3.1 Ethical Statement
The study was approved by the Animal Welfare and Ethical Review Body at Newcastle University (Project ID #702), and all methods complied with UK regulations regarding the treatment of animals. Birds were housed and managed according to RSPCA-Assured standards and DEFRA guidelines on farm, and the Home Office Code of Practice while at the University. A Home Office Schedule 1 method of euthanasia was used. Reporting of the study follows the recommendations in the ARRIVE guidelines.
3.2 Adult Hens
From one day old, H&N and Hy-Line Brown pullets (Gallus gallus domesticus) were reared in two floor-based systems with litter at commercial farms in the Midlands, UK. Both sites were operated by the same pullet rearing company, Country Fresh Pullets, according to RSPCA-Assured standards, and arrived at an egg production farm, operated by a different company, in northern England in October 2017. The H&N pullets were 16 weeks of age (WOA) when introduced to a multi-tier free range adult housing system, whilst the Hy-Line birds were 17 WOA upon introduction to an enriched cage system on the adjacent site. At the time of sampling in October 2018, both groups had spent almost a year living in their respective systems, and hens were 65 WOA in the multi-tier system and 66 WOA in the enriched cage system.
The multi-tier housing unit consisted of 16,000 birds, divided into four internal colonies of 4,000 hens. However, as these birds all shared a range, individuals could move between the colonies during the day by accessing different popholes. Within the shed, the floor was covered with wood shavings litter and there were three additional tiers, the top of which was located 2.4m above the floor. The system provided round metal perches and nest boxes shaded by orange plastic dividers. Internal stocking density was 9 birds/m2 of usable area, where usable area is made up of the ground surface of the building accessible to the hens and additional raised areas and platforms at least 30 cm wide. Water was provided through nipple-drinkers, with one nipple for every 10 birds. Layer feed was circulated through a conveyor belt system, with a frequency of eight times per day. The average temperature inside the shed was 19 – 22oC. Birds received 15.75 hours of artificial light per day, from 06:45 to 22:30. Popholes opened at 09:00 daily and were closed 30 minutes after twilight ended. The grassy range had a dimension of 8.1 hectares and contained several two-tiered, covered wooden shelters, with ramps to the upper tier. No cover was provided by vegetation.
The enriched cage housing unit contained 33,120 Hy-Line birds, with a stocking density of 15 birds/m2. Each cage was 300 (l) x 150 (w) x 55 (h) cm in dimension and held 50 birds. Cages were arranged into four banks of three vertical tiers with 22 cages per tier, repeated over two floors. Enrichment provided in each cage consisted of perches, a nest box, a scratch mat and a grit auger to drop feed onto the scratch mat. The average temperature in the unit was 19 – 23oC. Birds received 15.5 hours of artificial light per day, from 01:45 to 17:15. Water was provided through nipple-drinkers, with one nipple for every 10 birds. Layer feed was circulated six times a day through a conveyor belt system. Birds in both systems experienced an identical programme of vaccinations. Hens in the multi-tier system were given a wormer at four intervals during lay. Due to poor laying performance, enriched cage birds alone were given a course of antibiotics at 39 WOA (Denegard; for the treatment and prevention of chronic respiratory disease and air sacculitis caused by Mycoplasma gallisepticum and Mycoplasma synoviae). Feed intake was typically 20g per bird higher per day in multi-tier system than in the enriched cages. At 65 WOA for comparability, average bird mass was 1544g in the multi-tier aviary compared to 1967g in the enriched cage system. Production rates were 90.3% in the multi-tier and 81.6% in the enriched cage system, and cumulative mortality had reached 2.50 and 2.74% respectively.
Whilst the multi-tier free-range and enriched cage housing systems were operated by separate personnel, they were overseen by the same Production Manager, whose role was to promote the efficient and sustainable production of eggs for retail sale.
