A Genome-Wide mRNA Screen and Functional Analysis Reveal FOXO3 as a Candidate Gene for Chicken Growth

Chicken growth performance provides direct economic benefits to the poultry industry. However, the underlying genetic mechanisms are unclear. The objective of this study was to identify candidate genes associated with chicken growth and investigate their potential mechanisms. We used RNA-Seq to study the breast muscle transcriptome in high and low tails of Recessive White Rock (WRRh, WRRl) and Xinghua chickens (XHh, XHl). A total of 60, 23, 153 and 359 differentially expressed genes were detected in WRRh vs. WRRl, XHh vs. XHl, WRRh vs. XHh and WRRl vs. XHl, respectively. GO, KEGG pathway and gene network analyses showed that CEBPB, FBXO32, FOXO3 and MYOD1 played key roles in growth. The functions of FBXO32 and FOXO3 were validated. FBXO32 was predominantly expressed in leg muscle, heart and breast muscle. After decreased FBXO32 expression, growth-related genes such as PDK4, IGF2R and IGF2BP3 were significantly down-regulated (P < 0.05). FBXO32 was significantly (P < 0.05) associated with carcass and meat quality traits, but not growth traits. FOXO3 was predominantly expressed in breast and leg muscle. In both of these tissues, the FOXO3 mRNA level in XH was significantly higher than that in WRR chickens with normal body weight (P < 0.05). In DF-1 cells, siRNA knockdown of FOXO3 significantly (P < 0.01) inhibited the MYOD expression and significantly up-regulated (P < 0.01 or P < 0.05) the expression of growth-related genes including CEBPB, FBXO32, GH, GHR, IGF1R, IGF2R, IGF2BP1, IGF2BP3, INSR, PDK1 and PDK4. Moreover, 18 SNPs were identified in FOXO3. G66716193A was significantly (P < 0.05) associated with growth traits. The sites C66716002T, C66716195T and A66716179G were significantly (P < 0.05) associated with growth or carcass traits. These results demonstrated that FOXO3 is a candidate gene influencing chicken growth. Our observations provide new clues to understand the molecular basis of chicken growth.


Introduction
Chicken growth, an important economic trait, is determined by genetic, nutritional and environmental factors. Heritability estimates showed that chicken growth could be enhanced by genetic improvement [1,2]. This trait is controlled by multiple genes. At present, many studies have been performed to find genetic factors affecting growth. Candidate genes and quantitative trait loci (QTLs) such as GH, IGFBP2 and GHSR have been identified [3,4]. Recently, genomewide associate study (GWAS) analysis found that two FOXO1A single-nucleotide polymorphisms (SNPs) were strongly associated with chicken growth [5]. However, the genetic mechanisms of chicken growth are unclear, and a more systematic picture of the genes responsible for this trait is needed. Recently, next generation sequencing provided an important opportunity for the genome-wide characterization of genes and pathways involved in growth [6][7][8].
In this study, two chicken breeds, Recessive White Rock (WRR) and Xinhua (XH), were used for RNA-Seq. WRR is a typical fast-growing breed that is known for its large body size and thick myofibers. XH is a Chinese native slow-growing breed which is characterized by small body size. The different growth speeds of these two breeds led to distinct growth performance at 7 weeks of age. Both breeds were used for studies on chicken growth and fat deposition traits [1,9]. With the use of a population derived from reciprocal crosses between these two breeds, a quantitative trait loci (QTL) on chromosome 1 was identified to be related to chicken growth traits by Genome-wide association study [5]. Previous study by Methylated DNA immunoprecipitation-sequencing showed that growth-related genes exhibited altered DNA methylation between WRR and XH [10]. Therefore, in the present study, we intended to use two-tail samples of these two breeds at 7 weeks of age to study the gene expression differences between fast-and slow-growing broilers in a genome-wide level and then to identify candidate genes for chicken growth. In our study, we performed RNA-Seq and differentially expressed genes (DEGs) were randomly selected to conduct qPCR experiments to validate the RNA-Seq results. Gene ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway and gene network analyses were performed on the DEGs. Subsequently, the chicken FOXO3 and FBXO32 genes were selected for in vivo and in vitro studies to investigate their potential mechanisms functioned on growth.

