Behavioural studies
Double-choice tests
Body weight at 3 weeks of age was used to select FG and SG chickens from a commercial flock of Ross 308 birds (Darwalla Group, Qld, Australia). The weight of 100, randomly selected, birds was used to identify those birds weighing <800g as SG and those weighing >1,000g as FG, with each category representing approximately 10% of the flock. Birds were randomly selected from the flock until there were 36 in each of the FG and SG categories. The selected birds were transferred to individual cages (45×35×35cm) equipped with one nipple drinker and two identical feeders (16×10×7cm) placed side-by-side. The 3-week-old chickens were kept for another 3 weeks for the experimental phase. They were offered a commercial standard grower feed for the first week of adaptation period, and a commercial standard finisher feed for the following two weeks as part of the double-choice (DC) testing period. The ingredient and nutrient composition of the two standard commercial feeds (Darwalla Group, Mt Cotton, Queensland, Australia) feeds can be found in Table S1 in Additional file 1.
Table 4
Description of the treatments. Content of essential (Met, Lys, Thr) and non-essential (Ala, Asp and Asn) amino acids in the controla and supplemented feedsb offered in double-choice (DC) tests (T1, T2 or T3)c performed in slow (SG)- and fast-growing (FG) chickens.
Amino Acid
|
Type
|
T1: Control feed (%)a
|
T2: EAA Supplemented feed (%)c
|
T3: NEAA Supplemented feed (%)c
|
Methionine (Met)
|
EAA
|
0.428
|
0.740
|
-
|
Lysine (Lys)
|
EAA
|
1.626
|
2.813
|
-
|
Threonine (Thr)
|
EAA
|
1.161
|
2.009
|
-
|
Alanine (Ala)
|
NEAA
|
1.427
|
-
|
2.468
|
Aspartic acid (Asp)
|
NEAA
|
2.735
|
-
|
4.732
|
Asparagine (Asn)
|
NEAA
|
1.368
|
-
|
2.366
|
Acronyms: EAA=essential amino acids; NEAA=non-essential amino acids
aThe control feed was a standard commercial broiler feed (Darwalla Group, Esk, QLD, Australia). See Table S1 in Additional file 1.
bThe EAA or NEAA supplemented feeds consisted of the control diet supplemented with a mix of Met, Lys and Thr (EAA) or Ala, Asp and Asn (NEAA), respectively. All amino acids were supplemented to reach 73% excess of the control diet, the maximum level with no significant impact on feed intake (Baker and Han, 1994; Mack et al., 1999).
cThe double-choice (DC) tests T1, T2 and T3 consisted of a control DC offering two identical control feeds (T1) or a DC between the control feed and either the EAA (T2), or the NEAA (T3) supplemented feeds.
|
The DC tests were conducted to study feed preference based on EAA or NEAA supplementation by offering ad libitum access to the two feeders one containing the control finisher feed and the other the experimental (supplemented) feed. The preparation of the experimental feeds consisted in spraying the base commercial finisher feed with water (control), or water containing a mix of three EAA (methionine, Met; lysine, Lys; and threonine, Thr) or three NEAA (alanine, Ala; aspartic acid, Asp; asparagine, Asn). The content of amino acids of interest in the control and the EAA or NEAA supplemented diets is presented in Table 4. The added EAA or NEAA were 73% above the amounts present in the control feed because this level of excess was found as the maximum inclusion of EAA not having a significant (P>0.05) effect on feed intake or growth compared to optimum levels in broiler chickens [17, 18]. The amount of each feed consumed was measured every 24 hours. After every measurement, feeders’ position was switched to avoid errors due to side preference. Preference (%) for a feed was determined as a percentage of the experimental feed consumed divided by the total amount of both feeds consumed.