3.3 Sampling
Within each housing system, birds with good and poor external indicators of physical body condition were selected. Hens were chosen by the farm’s Production Manager, who was familiar with the conditions of the birds. The a priori criteria employed were: i) body size, ii) feather coverage, and iii) redness of the comb, wherein a high level of each factor represented good physical condition. Because keel bone damage was previously found to be associated with reduced AHN42, candidate hens were palpated by the same researcher (EAA), trained to an agreed standard63, and those displaying signs of damage were rejected. A total of 12 hens with good body condition and 12 hens with poor body condition were sampled from both the multi-tier and enriched cage system, equating to a total sample size of 48 birds. This group size has proved sufficient to detect experience-based differences in our primary outcome measure (DCX+ cell densities) in previous studies41,42. Sampling occurred over four successive days, upon each of which 12 birds (3x multi-tier: good condition, 3x multi-tier: poor condition, 3x enriched cage: good condition, 3x enriched cage: poor condition) were selected from the farm and transported in carry boxes to Newcastle University for processing on the same day. In order to capture extremes of experience in the multi-tier system, good condition birds were selected directly from the range, while poor condition birds were selected from the top inside tier. This meant that the sampled good condition birds ventured outside at least some of the time, whilst the sampled poor condition birds may or may not have chosen to range. To allow birds time to go outside, there was a 30-minute delay between the opening of pop-holes in the morning and sampling of birds from the range. Hens were sampled from a different internal colony of the shed on each of the four days, along with the area of the range proximal to these colonies. To ensure independence in the enriched cage system, no two birds were sampled from the same cage, but to avoid a potential confound of cage height, a good and poor condition bird were always selected from different cages on the same tier. To collect a sample representative of the whole housing system, birds were sampled from each tier (top, middle and bottom) in both an inside and outside bank on each of the two floors. Representative images of good and poor physical condition hens sampled from each housing system are displayed in Figure 1.
3.4 Sampling of Representative Pullets at the Rearing Stage
As different strains of brown hens were sampled from the two adult housing systems, thereby confounding strain and housing type, a sample of pullets of the same genotypes were also taken directly from the rearing farm, to provide a baseline for possible genetic influences on the other measures taken. This sampling occurred after that of the adult hens, in January 2019. H&N and Hy-Line pullets were housed from one day-old in adjacent barns of a commercial rearing farm in Shropshire, England, operated by Country Fresh Pullets according to RSPCA-Assured standards64. The chicks originated from separate (hybrid-specific) hatcheries, each in the west of England. Both housing sheds contained litter in the form of wood shavings (Easichik, UK), which were bedded to a depth of 7.6 cm at the point chicks were introduced. Raised slatted areas were provided for perching, with access assisted by ramps placed every 7.6 m along their length. The H&N shed had a total area of 1,187.92 m2 (including the floor and raised slatted areas) and contained 17,085 pullets, which equated to a stocking density of 14.4 birds/m2. Though similar in design, the Hy-Line shed had a larger total area of 1,722.12 m2 and housed 25,236 pullets, with a stocking density of 14.7 birds/m2. Water was provided via nipple drinkers, with 12.2 and 12.4 birds per drinker in the H&N and Hy-Line sheds respectively. Both sets of birds had access to feed at all times via a chain feeder, but the quantity of feed in the feeder and the number of times it was topped up per day was adjusted throughout rearing, in order to maintain target body mass for each flock. The H&N shed housed 39 birds per metre of chain feeder, compared to 37 birds in the Hy-Line shed. Both strains received 10 hours of artificial light per day from six weeks of age onwards, with a light intensity of 10 lux at bird height. Ambient temperatures were 32-33oC for at the introduction of day-old chicks and were gradually reduced by 0.5oC per day, before being maintained at 20oC. Both strains experienced an identical program of vaccinations, administered from day old to 13 weeks. At the time of sampling, average bird mass was 1203g for the H&N flock and 1239g for the Hy-Line flock. Both flocks had a cumulative mortality of 1.9%. Sampling of pullets occurred during a single day, when both strains of bird were 14 weeks and 3 days old. Twelve birds of average size and body condition were selected from each rearing barn by a senior Production Manager (total n=24) and manually palpated by EAA to determine if keel bone fractures were present. Individuals exhibiting damage were avoided. Animals were placed in carry boxes and transported to Newcastle University, where they were housed in two pens with ad libitum feed and water overnight (one HyLine, one H&N), prior to tissue collection over the following two days (12 birds/day).
