Salmonella enterica serotype Enteritidis FSL S5-483 was used as the bacterial host for the phages in this study. The Salmonella strains used in the host range study are listed in Table 1. All strains were maintained at -80°C in Brain-Heart-Infusion broth (BD/Difco, East Rutherford, NJ, United States) supplemented with 20% glycerol. Working stocks were maintained on tryptic soy agar (TSA; BD/Difco, East Rutherford, NJ, United States) at 4°C for a maximum of 1 month. Prior to each experiment, fresh overnight cultures were prepared by inoculating an isolated colony into 10 ml tryptic soy broth (TSB; BD/Difco, East Rutherford, NJ, United States). Cultures were then incubated for 16 h at 37°C with gentle shaking at 170 rpm.

Bacteriophages were isolated from irrigation water (n = 15), cattle feces (n = 9) and sediment obtained from the bottom of irrigation ditches (n = 8) from Greater Vancouver, British Columbia, Canada. S. Enteritidis FSL S5-483 was used as an indicator organism for phage isolation. Of the phages recovered, four broad-host range phages were isolated: SI1 (from irrigation water), SF1 (from cattle feces) and SS1 and SS4 (from sediment). Phage SI1 was isolated following direct plating by mixing 10 g of sample with 90 ml of salt-magnesium (SM) buffer (0.05 M Tris-HCl; 0.1 M NaCl and 0.01 M MgSO₄; pH 7.5), followed by passage through a 0.45 μm membrane (Pall Corporation, Port Washington, NY, United States). Then, 100 μl of filtrate was mixed with 300 μl of 1:10 diluted S. Enteritidis grown to 16 h and four ml 0.7% TSA top agar, according to the double agar overlay method (Adams, 1959).

Phages SF1, SS1, and SS4 were isolated following enrichment by mixing 10 g of water, cattle feces or sediment samples, 90 ml TSB and 1 ml of S. Enteritidis cells grown to 16 h, followed by incubation at 37°C for 22 ± 2 h. The enriched samples were then spun at 4,000 × g and the supernatant subsequently passed through a 0.45 μm filter membrane (Pall Corporation). Then, 100 μl of filtrate was mixed with 300 μl of 1:10 diluted S. Enteritidis grown to 16 h and four ml 0.7% TSA top agar, according to the double agar overlay method (Adams, 1959).

Plates were incubated at 37°C for 18 ± 2 h for visualization of plaques. Plaques were lifted from the agar surface using a truncated pipette tip, suspended in 200 μl SM buffer, and rested for at least 6 h at room temperature. Double agar overlays were prepared with the suspension as described previously (Adams, 1959). Three single plaque isolations were carried out to obtain a pure phage lysate. Finally, phages were concentrated and stored at 4°C for further analyses.

Prior to host range determination, phage lysates were standardized to a concentration of 10⁹ PFU/ml as recommended by Khan Mirzaei and Nilsson (2015). The host ranges of the phages were tested by spotting 5 μl of lysate, in duplicate, on a lawn of Salmonella host cells grown to 16 h (n = 61, Table 1). To test for the presence of lysis from without (LO), successive 10-fold dilutions of the phage were also prepared in sterile SM buffer and 5 μl of each dilution spotted in duplicate on the Salmonella lawn. Plates were incubated at 37°C for 18 ± 2 h. Zones of clearing were characterized with a scaling system as described by Kutter (2009), where 0 indicated a zone with complete turbidity (no lysis) and +4 indicated a completely clear zone with no turbidity (Table 1). LO was detected where complete lysis (+4 ranking) was observed at low dilutions (i.e., 10^-1 PFU/ml), yet no lysis was observed at further dilutions (Kutter, 2009).

High titer phage lysates (10⁹ – 10¹¹ PFU/ml) were chosen for transmission electron microscopy (TEM) and prepared for imaging as described previously (Deveau et al., 2006), with modifications. Briefly, one ml of lysate was spun at 4°C for 1.5 h at 21,000 × g. The supernatant was subsequently discarded and the last 100 μl was saved. Consequently, 1 ml of 0.1 M ammonium acetate (Amresco, Solon, OH, United States) was added and the suspension subsequently spun again at 4°C for 1.5 h at 21,000 × g. This purification was repeated twice, with the last 100 μl reserved for TEM.

