidence of the poor attachment capacity of S. enterica on lettuce. We conclude that CI differences are reflective of differences in the bacteria's initial ability to attach to the seedling and the inability of S. enterica to grow in the phyllosphere of these crops (3, 15, 16).

TABLE 1.

S. enterica CI of the phyllosphere of plants seeded in contaminated soil

Plant	% of CIa
Mixed plants
Radish (Raphanus sativus)	98.1 a
Turnip (Brassica rapa var. rapa)	97.2 a
Broccoli (Brassica oleracea var. botrytis)	94.0 a
Carrot (Daucus carota subsp. sativus)	73.9 b
Lettuceb (Lactuca sativa)	60.5 b
Tomatoc (Lycopersicon esculentum)	46.2 b
Mixed greens
Radicchio (Cichorium intybus)	86.6 a
Endive (Cichorium endivia)	79.6 a
Cilantro (Coriandrum sativum)	66.7 b
Parsley (Petroselinum crispum)	60.6 b
Spinach (Spinacea oleracea)	56.8 bc
Lettuce (Lactuca sativa)	51.1 c
Lettuce cultivars
Balady Aswan	59.1 a
Salinas 88	58.3 a
Little Gem	53.8 a
PI251246	50.0 a
Pavane	48.0 a
Valmaine	44.0 a
Iceburg	39.4 a
La Brillante	36.0 a
Paris Island	35.7 a
Parade	23.5 a
Calmar	21.4 a
Tomato cultivars
Brandywine	61.1 a
Amish Paste	43.8 ab
Money Maker	41.2 ab
Rose	41.2 ab
Soldacki	28.6 ab
Stupice	22.2 ab
Green Grape	23.5 ab
San Marzano	17.6 ab
Nyarous	11.1 b
Yellow Pear	6.3 b

^aCI was recorded as the percentage of Salmonella recovered (1) or not (0) from each plant, with no fewer than 10 plants sampled per experiment. Data include results from all experiments. Letters a, b, and c represent statistical significance from a one-way analysis of variance and Tukey-Kramer multiple comparisons test. Similar letters indicate no significant differences, whereas different letters represent a P value of <0.05.

^bLettuce in the mixed plant and greens experiments was cultivar Paris Island.

^cTomato in the mixed plant experiments was Money Maker.

Due to the low CI of tomato, we also determined whether there might be differences among tomato cultivars. Of the 10 cultivars tested, most had CIs of between 43.8 and 17.6%; however, cultivar Brandywine (61.1%) had a significantly higher incidence than Nyarous and Yellow Pear (11.1% and 6.3%, respectively [Table 1; P = 0.0092]). Of the tomato seedlings contaminated with S. enterica, at least half had populations large enough to enumerate, ranging from 1.35 to 2.31 log CFU/g plant (data not shown); these populations were not significantly different.

In contrast to the low incidence and phyllosphere populations observed with tomato, lettuce had a significantly low CI (60.5%) compared to that of the Brassicaceae members (≥94%) but phyllosphere populations that were similar (Table 2). Furthermore, lettuce had the highest phyllosphere populations of the mixed leafy greens, although they were only significantly higher than the Apiaceae members. These data suggest that lettuce may not be naturally contaminated via soil, especially in a more realistic environment with nonhomogeneous soil contamination, such as a flood event or infected animal excretion and populations lower than 10⁴ CFU/g soil. However, once S. enterica attaches, high populations may be attained in the phyllosphere (3.06 log CFU/g plant tissue), which could be consumed and cause disease. Others have reported bacterial population sizes and CI for lettuce that were higher with soil contaminated with E. coli O157:H7 than with S. enterica (10), but a direct comparison of S. enterica contamination among agricultural crops has been lacking in the scientific literature. These results may suggest that United Kingdom salmonellosis outbreaks associated with lettuce were not the result of preharvest contamination (1, 13, 21, 22) or that soil was not the contamination route, as reported in the 2005 lettuce salmonellosis outbreak whose crop was irrigated with sewage (12). Although S. enterica can survive in soil for extended periods (17), preharvest contamination of leafy greens may be more likely from other routes, e.g., irrigation water.

TABLE 2.

S. enterica population of the phyllosphere of plants seeded in contaminated soil

Population	Plant species	Log population CFU/g plant tissue (SD)a
Mixed plants	Radish	3.3 (0.9) a
	Broccoli	3.2 (0.8) a
	Lettuce	3.3 (1.0) ab
	Carrot	3.0 (0.7) ab
	Turnip	2.6 (0.9) b
	Tomato	1.8 (0.7) c
Mixed greens	Lettuce	3.1 (1.2) a
	Radicchio	2.7 (0.9) a
	Spinach	2.7 (1.0) ab
	Endive	2.6 (1.1) ab
	Cilantro	2.3 (0.8) b
	Parsley	2.0 (0.3) b

^aValues in parentheses are standard deviations (SD). Letters represent the statistical significance from a one-way analysis of variance and Tukey-Kramer multiple comparisons test of the log populations. Similar letters indicate no significant differences, whereas different letters represent a P value of <0.05.

Our study is the first to report the poor natural attachment capacity and subsequent contamination via soil by Salmonella of lettuce and tomato compared directly to that of other agricultural crops. Since many salmonellosis outbreaks have been associated with consumption of tomato, these results support our earlier hypothesis that soil may not be the route of preharvest contamination (3). This report also highlights the possible existence of plant characteristics among tomato cultivars which appear to deter Salmonella attachment or survival in the phyllosphere and might be exploited to reduce the risk of contamination.