3.5 Tissue Collection & Processing
Collection of tissue from the adult hens and pullets occurred in two separate phases (in October 2018 and January 2019, respectively). Animals were weighed before being killed with an intravenous injection of pentobarbital (Euthatal, 0.5 ml/hen), according to a schedule that alternated between housing system and body condition for the adult hens (n=48) and strain for the pullets (n=24). Blood samples were collected for analysis of DNA methylation (results to be reported elsewhere). The spleen was removed and weighed before a sample was placed on ice in a tube containing 1ml RNAlater® Stabilization Solution (Thermo Fisher Scientific, Loughborough, UK). The caecum was dissected from the gut and placed into a 15ml Falcon tube before freezing on dry ice. Simultaneously, the brain was removed from the skull, placed into 0.1 M phosphate-buffered saline (PBS) in a Petri dish and divided along the longitudinal fissure with a scalpel. The forebrain hemisphere collected for immunohistochemical analysis alternated between left and right, in a manner that was balanced within groups of hens of each body condition and from each housing system. This tissue was immersion fixed for 44-48 h in 4% paraformaldehyde in 0.5 M PBS at 4°C. Samples were then cryoprotected in a solution of 30% sucrose in 0.5 M PBS, before being embedded in OCT (4583, Electron Microscopy Sciences - USA). Coronal sections (50 μm) were cut on a cryostat (HM 550, Microm – Germany) and stored in cryoprotectant solution (30% glycerol, 30% ethylene glycol, 0.1M PBS) at -20oC. Serial sections taken at 400μm intervals were processed for immunohistochemistry.
3.6 Immunohistochemistry & Quantification of AHN
As previously41,42, hippocampal formation (HF) tissue sections were stained using an antibody to doublecortin (DCX), to allow quantification of currently differentiating immature neurons generated through AHN. Free-floating sections from the adult hens were stained over six batches, each of which contained tissue from eight birds (2x multi-tier: good condition, 2x multi-tier: poor condition, 2x enriched cage: good condition, 2x enriched cage: poor condition). Sections from the pullets were stained over three batches, each of which contained tissue from eight birds, wherein four were of the H&N strain and four were Hy-Line. Staining was conducted according to the protocol detailed in Armstrong et al.42. The primary antibody was rabbit polyclonal to doublecortin (Abcam Cat# ab18723, RRID:AB_732011), incubated at a concentration of 1:1000 for 18 hours (4oC). Secondary antibody incubation utilised biotinylated antirabbit IgG secondary antibody made in goat (Vector Labs, BA-1000), at concentration 1:500 for 120 minutes (room temperature). 1:250 Horse Radish Streptavidin (Vector Labs, SA-5004) was used for conjugate enzyme incubation, and 3,3’-Diaminobenzidine (DAB) SIGMAFAST tablets were dissolved in pure water (final concentration 0.35mg/ml) for chromogen incubation.
