Identification of differentially expressed genes and pathways for intramuscular fat deposition in pectoralis major tissues of fast-and slow-growing chickens

1During the past decades, meat poultry breeding has been predominantly focused on accelerating growth rate and yields of breast and thigh meat. The impressive progress made in these traits, however, has been accompanied by deterioration of taste quality of the broiler meat and, in some markets, decreased acceptability of the meat by consumers.

2Intramuscular fat (IMF), located in most species in the epimysium, perimysium and endomysium, is an important determinant of meat quality. A certain amount of IMF can enhance meat quality traits such as the flavor, juiciness, water holding capacity and tenderness. Additionally, IMF can improve meat quality by reducing the drip loss and cooking loss.

3Compared to beef, chicken meat does not present a marbled aspect and intramuscular fat is not visible and not anatomically separable; a higher proportion of IMF is represented by polar lipids, presumably within membranes. Previous studies found that the IMF content of chicken meat increased with age from d 28 to d 90 and may increase further after that (d 90-d 140), along with flavor and taste of the meat.

4For livestock IMF, studies of the molecular mechanisms underlying IMF deposition have revealed large numbers of DEGs and signaling pathways including ADAMST4-signaling and insulin-signaling pathways. Compared to mammals, where most de novo synthesis of fatty acids occurs in adipose tissue, chicken differs in lipid metabolism because little fatty acid synthesis occurs in adipose tissues in this species. Until now, no systematic studies have been reported on the IMF development in chicken at the molecular level, although expression profiles in abdominal fat tissue, liver or muscle cells have been performed. This study provides a comprehensive analysis of gene expression profiles of chicken breast using both fast- and slow-growing breeds.

5To identify potential candidate genes affecting chicken IMF deposition, gene expression profiles in breast muscle of both Beijing-you (BJY, a slow-growing Chinese breed) and Arbor Acres (AA, commercial fast-growing broiler) chickens at different developmental stages were examined using Agilent cDNA microarray technology. Divergence of breast growth rates in BJY and AA chickens are shown in Table 1. Of the target traits measured, IMF contents and muscle weights at later ages (d 21, d 42, d 90 and d 120 just for BJY) were all higher than the content and weight at d 1 (Table 1). Thus, for each breed, the gene expression profile at d 1 was used as the control and the DEGs analyses were carried out as comparisons with d 1 (21 vs 1, 42 vs 1, 90 vs 1 and 120 vs 1, just for BJY). For AA chickens, 4255 genes (1310 known) were detected as DEGs with consistent fold changes ≥2.0, in all comparisons. In BJY chickens, 3182 genes (1080 known) were detected as DEGs (Table 2). There were 1746 DE genes (595 known) shared by the two breeds, (Additional file 1, Figure 1).

6Based on the known DE genes, GO analysis was performed in each breed, and the enriched GO-terms (P < 0.05) in the ontology classification “Biological Process” were selected and are presented in Additional file 2. The results showed the consistency of enriched GO-terms between the two breeds, mainly including the following processes: muscle system, lipid metabolism, cell cycle, protein metabolism, hormone metabolism, trans-membrane transport, oxidation-reduction, regulation of cell differentiation, regulation of immune system, blood circulation, regulation of apoptosis and ATP biosynthesis.

7To validate the microarrays, normal distribution analysis was performed with the results of the nine microarrays, and all of their coefficients were <0.5. Based on the 595 known genes that were shared as DEGs in both breeds, cluster analysis of all microarrays was performed (Figure 2) using the Cluster 3.0 software. The results demonstrated that expression profiles of samples at different ages of the same breed were polymerized together; the expression patterns of genes at d 1 differed more than those at other ages.

8To further validate the results of microarray testing, q-PCR was used to examine the relative expression of 9 DE key genes selected in each breed at different ages. The results showed acceptable consistency between the results of q-PCR and the fold-change of DE genes in the microarrays (Table 3).