Omics studies
Animal and tissue sampling
The 48 lightest and 48 heaviest birds at 3 weeks of age from a batch of 580 male (feather-sexed) Ross 308 chickens (Darwalla Group, Qld, Australia) were selected as SG or FG chickens, respectively. Chickens showing any sign of pathology were not included. The birds were transferred to individual cages similar to those used for the DC experiments and offered a commercial diet until 6 weeks old. Individual cages allowed for individual feed intake and body weight recording. The five lightest and five heaviest birds were selected for the metabolic studies and euthanised using cervical dislocation. Cervical dislocation is a standard method of euthanasia commonly used in poultry research experiments. It consists of handling the chicken while pulling the bird’s neck quickly (3-4 seconds) and firmly to dislocate the first cervical vertebrae from the skull resulting in severing the spinal cord and carotid arteries. This method does not involve the use of any chemical or equipment. Every bird was confirmed to be male using post-mortem and DNA sexing (MDS Australia Pty Ltd, Queanbeyan, Australia). A sample (400-500mg) of the proventriculus taken from each bird was placed in 1ml RNA-later solution (ThermoFisher Scientific, Waltham, USA) and stored at -80ºC.
RNA sample preparation and sequencing
RNA was extracted from 30-40mg of tissue using the Maxwell® 16 LEV Simply RNA Purification Kit manufactured by Promega Corporation, Madison, USA. The quality and quantity of the RNA samples were examined using a NanoDropTM spectrophotometer (ThermoFisher Scientific, Waltham, USA) and Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, USA). RNA samples with an Integrity Number ≥ 8, a 260/280 ratio between 2 and 2.1, and a 260/230 ratio between 2 and 2.2 were used for sequencing. An RNA sample of 500-4000ng from each SG and FG bird was sequenced using high throughput sequencing (Queensland Brain Institute, University of Queensland, Australia). A TruSeq Stranded mRNA Library Prep Kit was used for library construction. Sequencing and coverage were based on 2×125bp Paired-End Dual indexed reads and v4 illumina SBS chemistry (Illumina, San Diego, USA), with HiSeq 2000 used to produce an average of ~27 million sequence reads per sample.
Protein Sample preparation
Approximately 20mg of tissue was used for protein extraction and Mass Spectrometry. The tissue was placed in ~300µl of lysis buffer (6 M Guanidine Chloride, 50mM Tris pH 8, and 10 mM DTT) and sonicated for 10 seconds before being vortexed at 30°C for 1 hour. Acrylamide (25mM) was added and incubated for 1h before adding 5mM DTT. Protein was precipitated by adding 4 volumes of 1:1 methanol acetone and stored overnight at -20°C before centrifuged to remove the solvent, and the pellet suspended in 0.1% SDS (Sodium dodecyl sulfate). The protein concentration of each sample was measured using NanoDrop (ThermoFisher Scientific, Waltham, USA). Samples were digested with trypsin overnight at 37°C using an Amicon column and ammonium bicarbonate (50mM). Samples were desalted by C-18 Zip-tip (adapted from Millipore procedure) and analyzed by liquid chromatography electrospray ionization tandem mass spectrometry (LC-ESI-MS/MS).
Mass Spectrometry and data analysis
LC-MS/MS analysis was performed using a Prominence nanoLC system (Shimadzu, Kyoto, Japan) and TripleToF 5600 mass spectrometer with a Nanospray III interface (SCIEX, Toronto, Canada). A 70 min Liquid chromatography (LC) gradient was used to separate the peptides. The database of Protein Pilot (Uniprot, www.uniprot.org) was used to identify peptides and proteins. Sequential window acquisition of all theoretical mass spectra (SWATH) was performed on all samples. An information-dependent acquisition library (IDA) was prepared from one randomly chosen slow grower sample and one randomly chosen fast grower sample for each tissue.
Transcriptomic and genomic analysis
For transcriptomic analysis, sequenced reads were first put through a quality check (QC) using FastQC version 0.72 (Babraham Bioinformatics, Babraham Institute, Cambridge, UK) in Galaxy Australia platform with default parameters [19]. Then, the reads were mapped to Galgal.5 reference assembly using aligners HISAT2 and Salmon with default settings (20, 21). InteractiVenn, JMP, IGV, and MeV were used to visualize and perform cluster and Principal Component Analyses (PCA) [22, 23].
