The number of foodborne illness outbreaks linked to fresh produce has increased in the last years (Beuchat, 1996; Lynch et al., 2009). Pathogens most commonly associated with the consumption of fresh produce are viruses, bacteria and protozoa (DeWaal et al., 2007). These microorganisms cause approximately 76 million new cases of food-related illness in the United States each year, resulting in 325,000 hospitalizations and 5,000 deaths, while the economic cost generated by these events is around 152 billion dollars (Scharff, 2010). Consumption of tomatoes and peppers has been recently responsible of foodborne outbreaks caused by Salmonella serovars, which have involved from sporadic cases to multistate outbreaks, implicating severe problems in public health (Hedberg et al., 1999; Behravesh et al., 2011). Fresh produce contamination can occur from the field to the table; according to Lynch et al. (2009) the probability of contamination increases in specific production steps, including field work, packinghouse, and also during final processing in the kitchen. Salmonella is the main bacteria responsible for fresh produce outbreaks (Hanning et al., 2009; Fatica and Schneider, 2011), being the leading cause of hospitalization and deaths with a 35 and 28%, respectively (González and Rojas, 2005). Outbreaks of Salmonella Saint Paul occurred in 2008 and 2013 in the United States sickened 1,442 people in 43 states, and 81 people in 18 states, respectively. These outbreaks were associated (as suspects) with the consumption of jalapeño pepper and cucumber (CDC, 2008, 2013a). The occurrence of outbreaks evidences the failure of physical and chemical interventions methods (Parish et al., 2003).

In this regard, the use of bacteriophages as biocontrol, individually or combined with chemical products, has been proposed (Xu, 1999). Previous reports have indicated the efficacy of bacteriophages to eliminate Salmonella from meat, fresh produce and processed foods. Leverentz et al. (2001) applied a mixture of 4 lytic bacteriophages (SCPLX-1) specific for Salmonella Enteritidis on pre-cut melons and apples stored at temperatures of 5, 10 and 20°C in a period of 168 h achieving bacterial reductions of 3.5 Log₁₀CFU/melon at 5 and 10°C, and 2.5 Log₁₀CFU/melon at 20°C. Another study showed the effect of bacteriophages over S. Enteritidis suppression in Cheddar cheese produced from raw and pasteurized milk and stored at 8°C; S. Enteritidis did not survive in pasteurized milk cheese after 89 days in the presence of the phage. However, S. Enteritidis was able to survive around 50 CFU/g in raw milk cheese containing phage even after 99 days of storage (Modi et al., 2001). Waseh et al. (2010) orally administered P22 phage tailspike protein in chicken infected with Salmonella and found a reduction in bacterial colonization. These findings coincide with the ability of bacteriophages to reduce bacterial populations from foods.

Considering that S. enterica is a major foodborne pathogen generally involved in fresh produce-borne outbreaks (CDC, 2013b) and that tomatoes have been implicated in recent illness cases (CDC, 2013c), in this study we determined the efficacy of bacteriophage P22 to reduce S. enterica subsp. enterica serovar Typhimurium on tomato surface in order to demonstrate the feasibility of successful biocontrol against Salmonella.

Bacterium inoculum

Host strain: Salmonella Typhimurium LT2 (ATCC^® 19585™) inoculum was obtained by the addition of a loop of the bacterium to 6 L of trypticase soy broth (TSB, Bioxon, México) and incubated for 24 h at 37°C. Bacteria were concentrated to 500 ml using an ultra-filtration system (Hernandez-Morga et al., 2009). The bacterial concentrate was then centrifuged at 13,800×g for 10 min at 4°C (Thermo IEC Multi RF, 8466, USA). The pellet was washed two times with 100 ml of sterile monobasic phosphate buffered solution (PBS) [0.01 M, pH 7.2, Sigma-Aldrich, Inc. Saint Louis, MO, USA] and resuspended in 6 L of PBS. The cell suspension was adjusted to a final OD₆₀₀ of 1 (Eppendorf, Hamburg, Germany), according to Leverentz et al. (2001). The bacterial inoculum was quantified using the spread plate technique to obtain a final concentration of viable bacteria of 10⁹ CFU/ml (APHA, 1998).

Bacteriophage P22 propagation

Bacteriophage P22 was kindly provided by Dr. Charles P. Gerba (University of Arizona). In order to propagate the bacteriophage, five millilitre of the P22 stock and 50 ml of S. Typhimurium (10⁹ CFU/ml) were added to two litters of sterile PBS, pH 7.2 (Whichard et al., 2003), and incubated in a water bath overnight at 37°C and 110 rpm (Shak R Bath 3580 R, Lab-Line Instruments, Inc, USA). The mixture was then concentrated to 500 ml with an ultra-filtration system adapted from Hill et al. (2005) and Hernández-Morga et al. (2009), followed by centrifugation at 13,800×g for 10 min at 4°C. The supernatant was filtered through sterile cellulose acetate 0.22 µm membrane filters (Corning^®, NY, USA). The bacteriophage titer was determined by the soft agar overlay technique (APHA, 1998). The final bacteriophage concentration was 10¹¹ PFU/ml.

