Selected Lactic Acid-Producing Bacterial Isolates with the Capacity to Reduce Salmonella Translocation and Virulence Gene Expression in Chickens

Background Probiotics have been used to control Salmonella colonization/infection in chickens. Yet the mechanisms of probiotic effects are not fully understood. This study has characterized our previously-selected lactic acid-producing bacterial (LAB) isolates for controlling Salmonella infection in chickens, particularly the mechanism underlying the control. Methodology/Principal Findings In vitro studies were conducted to characterize 14 LAB isolates for their tolerance to low pH (2.0) and high bile salt (0.3–1.5%) and susceptibility to antibiotics. Three chicken infection trials were subsequently carried out to evaluate four of the isolates for reducing the burden of Salmonella enterica serovar Typhimurium in the broiler cecum. Chicks were gavaged with LAB cultures (106–7 CFU/chick) or phosphate-buffered saline (PBS) at 1 day of age followed by Salmonella challenge (104 CFU/chick) next day. Samples of cecal digesta, spleen, and liver were examined for Salmonella counts on days 1, 3, or 4 post-challenge. Salmonella in the cecum from Trial 3 was also assessed for the expression of ten virulence genes located in its pathogenicity island-1 (SPI-1). These genes play a role in Salmonella intestinal invasion. Tested LAB isolates (individuals or mixed cultures) were unable to lower Salmonella burden in the chicken cecum, but able to attenuate Salmonella infection in the spleen and liver. The LAB treatments also reduced almost all SPI-1 virulence gene expression (9 out of 10) in the chicken cecum, particularly at the low dose. In vitro treatment with the extracellular culture fluid from a LAB culture also down-regulated most SPI-1 virulence gene expression. Conclusions/Significance The possible correlation between attenuation of Salmonella infection in the chicken spleen and liver and reduction of Salmonella SPI-1 virulence gene expression in the chicken cecum by LAB isolates is a new observation. Suppression of Salmonella virulence gene expression in vivo can be one of the strategies for controlling Salmonella infection in chickens.


Introduction
Salmonella is a zoonotic pathogen. Its infection in human and domestic animals remains a serious and worldwide issue in public health, food safety, and livestock economies [1]. Consumption of Salmonella-contaminated livestock and their products, particularly poultry and poultry products, has often led to salmonellosis in humans. In addition, the increased level of antibiotic resistance in Salmonella recovered from animals potentially erodes the therapeutic effectiveness in antibiotic treatment of animal and human disease. Thus, control of Salmonella infection in animals is still urgently required.
Salmonella enterica serovar Typhimurium is one of three serotypes, sometimes responsible for 50% of the salmonellosis cases in humans [2]. The susceptibility of humans and animals to Salmonella infection is generally determined by both Salmonella virulence and host factors, such as age and genetics. The infection normally starts from the intestinal phase, through which Salmonella can enter the blood circulation and spread to various organs such as the liver and spleen, leading to systemic infection. There are two pathogenicity islands in Salmonella that are closely associated with virulence [3]. Salmonella pathogenicity island 1 (SPI-1) is critical in causing the intestinal phase of infection [3], while Salmonella pathogenicity island 2 (SPI-2) is essential for intracellular multiplication, which is necessary for systemic infection [4]. The SPI-1 is composed of a 40-kb DNA region coding for the type III secretion system (T3SS), including the structural, effector, and regulatory proteins.