9As IMF is located throughout skeletal muscle, and not as discrete adipose deposits in chicken, it is reasonable to assume that DEGs related to muscle development or lipid metabolism in current study would contribute to its process of deposition. The GO-term analysis showed that known DE genes related to muscle development (34 in BJY, 39 in AA) and genes related to lipid metabolism (59 in BJY, 70 in AA) were found (Additional file 3, Additional file 4). The 19 DE genes related to muscle development that were shared by the two breeds included FGF4, MYH4, MYBPC1, MYH7B, MYL2, MYL10, MYL3, MYO15A, TGFA and WNT4 (Additional file 5); these might be key genes. The 33 DEGs affecting lipid metabolism that were shared by BJY and AA breeds included ACSL1, ACSS2, APOH, FABP1, FABP3, CETP, RXRA, PPARGC1A, SNX4 and SNX30 (Additional file 5).

10Genes for which the expression was significantly correlated with changes in muscle development and lipid deposition were analyzed as these genes might be directly involved in IMF regulation. The changes in muscle weight, the content of IMF, triglyceride (TG) and phospholipids (PLIP) in breast tissue of the two breeds at different ages are provided in Table 1. The striking difference between the breeds in the rate of growth of the breast muscle is apparent but growth continues across all ages sampled.

11Based on the trait measurements, relationships between expression of DEGs and muscle development and lipid deposition (IMF, TG, PLIP contents) were examined. The expression of GDF3 and CAPNS1 in AA, and ANKRD1 and PLG in BJY were positively correlated (p < 0.05) with muscle development. The expression of BMP2K and MYBPC1 in AA, and CENPF, ELN, FGF7, FGFR1, MYBL2, MYCN and MYBPC1 in BJY were negatively correlated (p < 0.05) with that of muscle development (Table 4). For lipid deposition, CH25H, RCJMB04_10b24, RCJMB04_13o20, SNX3, and CETP in AA, and YWHAH, NR3C2 and CETP in BJY were positively correlated (p < 0.05) with PLIP contents while NACA, RBP7 and GLTPD1 in AA chickens, and LOC416618, ETFDH and GLTPD1 in BJY chickens were negatively correlated (p < 0.05) with PLIP contents. In the case of TG deposition, HMGCLL1, THBS1, UCP3 and SNX4 in AA, and EHHADH and SNX4 in BJY were positively correlated (p < 0.05) with TG contents and SGPL1, SH3PXD2B and THRSP in just the AA chickens were negatively correlated (p < 0.05) with contents of TG or IMF (Table 4). The expression of 1 DEG (MYBPC1), common to both breeds, was positively correlated with breast tissue weight across ages while 3 DEGs (CETP, GLTPD1 and SNX4), common to the two breeds, were positively or negatively correlated with IMF, TG or PLIP contents (Table 4).

12The relative abundances of MYBPC1, GLTPD1, CETP, SNX4 transcripts in BJY chickens and in AA chickens were further measured by q-PCR. Consistent correlation between mRNA expression of these genes and muscle development and lipid deposition (IMF, TG, PLIP contents) were found and are shown in Figure 3.

13The regulation of IMF is possibly a function of complex pathway interactions involving muscle, fat and connective tissue, so examining regulatory networks is the preferred method of analysis. After KEGG pathway analysis of the known DEGs related to muscle development and lipid metabolism, 24 metabolic pathways were identified in each breed, with 15 being shared by BJY and AA chickens (Additional file 6). Well known pathways affecting lipid metabolism (MAPK- and PPAR-signaling) were enriched in both breeds; the ErbB signaling pathway probably plays a role upstream of MAPK signaling. These analyses also demonstrate that pathways related to cell junctions (tight junction, ECM-receptor interaction, focal adhesion, regulation of actin cytoskeleton) were also enriched and might form a network with pathways related to lipid metabolism to influence the deposition of IMF (Figure 4).

14Fat deposition in chickens takes place mainly in visceral adipose tissue and in muscle. The latter, IMF, is very important for sensory aspects of meat quality and there is increasing interest in improving quality, perhaps using marker-assisted selection for IMF. Although global gene expression surveys have been performed on visceral tissues, this study is the first to systematically explore gene expression profiles in breast tissue using two distinct breeds across their development. The present objective was to identify global genes and pathways affecting chicken IMF deposition.