Assemble and blast analysis of reads from RNA-Seq
From RNA-Seq, we obtained 44139971, 36937542, 39046772 and 69669990 Illumina reads for WRR h , WRR l , XH h and XH l , respectively, giving rise to total residues of 4286130490, 3586054428, 3758078838 and 6698804887 bp, respectively (Table 1). All sequencing data have been submitted to NCBI Gene Expression Omnibus (GEO) database with the accession number GSE72424 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE72424). More than 70.5% of the total reads were mapped to the chicken genome. In total, 13828 genes were detected in the four samples, including 12848 in WRR h , 12818 in WRR l , 12419 in XH h and 12915 in XH l (Fig 1A). Of these genes, 11706 genes were identified in all four samples, while 201, 209, 125 and 249 genes were found exclusively in WRR h , WRR l , XH h and XH l , respectively ( Fig 1A). The sequence length distribution showed that 83% of genes identified in the four samples had a length less than 4000 bp, while no more than 0.5% of genes were longer than 10000 bp (S1 Fig).

DEGs among the four groups
A comparison of gene expression among the four samples showed that there were 60 DEGs (fold changes 2; q value < 0.05) between WRR h and WRR l (WRR h vs. WRR l ), 23 DEGs between XH h and XH l (XH h vs. XH l ), 153 DEGs between WRR h and XH h (WRR h vs. XH h ), and 359 DEGs between WRR l and XH l (WRR l vs. XH l ) ( Fig 1B and S1 Table). LAC_CHICK (Ig lambda chain V-1 region) was found in all four comparisons. LAC_CHICK and TTR (transthyretin) were commonly identified in WRR h vs. WRR l and XH h vs. XH l (S2 Table). Moreover, 84 genes, including some crucial to chicken growth such as FBXO32, FOXO3, MYOD1, PDK4, PNPLA2 and SMYD1, were commonly identified in WRR h vs. XH h and WRR l vs. XH l (S3 Table). Among all these DEGs, 6, 10, 11 and 12 genes were uniquely expressed in one of the two samples in each comparison of WRR h vs. WRR l , XH h vs. XH l , WRR h vs. XH h and WRR l vs. XH l , respectively (S4 Table). DEG directionality analysis showed that the number of up-regulated genes was higher than the number of down-regulated genes in both WRR h vs. WRR l and XH h vs. XH l , while there were a greater number of down-regulated genes than of up-regulated genes in both WRR h vs. XH h and WRR l vs. XH l (S2 Fig).

qPCR validation of DEGs obtained from RNA-Seq
To confirm the DEG results obtained from RNA-Seq, four genes (RPL29, PDK4, FOXO3 and LAPTM5) were randomly selected to carry out qPCR using the same RNA samples used for RNA-Seq. The qPCR results showed general agreement with the RNA-Seq results in terms of the direction of expression in each comparison (Table 2).