For grid preparation, 3 μl of purified lysates were placed on carbon coated copper grids (Ted Pella, Redding, CA, United States) following glow-discharge. The phage preparations were subsequently negatively stained with 2% phosphotungstic acid (Ted Pella). A Hitachi H-7600 transmission electron microscope was used for acquiring the images at the University of British Columbia Bioimaging Facility. An accelerating voltage of 80 kV was used for imaging.

Prior to nucleic acid extraction, DNase I (Invitrogen, Carlsbad, CA, United States) and RNAse A (Invitrogen) were added to high titer phage lysates (10⁹ – 10¹¹ PFU/ml) to final concentrations of 10 and 55 μg/ml, respectively, for degradation of host nucleic acid (Merabishvili et al., 2014), followed by incubation at 37°C for 30 min. Phage nucleic acid was then extracted with the PureLink Viral RNA/DNA Mini Kit (Thermo Fisher) as per the manufacturer’s instructions. The concentration and quality of the extracted nucleic acid was determined with a Nanodrop spectrophotometer (Thermo Fisher), where an A260/280 ratio of ∼1.8 and A260/230 ratio of ∼2.0 were considered as pure.

Restriction enzyme analysis was conducted to confirm the unique identities of the phages. Nucleic acid was digested with EcoR1 (New England Biolabs, Ipswich, MA, United States) according to the manufacturer’s instructions. Subsequently, 10 μl volumes of the nucleic acid digests were loaded onto a 1% agarose gel (Amresco) and electrophoresed in 1X TAE buffer (Thermo Fisher) at 80 V for approximately 1 h. Band patterns were visualized using the ChemiDoc MP System (Bio-Rad Laboratories).

Single step growth curves were constructed to determine the phage burst sizes and burst times, according to Park et al. (2012), with modifications. Cultures of S. Enteritidis were grown for 16 h in TSB (37°C, 170 rpm). One ml of culture was then added to 9 ml of fresh TSB and incubated at 37°C at 170 rpm until an optical density at 600 nm (OD₆₀₀) of 1.0 (∼10⁹ CFU/ml; stationary phase) was attained. Phages were then individually added at MOI of 0.01 and allowed to adsorb for 5 min at room temperature. To remove excess phage particles, the co-culture was spun at 4,000 × g at 4°C and the supernatant discarded. The pellets were resuspended in 10 ml of fresh TSB and incubated at room temperature with gentle agitation. Subsequently, 50 μl aliquots were collected every 5 min for a total duration of 60 min, immediately serially diluted in SM buffer, and spotted in duplicate on a host agar lawn of S. Enteritidis grown to 16 h for titer determination. Plates were incubated at 37°C for 18 ± 2 h for visualization of plaques. This experiment was independently conducted three times for each phage.

Resistant colonies of S. Enteritidis in the centers of spot assays (n = five colonies per phage) were selected to test for lysogeny. First, isolated colonies were serially re-streaked five times on TSA to reduce phage carry-over. On the fifth streak, a random colony was selected for polymerase chain reaction (PCR) to confirm their Salmonella identity by using primers specific to invA, according to Fong and Wang (2016). Briefly, a single colony was suspended in 200 μl of sterile de-ionized water and lysed in a microwave for two min. PCR detection was then carried out with the TopTaq Master Mix Kit (Qiagen, Valencia, CA, United States) with primers specific for the invA gene (forward: 5′-TCA TGG CAC CGT CAA AGG AAC C-3′ and reverse: 5′-GTG AAA TTA TCG CCA CGT TCG GGC AA-3′) (Li et al., 2012). PCR cycling conditions were as follows: initial denaturation (3 min, 94°C); three-step cycling, including denaturation (30 s, 94°C), annealing (30 s, 56°C), and extension (1 min, 72°C); followed by a final extension (10 min, 72°C). Sizes of the PCR products were confirmed with electrophoresis on 2% agarose (Amresco, Solon, OH, United States) with 1X Tris-acetate-EDTA (TAE) buffer (Thermo Fisher, Waltham, MA, United States). PCR products were visualized using the ChemiDoc MP System (Bio-Rad Laboratories, Hercules, CA, United States).

