Experimental design
A total of 370 one-day-old male broiler chickens (Ross 308) were purchased from a commercial hatchery (Lagerwey Hatchery, the Netherlands). All chicks were derived from the same 42 week old broiler breeder flock. At the hatchery the chicks were randomly allocated to two different experimental facilities (H1 and H2 + H3). The chicks were transported to these two facilities in the same truck. After a 30 min (H2 & H3) and 50 min’ (H1) drive, the day-old broilers arrived (day 0 of the experiment) and were placed in three different housing conditions (Fig. 8), i.e. H1, a grow-out feed trial facility, H2, a floor stable, and H3, isolators (Additional file 1: Figure S6).
H1 is a research facility at Cargill Animal Nutrition Innovation Center (Velddriel, the Netherlands), and consisted of standard grow-out pens used for broiler feed experiments. A total of 170 chicks were randomly allocated to 10 pens (2.26 m (b) × 0.90 m (w), 2.03 m2) (Fig. 8a). The distance between the two blocks separated by an aisle was 1.96 m. In each pen, 10 out of 17 broilers were followed individually throughout the grow-out period. Between the pens steel mesh panels were used as dividers, and the raised metal floor was covered with paper and a 2 cm layer of wood shavings. In this facility a downtime period of two week between experimental rounds was used, and between rounds the facility was cleaned and disinfected with a product with quaternary ammonium compounds and glutaraldehyde (MS Megades, Schippers, the Netherlands).
H2 and H3 were located at the Faculty of Veterinary Medicine of Utrecht University (Utrecht, the Netherlands). Broilers were randomly allocated to H2 or H3. In H2, 100 broilers were randomly distributed over 10 floor pens (1.00 m × 1.00 m, 1 m2), in one research unit (Fig. 8b). Adjacent pens were separated by solid wooden panels with only a mesh panel at the front of the pen. A single pen was present in four corners, and in the middle six pens were connected with each other (Fig. 8b). The floor of each pen was covered with a 2 cm layer of wood shavings. Before placement of the chicks, H2 was extensively cleaned and disinfected with vaporized hydrogen peroxide. In H3, 100 broilers were randomly distributed over 10 negative pressure HEPA filtered isolators (0.65 m × 1.5 m, 1 m2; Fig. 8c). All materials entering or leaving the isolators were passed through a chlorine tank sealed off with a removable lid. The floor consisted of a box (0.65 m × 0.65 m, 0.42 m2) filled with wood shavings to the same amount and from the same batch as in H1 and H2. The other 0.58 m2 consisted of a plastic mesh floor. All 10 isolators in H3 were extensively cleaned and disinfected with vaporized hydrogen peroxide before the experiment. There was a downtime period of six weeks with the previous experimental flocks for H2 and H3.
Between H1, H2 and H3 bird management conditions were kept as equal as possible. The wood shavings were transported from H1 to H2 and H3 three weeks before the start of the experiment and were stored under comparable conditions. Although the sizes of the pens were slightly different, the chick densities were the same. After 5 days, the number of chicks was reduced to 15 chicks per pen in H1 and 8 chicks per pen in H2 and H3 this resulted in a stocking density of 7.5 birds per m2 for H1, and 8 birds per m2 for H2 and H3. In the pens of H1 and H2 artificial lighting was set at 100 lx for 23 h/day (h/d) from day 0–3, 20 h/d from day 4–6 and 18 h/d from day 7–35. In the isolators artificial lighting was set with the same schedule, but with a light intensity of 200–400 lx. Temperature gradually decreased from 34 °C at day 0 with 2.5 °C per week to 20 °C at day 35. Temperature was monitored twice a day and corrected when needed. The birds were observed twice a day, and presence of clinical signs, abnormal behavior and mortality were recorded. At day 7 all birds in all facilities were vaccinated against Newcastle Disease virus (Avinew® Neo, Boehringer Ingelheim, Germany) with the same battery-operated backpack sprayer (H1 and H2) or handheld garden sprayer (H3).