15Chicken microarrays were employed, each using pooled RNA samples (n = 6 birds, within each of two breeds and at 4 or 5 ages; 9 arrays in all). Such a pooling strategy can dramatically improve accuracy when only one array is available in each biological condition. Potential candidate DEGs related to IMF deposition were rigorously defined, requiring their expression to differ across all comparisons conducted (21 vs 1, 42 vs 1, 90 vs 1 and 120 vs 1, just for BJY). To confirm results from the microarrays, more than 100 tests were done with q-PCR, involving 12–13 genes in breast tissue at 4 or 5 sampled ages in each breed. As shown in Figure 5, fold-changes in gene expression between the two methods were correlated in both BJY (r2 = 0.85) and AA (r2 = 0.72) chickens.

16The data obtained here indicate that three key genes related to lipid metabolism were shared by the two breeds (CETPGLTPD1 and SNX4) and showed consistent trends with the changes in IMF, TG or PLIP contents across the ages examined (Table 4). Cholesteryl ester transfer protein (CETP) functions in the reversible transport/exchange of cholesteryl esters from high-density lipoproteins (HDL) and triglycerides from very-low-density (VLDL) or low-density (LDL) lipoproteins. The SNX4 gene encodes a member of the sorting nexin family which associates with a variety of receptors, including those for insulin, EGF and leptin. Glycolipid transfer protein D1 (GLTPD1) is a cytosolic protein that transfers glycolipids between different intracellular membranes; its precise biological function is not known.

17For key candidate genes affecting muscle development, one gene common to the two breeds (MYBPC1) changed consistently with the changing patterns of breast tissue weight (Table 4). The MYBPC1 gene encodes a slow form(s) of MyBP C found in skeletal muscle and functions in the assembly and stabilization of sarcomeric M- and A-bands and regulates the contractile properties of actomyosin filaments. In chickens, increased levels of MyBP-C slow are found in dystrophic skeletal muscles. Further research is required to understand the molecular mechanisms that lead to the anticipated effects of this candidate gene on lipid metabolism and muscle development in chickens.

18GO-term analysis was used to explore the function of DEGs and KEGG pathway analysis was used to explore the regulatory network underlying chicken IMF deposition. As expected, several well-known pathways related to lipid metabolism were found, including the MAPK and PPAR signaling pathways. Large numbers of DEGs involved in PPAR signaling pathways here have been proven to be functional in lipid metabolism, such as FABP family genes (FABP1, FABP5, FABP6), ACSL4, CD36, PLTP. This is partially consistent with our previous studies where mRNA expression of adipocyte FABP and heart-type FABP genes and SNP markers from these genes were found to be associated with IMF content in Chinese chickens. Several DEGs (FGF1, FGF4, FGF7, FGF16, FGFR1 and FGFR2) belonging to the FGF family and receptors were reflected in the MAPK signaling pathway.

19Of special interest, these GO and KEGG analyses provide the first demonstration that a series of pathways related to cell junctions might contribute to the deposition of IMF. DEGs related to muscle development included well-known genes (MYBPC1, MYBPC2, MYH4, MYH7BMYL10, MYL2, MYL3, MYO15A and MY01F) related to the biosynthesis of myosin, of which MYL10 and MYL2 encode proteins involved in the formation of tight junctions, focal adhesions and regulation of the actin cytoskeleton. Additionally, the CTNNB1 gene encodes β-catenin, one of the proteins constituting adherens junctions, and anchoring the actin cytoskeleton; ACTB encodes beta-actin, one of the non-muscle cytoskeletal actins playing a central role in shape determination, cytokinesis, and cell motility, along with cell-cell and cell-matrix interactions. In addition, DEGs of the FGF family and their receptors (FGF1, FGF4, FGF7, FGF16, FGFR1 and FGFR2) were reflected in the regulation of the actin cytoskeleton and may modulate morphogenetic processes involving cellular rearrangements and tissue remodeling. The genes THBS1 and CD36 are involved in extracellular matrix (ECM)-receptor interaction and influence, directly or indirectly, cellular activities such as adhesion and migration. In support of this, previous studies have shown that changes in cytoskeletal organization and its contacts with the ECM are essential in the morphogenesis of fibroblastic preadipocytes to rounded, mature adipocytes, while the expression of actin, integrins and several cytoskeletal proteins is down-regulated during adipogenesis. Taken together, cell junctions including the interaction of the ECM and cytoskeleton might participate in accumulation of IMF during chicken development.