GO and KEGG pathway analysis for DEGs
DEGs were then used for GO analysis to uncover enriched (P < 0.05) biological processes terms in each comparison. A total of 142 biological process terms, including 21 in WRR h vs. WRR l , 6 in XH h vs. XH l , 47 in WRR h vs. XH h and 119 in WRR l vs. XH l , were identified in our study (S5 Table). Of these, 99 were unique terms that appeared only once in all of the four comparisons. Many of the repeated biological process terms were focused on developmental process, regulation of biological process, cell differentiation and cell adhesion. A total of 29 DEGs involved in cell differentiation and proliferation were observed in the four comparisons, including well-known genes affecting chicken growth such as CEBPB, MYH11, MYOD1, NOTCH2 and TGFBR2 (Table 3). Moreover, in comparing XH h vs. XH l and WRR h vs. XH h , the following processes related to muscle development were found: skeletal muscle development, muscle organ development, muscle cell differentiation and muscle tissue development (S5 Table). Six DEGs were included in those processes: ACTC1, FOXP2, LGALS1, MYOD1, XIRP1 and ZFAND5. These genes might be crucial to muscle development. DEGs we identified were significant enriched (P < 0.05 and Benjiamini adjusted P < 0.1) in eight KEGG pathways, with the most influenced pathway being lysosome (Table 4). Furthermore, cell junction-related pathways such as focal adhesion, extracellular matrix (ECM)-receptor interaction and cytokine-cytokine receptor interaction were included. There were 45 DEGs in these four pathways, including TGFBR2 and ITGAV (Table 5).  . The results showed that the top gene network was cellular movement (Fig 2A). Among the DEGs, FABP4, LGALS3, LYN and SPI1 were central of this network. Network analysis for all DEGs in the cross-breed comparisons (WRR h vs. XH h and WRR l vs. XH l ) revealed that the top gene network was skeletal and muscular system development and function ( Fig 2B), with MYOD1 and FBXO32 as node genes. The results of the GO, KEGG pathway and gene network analyses indicated that CEBPB, FBXO32 and MYOD1 might be the key genes related to chicken growth at 7 weeks of age. Further investigation of gene networks involving these three genes showed that FOXO3 might be the crucial gene, interacting with CEBPB, FBXO32, and MYOD1 and to affect growth ( Fig 3A  and 3B, which show networks compared for WRR h vs. XH h and WRR l vs. XH l , respectively). In WRR h vs. XH h and WRR l vs. XH l , the CEBPB, FBXO32, FOXO3 and MYOD1 mRNA levels were all higher in slow-growing chickens than in fast-growing chickens (Table 6). Furthermore, FBXO32 and FOXO3 showed relatively high fold changes in both of these two comparisons. Therefore, FBXO32 and FOXO3 were selected for functional analysis in this study.    The chicken FOXO3 showed more than 85.4% identity with its human, mouse, rat, elephant, whale, pig, dog, sheep and cattle, counterparts. However, much lower homology with cFOXO3 was found for fish and xenopus, which were 47.4% and 77.1%, respectively.

FOXO3 and FBXO32 gene mRNA expression in different tissues and between different breeds
The mRNA expression analysis in various tissues showed that the chicken FOXO3 gene was predominantly expressed in breast muscle and leg muscle tissues, followed by heart, lung and pituitary ( Fig 4A). FOXO3 expression was compared between WRR and XH chickens with normal body weight (BW) at seven weeks of age. In breast muscle, its mRNA level was significantly higher in XH chickens (P < 0.01) than in WRR chickens ( Fig 4B). In leg muscle, expression in XH was significantly higher (P < 0.05) than in WRR ( Fig 4B). The chicken FBXO32 gene was predominantly expressed in leg, heart and breast muscle ( Fig 5A). The mRNA level of FBXO32 was significantly higher in breast and heart muscle than in XH (P < 0.05) but lower in leg muscle (P < 0.05) ( Fig 5B).
The number indicated the position on the chromosome 3.