Colonies arising from the fifth streak were cultured to test for phage lysogeny as previously described (Petty et al., 2006). Colonies (n = 10) were suspended in 10 ml TSB and incubated at 37°C at 170 rpm for 20 h. Cultures were subsequently spun at 4,000 × g at 4°C to sediment the bacteria, and the supernatant tested for the spontaneous release of phage particles by spotting 5 μl in duplicate onto prepared S. Enteritidis agar overlays. Plates were incubated at 37°C for 18 ± 2 h for visualization of plaques. Supernatants from S. Enteritidis infection by Felix-O1, a strictly lytic phage, were used as a negative control.

Phage lysates were diluted in TSB to an initial concentration of ∼10⁷ PFU/ml and subsequently stored at a range of temperatures (-20, 4, 22, and 37°C) for determination of relative temperature stabilities. Controls (no phage) were included for each temperature. The samples stored at -20°C were prepared in single-use 15 μl aliquots to prevent multiple thawing and freezing events throughout the assay. Temperature stability experiments were conducted three times for each phage.

To test pH stability of the phages, phage lysates were diluted to ∼10⁸ PFU/ml in TSB at varying pH ranges of pH 4.0, 6.0, 8.0, and 10.0 (adjusted with 6 M HCl or 6 M NaOH) and subsequently stored at room temperature for further analyses. Blank controls (no phage) were included for each pH. These pH and temperature ranges were chosen based on previously reported similar assessments (Thung et al., 2017) and reflect the various pH and temperature conditions encountered in produce production chains (Park et al., 2012; Rombouts et al., 2016). pH stability experiments were conducted three times for each phage.

Phage titers were assessed on days 2, 4, 8, 10, 14, 16, 20, 25, and 30. Briefly, 10 μl volumes were serially diluted in SM buffer and 5 μl spotted in duplicate on prepared top agar of S. Enteritidis grown to 16 h. Enumeration of plaques were determined after incubation at 37°C for 18 ± 2 h.

Cultures of S. Enteritidis FSL S5-483, S. Agona FSL S5-513 and S. Typhimurium LMFS-JF-001 were prepared as described previously in section “Bacterial Strains and Growth Conditions.” Following incubation, cultures were spun at 4,000 × g and the cell pellets washed three times with fresh TSB. Then, the cultures were loaded into 96-well plates to a final concentration of 5 × 10⁴ CFU/ml and infected with phages SI1, SF1, SS1, and SS4 at MOIs of 1, 10, and 100 PFU/CFU. Plates were placed into a plate reader (SpectraMax M2, Molecular Devices, Sunnyvale, CA, United States) set to 25°C for determination of cell density at OD₆₀₀ every 30 min for 36 h. Each experiment was independently conducted three times.

The lysate of phage SI1 was tested for its efficacy to control S. Enteritidis on germinating alfalfa seed over 6 days. Six days was chosen as approximately 5–7 days are required for alfalfa seeds to sprout (Kramer and Lim, 2004). Cultures of S. Enteritidis grown to 16 h were spun at 4,000 × g for 10 min, washed three times with sterile potable water and serially diluted to a final volume of 35 ml of sterile water. The seed was inoculated by drop-wise addition of 15 ml diluted culture to 150 g seed to achieve an initial concentration of approximately log 3.5 CFU/g. Blank controls were processed similarly, but with sterile water only. The seed was placed in a biological safety cabinet at room temperature under continuous air flow for 2 h. Finally, the seed was transferred to sterile plastic boxes lined with a layer of sterile gauze pad and stored in the dark in a 22°C incubator. Three independent replicate experiments were performed.