20The KEGG analysis implicated the MAPK signaling pathway in processes involving tight junctions, focal adhesion and regulation of the actin cytoskeleton. This is also consistent with studies showing that activation of MAPK activity resulted in the disruption of tight junctions, and that inhibition of MAPK activation prevented this process. We suggest that processes related to cell junctions might interact with pathways related to lipid metabolism, mainly through MAPK activity, to influence the deposition of IMF. The proposed molecular regulatory network affecting IMF deposition during chicken development is presented in Figure 4. This novel suggestion of IMF regulation and its detailed mechanism through pathways related to cell junctions in addition to lipid metabolism needs further examination.

21The present approach has used gene expression profiling to elucidate molecular mechanisms of post-hatch IMF deposition in chickens. Possible regulation by translational mechanisms and posttranslational modifications may also contribute. A more complete understanding of IMF development in chickens should include further examination of the expression and function of the proteins encoded by the genes identified here in both embryonic and post-hatch stages of development.

22With aim of identifing global candidate genes and new pathways related to IMF deposition in chicken breast, Agilent cDNA microarray analyses were performed with both fast- and slow- growing breeds. Gene expression profiles of breast muscle sampled at different developmental stages of BJY and AA chickens were determined. Relative to d 1, breast muscle at d 21, d 42, d 90 and d 120 (only for BJY) contained 1310 DEGs in BJY and 1080 DEGs in AA. Several DEGs (MYBPC1, CETP, GLTPD1 and SNX4) may play key roles in IMF developmental processes because their expressions were correlated with the changing patterns of lipid content or breast weight across the ages sampled in both two breeds. In addition, the results of KEGG pathway analysis imply that IMF deposition in chickens is regulated and mediated not only by genes and pathways related to lipid metabolism and muscle development, but also by others involved in cell junctions with the function in maintaining the integrity of tissues and signal transduction. These findings establish the groundwork and provide new clues for deciphering the molecular mechanisms underlying IMF deposition in poultry. Additional studies of translational and posttranslational effects will be required to complement these mRNA expression analyses.

23All experimental procedures, using female Beijing-You chickens (BJY, the Institute of Animal Sciences, Chinese Academy of Agricultural Sciences, Beijing, China) and Arbor Acres (AA, Dadongliu broiler Company, Beijing, China), were performed in accordance with the Guidelines for Experimental Animals established by the Ministry of Science and Technology (Beijing, China). Individuals within each breed had the same genetic background. Birds (60 BJY and 48 AA) were reared in stair-step caging under continuous lighting using standard conditions of temperature, humidity and ventilation. Chickens used for sample collection at d 1 were not fed. The same diet was fed to all chickens and was formulated to be intermediate between recommendations for the two breeds. The starter ration (d 1 to d 21) with 20% crude protein and 2.87 MC/kg differed only slightly from that used in the grower (after d 22) phase; 19% crude protein and 3.0 MC/Kg. Feed and water were provided ad libitum during the experiment.

24At each sampling age, d 1, d 21, d 42, d 90, and d 120 (only for BJY), six birds of similar weight from each breed were sacrificed for tissue collection. Samples of the left pectoralis major muscle were excised, snap-frozen in liquid nitrogen and stored at −80°C. The entire right breast was collected and stored at −20°C for trait measurements.

25Intramuscular fat (IMF) content of breast muscle was determined by extraction with petroleum ether in a Soxhlet apparatus and expressed as percentages of the dry weight of the muscle.