In vitro assay for bacterial biocontrol

In order to determine the ability of bacteriophage P22 to lyse S. Typhimurium (ATCC^® 19585™), an experiment in liquid culture at 37°C for 48 h was performed. One colony of S. Typhimurium was grown in 10 ml of TSB at 37°C overnight. Then, aliquots of 1 ml were individually transferred to six flasks, each containing 49 ml of pre-warmed TSB and incubated at 37°C until the OD₆₀₀ was 0.5 (ca. 1.5×10⁸ CFU/ml). The bacteriophage P22 was individually inoculated into the bacterial cultures at a theoretical MOI of 100 and 1000 (in duplicated); additionally, two flasks were not inoculated (Control). Bacterial titer was measured in Hektoen agar plates at 2, 4, 6, 8, 12, 24 and 48 h after bacteriophage inoculation. The experiment was performed in duplicate.

Tomato preparation

Roma saladette tomatoes under physiological maturity were obtained from the greenhouse of the Centro de Investigación en Alimentación y Desarrollo, Culiacán Station. Homogeneous size and free of mechanical damage tomatoes were selected and disinfected by immersion into a sodium hypochlorite solution (Fermont, México) adjusted at 300 mg/L for 10 min. Then tomatoes were immersed in a 5% sodium thiosulfate solution (Fermont, México) for 10 min, and finally rinsed with sterile purified water and stored at 4°C for 12 h prior use. Tomatoes were evaluated for the absence of microbial contaminants after the disinfection procedure.

Biocontrol of S. Typhimurium on tomato surface

For this purpose, the immersion method was performed as previously described by Chaidez et al. (2007) with some modifications. Briefly, 72 tomatoes were divided into three groups: the first group consisted of 24 S. Typhimurium -inoculated tomatoes sprayed with bacteriophage P22; the second group had 24 S. Typhimurium -inoculated tomatoes sprayed with sterile PBS; and the third group had 24 tomatoes sprayed with P22 alone (control). For the first two groups, tomatoes were immersed for 15 min in 6 L of bacterial suspension (10⁹ CFU/ml) with constant agitation, whereas the last group (control) was not inoculated with anything. Then, all of tomatoes were dried for 1 h in two parallel sterile glass rods located in a biosafety cabinet and stored in sterilized bags (Ziploc^®, SC Johnson USA) at 4°C until use (Lang et al., 2004).

Bacteriophage P22 (10¹¹ PFU/ml) was sprayed onto S. Typhimurium -inoculated and control tomatoes for 45 s from a distance of 45 cm with a manual sprayer (SPRAY-33, Truper^®, México) to simulate packinghouse spray disinfecting conditions. Following bacteria and bacteriophage inoculation, one set of each group was stored at 10°C and another set was stored at 20°C. Tomatoes were analyzed every 24 h for 7 days. Two independent experiments were performed for each temperature with 3 replicates per experiment.

Recovery and enumeration of S. Typhimurium and bacteriophage P22

Treated tomatoes (n=3), as described above, were randomly chosen and evaluated every 24 h for 7 d for bacteria and phage recovery. Each tomato was placed in a sterile plastic bag containing 100 ml of PBS and manually rinsed by rubbing for 1 min to remove S. Typhimurium and bacteriophage from the surface (Hirotani et al., 2002). Samples were serially diluted and plated in duplicate on XLD agar (Bioxon, México). Plates were incubated at 37°C and enumerated after 24 h. Bacteriophage concentration was determined using the soft agar overlay as previously described (APHA, 1998). Data was expressed as Log₁₀CFU/tomato and Log₁₀PFU/tomato, for bacterial and viral quantification, respectively.

Statistical analysis

The survival of S. Typhimurium on tomato surface was determined by a design of two blocks completely randomized. The persistence of bacteriophage P22 on tomato surface was analyzed with a three complete randomized block design. Efficiency of bacteriophage P22 for S. Typhimurium reduction on tomato surface was measured with a three complete randomized block design. Analysis of variance and multiple mean comparisons were conducted through the Tukey test. Differences with P<0.05 were considered significant. Data analysis was performed using MINITAB version 14 (Minitab Inc., State College, PA).

