Experimental design, animals, housing and diets
Bacteriophages used in the present study was a commercial product from CJ Cheiljedang Corp. Seoul, South Korea, consisting of a mixture of phages targeting Salmonella gallinarum, Salmonella typhimurium, S. Enteritidis, Escherichia coli at the concentrations of 1.0x108 pfu/g each and Clostridium perfringens (1.0 x106 pfu/g). A total of 504 1-d-old male broilers (ROSS 308) with the initial BW 42.9 ± 1.0 g were used in a 35-d experiment. Chicks were randomly divided into the four experimental groups, and each group had 7 replicate cages, with 18 broilers per replicate cage. The treatment groups were as follows: i) CON group (Control/ basal diet without BP supplementation), ii) PC group (CON + 0.25 g antibiotics; AVILAMIX®/kg feed), iii) BP 0.05 group (CON + 0.5 g bacteriophage/kg feed), and iv) BP 0.10 group (CON + 1.0 g bacteriophage/kg feed). The bacteriophage cocktail concentrations used in the present study was based on the concentrations used in previous studies [14, 17-19] and was administrated by replacing the same amount of corn. Broiler chickens were raised in a temperature-controlled room in a three-tier stainless steel cages of identical size having 8 adjacent cages per level. The dimensions of each cage was 120 cm width x 40 cm length x 60 cm height and was equipped with 2 drinker nipples and 2 open trough feeders. Room temperature was maintained at 33 ± 1°C for the first 3 d, and then gradually reduced by 3°C a week until reaching 24°C and maintained for the remainder of the experiment and the relative humidity was around 60%. The basal diet was formulated to meet or exceed all the nutrient requirements of broilers as recommended by National Research Council , and supplied in mash form. The experiment was divided in two nutritional phases, including starter (1 to 21 d), and finisher phase (22 to 35 d), and the ingredients and analyzed nutrient composition of the basal diet are shown in Table 1. Artificial light was provided 24 h/d by the use of ﬂuorescent lights. All diets were fed in mash form with feed and water being provided ad libitum throughout the experimental period.
Sampling and measurements
Body weight and feed consumption were recorded at day 0, 7, 21 and 35. This information was then used to calculate body weight gain (BWG) average feed intake (FI), and feed conversion ratio (FCR).
Broilers were fed the respective diets containing 0.20% chromium oxide (Cr2O3) as an indigestible marker for 7 d prior to the total excreta collection period on day 35. Excreta samples were collected by placing a collecting tray under each replicate cage for the analysis of total tract apparent digestibility for dry matter (DM), gross energy (GE) and nitrogen (N). The representative feed and excreta samples were immediately stored at -20◦C until analysis. The excreta samples were dried for 72 h at 70◦C and finely ground to allow for passage through a 1-mm screen. The procedures utilized for the determination of total tract apparent digestibility for DM, GE and N were in accordance with the methods established by the AOAC International . Diets samples were analyzed for crude protein (N×6.25; method 988.05), crude fat (954.02), ash (method 942.05), calcium (method 984.01), phosphorous (method 965.17) and amino acids (method 982.30E) following the procedures established by AOAC, International . Chromium levels were determined via UV absorption spectrophotometry (UV-1201, Shimadzu, Kyoto, Japan) and the apparent total tract digestibility (ATTD) of DM, N, were calculated using indirect methods described by Williams et al. . Nitrogen was determined (Kjeltec 2300 Nitrogen Analyzer, Foss Tecator AB, Hoeganaes, Sweden), and CP was calculated as N × 6.25. Amino acid analyzer (Beckman 6300, Beckman Coulter Inc., Fullerton, CA) was used to measure lysine and methionine after acid hydrolysis for 24 h in HCl. Parr 6100 oxygen bomb calorimeter (Parr instrument Co., Moline, IL, USA) was used to determine gross energy by measuring the heat of combustion in the samples.