It is important to control Salmonella infection at the farm level in order to reduce Salmonella contamination in poultry and poultry products. Attempts to control Salmonella infection and carriage by vaccination and antibiotic treatment of chickens have not yet produced significant results [5,6]. An alternative approach has been the use of competitive exclusion and probiotics [7]. Competitive exclusion of Salmonella is achieved by enhancing the establishment of normal intestinal microbiota in newly hatched chickens and thus inhibiting intestinal colonization of Salmonella [8]. Probiotics can have different modes of action to control the infection of enteric pathogens, such as secretion of antimicrobial substances (for example bacteriocins) [9], enhancement of digestive functions [10], and modulation of the mucosal and systemic immune responses [11,12]. Control of Salmonella burden and colonization has been one of the common strategies. Previous studies have shown that the expression of gene encoding virulence factors of Salmonella can be modulated by probiotics in vitro [13,14,15,16,17]. Recently, we identified several lactic acidproducing bacterial (LAB) isolates that were able to protect Caenorhabditis elegans from death caused by S. Typhimurium infection [18]. In the present study, we have investigated the probiotic properties of these isolates and, moreover, if they are able to reduce Salmonella infection in broilers and also modulate Salmonella SPI-1 virulence gene expression.

Results
Resistance of selected lactic acid-producing bacterial isolates to low pH, high bile salt, and antibiotics determined by in vitro studies Among the 14 tested LAB isolates, eight (ALB2, ALB6, CL9, L3, LB1, LB2, LB4, and S64) have been reported previously to have antimicrobial activity towards S. Typhimurium DT104 [18]. The remaining six isolates (K16, K67, S8, S33, S66, and SG2) were also tested for their antimicrobial activity using the method of spot-on-the-lawn [19]. These isolates were able to inhibit the growth of S. Typhimurium DT104 with clear inhibition halos (diameter $16 mm). Table 1 shows the tolerance of 14 selected LAB isolates to low pH pretreatment. Six isolates (ALB2, CL9, L3, LB1, LB4, and SG2) had similar optical densities at 600 nm (OD 600 ) after growth regardless if the cultures had been pretreated at pH 2 or 5.6, indicating their tolerance to low pH. However, the growth of seven out of eight remaining isolates (K16, K67, LB2, S8, S33, S64, and S66) was inhibited by the low pH pretreatment (P,0.05). In the test to examine tolerance to high bile salt, four (ALB2, CL9, K16, and S33) out of the 14 isolates showed poor growth (OD 600 ,0.5) after 24 h incubation at 37uC at most concentrations of bile salt above 0.3% (v/v), suggesting that they are sensitive to bile acid ( Table 2). Eight isolates (K67, L3, LB1, LB2, LB4, S8, S66, and SG2) were able to grow with OD 600 values greater than 1.0 at concentrations of bile salt from 0.3% to 1.5% (v/v), indicating good tolerance to bile salt. The remaining two isolates (ALB6 and S64) had a moderate growth with the OD 600 values between 0.5 and 1.0 after 24 h incubation at bile salt concentrations above 0.3% (v/v). Table 3 shows the susceptibility of the 14 tested LAB isolates to eight different antibiotics. The isolates varied in their susceptibility to the antibiotics. Three isolates (LB1, LB2, and LB4) were susceptible to all tested antibiotics within the concentrations examined. The majority of remaining isolates grew at the highest concentration of one or two antibiotics. Based on the microbiological cut-off values set by the European Food Safety Authority [20] and the European Commission [21], the majority of the 14 isolates could be categorized as susceptible to ampicillin and erythromycin and about half of the isolates were resistant to gentamicin, tetracycline, and ciprofloxacin.
Effect of selected lactic acid-producing bacterial isolates on Salmonella burden in the cecum of chickens Isolates LB1 (Lactobacillus zeae), S8 (Lactobacillus plantarum), S64 (Lactobacillus reuteri), and S33 (Pediococcus pentosaceus) were selected for further in vivo studies after the in vitro characterization. The selection was based on their resistance to low pH, high bile salt, and antibiotics, their species identification, and previously reported ability to protect C. elegans from death caused by Salmonella infection [18]. In the in vivo studies, cecal digesta samples from all negative control groups of chickens (three chicken trials) had undetectable Salmonella counts that were below the detection limit of the plating method (2610 2 Colony Forming Units (CFU)/g digesta) [22]. In Trial 1, Salmonella counts in the cecal digesta from the chickens in the positive control group (challenged with Salmonella only) on day 4 post-infection (PI) were 5.3360.13 (Log 10 CFU/g fresh digesta). LB1, S8, S33, or S64 treatment groups had 5.7060. 12, 5.3860.12, 5.4760.13, or 5.7860.07 Log 10 CFU of Salmonella/g fresh digesta in their ceca. No significant difference among these five treatments (P.0.05) was found.