26Samples of the right pectoralis major muscle were homogenized using the method of Folch. The contents of triglyceride (TG) and phospholipids (PLIP) in the solvent phase, after centrifugation, were analyzed with TG and PLIP kits (Deliman Biochemical technology Co., LTD, Beijing, China).

27Total RNA was isolated from breast muscle samples using Trizol reagent (Invitrogen, USA) according to the manufacturer's instructions and dissolved in RNase-free water at a final 2.0 μg/μl concentration. RNA from Pectoralis major collected at each sampling age were extracted, and pooled within days and breeds for testing with microarrays. Microarray hybridization was carried out by GeneTech Biotechnology Limited Company (Shanghai, China) using Agilent Chicken Gene Chips (ID: 015068) with 42034 probes. Array scanning and data extraction were carried out following the standard protocol.

28The normal distribution of signal plots in every chip was determined. Clustering was performed based on the DE genes in each chicken breed, using un-centered Pearson correlations and average linkage cluster 3.0, and was displayed in TreeView. Normalized fluorescence intensity values of each dye-swapped experiment were averaged separately for sample and reference channels. Thereafter, for each probe, averaged sample and reference fluorescence values were log2-transformed. Average linkage hierarchical clustering was performed using the Euclidian metric. In heat-maps, the color of features (probes) was determined by log2 (reference/sample).

29The distribution of expressed genes was analyzed by JMP4.0 according to their expression level. If the flag of a gene was “A” by the scanner according to the data normalization and results of Agilent Microarray Suite 4.0 software, it was considered to be "not detected", and hence "not expressed" in this study. Similarly, the genes with “P” flags were considered to be “expressed transcripts”. Expressed transcripts were defined as being present in samples of at least one sampled age and were used for all following studies. The expression value of each probe set was normalized and calibrated using the RMA method.

30Screening of differentially expressed genes (DEGs) was performed on the basis of differences in the IMF, TG and PLIP contents at different ages in each breed. Expression at d 1 was used as the controls, and comparisons were made within each breed at d 21, 42, 90 and 120 (only for BJY). Genes were considered to be DEGs only when the fold-change in abundance for all comparisons exceeded 2.0.

31To avoid amplification of any residual genomic DNA, all PCR primers were placed at or just outside the exon/exon junctions and specificity was determined with BLASTN (Additional file 7). After a general reverse transcription reaction, PCR analyses were performed in 20 ul amplification reactions containing 10 ul of 2× SYBR Green PCR Master Mix (Tiangen Biological Technology Co., Ltd, Beijing, China), 20 ng cDNA and 0.5 μl (10 mM) of each primer using the following conditions according to the manufacturer's instructions: 95°C for 10 minutes for 1 cycle, 40 cycles at 95°C for 15 seconds and then at 63°C for 45 seconds.

32Quantification of the transcripts was determined using standard curves with 10-fold serial dilutions of cDNA (10-7 to 10-12 g). Melting curves were constructed to verify that only a single PCR product was amplified. Within runs, samples were assayed in triplicate with standard deviations of threshold cycle (CT) values not exceeding 0.5, and each q-PCR run was repeated at least twice. Negative (without template) reactions were performed within each assay.

33Gene Ontology enrichment analysis was performed for features corresponding to DEG in each breed using the GOEAST software toolkit. The significance level of GO term enrichment was set as FDR-adjusted p-value less than 0.05 by the Yekutieli method.

34Correlation analysis was performed between the key DE genes related to muscle development or lipid metabolism with changes in lipid content or breast muscle weight across the sampled ages within each breed. If the expression of a given gene was significantly correlated (p < 0.05) with breast tissue weight, or content of IMF, TG or PLIP, this gene was considered to be a key DE gene.