Excreta microbial counts
For excreta microbial counts, excreta samples were collected from all 7 replicate cages each treatment at day 35 by placing excreta collection trays under each cage. Fresh droppings (deposited within 2 h) were collected from each replicate cage per treatment and transferred into clean plastic containers. The excreta samples were immediately transferred to the laboratory in an ice box for the enumeration of Salmonella, Escherichia coli (E. coli), Clostridium spp. and Lactobacillus. The viable counts of bacteria in the excreta were then determined by plating serial 10-fold dilutions (in 10 g/L peptone solution) in respective media. The selective medium used for isolation of Salmonella was Salmonella Shigella (Difco, USA), for E. coli, Mac Conkey (Difco, USA), for Clostridia spp. Cooked Meat Medium (Oxoid, UK) and for Lactobacillus, Lactobacilli medium III (Medium 638, DSMZ, Braunschweig, Germany). The Lactobacilli MRS agar plates were incubated for 48 h at 39°C, and the MacConkey agar and Salmonella Shigella agar plates were incubated for 24 h at 37°C whereas Cooked Meat Medium agar plates were incubated at 30°C for 24 h under anaerobic conditions. The colony counts were then enumerated and results are presented as log10-transformed data.
Ileal mucosa microbiome
For gut microbiome analysis, ileal mucosal samples were collected at day 35 from randomly selected 6 broilers per treatment groups (CON, PC, BP 0.05 and BP 0.10). Briefly, birds were sacrificed by cervical dislocation and exsanguination. After autopsy, the intestinal tract was excised and the intestinal content was removed followed by washing the intestinal segment with distilled water. Then ileal segment (distal ileum) was cut about 10–15 cm proximally to caeca and separated from the intestine and then rinsed in PBS and the mucosal layer was scraped with a glass slide. Mucosal scrapings were collected into a 50 ml conical tube and stored in an ice box and then transferred to Macrogen Inc., (Seoul, Republic of Korea) for gene sequencing. Genomic DNA extraction from the mucosal samples and the preparation of library of amplicons consisting of 16S rRNA gene and sequencing was done by Illumina MiSeq platform at Macrogen Inc. (Seoul, Republic of Korea) using MiSeq sequencing including barcoded 16S rRNA amplicons.
The 16S rRNA gene sequences were processed using the Mothur software to remove low-quality sequences . Briefly, sequences that did not match the PCR primers were eliminated from de-multiplexed sequence reads. The sequences containing ambiguous base calls and sequences with a length less than 100 bp to were trimmed minimize the effects of random sequencing errors. Chimeric sequences were further deleted using the UCHIME algorithm implemented in Mothur. QIIME (Quantitative Insights into Microbial Ecology) software package (version 1.9.1) was used for de novo operational taxonomic unit (OTU) clustering with an OTU definition at an identity cutoff 97% . Taxonomic assignment was performed using the naïve Bayesian RDP classifier and the Greengenes reference database. Beta-diversity was measured using unweighted UniFrac distance metrics using QIIME. The unweighted UniFrace considers the community membership (presence or absence of OTUs) . Principal coordinate analysis (PCoA) plots were generated based on the unweighted UniFrac distance metrics.
For physicochemical properties of the breast meat, 10 birds (n = 10) per treatment selected randomly at day 35 were individually weighed and killed by cervical dislocation and exsanguinated. The breast muscle (pectoralis major), Bursa of Fabricius, liver, spleen, and abdominal fat were then removed and weighed. Organ weights were expressed as a relative percentage to the whole body weight. The breast muscle Hunter lightness (L*), redness (a*), and yellowness (b*) values were determined using a Minolta CR410 chromameter (Konica Minolta Sensing Inc., Osaka, Japan). The pH of the breast muscle sample was measured by a calibrated, glass-electrode pH meter (Testo 205, Testo, Germany). The water-holding capacity (WHC) was analyzed according to the methods described by Kauffman et al. . Drip loss was measured using approximately 2 g of meat sample according to the plastic bag method described by Honikel .
Data were analyzed using the GLM procedure of SAS (version 9.4; SAS Inst., Inc., Cary, NC) in a completely randomized design. The cage served as the experimental unit for growth performance excreta microbial counts and digestibility indices whereas for microbiome and meat analysis, individual bird served as experimental unit. Pre-planned contrast was used to test the following: 1) the individual eﬀect of CON versus PC diets 2) the overall eﬀect of Bacteriophage supplementation versus PC diet (PC vs BP 0.05, BP 0.10). Furthermore, linear and quadratic polynomial contrasts were used to examine responses to supplemental graded levels of Bacteriophage at 0%, 0.05% and 0.1%. Variability in the data was expressed as the standard error of means (SEM) and P ≤ 0.05 was considered to be statistically signiﬁcant and P < 0.1 as trends.
For gut microbiome, analysis of similarities (ANOSIM) to determine whether the microbial compositions between the treatment and control groups were significantly different was done using QIIME software package (version 1.9.1) and was based on the unweighted UniFrac distance metrics.