The lysate of phage SI1 was applied to the seed at 22 h post-inoculation. Briefly, 75 g of inoculated seed was aseptically removed and treated with phage SI1 in 35 ml sterile water to yield an MOI of approximately 110 PFU/CFU. The seeds were soaked for 2 h at room temperature with gentle agitation by shaking at 175 rpm. Inoculated seed and controls that received only sterile water were processed in tandem. A 2 h phage soak was chosen due to (i) the short latent period (25 min) and relatively high burst size (83 phages) of SI1, thereby facilitating approximately four cycles of productive infection; and (ii) the simulation of a logistically feasible decontamination step, performed within a time frame that could be adopted into commercial sprout production practices. Following the soak, the excess fluid was removed by straining through sterile filter paper. Seed samples were then aseptically transferred to sterile plastic boxes lined with a layer of sterile gauze pad. Treated seed was stored in the dark in a 22°C incubator.

The germinating seed was moistened with seven ml of sterile water every 24 h over 6 days. Simultaneously, 10 g of seeds were removed daily and mixed with 100 ml of sterile phosphate buffered saline (PBS; Amresco) in a sterile Whirlpak bag (Nasco, Fort Atkinson, WI, United States). The samples were placed in a Stomacher (Seward, Worthing, West Sussex, United Kingdom) and homogenized for 2 min at 230 rpm. Subsequently, 100 μl aliquots were serially diluted in PBS and spread over xylose-lysine deoxycholate (XLD; Amresco) agar in duplicate. XLD plates were incubated at 37°C for 22 ± 2 h for enumeration of Salmonella (red colonies with black centers). Phage titers were measured by spotting 5 μl in duplicate on TSA seeded with S. Enteritidis grown to 16 h. Plates were incubated at 37°C for 18 ± 2 h for enumeration of plaques.

To assess the impact of the phage treatment on the final sprout yield, 150 g of seed was artificially contaminated with S. Enteritidis and 75 g was withdrawn for phage treatment, as per the procedures described previously. Sprouting seeds were then weighed after 6 days. Three independent replicate experiments were performed, with two technical replicates taken for each measurement.

For the pH and heat stability assays, the final titer of the phages after 30 days of treatment was compared to the initial titer at the beginning of treatment with a Student’s t-test (α = 0.05). To compare the relative susceptibilities of the phages to each treatment, the log decreases in phage titer after 30 days of treatment was calculated (i.e., the difference in log PFU/ml at time zero and after 30 days of treatment). A one-way analysis of variance (ANOVA) was then implemented with a Tukey’s Honest Significant Difference post hoc test applied to all significant ANOVA results (α = 0.05).

For the sprouts biocontrol assay, the log differences in Salmonella counts at each sampling time point between untreated and treated alfalfa seed samples were analyzed with a Student’s t-test with a significance level of α = 0.05. The differences in weights between the control and treated alfalfa sprouts were also assessed using a Student’s t-test (α = 0.05).

All statistical analyses were performed using JMP version 11.1.1 (SAS Institute, Inc., Cary, NC, United States). A P-value of ≤0.05 was considered statistically significant for all analyses.

Results of host range analysis of the four phages suggested that they were able to lyse 31–37% of the Salmonella strains tested as indicated by a +3 or +4 spot test (Table 1). Susceptible strains encompassed (i) the serotypes causing the highest rates of salmonellosis (i.e., S. Enteritidis and S. Typhimurium); (ii) serotypes involved in North American outbreaks (i.e., S. Agona, S. Saintpaul, and S. Heidelberg); and (iii) emerging, uncommon serotypes (i.e., S. Berta and S. Canada) (Ellis et al., 1998).