Discussion
During postnatal growth, skeletal muscle increase occurs mainly due to muscle hypertrophy accompanied by satellite cell proliferation to incorporate new myonuclei into existing myofibers [11]. In this study, the comparison with the highest number of DEGs resulted in the most GO biological process terms, and terms of development process, cell differentiation and cell adhesion were present in multiple comparisons. Lysosome was the most significantly enriched pathway. Lysosomes are a membrane-bound cell organelles that contain acid hydrolase enzymes and are interlinked with endocytosis, phagocytosis and autophagy. Previous research showed that the autophagy-lysosome system was a key player in regulating protein degradation in skeletal muscle [12,13]. Moreover, other enriched pathways were cell junction-related pathways (focal adhesion, ECM-receptor interaction and cytokine-cytokine receptor interaction), suggesting that pathways critical to maintaining the integrity of tissues might be involved in chicken growth at 7 weeks of age. Focal adhesion complexes serve as mechanical linkages to the ECM and are the signaling centers of numerous intracellular pathways related to cell motility, proliferation and differentiation [14]. ECM components play integral roles in the formation of the muscle satellite cell niche, and their specific interactions with satellite cells can regulate cell adhesion, migration, differentiation, proliferation and self-renewal [15]. Cytokines such as IL-6 are key regulators of cell growth, proliferation, differentiation and apoptosis [16]. Some cytokines produced by myofibers and peritendinous tissue are, termed myokines [17]. Myokines activate of intracellular signaling cascades by binding to their specific receptors, thereby regulating metabolism in skeletal muscle [18]. Importantly, we identified, three crucial transcription factors, CEBPB, FBXO32 and MYOD1, that might be related to chicken growth. Previous research revealed that CEBPB regulated multiple genes in response to GH [19]. Moreover, CEBPB was an activator of adipogenesis and acted as an inhibitor of myogenesis [20,21]. As one well-known myogenic regulatory factor, MYOD1 has been extensively studied, and its effect on growth has been well-demonstrated [22,23]. It can trans-differentiate many cell types to muscle cells and is a key regulator of myogenesis [23][24][25]. Several recently published reports suggested that MYOD1 could modulate and facilitate the assembly of muscle enhancers [26,27]. Furthermore, CEBPB, FBXO32 and MYOD1 can interact with each other [20,28]. In this study, network analysis indicated that FOXO3 was a central gene interacting with CEBPB, FBXO32 and MYOD1.
FOXO3 is a member of the Forkhead box class O (FOXO) transcription factor family. Like other members of this family, FOXO3 was demonstrated to play a crucial role in many species, from lower animals to mammals. It performs a variety of cellular functions, including cell growth and differentiation, cell-cycle control, energy metabolism, DNA damage repair, response to oxidative stress, and apoptosis [29][30][31][32][33]. In mammals, FOXO3 was relatively ubiquitously expressed, consistent with our findings in chicken [34]. However, the mammalian FOXO3 was predominantly expressed in heart, brain, kidney and ovary, whereas the chicken FOXO3 was highly expressed in breast muscle and leg muscle tissues [34]. A previous study suggested that FOXO3 contributes to cell growth in striated muscle [33]. Our expression comparison between fast-growing and slow-growing breeds also confirmed its important function in skeletal muscle. It is possible that the higher expression of CEBPB, FBXO32 and FOXO3 in slow-growing XH chickens contributed to their lower growth performance, whereas the higher level of MYOD1 would partly counteract these effects. Moreover, a series of genes involved in the somatotropic axis, including GH, GHR, IGF1R, IGF2R, IGF2BP1 and IGF2BP3, was effectively up-regulated by FOXO3 knockdown after siRNA interference, suggesting that FOXO3 might play an important role in growth. In particular, some genes remarkably up-regulated by FOXO3 knockdown, including INSR, PDK1 and PDK4, were present upstream of the IGF1/ FOXO signal transduction pathway, which would further promote FOXO3 inhibition [33,34]. Recent studies suggested that MYOD was a direct target of FOXO3 in myoblasts [35]. In vivo and in vitro, FOXO3 was demonstrated to play an important role in activating MYOD transcription [35]. In this study, FOXO3 knockdown resulted in a significant down-regulation of MYOD, whereas expression of the other MRFs (MYF5 and MYF6) was unaffected, consistent with previously published studies [35]. These results indicated that FOXO3 could interact with MYOD but not the others MRFs. Conversely, previous research identified FOXO3 as a major activator of FBXO32 expression, a factor associated with