Introduction

Chicken, a pivotal source of animal high-quality protein in human diets, stands as one of the most widely consumed food products globally [1]. Over the past few decades, a substantial surge in body weight and pectoral muscle weight has been witnessed, mainly attributable to intensive genetic selection [2, 3]. Skeletal muscle, accounting for about 40% of body mass of animals, is comprised of elongated, multinucleated muscle fibers [4, 5]. Multinucleated muscle fibers are formed from the fusion of myoblasts during embryonic myogenesis [6]. The number of myofibers in animals typically remains unchanged after birth, with muscle growth primarily depending on myofiber hypertrophy [7]. The growth and development of skeletal muscle is a complex biological process that is regulated by diverse genes, transcription factors, and epigenetic factors [8]. Revealing the molecular mechanisms of animal muscle development has been a focus of research in the field of genetic and breeding research.

With the advancements in high-throughput sequencing and multi-omics technology, unraveling the intrinsic biological mechanisms of muscle development has become more convenient and cost-effective [9, 10]. An increasing number of genes or quantitative trait loci that regulate the growth traits of chickens have been characterized. A recent analysis from genome-wide association studies (GWAS) identified ADGRG6 as a major candidate gene responsible for carcass composition traits [11]. The genes NSUN3 and AGK were related to muscle development in broiler chickens. Despite the decreasing sequencing costs, GWAS remain costly, necessitating large sample sizes to gain sufficient SNP data [12, 13]. Challenges like linkage disequilibrium and population stratification complicate the accurate identification of causal variants in GWAS [14]. Further research at the transcriptome level could provide more comprehensive mechanistic insights [15, 16]. Comparison studies between modern broiler chickens and local breeds, conducted through RNA sequencing have revealed the upregulated genes such as IGF1, IGF1R, WFIKKN2, positively regulating chicken growth [17]. In another study using fast- and slow-growth meat chickens, DNA methylation analysis identified methylation sites in growth factors like IGF1R and FGF18, which are related to skeletal muscle development [18]. In 2013, the assay for transposase accessible chromatin with high-throughput sequencing (ATAC-seq) was developed for evaluating genome-wide chromatin accessibility and identifying transcription factor binding sites [19]. Integrating ATAC-seq with high-throughput paired-end RNA sequencing (RNA-seq) data can offer more precise information on the impact of possible regulatory differences in differentially enriched open chromatin regions [1: Table S2). The mapped reads were counted at gene level using featureCount (SUBREAD package; v.1.6.3) [27] with the chicken reference (GRCg7b) and normalized using TPM procedure to compute the TPM values for subsequent analysis steps. Principal component analysis (PCA) was performed using R package FactoMineR (v.2.9.0) [28]. Differentially expressed genes (DEGs) between different groups were identified by DESeq2 package (v.1.36.0) [29] in the R program. Genes with an adjusted P value < 0.05 and a log2 (Fold Change) > |1| were considered significant DEGs [30,1: Table S3). Subsequently, duplicate reads were marked and removed using Sambamba (v.0.7.1) [36]. The insert length was directly calculated from the aligned BAM file with Samtools (v.0.1.11) [37]. For peak calling, MACS2 (v.2.2.7.1) [38] was employed. IDR (v.2.0.3) was used to filter enriched regions at an IDR of less than 0.05 to generate high-confidence peaks [39]. Peak annotation and analysis were carried out using ChIPseeker (v.1.32.1) [40]. Promoter regions were defined as peaks overlap** a region that was ± 3 kb from the transcriptional start site (TSS). Differential peak analysis was performed using the diffBind package [41] with the false discovery ratio < 0.05 and log2 (Fold Change) > |0.5| in two comparison groups [42,43,44]. The Bedtools (v.2.30.0) [45] suite was employed to calculate overlap and enrichment between different intervals. The peaks of different groups were visualized in IGV tool (v.2.13.2) [46, 47].

Integrative analysis of RNA-seq and ATAC-seq data

Since chromatin accessibility is closely related to gene regulation, the VennDiagram package (v.1.7.3) [48] in R was used to obtain the overlap** genes of DEGs derived from RNA-seq and genes associated with differential accessible regions (DARs) identified in ATAC-seq. Subsequently, the clusterProfiler [33] was utilized for GO enrichment function analysis and KEGG pathway analysis of these overlap** genes. The genes of top 10 terms were extracted for frequency statistics. We combined the genes with higher frequencies into gene sets, which were used as input for signal pathway and network analysis on the GeneMANIA [49] and STRING online platforms [50].