Since the experimental design of Trials 2 and 3 was identical, data from these two trials were analyzed by a mixed model with trial as a fixed effect (Table 4). When the data from Trials 2 and 3 were pooled, among the four probiotic treatment groups, the low and high dose mixed culture groups and the high dose LB1 group had significantly higher Salmonella counts in the cecal digesta on days 3 and 4 PI than the positive control, indicating that the selected individual probiotic candidates (LB1, S8, and S64) or their combination at the tested level could not reduce the Salmonella burden in the cecum of chickens.
Effect of selected lactic acid-producing bacterial isolates on invasion of Salmonella in the spleen and liver of chickens In Trial 1, eight and seven birds out of the 12 birds from the positive control group were found to be Salmonella positive in the spleen or liver, respectively, on day 4 PI after the tissue samples were enriched for Salmonella. This suggests a successful Salmonella challenge at the dose of 1610 4 CFU/chick. As shown in Tables 5  and 6, the data on spleen and liver infection from Trials 2 and 3 were combined for analyses with trial as a fixed effect. No significant effect (P.0.05) of trial was observed. Compared with the positive control, the treatment with a low dose of mixed LAB cultures was able to attenuate Salmonella infection in the spleen (73% vs. 32%, P = 0.06) and liver (74% vs. 44%, P = 0.07) on days 3 and 4 PI. A similar result was also observed with the treatment of low dose LB1, which reduced Salmonella infection in the liver (74% vs. 47%, P = 0.09).
Effect of selected lactic acid-producing bacterial isolates on the SPI-1 virulence gene expression of Salmonella in the cecum of chickens In Trial 3, expression of 10 SPI-1 virulence genes (hilA, hilC, hilD, sopB, sopD, sopE2, sipA, sipC, avrA, and sptP) of Salmonella in the cecal digesta of chickens was investigated by RT-qPCR assays. Compared with the Salmonella cells from the chickens in the positive control group, application of LB1 or mixed LAB cultures did not significantly (P.0.05) affect expression levels of the 10 virulence genes on day 1 PI. Interestingly, expression of all the genes except hilC in the Salmonella cells from all the LAB treatment groups of chickens, regardless of dose and combinations, was down-regulated by 4 to 257 fold (P,0.05) on day 3 PI compared with the positive control group ( Table 7). The low dose group treated with mixed LAB cultures had the most dramatic decrease (18 to 257 fold) for all the genes, followed by the low dose LB1 group. Nevertheless, there was no significant (P.0.05) difference in the gene expression of Salmonella from these four groups of chickens that had been treated with LAB isolates. In vitro expression of SPI-1 virulence genes was also examined. Nine out of the 10 virulence genes in Salmonella were down-regulated when co-cultured with the extracellular culture fluid (ECF) from isolate LB1 cultures (Table 7). Only sipC was up-regulated.