As previously41,42, stained DCX+ cells were quantified in the rostral (interaural 5.68/0.50) and caudal (interaural 0.50/-0.50)65 HF. An optical microscope (Leica DM6B-Z, Germany) equipped with a digital video camera (Leica DFC450 C, Germany) and motorized stage system (Leica AHM, Germany) was connected to a computer running Stereo Investigator software (version 2018.1.1, MBF Bioscience, USA). HF borders were outlined at 2.5X magnification (0.07 numerical aperture) according to the chick stereotaxic atlas65, and cell counting performed at 100X magnification (0.65 numerical aperture) according to the Optical Fractionator method. Stereological parameters were set to an optical fractionator grid of 120 x 120 µm for rostral HF and 240 x 240 µm for caudal HF, a counting frame of 50x50 µm for both regions and a mounted thickness of 20 µm. For each animal, 4 to 6 hippocampal sections 800 μm apart were systematically analysed, starting with the rostral-most section bearing hippocampal tissue. Quantification was performed blind to housing and body condition groups. DCX+ cells of multipolar and bipolar (or fusiform) morphologies were counted separately (see Armstrong et al.42). The bipolar/fusiform neurons (medium-small sized cells, elliptical/oval cell bodies, ≤2 processes) are assumed to be younger and still migrating, while the multipolar neurons (medium-large sized cells, round or polygonal/angular cell bodies, ≥3 processes) are assumed to be more mature and settling66. Densities of DCX+ cells per cubic millimetre of sampled tissue were calculated by dividing the number of counted cells of each type by the area of the counting frame (2500µm), multiplying by both the number of counting sites sampled in that brain and the section thickness (50µm), and multiplying by 109. Densities for rostral and caudal HF were calculated separately.
3.7 Inflammatory Gene Expression
The methods used to extract RNA from the sampled spleen tissue, reverse-transcribe it into cDNA, and produce gene-specific templates for standard curves are detailed in Armstrong et al.67. Primer sequences for the quantified target genes involved in the inflammatory response (IL1β, IL6, IL8, IL10 & TGFβ) are displayed in Table 1. Quantitative PCR (qPCR) assays were conducted as previously67, with LBR utilised as a control gene for normalisation. The 48 samples from the adult hens were processed in a single batch. Analyses were performed blind to the group status of the samples. Splenic gene expression was not explored in the sampled pullets. Gene expression values were log(10)- transformed for statistical analysis.
Table 1
Primer sequences employed for quantification of inflammatory cytokine expression in splenic tissue.
Gene
|
Accession
|
Orientation
|
Primer Sequence (5’-3’)
|
Product Length (base pairs)
|
LBR
|
NM_205342
|
Forward Reverse
|
GGTGTGGGTTCCATTTGTCTACA
CTGCAACCGGCCAAGAAA
|
80
|
IL1β
|
NM_204524.1
|
Forward Reverse
|
TGCCTGCAGAAGAAGCCTCG CTCCGCAGCAGTTTGGTCAT
|
137
|
IL6
|
NM_204628 .1
|
Forward Reverse
|
TCGCCTTTCAGACCTACCTG CAGATTGGCGAGGAGGGATT
|
179
|
IL8
|
NM_204608 .1
|
Forward Reverse
|
TGTGAAGAGATCGCTGTGTG AGGCATCGCATTCCAGC
|
85
|
IL10
|
NM_001004414.2
|
Forward Reverse
|
GGGAGCTGAGGGTGAAGTTT TCTGTGTAGAAGCGCAGCAT
|
154
|
TGFβ
|
NM_001318 456.1
|
Forward Reverse
|
TTACTACGTGGGCCGGAATG CCCCCAAAAAGGGAACCATCT
|
193
|
3.8 Composition of the Caecal Microbiome
For each of the 48 adult hens, DNA was extracted from a 200mg section of intestinal tissue and caecal content using ZymoBIOMICS DNA minikits (Cambridge Bioscience, UK) according to the manufacturer’s instructions. The section was cut using a sterile scalpel blade to expose the mucosa and luminal contents to bead-beating with a Qiagen TissueLyser at 30 Hz for 10 minutes. At each extraction, two controls were included, a blank extraction kit to control for contamination and 75 μl of ZymoBIOMICS standard bacterial community (Cambridge Bioscience, UK) to control for variations in DNA extraction efficacy. Extracted DNA was quantified using a NanoDrop 2000 spectrophotometer (NanoDrop Technologies).