Results

Time-course growth performance analysis of broilers

We monitored the growth performance of CC and RR over a 42-d period, including the BW and PMW, and examined their morphology of muscle fiber from D7 to D42. As expected, the results showed that BW and PMW of the two groups gradually increased with advancing age while the weight of CC was significantly higher than RR at all growth stages (Fig. 1a and b, Additional file 1: Table S4). The BW and PMW of CC group were 2,986.70 ± 25.61 g and 493.76 ± 76.13 g at D42. The BW and PMW of RR were 2,460.93 ± 168.09 g and 369.03 ± 52.95 g at the same time. The histological examination revealed minor differences between CC and RR groups with CSA and MFD (P > 0.05) except the CSA at D21 and D35 (Fig. 1c and d) while the slice diagram revealed that the size of muscle fibers continued to increase with age. Morphology of the myofibers at all time point after hatching is shown here (Fig. 1e). The BW and PMW were positively correlated with CSA while negatively correlated with MFD of myofibers at D21 and D42 (Fig. 1f). When considering the weight results, it became evident that both BW and PMW showed a gradual increase before D21, followed by a linear increase after D21. These results suggested that D21 and D42 are crucial stages for broiler growth. To identify the genes that were differentially expressed between different groups, pectoral muscle tissue of CC and RR was collected at D21 and D42 for RNA-seq and ATAC-seq (Fig. 1g).

Fig. 1
figure 1

Growth performance of postnatal broiler chickens. a The changes in body weight (BW) of individuals with Cornish (CC) and White Plymouth Rock (RR) from the day of hatching (DOH) and day 7 after hatching (D7), D14, D21, D28, D35, D42. b The changes in pectoral muscle weight (PMW) of individuals with CC and RR from DOH to D42. c Cross-sectional area (CSA) of muscle fibers from D7 to D42. d Muscle fiber density (MFD) changes from D7 to D42. e Morphology of the myofibers stained by hematoxylin–eosin from D7 to D42. The corresponding scale is marked in the lower left corner of each image. Scale bars: 100 μm. f The scatter plot and Pearson’s correlation of broiler chickens at D21 and D42. g Pectoral muscles were collected at D21 and D42 for RNA-seq and ATAC-seq analysis. ** and * indicate P value less than 0.01 and 0.05, respectively

Full size image

Transcriptional profiling of pectoral muscle

The PCA plot exhibited a separation between the two stages of D21 and D42 along PC2 (Fig. 2a). To identify the genes that were DEGs between different groups, we performed a differential expression analysis with the four groups in the two dimensions of breeds and ages. The volcano map preliminarily showed the expression of DEGs, as shown in Fig. 2b. A total of 204, 219 D21BGs and 302, 404 D42BGs of CC and RR were identified, respectively (Additional file 1: Table S4–5). Additionally, there were 137, 58 CBGs and 143, 60 RBGs at D21 and D42 (Additional file 1: Table S5–6).

Fig. 2
figure 2

Analysis of RNA-seq data from the pectoral muscle. a PCA plot of RNA-seq data of CC and RR groups at D21 and D42. Each point represents a sample. b Volcano plot of differential expressed genes (DEGs) obtained through the comparison between pairwise groups. c The GO enrichment results of CC-biased DEGs (CBGs) and RR-biased DEGs (RBGs) at D21 and D42. d GO enrichment results of D21-biased DEGs (D21BGs) and D42-biased DEGs (D42BGs) of CC and RR

Full size image

In terms of differential gene functions between CC and RR groups at D21 and D42, various biological functions were enriched. At D21, CBGs were primarily enriched in terms of extracellular space and anchored component of membrane (Fig. 2c). Additionally, they were also enriched in Glycolysis/Gluconeogenesis, Fructose and mannose metabolism, Purine metabolism, and Cell adhesion molecules (Additional file 2: Fig. S1a). Among them, ALDOB were found to be related to both the extracellular space term and the Glycolysis/Gluconeogenesis pathway. RBGs at D21 were enriched in the terms such as circadian rhythm and negative regulation of cellular response to growth factor stimulus (Fig. 2c). The CBGs at D42 were enriched in ATP-dependent chromatin remodeling and WNT signaling pathway as shown in Fig. S1a. Furthermore, the pathway of Purine metabolism, including NPR1, was identified at both CBGs at D21 and D42. The RBGs at D42 were primarily enriched in cholesterol metabolic process term and amino acid metabolism pathways (Fig. 2c, Additional file 2: Fig. S1a).