Discussion
This study was to characterize our recently identified LAB isolates for controlling Salmonella infection in broilers, including Table 1. Growth (OD 600 ) of selected lactic acid-producing bacterial isolateswith or without pretreatment of low pH * . determination of the mechanism for the control in particular. We have demonstrated that orally administered LAB isolates at a low dose (10 6 CFU/strain/chick) could attenuate Salmonella infection in the spleen and liver of broilers, which coincided with downregulation of Salmonella SPI-1 virulence gene expression in the chicken cecum. In addition, our data on in vitro expression of Salmonella SPI-1 virulence genes also support our in vivo observations, suggesting a possible correlation between the attenuation of organ infection and down regulation of SPI-1 virulence gene expression. The LAB treatments were unable to lower cecal Salmonella burden in the present study. The reduction of Salmonella burden in the gastrointestinal tract by probiotics is subjected to multiple factors, among which antimicrobial substances (e.g. bacteriocins) produced by probiotic bacteria can have a direct impact [9]. In this regards, we speculate that the LAB isolates tested in the present study may not have produced antimicrobial substances. Further studies are required to determine the underlying reason. S. Typhimurium harbours more than 80 virulence genes. The ten genes examined in the present study are located in SPI-1 and have important roles in the intestinal invasion of Salmonella [4]. Expression of SPI-1 genes is controlled by 5 regulators (HilA, HilC, HilD, InvF, and SprB) with HilA being a master regulator [4]. In addition, these SPI-1 regulators can modulate SPI-2 gene expression that is essential for intracellular multiplication and lead to systemic infections. For instance, HilD is required for activating Table 4. Salmonella counts in the cecal digesta on days 3 and 4 post-infection in Trials 2 and 3 * .  the regulatory region of the ssrAB operon, a key regulator of SPI-2 genes [4,23]. In the present study, we have examined the gene expression of hilA, hilC, and hilD in response to LAB treatment. The expression of hilA and hilD genes was significantly downregulated by both single and mixed LAB isolates, particularly at the low dose. Translocation of the SPI-1 effectors (AvrA, SipA, SipC, SopB, SopD, SopE2, and SptP) into the host enterocytes via the type III secretion system (T3SS) is responsible, at least in part, for the intestinal invasion [24,25,26]. AvrA, a member of a family of acetyl transfereases, can dampen host inflammatory and   apoptotic defense [27,28]. SipA affects actin cytoskeleton, induces membrane shuffling, and promotes inflammation of the host cells. SipC is involved in actin bundling and nucleation. Through its phosphoinositide phosphatase activity, SopB participates in a broad range of events such as host membrane fission, Salmonellacontaining vacuole maturation, and inflammation [29]. SopD is encoded outside SPI-1 and appears to be involved in membrane fission and macropinosome formation. SopE2, a guanine nucleotide exchange factor for Cdc42 of host cells, is important in membrane ruffling, bacterial entry and inflammation induction [26,30]. SptP is involved in inflammation and cytoskeleton manipulation of host cells [26]. SipA, SopB, SopD, and SopE2 have also been indicated to play roles in the late stage of Salmonella infection in mice [31]. Expression of avrA in Salmonella has been shown to increase bacterial invasion in the jejunum and colon and translocation in the spleen, liver, and gallbladder in chronically Salmonella-infected mice [32]. In the present study, the expression of all of these seven SPI-1 effectors was significantly downregulated by both single and mixed LAB isolates. Compared with the corresponding high dose of LAB groups, the down-regulation was more evident at the low dose of LAB treatments except for the expression of sopE2 by LB1 treatment only ( Table 7). Modulation of virulence gene expression in bacterial pathogens can be a mechanism underlying probiotic effects to control the infection of enteric pathogens. Previous in vitro studies have shown that lactobacilli and Bifidobacterium bifidum were able to downregulate gene expression of hilA and hilD in S.Typhimurium and Enteritidis, respectively [13,14,15,16,17]. Similar in vitro observations have also been reported in E. coli O157:H7, in which Lactobacillus acidophilus repressed virulence gene expression and quorum sensing [33] and Lactobacillus rhamnosus reduced the expression of Shiga toxin genes (stx) [34]. The expression of eight Cag pathogenicity island genes was also down-regulated by Lactobacillus salivarius in a gastric epithelial cell line model [35]. Salmonella virulence gene expression can also be modulated by lactic acid and short-chain fatty acids (SCFAs). Lactate, propionate, and butyrate have been shown in vitro to down-regulate the expression of Salmonella virulence genes, such as hilA, hilD, invF, and sipC [36,37,38,39,40,41]. Acetate and formate were, however, reported to induce the in vitro expression of SPI-1 genes [36,42]. Some of our selected LAB isolates were able to produce lactic acid and SCFAs during fermentation (data not shown). It is unclear at present if the suppression of SPI-1 virulence gene expression of Salmonella in the chicken cecum described in the present study was caused by lactic acid and SCFAs produced by the isolates.
