Bacteriophage in-feed application: A novel approach to preventing Salmonella Enteritidis colonization in chicks fed experimentally contaminated feed

DESCRIPTION OF THE PROBLEM

In the United States, foodborne nontyphoidal salmonellosis affects approximately one million individuals, results in 378 deaths and costs approximately $3.3 billion USD annually (Scallan et al., 2011). Most of these infections are associated with animal meat and product consumption, with poultry or poultry products being the major source (Marin and Lainez, 2009). Therefore, eradication of Salmonella in the production cycle of poultry and in processing plants is important for food safety. Salmonella is ubiquitous and can infect poultry through numerous routes (Rajan et al., 2017). Salmonella can be transferred vertically from hen to chick or horizontally between chicks (Lilkebkelke et al., 2005). Environmental Salmonella in hatcheries and broiler houses is another route of infection (Rajan et al., 2017). In addition, Salmonella can contaminate feed production facilities and feed ingredients (Boyer et al., 1962; Maciorowski et al., 2007; Jones, 2011). In some instances, 1 human outbreaks of Salmonella infections were traced back to Salmonella-contaminated feed given to poultry (Pennington et al., 1968; Crump et al., 2002). Formaldehyde and organic acids have been traditionally added to feed to reduce Salmonella (Jones, 2011). Formaldehyde, while efficacious in decontaminating feed, is a hazardous material (Sheldon and Brake, 1991). Organic acids are not as hazardous but require higher inclusion rates (1%) and application for several days for equivalent efficacy (Smyser and Snoeyenbos, 1979). Therefore, a safe alternative solution for reducing Salmonella in feed would be beneficial.

One potential solution to feed contamination with Salmonella is bacteriophages (BP). Lytic BP bind to target bacteria with high specificity, inject, and replicate DNA. The host bacteria are destroyed when the daughter viral particles are subsequently released (Joerger, 2003). Unlike antibiotics, which kill both pathogenic bacteria and normal flora in the gut, BP are highly selective to their host bacteria (Wernicki et al., 2017). In the 1920s, BP were utilized to treat fowl typhoid in poultry (Pyle, 1925). Bacteriophages have been used in a wide variety of applications such as in the treatment of live birds (Borie et al., 2009), poultry products (Atterbury et al., 2003; Goode et al., 2003; Hungaro et al., 2013), and processing equipment (Gong and Jiang, 2017; Gong et al., 2017). Single BP or cocktail BP administered to chickens through aerosol spray or oral gavage have been demonstrated to treat bacterial infections and decrease mortality (Borie et al., 2009; Johnson et al., 2008; Miller et al., 2010). Although there is some research with in-feed BP application (Adhikari et al., 2017; Kim et al., 2013), there is no published research related to BP treatment of Salmonella-contaminated poultry feed. The purpose of this study was to evaluate the efficacy of a bacteriophage specific to Salmonella Enteritidis (S.E.) in reducing the colonization of S.E. in broilers fed S.E.-contaminated feed.

MATERIALS AND METHODS

Birds and Management

This study was approved by the Southern Poultry Research group Institutional Animal Care and Use Committee. Two hundred seventy day-of-hatch Ross 708 broiler chicks were obtained from a commercial hatchery (Blairsville, GA). Thirty chicks were placed into each of the 3 floor pens per treatment (90 birds/treatment) with fresh shavings in a modified conventional poultry house with solid-sides and concrete floors on day of trial (DOT) 1. The facility was HEPA filtered exhaust and thermostatically controlled with a heat pump. Birds were raised under ambient humidity and were provided a lighting program as per the primary breeder recommendations. The floor space per animal was 1.00 sq. foot/bird, with one tube and one Plasson drinker in each pen. Feed and water were given ad libitum. Birds were monitored twice daily for mortality, general flock condition, changes in environmental conditions, and all observations were recorded.

Treatments

Basal diets were based off corn-soy commercial-type broiler diets in accordance with the NRC guidelines (1994). The treatments were 1) control, 2) 1 kg BP/metric ton of feed, 3) 1.5 kg BP/metric ton of feed from DOT 1 through DOT 28. All diets contained Amprolium (Huvepharma, Peachtree City) at 113.5 g/ton. A broiler starter mash diet was fed from DOT 1 to 8 and then on day 8 an S.E.-contaminated starter mash was fed until DOT 14. From DOT 14 to 28, a starter mash with no S.E. was fed.

Bacteriophage

The dietary BP used in this experiment (CTCBIO Inc., 93, Ogeum-dong, Songpa-gu, Seoul, Republic of Korea) was a lytic phage that specifically targeted S.E. (KCTC 12012BP). Bacteriophage was present at a concentration of 108 plaque-forming unit (pfu) per gram. These pfu were quantifiable plaques formed when live viral particles infect host cells within a cell monolayer.