Extracted DNA was sent for paired-end sequencing of the 16S rRNA gene at the Centre for Genomic Research (University of Liverpool) using an Illumina MiSeq run. The V4 hypervariable region (515F/R806) was amplified to yield an amplicon of 254 base pairs68. Library preparation was performed using a universal tailed tag design with subsequent amplification performed using a two-step PCR with a HiFi Hot Start polymerase (Kapa)69. The first round of PCR was performed using the primers 5’ ACACTCTTTCCCTACACGACGCTCTTCCGATCTNNNNNGTGCCAGCMGCCGCGGTAA-3’ (forward) and 5’-GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCTGGACTACHVGGGTWTCTAAT-3’ (reverse)69. The raw Fastq files were trimmed for the presence of Illumina adapter sequences using Cutadapt version 1.2.1. The reads were further trimmed using Sickle version 1.200 with a minimum window quality score of 20. Reads shorter than 10 base pairs after trimming were removed.
QIIME2 version 2020.2.0 was used for analysis of the Illumina data70. Amplicon sequence variant (ASV) assignment was completed using the dada2 plugin71 and a feature table produced using the feature-table plugin72. Taxonomy was assigned using a pre-trained NaiveBayes classifier based on the SILVA 132 database of the 515F/R806 region of the 16S rRNA gene73, available for download at https://docs.qiime2.org/2018.11/data-resources/, using the q2-feature-classifier plugin74.
3.9 Data Analysis
Analysis of body and spleen mass, AHN and splenic mRNA expression was performed in IBM SPSS Statistics (v.25). Data was confirmed to meet the assumptions of the statistical approaches employed (e.g. by assessing normality of residuals). In the sample of pullets, body masses of the two strains were compared through an independent samples t-test. For spleen mass, univariate ANOVAs were conducted with body mass as a covariate and strain as a between-subject fixed factor. In the adult hens, a univariate ANOVA was used to compare body mass, with housing system and body condition as fixed factors and together in an interaction term. Spleen mass was explored in a similar model, but which also included body mass as a covariate. To explore AHN in the pullets, separate linear mixed models (LMMs) with unstructured covariance were conducted for raw DCX+ multipolar and bipolar cell densities. These included staining batch as a random factor, HF subregion (rostral/caudal) as a repeated fixed factor, and strain (H&N/Hy-Line) as between-subject fixed factor. The interaction between subregion and strain was also included. To analyse cell counts for the adult hens, separate LMMs were again conducted with DCX+ multipolar and bipolar cell densities as the dependent variables. Models included staining batch as a random factor, HF subregion as a repeated fixed factor and housing system (multi-tier free range/enriched cage) and body condition (good/poor) as between-subject fixed factors. All interactions between HF subregion, housing system and body condition were included. For the purpose of figures, cell densities were normalised (z-scored) within their staining batch. For each inflammatory gene quantified in the spleen, measured molar mRNA was log-transformed and converted to a ratio of LBR mRNA expression in the same sample. A series of univariate ANOVAs were conducted with target transcript expression ratios (IL10, IL1B, IL6, IL8 or TGFB / LBR) as dependent variables, with housing system and body condition as between-subject fixed factors and in an interaction term.
Alpha and beta diversity analyses of extracted caecal DNA were performed at a sampling depth of 13,000 using the alignment75, phylogeny76 and diversity (https://github.com/qiime2/q2-diversity) plugins. Alpha diversity was measured using Faith’s phylogenetic diversity (FPD) index77 to assess species richness and a Shannon diversity (SD) index to assess species evenness. Alpha diversity was compared between sample groups using a Kruskal Wallis test with a false discovery rate (FDR) correction. Taxa plots were produced using the q2-taxa plugin (https://github.com/qiime2/q2-taxa). Beta diversity, a metric used to compare species diversity and abundance between samples, was calculated with a robust Aitchison PCA metric78 using the DEICODE plugin (https://library.qiime2.org/plugins/deicode). The beta diversity matrix was used to draw principal coordinate analysis (PCoA) plots and an ANOSIM test over 999 permutations were used to determine the significance of differences in beta diversity between groups. Songbird was chosen to analyse differential abundance of ASVs between groups (--p-formula “Body_Condition + Housing_System”) since it overcomes challenges created by the compositional nature of microbiota data79. Results from both DEICODE and Songbird were visualised through Qurro to identify the taxonomy of ASVs (features) contributing to differences between housing system and body condition groups80.