There is a lack of in vivo evidence on the reduction of Salmonella virulence gene expression by probiotics for control of the pathogen. The present study has generated a new piece of evidence demonstrating that the expression of intestinal infectionassociated genes in S. Typhimurium can be down-regulated by selected LAB isolates when the pathogen was in the chicken cecum. This phenomenon coincided with an attenuation of Salmonella infection in the chicken spleen and liver. The evidence is further supported by our in vitro data, which revealed that coculturing of ECF of L.zeae LB1 cultures with Salmonella resulted in a significant reduction in the expression of genes associated with Salmonella intestinal infection. These observations can thus lead to a hypothesis that metabolites from the LAB cultures are, at least in part, responsible for regulating the virulence gene expression in Salmonella, influencing Salmonella intestinal infection of chickens. Identification of the functional metabolites from LAB cultures would be the initial step to address the hypothesis, which can be facilitated by using in vitro assays for the virulence gene expression. Furthermore, identification of a dose response to the functional metabolites in down-regulating the virulence gene expression in Salmonella in vivo and in attenuating the organ infection of chickens remains critical in supporting the hypothesis.

In vitro characterization of lactic acid-producing bacterial isolates
Fourteen LAB isolates, including eight (ALB2, ALB6, CL9, L3, LB1, LB2, LB4, and S64) reported previously [18] and six used in the present study (K16, K67, S8, S33, S66, and SG2), were evaluated in triplicate in vitro for their tolerance to low pH and high bile salt and antibiotic resistance. The stock culture of each strain, stored at 280uC in 20% glycerol, was plated onto DeMan, Rogosa, and Sharpe (MRS) plates. Individual colonies were selected and inoculated into MRS broth, cultured at 37uC under anaerobic conditions (85% N 2 , 10% CO 2 , and 5% H 2 ) for 18 h, and the species identity of the six isolates were determined by sequencing 16S rRNA genes and comparison with the GenBank database using BLAST.
The susceptibility of the isolates to penicillin G, ciprofloxacin, tetracycline, erythromycin, ampicillin, and gentamicin was evaluated on MRS agar with MIC (Minimum Inhibition Concentration) Evaluator Strips (Oxoid, Basingstoke, UK) through gradient diffusion tests. The inhibitory zones generated by chloramphenicol and lincomycin were determined on the MRS agar using Oxoid Antimicrobial Discs (Oxoid, Basingstoke, UK). The range of antibiotics concentrations used in the test is listed in Table 3.
Tolerance to low pH was evaluated by incubating bacterial cells in simulated gastric fluid [43] with the pH values of 2.0 and 5.6, for 2 h prior to anaerobic incubation in the MRS broth at 37uC for 20 h. Tolerance to bile salt was studied by incubating the isolates in the MRS containing various concentrations (0, 0.3, 0.6, 1.5, and 1.5%) of bile salt (Catalog Number LP005, Oxoid, Nepean, ON, Canada) under anaerobic condition at 37uC for 24 h. The optical density (OD) of a sample was measured at a wavelength of 600 nm. The selected range of bile salts was based on the physiological concentration in the small intestine [44].
The effect of the ECF from a LAB culture (L. zeae LB1) on in vitro expression of Salmonella SPI-1 virulence genes was examined. To prepare the ECF, L. zeae LB1 was grown in MRS broth anaerobically at 37uC for 16 h and the ECF was recovered from the cultures (n = 3) by centrifugation (6,000 g, 10 min). After neutralized to pH 6.8-7.0 with NaOH (1 N), the ECF was sterilized through membrane filtration (0.2 mm pore size) and then mixed with the Luria-Bertani (LB) broth (1:1 ratio) that had double concentrations of each medium component. The ECF-LB mixtures were inoculated (1%) with S. Typhimurium P193, a nalidixic acid-resistant strain, and incubated at 37uC for 14 h prior to the treatment of Salmonella cells with RNAlater (Life Technologies Inc., Burlington, Canada).