skeletal muscle protein degradation [36,37]. In contrary, we found that FOXO3 knockdown strongly increased FBXO32 expression. Nevertheless, our data revealed that FOXO3 down-regulation induced a dramatic down-regulation of CEBPB, which is an inhibitor of myogenesis [20]. The up/down-regulation of FBXO32 and CEBPB was inconsistent with the reported inhibitory roles of FOXO3 on muscle growth. Therefore, further study is needed to investigate protein expression changes. The human FOXO3 contains a sequence of 95 amino acids forming a forkhead DNA-binding domain motif (148-257 amino acids). It also possesses a nuclear localization sequence (NLS, 249-251 and 269-271 amino acids), a nuclear export sequence (NES, 386-396 amino acids) and a conserved C-terminal transactivation domain [38]. The sequences of the chicken FOXO3 gene showed high identity with the previously cloned mammalian FOXO3. The chicken FOXO3 had the same amino acids as the human FOXO3 in the NLS and NES. It shared high homology with human FOXO3 in the forkhead DNA-binding domain, with a six amino acid difference. FBXO32, also known as Atrogin 1 or MAFbx, is a skeletal and cardiac muscle-specific Fbox containing protein that was shown to be associated with the maintenance of muscle mass [22,25]. FBXO32 is a muscle-specific gene that has a key function in muscle atrophy [39]. Previous research on feed deprivation showed that as the muscle degraded, FBXO32 expression increased [40]. Although we did not find any association between SNPs of FBXO32 and chicken growth traits, our findings showed that it was highly expressed in chicken leg muscle, heart and breast muscle tissues, which is in accordance with its role. FBXO32 was regulated by multiple transcription factors. A previous study suggested that FOXO3 can act on the promoter of FBXO32, mediating FBXO32 transcription and muscle cell atrophy [41]. After FBXO32 expression significantly decreased, growth related genes including PDK4, IGF2R and IGFBP3 were significantly down-regulated. Previous published reports showed that FBXO32 and PDK4 may interact in skeletal muscle metabolism [42,43]. The expression levels of GH and GHR had tendency to increase, indicating that FBXO32 may have influence on muscle growth at the mRNA level.
Thus far, many SNPs and QTLs were reported to correlate with growth [3,4]. In humans, one FOXO3 promoter SNP was associated with human body mass index [44]. In pig, FOXO3 SNPs were closely associated with growth and development traits [45]. To date, this study is the first to scan variations in the chicken FOXO3 gene and then evaluate their effects on chicken growth. In humans, FOXO3 was phosphorylated and acetylated on multiple sites, including K242, K245, K259, K271, K290, K569, S207, S253, S295, S315, S318, S321, S325, S345, S399, S413, S426, S555, S588, S626, S644, T32 and T179 [33,38,46]. In this study, we scanned the entire exon 2 of chicken FOXO3, and a total of 11 SNPs were identified in the 1,432 bp exon 2 fragment. These variation sites corresponded to R222, P238, A251, S286, N316, D339, P361, T404, F459, T475 and N530 in human FOXO3. Although these variations were not present in phosphorylation or acetylation sites, C66716002T (corresponding to A251 in human FOXO3), was located in the forkhead DNA-binding domain, which is crucial in regulating the transcription of target genes. This SNP did not cause an amino acid change, but might affect mRNA splicing, stability, or protein folding, and thereby alter protein functions such as DNA-binding [47]. Moreover, it is located in the NLS region. Our association analysis demonstrated that this mutation had a significant effect on growth traits. Among the mutations identified in exon 2 of chicken FOXO3, G66715314C was a non-synonymous variation resulting in a change in the amino acid sequence. It was previously shown that non-synonymous SNPs affect protein functions and protein-protein interactions [48,49]. Therefore, the G66715314C mutation may have important effects on economic traits, and its association with chicken growth traits needs further study. Conversely, four SNPs found in this study were located near the exon-intron boundaries, which are special regions that usually had important functional roles in protein [50]. Three of these four SNPs were associated with growth traits. Importantly, the G66716193A site, which was 60 bp away from the intron-exon boundary, was significantly associated with most growth traits.
In summary, this study provided a comprehensive transcriptome analysis of breast muscle between fast-and slow-growing chickens. CEBPB, FBXO32, FOXO3 and MYOD1 may play key roles in chicken growth at seven weeks of age. Further expression analysis, siRNA analysis and association analysis demonstrated that chicken FOXO3 is a candidate gene involvement in chicken growth. Our observations provide new clues to understand the molecular basis of chicken growth.