SI1, SS4, SF1, and SI1 were classified as broad host-range phages as they infected the largest proportion of the tested Salmonella strains, compared to other Salmonella phages recovered and characterized (n = 44). Of the 61 Salmonella strains tested, 23 were susceptible to infection by SI1, 22 were susceptible to SS4, 20 were susceptible to SF1 and 19 to SS1, as indicated by a +3 or +4 clearing (Table 1). The similarity between host ranges indicate that these four phages may recognize similar host receptors (Kalatzis et al., 2016). It should be noted that strains of S. Typhimurium (n = 8), responsible for the highest proportion of foodborne salmonellosis, demonstrated a high degree of susceptibility to these phages; seven were susceptible to infection by SI1 and SS4 and six were susceptible to SF1 and SS1. LO was not observed in any of the characterized phages, indicating the presence of obligate productive infection with all S. enterica strains tested.

The relative abilities of the four phages to suppress S. Enteritidis FSL S5-483, S. Agona FSL S5-513 and S. Typhimurium LMFS-JF-001 in TSB were assessed at an MOI of 1, 10, and 100 PFU/CFU at 25°C. At MOI = 1, both SI1 and SS4 suppressed the growth of S. Enteritidis over a 36 h period. Growth of S. Enteritidis was also suppressed by SF1 and SS1, but growth resumed at 13 and 17 h, respectively, after initial infection (Figure 6). Growth of S. Enteritidis appeared to recover at 31 h following treatment with SS4. An MOI of 10 prolonged the suppression of S. Enteritidis to 19 and 25 h when infected with SF1 and SS1, respectively. Further, application of SS4 at an MOI of 10 caused complete inhibition of S. Enteritidis growth for 36 h. Finally, phage treatment at an MOI of 100 suppressed the growth of S. Enteritidis for the 36 h duration (Figure 6).

Suppression of S. Agona also occurred with all MOIs tested (Figure 7), but the extent was not as pronounced as with the host, S. Enteritidis. Instead, considerable suppression did not occur until an MOI of 100 was evaluated. At this MOI, it appeared that SS1 was the least effective in controlling S. Agona as growth resumed 17 h after the initial infection. In contrast, SI1 and SS4 were the most effective in suppressing growth, although S. Agona appeared to recover at 28 and 32 h after infection with these phages, respectively.

Lastly, S. Typhimurium was tested for its susceptibility to the phages in TSB (Figure 8). Again, the extent of suppression after phage infection was not as pronounced as with S. Enteritidis, but did occur at all MOIs. The most dramatic reduction in growth occurred at MOI 100, although growth was not suppressed entirely for the 36 h duration. At this MOI, SI1 was the most effective in attenuating growth (Figure 8).

Although not all MOIs were effective in controlling growth of Salmonella, nor did all Salmonella strains show similar susceptibilities to the phages, it should be noted that phage infection at all MOIs, across all strains, resulted in an extended lag phase [defined as OD₆₀₀ < 0.2 (Wang et al., 2009)], indicating that the phages had a suppressive effect on Salmonella. The ability of the Salmonella strains to recover after initial infection is likely due to the emergence of phage-resistant mutants (Guenther et al., 2012). It may be possible to prolong the duration of phage sensitivity by infection with a mixture of phages (Chan et al., 2013), though it was not evaluated in this present study.

With all strains, infection with an MOI of 100 proved to be the most effective and represents the MOI used for many food processing applications (Silva et al., 2014). However, with the phage host, S. Enteritidis, growth was completely suppressed at all MOIs with phage SI1, underlining its remarkable efficacy in controlling S. Enteritidis in vitro. Mechanistically, SI1 may require multiple attachment sites on the bacterial cell membrane for adsorption and/or SI1 receptor sites may be essential for cellular metabolic processes – both of which would contribute to the attenuation of phage resistance by the host (Rakhuba et al., 2010; Kong et al., 2011).

The ability of SI1 to control Salmonella on sprouting alfalfa seeds was assessed. SI1, in particular, was selected for this study as it caused complete inhibition of S. Enteritidis in TSB at all tested MOIs and possessed the greatest burst size (approximately 83 phages) and possessed one of the shortest latent periods (25 min) (Figure 2). Moreover, S. Enteritidis is a serotype previously linked to North American sprout outbreaks (Centers for Disease Control and Prevention [CDC], 2017) and further, has been implicated in the highest number of salmonellosis outbreaks worldwide (Mattick et al., 2001). On the basis of these factors, they were selected for use in this biocontrol study.