Animal experiments
Newly hatched commercial female broiler chicks were obtained from Stratford Chick Hatchery (Stratford, ON, Canada). Birds were housed in floor pens containing wood shavings at the Isolation Unit of Ontario Veterinary College (University of Guelph, ON, Canada). Chickens had free access to water and a commercial starter diet without supplementation of antibiotics. The research protocol was approved by the University of Guelph Animal Care Committee and the guidelines of the Canadian Council for Animal Care were followed.
S. Typhimurium P193 was used to infect chickens according to the procedure described previously [45]. The procedure was originally introduced by Mead et al. (1989) [46] for assessment of probiotic efficacy. Salmonella was cultured in LB broth at 37uC for 16 h, followed by dilution with sterilized phosphate-buffered saline (PBS, pH 7.2) to a final cell count of 4610 4 CFU per ml for chicken infection experiments. To prepare LAB inoculum cultures, each isolate was propagated twice in MRS broth and then diluted to the desired concentrations with sterilized PBS. The desired concentrations were confirmed by plating on MRS plates after a series dilution.
To follow up our observation on Salmonella infection to chicken spleens and livers in Trial 1, Trials 2 and 3 were designed to investigate the effect of Lactobacillus isolates on protecting broilers from organ infection by Salmonella, including the influence of the isolates on virulence gene expression of Salmonella in the cecum. There were six treatments in each trial: 1) negative control (no LAB + no Salmonella), 2) positive control (no LAB + Salmonella 1610 4 CFU/chick), 3) low dose LB1 (isolate LB1 1610 6 CFU/ chick + Salmonella 1610 4 CFU/chick), 4) high dose LB1 (isolate LB1 1610 7 CFU/chick + Salmonella 1610 4 CFU/chick), 5) low dose mixed isolates (isolates LB1, S8, and S64 1610 6 CFU/ strain/chick + Salmonella 1610 4 CFU/chick), and 6) high dose mixed isolates (isolates LB1, S8, and S64 1610 7 CFU/strain/ chick + Salmonella 1610 4 CFU/chick). Each treatment had 14 and 28 chickens in Trials 2 and 3, respectively. The maintenance of chickens, gavage of Lactobacillus cultures, and challenge of Salmonella were similar to those in Trial 1. Cecal digesta, livers and spleens were sampled aseptically on days 3 and 4 postinfection (PI) for Salmonella analysis in Trials 2 and 3. In addition, cecal digesta samples were aseptically collected on days 1 and 3 PI in Trial 3 and kept immediately in RNAlater (Life Technologies Inc., Burlington, Canada) for analysis of virulence gene expression of Salmonella.

Detection of cecal colonization and organ infection of Salmonella
On the same day of sample collection, cecal digesta samples were processed by a series 1:10 dilution with sterile solution of 0.1% peptone and 0.1% Tween 80 in water and plating on Brilliant Green Sulfa Agar (BGS; BD, Sparks, MD, USA) supplemented with 200 mg/ml nalidixic acid to facilitate selection of the antibiotic-resistant challenge organism. Salmonella colonies are pink to red on this selective media. The plating was accomplished by an Eddy Jet Spiral Plater (Neu-tec Group, Farmingdale, NY, USA). For detecting organ infection of chickens, about 0.1 g of spleen or liver samples were homogenized with PowerLyzer (MO BIO Laboratories, Carlsbad, CA,USA) by bead beating at homogenization level 2500 for 30 sec in a tube containing 1 ml of 0.1% peptone water and 0.5 ml of beads and incubated for 16 h at 37uC in 4 ml of Selenite Brilliant Green Sulfa Enrichment (Difco) broth (Sparks, MD, USA), followed by plating onto BGS supplemented with 200 mg/ml nalidixic acid to examine the presence or absence of Salmonella colonies. The recovery of Salmonella is reported as the ratio of the number of Salmonella positive samples to the number of total samples tested. The sensitivity of sample preparation (including both digesta and organ samples) and plating was assessed by including a known amount of Salmonella cells in the samples. The detection limit of our sample preparation and plating system was determined to be 100 CFU/g sample if the enrichment procedure was not applied.