Challenge Protocol and Sampling

Nalidixic acid–resistant SE was cultured in meat and bone meal (MBM) to a final concentration of 10⁷ CFU/g of MBM. Then the MBM was mixed into the feed. In addition, a chromium marker was mixed in with the MBM to monitor distribution of the MBM/SE throughout the feed. The S.E.-contaminated feed was fed from DOT 8 until DOT 14. The measured S.E. level after 1 wk was approximately 10⁴ CFU/g feed. On DOT 14, samples were taken from each treatment and S.E. was enumerated from the feed. The control had 1 x 10⁵ CFU, whereas both BP treatments had 1 x 10⁴ CFU. The S.E. strain used was a field strain which has been made nalidixic acid (Young et al., 2007).

Cloacal swabs were taken from ten (10) birds per pen on DOT 14 and 21 for Salmonella presence and enumeration. Cecal samples were aseptically taken from 25 birds from each pen on DOT 28. Spleen and liver from 10 birds were taken from each pen and cultured for S.E. prevalence on DOT 28.

Salmonella Enumeration

All samples taken were placed on ice in sterile bags. Organ samples were weighed and tetrathionate broth was added to each ceca and combined liver and spleen at 1 part to 9 parts broth (1:10 wt/vol dilution) (Difco Laboratories, Detroit, MI). Samples were mixed with a stomacher. A tetrathionate broth solution was added, a 1 mL aliquot was removed for most probable number method (MPN) analysis, and samples were incubated overnight at 41.5°C. A loopful of sample was struck onto xylose lysine tergitol-4 agar (XLT-4) (Difco Laboratories, Detroit, MI) plates which were incubated overnight at 37°C. Up to 3 black colonies were selected and confirmed as Salmonella positives using Poly-O Salmonella Specific Antiserum (MiraVista, Indianapolis, IN).

Salmonella Enumeration via Most Probable Number Method

Salmonella in ceca and organ (liver and spleen) samples were enumerated using the MPN of Berghaus et al. (2013). Briefly, a 1-mL aliquot of the homogenized organ/tetrathionate broth sample was transferred to 3 adjacent wells in the first row of a 96-well 2 mL deep block. A 0.1 mL aliquot of sample was transferred to 0.9 mL of tetrathionate broth in the second row, repeated process for remaining rows to produce five 10-fold dilutions, and incubated blocks for 24 h at 42°C. One mL of each well was transferred onto XLT-4 agar containing nalidixic acid, and were incubated at 37°C for 24 h. Final dilutions were recorded, and MPN calculations were performed (Blodgett, 2010). Suspect Salmonella isolates were confirmed by poly-O Salmonella-specific antiserum.

Statistical Methods

Salmonella Enteritidis prevalence’s in cloacal swabs, liver/spleen samples, and ceca were compared between treatment groups using generalized estimating equations logistic models adjusted for clustering by pen. Salmonella Enteritidis MPN in cloacal swabs and ceca were compared between treatment groups using a mixed-effects Tobit censored regression model to account for the distribution of samples that were above or below the limits of the MPN assay. Tobit models included a random effect for pen, with culture-negative cloacal swab samples being censored at a lower limit of −0.05 log₁₀ MPN/swab, and culture-negative ceca samples being censored at a lower limit of −0.5 log₁₀ MPN/g. Censoring occurs when the dependent variable is unobservable below a certain range and it is only known that the value is below the lower limit of detection. Briefly, the Tobit models attempt to estimate the true mean MPN based on the distribution of MPN in the culture-positive samples as well as the proportions of culture-negative samples in the different treatment groups (Tobin, 1958; Long, 1997). The MPN values were log-transformed before statistical analysis. Post hoc pairwise comparisons between treatments were performed using the Bonferroni procedure to limit the type I error rate to 5% over all comparisons. All statistical testing assumed a two-sided alternative hypothesis, and P < 0.05 was considered significant. Analyses were performed using commercially available statistical software.

RESULTS AND DISCUSSION

In this study, both BP treatments were effective in lowering colonization in broilers fed S.E. contaminated feed as compared with the control birds. Although other researchers have tested efficacy of in-feed BP, the Salmonella challenge was via oral gavage as opposed to in feed (Kim et al., 2013; Wang et al., 2013b; Adhikari et al., 2017). Therefore, this trial also demonstrated an alternative method for Salmonella challenge in poultry.