35KEGG pathway information was used in this analysis. Probeset IDs of each category were first mapped to NCBI Entrez gene IDs according to the Agilent Chicken Array annotation file, and then were mapped to KEGG gene IDs according to the KEGG gene cross-reference file. Pathways that were significantly enriched with DEGs were identified by a hypergeometric test using R packages (p < 0.01, FDR adjusted). Pathways with < 3 known chicken genes were discarded. Graphical pathway maps were downloaded from the KEGG FTP server, and DEGs were then highlighted in them according to the coordinate description in the XML files at the KEGG FTP server, using Perl GD, XML:Parser and XML:LibXML modules.

36ACSF3, Acyl-CoA synthetase family member 3; ACSL1, Acyl-CoA synthetase long-chain family member 1; ACSS(1&2), Acyl-CoA synthetase short-chain family member; ACTB, Actin, beta; ADAMST4, A disintegrin and metalloproteinase with thrombospondin motif 4; ANKRD1, Ankyrin repeat domain 1; APOH, Apolipoprotein H; BMP(4,7), Bone morphogenetic protein; BMP2K, BMP-2-inducible protein kinase; CAMs, Cell adhesion molecules; CAPNS1, calpain, small subunit 1; CD36, Cluster of Differentiation 36; CD44, CD44-like protein; CENPF, Centromere protein F; CETP, Cholesteryl ester transfer protein; CH25H, Cholesterol 25-hydroxylase; CTNNB1, Catenin (cadherin-associated protein), beta 1; CYP17A1, Cytochrome P450, family 17, subfamily A, polypeptide 1; DCK, Deoxycytidine kinase; DGAT2, Diacylglycerol O-acyltransferase homolog 2; ECM, Extracellular matrix; EGF, Epidermal growth factor; EHHADH, Enoyl-Coenzyme A, hydratase/3-hydroxyacyl Coenzyme A dehydrogenase; ELN, Elastin; ErbB, Erythroblastic Leukemia Viral Ongene Homolog; ETFDH, Electron-transferring-flavoprotein dehydrogenase; FABP(13, 5, 6, A/Adipocyte, H/Heart), Fatty acid binding protein; FGF(47, 16), Fibroblast growth factor; FGFR, Fibroblast growth factor receptor; GDF3, Growth differentiation factor 3; GLTPD1, Glycolipid transfer protein domain containing 1; HMGCLL1, 3-hydroxymethyl-3-methylglutaryl-Coenzyme A lyase-like 1; KEGG, Kyoto Encyclopedia of Genes and Genomes; LDLR, Low density lipoprotein receptor; LOC416618, NAD(P) dependent steroid dehydrogenase-like,; MAP2K3, Mitogen-activated protein kinase kinase 3; MAPK, Mitogen-activated protein kinase; MBOAT2, Membrane bound O-acyltransferase domain containing 2; MYBL2, V-myb myeloblastosis viral oncogene homolog (avian)-like 2; MYBPC1, Myosin binding protein C, slow type; MYCN, V-myc myelocytomatosis viral related oncogene, neuroblastoma derived; MYF5, Myogenic factor 5; MYH(4, 6, 7B), Myosin, heavy polypeptide; MYL(23, 4, 10), Myosin, light chain; MYO15A, Myosin XVA; NACA, Nascent polypeptide-associated complex alpha subunit; NR3C2, Nuclear receptor subfamily 3, group C, member 2; PGK1, Phosphoglycerate kinase 1; PLTP, Phospholipid transfer protein; PLG, Plasminogen; PPAR, Peroxisome proliferator-activated receptor; PPARGC1A, Peroxisome proliferator-activated receptor gamma, coactivator 1 alpha; RBP7, Retinol binding protein 7; RCJMB04_13o20, NSFL1 (p97) cofactor (p47); RXRA, Retinoid X receptor-alpha; SGPL1, Sphingosine-1-phosphate lyase 1; SH3PXD2B, SH3 and PX domains 2B; SNX(34, 17, 30), Sorting nexin; TGFA, Transforming growth factor, alpha; THBS1, Thrombospondin-1; THRSP, Thyroid hormone responsive, SPOT14 homolog; UCP3, Uncoupling protein 3; YWHAH, Tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein; WNT4, Wingless-type MMTV integration site family, member 4.