Ethics statement
All animal experiments were performed in accordance with and were approved by the Animal Care Committee of South China Agricultural University (Guangzhou, China) (approval ID: SCAU#0011). All efforts were made to minimize animal suffering.

Animals and tissues
Two chicken breeds WRR and XH were used for RNA-Seq in this study. WRR is a breed with a fast growth rate, while XH is a Chinese native breed with a slow growth rate. WRR and XH were provided by Guangdong Wens Foodstuff Company Ltd, Guangdong, China and Zhicheng Avian Breeding Company Ltd, Guangdong, China, respectively. Briefly, birds were fed ad libitum with 16.5% CP and 2800 kcal of ME/kg, with free access to water. BW was measured at seven weeks of age and three female chickens from each of the two-tail samples of WRR and XH were selected based on it. The four groups WRR with high weight (WRR h ), WRR with low weight (WRR l ), XH with high weight (XH h ) and XH with low weight (XH l ) were generated. Their average BW7 values were 1064.0 ± 11.1, 695.0 ± 24.4, 305.8 ± 23.3 and 207.6 ± 11.1 g, respectively. At 7 weeks of age, chickens were killed by stunning followed by exsanguination. Breast muscles were collected and stored at -80°C until RNA extraction.
Five XH chickens at 7 weeks of age were used for expression analysis of FOXO3 and FBXO32 in different tissues. A total of 19 tissues, cerebrum, cerebellum, pituitary, hypothalamus, heart, liver, spleen, lung, kidney, breast muscle, leg muscle, muscular stomach, glandular stomach, ovary, duodenum, subcutaneous fat, abdominal fat, thymus and bursa of Fabricius, were collected from each chicken.
Four XH and four WRR chickens with normal BW were used for the expression analysis between different breeds by qPCR. Breast muscle and leg muscle tissues were dissected from those 8 birds. All those birds were obtained from the Chicken Breeding Farm of South China Agricultural University, Guangdong, China.
An F2 full sibling hybrid population, from a WRR and XH cross as described previously, was used for association analysis between SNPs and chicken growth or carcass traits [51]. Genomic DNA was extracted from EDTA-anticoagulated blood and then used for genotyping.

RNA extraction and RNA-Seq
Total RNA from various tissues and DF-1 cells was isolated by TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and then treated with DNase (Promega, Madison, WI, USA) following the manufacturer's instructions. RNA quality and concentration were evaluated by an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). Subsequently, three breast muscle RNAs of the same group were mixed in equal amounts. In this way, four pooled samples (WRR h , WRR l , XH h and XH l ) were generated and then were sent to Shanghai Majorbio Bio-pharm Biotechnology Co., Ltd. (Shanghai, China) for RNA-Seq. cDNA libraries were constructed according to Illumina's protocols and then each library was sequenced on a single line of Illumina Hiseq 2000 (Illumina, San Diego, CA, USA) to obtain paired-end 101-bp reads.

Bioinformatic analysis of RNA-Seq
For raw data from RNA-Seq, we first removed reads containing adaptors, unknown bases and low quality bases to obtain high quality reads. Then, the clean reads of the four samples were aligned to the chicken reference genome (http://asia.ensembl.org/info/data/ftp/index.html) by TopHat software, with no more than 2 bp mismatches. The mapped reads were used for further transcriptome annotation and expression calculation with using FPKM (Fragments Per Kilobase of transcript per Million mapped reads). All expressed genes were subjected to GO analysis by AmiGO2 (http://amigo.geneontology.org/amigo), with Bonferroni-adjusted P value < 0.05. With the use of Cuffdiff (http://cufflinks.cbcb.umd.edu/), genes with greater than 2-fold changes between two samples and a q value < 0.05 were regarded as DEG. All those DEGs were subjected to GO analysis and KEGG pathway enrichment analysis with the DAVID Functional Annotation Tool (http://david.abcc.ncifcrf.gov/) using a 0.05 cutoff for the P value and Benjiamini adjusted P < 0.1 cutoff for the Benjamini adjusted p-value. Moreover, all DEGs underwent gene network analysis by IPA.

cDNA synthesis and qPCR
Primers P1 to P5 (S6 Table) were used for the validation of RNA-Seq data by qPCR. For the FOXO3 expression analysis in different tissues and breeds, total RNA was reverse transcribed using a PrimeScript RT reagent Kit with gDNA Eraser (Takara Co., Japan) to synthesize the first-strand cDNAs. qPCR was conducted in a total volume of 20 μL: 10 μL Bestar Real time PCR Master Mix (SYBR Green) (DBI Bioscience, Germany), 0.5 μL of each primer (10 μM), 8.0 μL of RNase-free water and 1 μL of cDNA on a BIO-RAD CFX96 system (Bio-Rad, USA). P4 was used for qPCR of FOXO3 and β-actin (P1) was used as the internal control (S6 Table). All reactions were run in triplicates. The relative expression level was calculated by the 2 -ΔΔCt method. Where ΔΔCt corresponded to the difference between ΔCt measured for the mRNA level of each sample and ΔCt measured for the mRNA level of the reference sample, ΔCt = Ct target gene −Ct reference gene .