Bacteriophage in-feed application: A novel approach to preventing Salmonella Enteritidis colonization in chicks fed experimentally contaminated feed - Image 1

Salmonella prevalences in cloacal swabs are summarized in Table 1. In a factorial analysis, there was a significant effect of treatment (P < 0.001). Both BP groups had a significantly lower S.E. prevalence than the untreated group, and the 1 kg/ton group had a significantly lower prevalence than the 1.5 kg/ton group. There was also a significant effect of day (P = 0.014), with the prevalence being lower on DOT 21 compared with DOT 14. There was no significant interaction between the effects of treatment and day (P = 0.375). All Salmonella isolates obtained from cloacal swabs were identified as belonging to serogroup D, which was consistent with the S. E. challenge strain. Salmonella MPN in cloacal swabs based on the Tobit censored regression model are summarized in Table 2. There was a significant effect of treatment (P = 0.010), with the 1 kg/metric ton group having a significantly lower estimated mean log₁₀ MPN/swab than the untreated group, whereas the 1.5 kg/metric ton group was intermediate and did not differ from either of the other 2 groups. There was also a significant effect of day (P = 0.001), with the estimated mean log₁₀ MPN/swab being higher on day 14. There was no significant interaction between the effects of treatment and day (P = 0.855).

There were no significant differences between treatments with respect to Salmonella prevalence in liver/spleen samples (P = 0.115). Salmonella prevalences in the liver/spleen samples collected on DOT 28 were: 7 of 30 (23.3%) in the untreated group; 8 of 30 (26.7%) in the 1 kg/metric ton group; and 5 of 30 (16.7%) in the 1.5 kg/metric ton group. All Salmonella isolates obtained from liver/spleen samples were identified as belonging to serogroup D. Cecal S.E. prevalence is summarized in Table 3 and Figure 1. There was a significant effect of treatment (P < 0.001). The S.E. prevalence in both BP groups was significantly lower than that in the controls, whereas the 2 BP groups did not differ significantly from one another. All Salmonella isolates obtained from ceca samples were identified as belonging to serogroup D. Salmonella MPN in ceca based on the Tobit censored regression model are summarized in Table 4. There was a significant effect of treatment (P = 0.004). Both BP treatments had a significantly lower estimated mean log₁₀ MPN/g than the controls, whereas the 2 BP-treated groups were not significantly different from one another.

Bacteriophage feed administration is practical in commercial poultry settings for mass application (Wernicki et al., 2017), as demonstrated by the use of Campylobacter BP administered in poultry feed more rapidly reduced colonization when compared with oral gavage (Sillankorva et al., 2010). Researchers have had varying levels of success with reducing of Salmonella with BP application (Fiorentin et al., 2005; Toro et al., 2005; Higgins et al., 2007). Because of the bacterial serovar specificity of BP, targeting BP activity against field isolates of Salmonella is important. Our study successfully reduced S.E. with a single BP.

It is important to consider the impact of new feed additives on performance and safety in poultry. There is no current evidence of negative impacts of BP on weight gain of feed conversion or egg production under nonchallenge conditions in broilers and layers (Atterbury et al., 2007; Zhao et al., 2012; Wang et al., 2013a; Kim et al., 2015). In fact, a study by Wang et al. (Wang et al., 2013b) demonstrated that supplementing phages specific for Salmonella to broiler chickens significantly increased the concentration of beneficial Lactobacillus bacteria in the excreta vs. nonsupplemented controls and bacitracin supplemented birds. This alteration in gastrointestinal bacterial populations may provide benefit to BP administration, especially when coupled with the reduction of pathogenic bacteria. From a safety perspective, the specificity of BP means that they are only biologically active when their targeted bacteria are present in the environment (Sharma et al., 2017). Furthermore, BP application would not give rise to drug residues, making it an ideal antibiotic alternative (Golkar et al., 2014).

Although the levels of S.E. used in this challenge model are higher than normally cultured in poultry feed, researchers have demonstrated that less than one Salmonella organism per gram of feed is sufficient enough to colonize chicks (Schleifer et al., 1984). Therefore, reducing in-feed Salmonella is important to mitigating Salmonella infections in poultry. The tested BP could be used as an effective additive for the reduction of Salmonella in feed and incorporated into a comprehensive Salmonella management program.

CONCLUSIONS AND APPLICATIONS

1. Application of S.E. contaminated feed was a successful method of colonizing broilers with S.E.

2. The BP tested was successful in reducing S.E. contamination of feed and could be incorporated into an antibiotic-free Salmonella control plan.

This article was originally published in 2020 J. Appl. Poult. Res. 29:930–936. https://doi.org/10.1016/j.japr.2020.09.003. This is an Open Access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).