Treatment with SI1 (MOI = 100) resulted in a significant (p < 0.05) 2.51 ± 0.24 log CFU/g reduction of S. Enteritidis, 2 h after treatment (Figure 9), corresponding to a decrease of 38.3 ± 3.0% of the initial viable population. This was accompanied by a 1.02 ± 0.33 log PFU/g increase in phage titer (Figure 10). In contrast, previous work by Kocharunchitt et al. (2009) reported a one log CFU/g decrease in S. Oranienburg populations following application of phage SSP6 onto alfalfa seeds at the beginning of germination. Similarly, a 1.37 log CFU/g reduction of Salmonella populations was observed on mustard seeds at 24 h following phage treatment (Pao et al., 2004). Sprout production standards, as set by Health Canada, recommend that sprout decontamination methods achieve a minimum three log reduction in pathogen counts (Canadian Food Inspection Agency [CFIA], 2014). Further validation of SI1 infectivity across a range of potential bacterial contaminants and at various stages throughout the sprouting process is therefore required to ensure complete compliance with Health Canada standards, although a >2.5 log CFU/g reduction of S. Enteritidis is promising. Additionally, the final weight of the seeds treated with Salmonella only (82.40 ± 2.83 CFU/g) was not significantly different (p > 0.05) than that of the phage-treated sprouts (80.64 ± 1.41 CFU/g), further demonstrating its potential suitability for use in industry.

On days 2–6 following phage treatment, Salmonella cell densities on treated alfalfa seeds were also reduced, but this was not significant (p > 0.05) (Figure 9). In line with this observation, phage titers increased the day of seed treatment, indicating its replication. However, the initial increase was followed by a stagnation of growth and small decreases in titer (Figure 10). Notably, it is presumed that the emergence of phage-resistant Salmonella may have contributed to the diminished effectiveness in the days following treatment. Indeed, phage-resistant Salmonella has been identified in both in in vitro systems (Vipra et al., 2013) and foods treated with phage (Kocharunchitt et al., 2009; Guenther et al., 2012). Emergence of bacterial mutants resistant to phage is particularly apparent when MOI values are high, as this enhances the selective pressure to resist infection (Vipra et al., 2013). It has been reported, however, that phage-resistant mutants possess attenuated pathogenicity and diminished fitness (Kong et al., 2011). A possible remedy to control the emergence of such mutants is through the use of a phage cocktail (Spricigo et al., 2013; Pereira et al., 2016a), which may additionally extend the spectrum of lysis to include other Salmonella strains (Chan et al., 2013).

Although the present results are not fully consistent with the data obtained in vitro, it is hypothesized that the simplicity of an in vitro system represents an ideal scenario for phage infection and multiplication. The nature of a food matrix presents with various complicating factors. For instance, possibilities include biofilm production on alfalfa sprouts (Kocharunchitt et al., 2009), which could hinder phage adsorption (Sutherland et al., 2004); growth of endogenous microbiota naturally present on sprout seeds, which may provide alternative adsorption sites (Ye et al., 2010); or internalization of Salmonella into the sprouts itself (Erickson, 2012), rendering them unavailable for phage attack. These factors could account for the diminished efficacy of the phage and also its inconsistent increases in phage titer throughout this assay. It is possible that additional phage treatments throughout the sprouting process, or phage treatment in combination with other treatments (e.g., chlorine or organic acid washes), would further reduce the viable Salmonella populations on alfalfa seeds. It should also be noted that the high initial load of Salmonella used in this assay is unrepresentative of real world situations, yet is important from a technical perspective to determine the log kill. Ye et al. (2010) reported a six log CFU/ml decrease of Salmonella on artificially contaminated mung bean sprouts upon treatment with a combination of six Salmonella phages and Enterobacter asburiae, a naturally competitive microorganism. Interestingly, this combination treatment was significantly more effective than treatment with phage or E. asburiae alone.