Polymorphism identification in FOXO3 and FBXO32 and association analysis with growth traits
With the use of P9, P10 and P11 (S6 Table), ten chickens from the WRR and XH F2 full sibling hybrid population were selected to identify polymorphisms in the coding region and intron region near the exon-intron boundaries of the FOXO3 gene through direct sequencing. Primers P30 (S6 Table) were used for identify polymorphisms in the 5' flank region of the FBXO32 gene. Another ten chickens from the WRR & XH F2 full sib hybrid population were selected to sequence. Sequence analyses were conducted with the software DNASTAR V 3.0 (http://www. biologysoft.com/; Steve ShearDown, 1998-2001 version reserved by DNASTAR Inc., Madison, Wisconsin, USA). Only polymorphisms occurring more than twice were regarded as variations. P10 and P30 was also used for the following association analyses in this population by direct sequencing. Association analyses of polymorphisms with chicken growth and carcass traits were conducted using SAS 8.0 software (SAS Institute Inc, Cary, NC, USA) using the following GLM model: where Y is an observation of the traits, μ is the overall population mean, G is the fixed effect of genotype, D is the random effect of dam, H is the fixed effect of hatch, S is the fixed effect of sex, and e is the residual error.
The chicken fibroblast cell line DF-1 was obtained from the Cell Bank of Committee on Type Culture Collection of Chinese Academy of Sciences and was cultured in Dulbecco's Modified Eagles Medium (DMEM) (GIBCO BRL, Life Technologies, Invitrogen Corporation, Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS) (Hyclone Laboratories, Logan, Utah), 100 U/ml penicillin, and 100 μg/ml streptomycin. Cells were cultured and maintained in humidified air at 37°C with 5% CO 2 . When the DF-1 cells were grown to approximately 50-80% confluence, independent transfections of the three siRNAs and the control NC siRNA were carried out using the transfection reagent Lipofectamine 3000 (Invitrogen Corporation, Carlsbad, CA) following the manufacturer's protocol. For each siRNA targeting FOXO3, a plasmid concentration gradient (25 nM, 50 nM and 100 nM) was set to investigate which concentration ratio had the highest transfection efficiency. Here, experiments for each gradient were conducted in triplicates. For each siRNA targeting FBXO32, the recommended concentration of Lipofectamine 2000 (50 nM) was transfected into DF-1 cells. At 48 h posttransfection, RNA was extracted from cells to perform qPCR. For each siRNA, the expression of target genes at each concentration was compared to that in cells transfected with NC siRNA. P12 was used for qPCR for FOXO3, P14 was used for qPCR for FBXO32, and β-actin (P1) was used as the control (S6 Table). Eventually, we found that siRNA-1082 at 100 nM had the highest relative efficiency of transfection and it was selected for the following experiments. siRNA-002 for FBXO32 had the highest relative efficiency of transfection and was used for transfecting DF-1 cells. When cells were grown to 50-80% confluence, siRNA-1082 and NC siRNA at 100 nM and siRNA-002 and NC siRNA at 50 nM were transfected into DF-1 cells. RNA was collected at 48 h post-transfection. With the use of qPCR, the expression change was analyzed for growth-related genes including CEBPB, FBXO32, GH, GHR, IGF1R, IGF2R, IGFBP2, IGFBP3, IGF2BP1, IGF2BP3, INSR, MYF5, MYF6, MYOD, MYH10, PDK1 and PDK4 (primers are shown in S6 Table). XH l indicate the comparisons between WRR h and WRR l , between XH h and XH l , between WRR h and XH h and between WRR l and XH l , respectively. In each comparisons, up-regulated indicates that the expression in the second group was higher than that in the first group, while down-regulated indicates that the expression in the first group was higher than that in the second group.