Experimental feeds
All broilers had ad libitum access to feed and water throughout the experimental period. To establish differences in intestinal microbiota, feeds with and without MCFA (+MCFA and –MCFA) were used [22]. The two different feeds were formulated to meet the nutrient requirements of broilers and based on digestibility and nutrient data provided by the Feed Tables from the Dutch Central Bureau of Livestock Feeding (CVB, 2016). For each feeding phase, a starter and grower basal feeds were produced. Starter and grower feeds contained 2,850 and 2,925 kcal of apparent metabolizable energy (AME)·kg− 1 and 10.48 and 9.87 g·kg− 1 apparent fecal digestible lysine. The feed was wheat-corn soybean meal based and in the +MCFA feed, a blend of 0.3% C10:0 capric acid and 2.7% C12:0 lauric acid (Sigma-Aldrich, the Netherlands) was added. Diets were kept isocaloric by exchanging the MCFA blend with soybean oil and animal fat based on the ingredient energy values. Diets were produced at Research Diet Services (the Netherlands) and pelleted using steam addition (approximately 80 °C) at 2.5 mm (starter feeds; 0 to 14 days of age) and at 3.0 mm (grower feeds; 14 to 35 days of age). Diets did not contain antimicrobial additives.
Data collection
After cleaning and disinfection of the three housing units (H1–3), a hygienogram was made with Replicate Organism Detection And Counting (Rodac) plates. The Rodac plates (55 mm diameter) contained medium with 16 g/l agar, 1 ml/l tween 80, 1 g/l ammonium carbonate, 2 g/l lecithin, 1 g/l l-histidine, 5 g/l sodium chloride, 10 g/l meat extract, 10 g/l peptone (tryptone + meat peptone) and 0.5 g/l sodium thiosulphate (5H2O) (GD Animal Health, the Netherlands). In every pen at least one Rodac plate was pressed gently on a surface for 30 s. After incubation for 24 h at 38 °C the number of colonies was counted to determine the number of colony forming units (CFU).
In all three housing conditions, individual broiler weights were recorded at the start of the experiment (day 0) and at days 14 and 35. In addition to body weights, also feed consumption was recorded for each pen. The gain to feed ratio was calculated as kg of weight gain/kg of feed consumed for each time period (0 to 14, 14 to 35 and 0 to 35 days of age). On day 35, all broilers in the different housing units were euthanized, using carbon dioxide (H1) or electrocution followed by cervical dislocation (H2 and H3). A pen from H2 and H3 were selected alternately for euthanasia, to avoid a sampling effect due to time differences between housing conditions. Cecal content of each broiler was gently squeezed into sterile cryotubes and snap frozen on dry ice and stored at − 80 °C for microbial genomic DNA extraction. Between sampling of each broiler sterile gloves were changed, and the table, scissors and tweezers were cleaned with 70% ethanol to prevent cross contamination between samples.
DNA extraction
DNA was extracted, from 0.25 g cecal content, using 700 μl Stool Transport and Recovery (STAR) buffer (Roche Diagnostics Nederland BV, the Netherlands). The cecal sample was transferred to a sterile screw-capped 2 ml tube (BIOplastics BV, the Netherlands) containing 0.5 g of zirconium beads (0.1 mm; BioSpec Products Inc., USA) and 5 glass beads (2.5 mm; BioSpec Products Inc., USA). The samples were treated in a bead beater (Precellys 24, Bertin technologies, France) at a speed of 5.5 ms− 1 for 3 × 1 min, followed by incubation at 95 °C with agitation (15 min and 300 rpm). The lysis tube was centrifuged (13,000 g for 5 min at 4 °C), and the supernatant was transferred to a 2 ml microcentrifuge tube. Thereafter, the above described process was repeated with 300 μl of STAR buffer. An aliquot (250 μL) of the combined supernatants from the sample lysis was then transferred into the custom Maxwell® 16 Tissue LEV Total RNA Purification Kit cartridge. The remainder of the extraction protocol was then carried out in the Maxwell® 16 Instrument (Promega, the Netherlands) according to the manufacturer’s instructions. DNA concentration was measured with a NanoDrop ND-1000 spectrophotometer (NanoDrop® Technologies, USA), and DNA was stored at − 20 °C until further use.