Extraction of bacterial total RNA and reverse transcription/quantitative PCR (RT-qPCR) Bacterial total RNA was extracted using the Ambion RiboPure-Bacteria Kit (Life Technologies Corporation, Carlsbad, CA, USA) according to the manufacturer's instructions. Briefly, samples of cecal digesta or Salmonella pure cultures that had been treated in RNAlater were washed with 2 volumes of PBS. After centrifugation (1500 x g) at 4uC for 10 min, the pellet was transferred to a new tube containing 0.5 ml beads and 0.8 ml RNAwiz provided with the Ambion RiboPure-Bacteria Kit and subjected to bead beating with PowerLyzer (MO BIO Laboratories, Carlsbad, CA, USA) for 2 min each at homogenization level 2500, repeated once. The tubes were chilled on ice for 2 min between bead beatings. Total bacterial RNA was then extracted, washed, and eluted following the manual of the kit. The RNA quantity was measured using an ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA).
To remove genomic DNA contamination, total bacterial RNA samples were treated with DNaseI from the Ambion TURBO DNA-free kit (Life Technologies Corporation, Carlsbad, CA, USA) according to the manufacturer's protocol. Two micrograms of total RNA was then used for synthesis of cDNA with 100 ng of random hexamers and SuperScript II reverse transcriptase from the Invitrogen SuperScript II first-strand cDNA synthesis kit (Life Technologies Corporation, Carlsbad, CA, USA) according to the manufacturer's instructions.
The qPCR assays were performed using a Stratagene MX3005 thermal cycler (Stratagene, La Jolla, CA, USA) with the following thermal cycle profile: 95uC for 3 min for initial activation of the DNA polymerase; 40 cycles of 95uC for 15 sec (denaturation), specific temperature listed in Table 8 for 30 sec (annealing) and 72uC for 30 sec (extension). A thermal melt curve was generated by heating at 95uC for 1 min, 55uC for 30 sec, and ramping back to 95uC in 0.5uC increments. The 25 ml reaction mixture contained 1.0 ml of 3-fold diluted sample cDNA, 12.5 ml of 26 iTaq SYBR Green Supermix (Bio-Rad, Hercules, CA, USA), and forward and reverse primers ( Table 8). The expression levels of virulence genes were normalized using 16S rDNA as an internal housekeeping control, and fold change of target genes was calculated by the delta-delta-Ct method [47]. DCt represents the difference between the Ct value with the primers to a target gene and the Ct value to the housekeeping gene (16S rRNA gene). DDCt represents the difference in DCt values between the treatment and control groups. The values derived from 2 2DDCt represent fold changes of samples in abundance relative to the reference samples. The reference samples had the 2 2DDCt value of 1. The samples from the positive control chicken group in the chicken trial and the samples not subjected to the ECF treatment in the in vitro experiment served as the reference samples, respectively.

Statistical analysis
Values were expressed as mean 6 standard error. Tolerance to low pH and high bile salt and gene expression data were analyzed by the SAS GLM procedure and Tukey multiple comparisons. Salmonella counts in cecal contents were log transformed and subjected to analysis of variance and Dunnett's test for multiple comparisons. Logistic regression was used for analysis of data on Salmonella infection in the liver and spleen. When data from animals in Trials 2 and 3 were pooled, mixed model was employed with trial as a fixed effect.