In vitro assay for bacterial biocontrol

The bacteria/bacteriophage challenge in a liquid culture is a passive approach (in vitro) to demonstrate the lytic effect of phages against target bacteria. In this case, high multiplicity of infection (MOI) of 100 and 1000 were used to ensure a bacteriophage/bacteria ratio in the system sufficient to infect all target bacteria by at least one phage particle, and induce bacteria lysis in a short period of time. This approach may predict the challenge behavior in vivo. Kinetics of Salmonella growth is shown in Figure 1. Salmonella infected with P22 at the highest MOI ratio of 1,000 showed undetectable levels of bacteria between two and four hours after infection; nonetheless, measurable bacteria were observed after six hours with a peak value of 5.5 Log₁₀CFU/ml at 48 h after infection (Figure 1, P<0.001). Meanwhile, at the MOI of 100, bacteriophage P22 was unable to significantly reduce Salmonella growth, however somewhat delay was observed in bacteria growth between two and four hours after infection (Figure 1, P<=0.005). The bacterial control concentration was 7.48±0.2 Log₁₀CFU/ml after 48 h of incubation. These results indicate that phage P22 is able to lyse Salmonella Typhimurium at a higher MOI, however Salmonella regrowth at low and high MOI suggests that Salmonella may develop Bacteriophage Insensitive Mutants (BIM´s) or a lysogenic cycle due to experimental conditions.

S. Typhimurium survival on tomato surface

To evaluate Salmonella survival on tomato surface, tomatoes were inoculated with S. Typhimurium and stored at 10 and 20°C for seven days. Tomatoes maintained at 10°C exhibited the lowest Salmonella concentrations during storage, with a mean concentration of 6.29 Log₁₀CFU/tomato and maximum/minimum concentrations of 6.5 and 5.83 Log₁₀CFU/tomato, respectively, at day seven. On the other hand, tomatoes stored at 20°C exhibited higher S. Typhimurium concentrations than tomatoes stored at 10ºC showing an increasing tendency as the contact time increased (from 7.1 Log₁₀ CFU/tomato at time zero to 8.32 Log₁₀ CFU/tomato at day seven), with a mean concentration of 7.69 Log₁₀CFU/tomato, which suggest a relationship between temperature and contact time.

Bacteriophage P22 survival on tomato surface

Phage survival was assessed on tomato surface sprayed with P22 alone or in presence of S. Typhimurium and stored at 10 and 20°C during the seven days. Both conditions showed decrease on phage survival after the storage period. In tomatoes simultaneously inoculated with Salmonella and P22 at 10°C, P22 was reduced up to 7.23 Log₁₀ PFU/tomato (3.79 Log₁₀ PFU/tomato reduction), as compared to initial phage concentration (11.02 Log₁₀PFU/tomato); on the other hand, tomatoes inoculated only with P22, phage was reduced 4.03 Log₁₀ PFU/tomato between day 0 and day seven after inoculation.

High levels of P22 were detected when tomatoes were sprayed with S. Typhimurium along with P22 and stored at 20°C. However, a reduction of 3.24 Log₁₀PFU/tomato was observed on day seven with a final concentration of 8.36 Log₁₀PFU/tomato. In contrast, tomatoes inoculated only with P22 showed a reduction of 4.11 Log₁₀PFU/tomato with a final concentration of 6.9 Log₁₀PFU/tomato on day seven. These results show that phage had a reduction in their concentration, regardless of the experimental condition in all cases.

Biocontrol of S. Typhimurium on tomato surface

Bacteriophage P22 lytic effect on S. Typhimurium inoculated onto tomato surface stored at 10 and 20°C during seven days is shown in Figure 2. Control groups for both temperatures where inoculated only with Salmonella. P22 treated-tomatoes stored at 10°C registered a constant decrease of S. Typhimurium concentration (Figure 2B). The highest reduction was detected on day seven with 3.02 Log₁₀CFU/tomato, compared to the baseline concentration, while the mean reduction concentration during the seven-day experiment was 1.38 Log₁₀CFU/tomato. P22 treated-tomato and stored at 20°C presented a mean bacterial reduction of 0.7 Log₁₀CFU/tomato seven days after inoculation (Figure 2C). Control groups showed higher S. Typhimurium concentrations as compared to tomatoes that received bacteria and P22 (Figure 2B and C).

The use of bacteriophages to reduce contamination of S. Typhimurium has several advantages over the use of chemical sanitizers. Firstly, bacteriophages are prokaryotic cells specific and, therefore, phages possess no harm to plants and/or animal cells. Secondly, within prokaryotic cells, bacteriophages will only lyse target bacteria, leaving fresh produce microbiota intact (Ackermann and DuBow, 1987; Gill and Abedon, 2003).