Microbiota composition
Extracted DNA was diluted to 20 ng/μL in nuclease free H2O. All PCR plastics were UV irradiated for 15 min before use. For 16S rRNA gene-based microbial composition profiling, barcoded amplicons covering the variable regions V5-V6 of the 16S rRNA gene were generated by PCR using the 784F and 1064R primers [42].
Each sample was amplified in duplicate using Phusion hot start II high fidelity polymerase (Finnzymes, Finland), checked for correct size and concentration on a 1% agarose gel and subsequently combined and purified using CleanNA magnetic beads (CleanNA the Netherlands). The 50 μl PCR reactions contained 36.5 μL nucleotide free water (Promega, USA), 0.4 μL of 2 U/μl polymerase, 8 μL of 5 × HF buffer, 1 μl of 10 μM stock solutions of each of the forward (784F) and reverse (1064R) primers, 1 μL 10 mM dNTPs (Promega) and 1 μL template DNA.
Reactions were held at 98 °C for 30 s and amplification proceeded for 25 cycles at 98 °C for 10 s, 42 °C for 10 s, 72 °C for 10 s and a final extension of 7 min at 72 °C. Synthetic communities of known composition were added as positive controls [42], and samples with nuclease free water were added as no-template negative controls to ensure high quality sequencing data. A composite sample for sequencing was created by combining equimolar amounts of amplicons from the individual samples, followed by a final purification step with magnetic beads to remove any remaining contaminants. The resulting libraries were sent to GATC Biotech (Germany; now part of Eurofins Genomics Germany GmbH) for sequencing on an Illumina Hiseq2500 instrument.
Data was analyzed using NG-Tax [42]. In short, paired-end libraries were filtered to contain only read pairs with a perfect match to the primers and perfectly matching barcodes, to demultiplex reads by sample. OTU were defined as unique sequences. The OTU picking strategy was based on an open reference approach. First, reads were sorted by abundance per sample and OTUs with an abundance of < 0.1% were discarded. In a second step the remaining reads were matched to the first set of OTUs allowing for one mismatch. Taxonomy was assigned using SILVA 128 16S rRNA gene reference database [43].
High-performance liquid chromatography (HPLC)
After DNA extraction, from the same 2 ml cryo tubes, 100 mg of cecal content was diluted in 900 μl Milli Q, and centrifuged (13,000 g for 15 min at 4 °C). Supernatant was stored at − 20 °C until HPLC analysis. Crotonate was used as internal standard, and the external standards were acetate, butyrate, isobutyrate, lactate and propionate. Substrate conversion and product formation were measured with a Spectrasystem HPLC (Thermo Scientific, the Netherlands) equipped with a Shodex SUGAR SH1011 column with guard column SUGAR KS-G 6B (Agilent, the Netherlands) for the separation of organic acids and carbohydrates. Measurements were conducted at a column temperature of 45 °C with an eluent flow of 0.8 ml min − 1 flow and the detector RID 20a.
Statistical analysis
All statistical analyses were performed in R version 3.4.2 (R Foundation for Statistical Computing, Austria [44]), using the packages: Phyloseq, Microbiome, and Vegan [45,46,47]. To test for differences in relative abundance of genera between two groups, we used a Wilcoxon rank-sum test and corrected for multiple testing with Benjamini-Hochberg (BH). Alpha diversity (within sample) was determined using phylogenetic diversity [48], Shannon, Inverse Simpson and Fisher. Faiths phylogenetic diversity not only takes into account the numbers of bacteria, but also the phylogenetic relatedness of those bacteria [48]. Beta diversity (between samples) was determined using Jaccard, Bray-Curtis, weighted and unweighted UniFrac metrics [49,50,51]. Differences in alpha diversity between treatment groups were tested with a Kruskal-Wallis test and pairwise comparisons were tested using a Wilcoxon rank-sum test. Multivariate microbiota data were visualized using principal coordinates analysis (PCoA), and non-parametric permutational analysis of variance (PERMANOVA) tests were used to analyze group differences within multivariate community data [52]. Growth performance data (body weight, average daily gain, average daily feed intake and gain to feed ratio) and concentrations of butyrate, acetate, propionate and lactate concentrations were analyzed with ANOVA test with Tukey’s post-hoc test, using pen as experimental unit.