As for genomics, Galaxy Australia was used to perform the analyses [19]. RNA sequences were mapped on the reference genome (Galgal.5) using Bowtie2, version 2.3.4.1 [24]. Genomic variants between the 5 SG and 5 FG chickens were identified using FreeBayes, which is a Bayesian genetic variant detector designed to find small polymorphisms, specifically SNPs (single-nucleotide polymorphisms), indels (insertions and deletions), MNPs (multi-nucleotide polymorphisms), and complex events (composite insertion and substitution events). The list of candidate genes used for the study of genomic variants is presented in Table S2 in Additional file 1. The FreeBayes output was first filtered for Phred score ≥30 to attain accuracy of 99.9%. Phred score is an index of variant quality score indicating the confidence level in detecting a particular variant accurately [25]. Next, the FreeBayes output was filtered for SNPs considered of interest in relations to metabolic functions of chickens growing at different rates. These SNPs were further filtered for a SIFT (Sorting Intolerant From Tolerant) score of <0.05, which predicts the extent to which the polymorphism effects gene function, with 0.0 being deleterious and 1.0 being tolerable in Variant Effect Predictor (VEP) analysis.
Functional enrichment analysis
Differentially expressed genes (DEG) and differentially abundant proteins (DAP) between the FG and SG groups were used as input for metabolic pathway enrichment analyses. Enrichment analyses were performed in DAVID 6.8 [26]. Enriched metabolic pathways and terms in different databases including Gene Ontology, GO [27], Kyoto Encyclopedia of Genes and Genomes, KEGG [28], and REACTOME [29] were used to attain insight into the function of DEGs and DAPs.
Statistical analyses
Preference tests
A 2×3 factorial design was used including two growth rates for SG or FG chickens and three preference comparisons of feed vs feed plus amino acids (Table 4). Preference for each feed in a DC test was compared to 50%, which is the neutral value, using t-student test. Main effects and interactions were analysed using the general linear model (GLM) procedure of SAS9.4 (SAS Institute, Cary, North Carolina, United States), with P < 0.05 as the level of significance in Tukey test. The statistical model to compare the preference and individual feed intakes was:
yijk = μ + Ai + Bj +(AB)ij + εijk i = 1-2; j = 1-3; k = 1-6
yijk = observation k in level i of group of chicken and level j of treatment; μ = the overall mean; Ai = the effect of level i of group of chicken (slow vs fast grower); Bj = the effect of level j of treatment; (AB)ij = the effect of the interaction of level i of chicken group with level j of treatment; εijk = random error; i = Number of levels of chicken group; j = Levels of treatment; k = biological replicates
Total feed intake was used as a covariate for comparisons between treatments and controls and for amino acids consumed. The covariate allowed comparison between SG and FG chickens while adjusting the treatment effect for the variability in the feed intake.
Randomized design with a covariate was also run in GLM procedure of SAS9.4 according to the statistical model below.
yijk = β0 + β1xijk + τi + gj + εijk i = 1-3; j = 1-2 ; k = 1-6
yijk = observation k in treatment i group j; β0 = the intercept; β1 = the regression coefficient; xijk = a continuous independent variable of total feed intake (covariate); τi = the fixed effect of treatment (T1-3); gj = the fix effect of group (slow vs fast grower); εijk = random error
Metabolic studies
For transcriptomics, read count and differential expression analysis were performed using Limma [30]. Limma provides a linear model capable of multiple RNA comparisons in the context of multifactor design experiments to identify differentially expressed genes (DEG). The basic statistics used for DEG identification was t-statistics moderated across genes using Baysian model. This will increase the reliability of the results. Significance threshold was set as a P-value of 0.05 or less.
As for proteomics, sequential window acquisition of all theoretical mass spectra (SWATH)-MS relative quantitative data were analysed using PeakView v2.1 (SCIEX) [31]. To identify differentially abundant proteins (DAP), statistical analyses were performed using MSstats in R as previously described [32, 33]. Values less than 0.05, adjusted for multiple testing, were considered as significant P-values.
Pathway enrichment analyses were performed on both DEG and DAP lists. Pathways that were statistically (P < 0.05) over-represented in the DEG and DEP lists were defined as significantly enriched. This was done through Fuzzy clustering algorithms and Fisher’s exact test built in DAVID bioinformatics platform [26].
Regarding genomic variants, allelic frequencies were compared between SG and FG chickens using Fisher’s exact test in SAS 9.2 (SAS Institute, Cary, North Carolina, United States). P-values of 0.05 or less were considered as significant for all allelic frequency comparisons.