However, one of the biggest concerns of using bacteriophages as biocontrol is that they might shift from lytic to lysogenic cycle. During this cycle, phage inserts its nucleic acid into the bacterium and may transfer resistance or virulence genes (if phage contained), resulting in more virulent and resistant bacteria. However, reports have suggested that bacteriophage P22 showed low frequency of lysogenization cycle, which may be attributed to Salmonella's deficiency of adenilatecyclase (cya gene) or to the receptive protein (crp gene) production (Hong et al., 1971). Furthermore, Steinberg and Gough (1976) demonstrated that S. Typhimurium infection by wild type phage P22 causes a lytic response to low multiplicities of infection and a lysogenic response to high multiplicities of infection (MOI >5 more than 90% are lysogenized). Merril et al. (1996) reported the potential use of a mutated variant of P22 as therapeutic agent for bacteremia. In our results, plaques developed by P22 were clear and with 0.5-0.6 mm in diameter (data not shown), which suggests that P22 may have developed lytic cycle in this experimentation (Hong et al., 1971).

The use of high phage MOI has been described to lead an early bacterial lysis because high virion particles are adsorbed on the bacterial surface and cause the weakened of the cell wall and consequently bacterial lysis, without bacteriophage particles production. Abedon (2011) describes this phenomenon as “Lysis from without (LO)”, which has been demonstrated in various bacteriophages, such as like T4 phages. Although not all phages develop LO, bacterial lysis at high MOI in our results would be explained by the LO phenomenon.

The in vitro infection assay, showed that P22 was able to reduce Salmonella to undetectable levels 4 h after phage inoculation; however, after this period the bacterium was able to re-grow, reaching levels close to the control. This effect is most likely associated with the low proportion of P22 in a lysogenic state, which allows Salmonella suppression at initial stages; however, lysogenized bacteria could re-grow during long-term incubation (Gama et al., 2013). It is also important to emphasize that enrichment step (rich culture medium and optimal temperature) provide ideal conditions for bacterial growth, whereas in vivo conditions might present low nutrient availability or non-permissive temperatures for Salmonella growth.

According to our data, S. Typhimurium (without P22 treatments) was able to survive in tomato surface at storage temperature of 10°C for at least seven days, showing a marginal reduction of 0.67 Log₁₀ CFU/tomato at the end of experimentation. Contrary to what it was observed to tomatoes stored at 20°C, where an increased concentration of Salmonella (1.22 Log₁₀ CFU/tomato) was observed at seventh day compared to day zero, showing an increasing tendency as increases time; which suggests that Salmonella may be growing under these conditions. Other studies also described an increase in bacterial populations when fruits were stored at high temperatures (20 and 25°C) compared to low temperatures (10 and 4°C) where they recorded reductions up to 2 Log₁₀ on fresh-cut fresh produce (Leverentz et al., 2001; Castro-del Campo et al., 2004). All microorganisms have an optimal growth temperature at which they exhibit their highest growth rate.

 In this sense, the increased concentration of Salmonella could be attributed to the mesophilic nature of the bacteria. Zhuang et al. (1995) suggested that viability of Salmonella on the surface of tomatoes at 10°C might be a potential survival temperature during transport and storage prior to ripening and consumption; and even more, our results showed that at 20°C (marketing conditions) Salmonella was able to grow and increases its concentration, what hypothetically could represent a risk of salmonellosis outbreak if a scenery similar occurs in real conditions; therefore, suitable storage temperature of fruits and vegetables is fundamental to maintain low microbial levels and prevent pathogenic microorganism’s growth (Leverentz et al., 2001).

In this investigation, bacteriophage P22 showed a positive lytic effect over S. Typhimurium, achieving a maximum reduction of 3.02 Log₁₀CFU/tomato in seven days at 10°C, demonstrating its effectiveness to reduce populations of a pathogenic microorganism, as reported by Sklar and Joerger (2001); Modi et al. (2001); Pao et al. (2004) and López-Cuevas et al. (2012). Listeria monocytogenes was reduced in high concentrations (5.7 Log₁₀CFU/ml) in fresh-cut honeydew melon, and higher reductions were achieved through a combination of bacteriophages and bacteriocins (Leverentz et al., 2003). Several studies explore bacteria reductions in fresh-cut produce with phages. For the best of our knowledge, the present study is one of the few studies to be conducted in whole fruit.

This research provides data about the effectiveness of the use of bacteriophages to reduce or eliminate pathogenic microorganisms present on fresh produce. The US-EPA determined the reasonable standard of efficiency of a disinfectant in a fresh produce surface that at least should achieve a 2 Log₁₀CFU reduction of the pathogen population (EPA, 1997). Thus, the use of bacteriophages is a natural alternative, which could benefit both conventional and organic agriculture.

This research data suggests that bacteriophage P22 would provide a protective effect to fresh produce through the production chain as a novel, environmentally safe alternative to control Salmonella enterica subsp. enterica on fresh produce.

The authors have not declared any conflict of interests.

The authors thank the National Food Safety Laboratory Research personnel (LANIIA-CIAD). We also thank Célida Martinez-Rodríguez for her technical assistance.