A total of 137, 58 CBGs and 143, 60 RBGs at D21 and D42 were identified, respectively (Additional file 1: Table S7–8). Significant enrichment was observed in terms associated with apical plasma membrane and actin-based cell projection of D21BGs of CC (Fig. 2d). The D42BGs of CC were mainly involved in cell differentiation and muscle development, including FOS, MUSTN1, TGFB3, WNT9A, and GREM1 (Fig. 2d). Additionally, the D21BGs of RR were enriched in terms such as myosin complex and cytoskeletal motor activity that included MYH1D, MYO1A, and ACTC1. The D42BGs in RR exhibited enrichment in defense response and immune system processes, as well as pathways such as cytokine-cytokine receptor interaction and cell adhesion molecules (Fig. 2d, Additional file 2: Fig. S1b). Notably, D21BGs and D42BGs of CC and RR were involved in steroid biosynthesis and cell–cell interactions (Additional file 2: Fig. S1b).

Analysis of chromatin accessibility in pectoral muscle

In order to better understand the mechanisms underlying gene expression alterations, we conducted a comprehensive examination of chromatin openness using ATAC-seq across the four experimental groups. The ATAC-seq data displayed approximately 200 bp fragment insert size (Fig. 3a, Additional file 2: Fig. S2a–c), and the PCA revealed a separation tendency among the four groups (Fig. 3b), indicating the reliability of the ATAC-seq data. For the 12 pectoral muscle samples, a total of 75,149 ATAC-seq peaks were identified (Additional file 2: Fig. S2d). These peaks were annotated in their promoter, exon, intron and intergenic regions, and a higher abundance observed in promoter and intronic regions (Fig. 3c). Contrasting the peaks in terms of breed and time dimensions revealed a comparable number of intersecting peaks between the two groups. Specifically, the intersection of peaks between CC and RR at D21 and D42 were 11,418 and 11,151, respectively (Fig. 3d). Furthermore, the intersection of peaks between pectoral muscle samples of D21 and D42 in the CC and RR groups were 11,681 and 10,843, respectively (Fig. 3d).

Fig. 3
figure 3

Analysis of chromatin accessibility from the pectoral muscle. a Fragment insert length distribution plots of ATAC-seq samples. b PCA plot of ATAC-seq data of CC and RR groups at D21 and D42. Each point represents a sample. c The feature distribution of ATAC-seq peak dataset. d Overlap peaks between two groups. e Heatmaps of differential accessible regions (DARs)

Full size image

Subsequently, a pairwise comparative analysis based on the ATAC-seq data was conducted to identify DARs. A total of 3,283, 24,417, 7,916, and 15,553 significant DARs were identified in D21, D42, CC and RR development (Additional file 1: Table S9–12). The genes nearest DARs were annotated, revealing 1,486, 8,462, 3,028, and 5,615 genes at D21, D42, CC, and RR, respectively. The heatmaps revealed that these significant DARs showed very clear specific patterns in each group (Fig. 3e, Additional file 2: Fig. S2e).

Integration analysis of ATAC-seq and RNA-seq

To further determine the relationship between gene expression and open chromatin regions, we examined the association between gene accessibility and gene expression during pectoral muscle development by integrating RNA-seq and ATAC-seq data (Additional file 1: Table S13). As the Venn diagrams shown, we found 6 upregulated CBGs carrying DARs, including TMEM164, FSTL4, ZNF692, KIF14, APOA5, and KIFC1, and 16 RBGs carrying DARs related to muscle development, such as MYO7B at D21 (Fig. 4a). In addition, 12 upregulated and 1 downregulated DEGs contained DARs were identified in the CC vs. RR at D42. Among these, the RBG was identified as IRF9, which was associated with the inflammatory response. Focusing on the impact of age on chromatin accessibility, we identified 69 D42BGs carrying DARs of CC group, including WNT9A, TGFB3 that related to muscle development. Additionally, we found 25 D21BGs containing DARs in CC groups, which included NRG1, a gene related to lipid metabolism, and TMEM164, a member of the transmembrane protein family. Moreover, 5 upregulated DEGs carrying DARs were found between D21 and D42 of RR, including IRF7, DCSTAMP, RGCC, TTC7A and NR1D1. In this comparison, there were 65 DEGs with DARs were downregulated at D42. As expected, a positive correlation was found between accessibility signatures and gene expression patterns, based on the calculated fold changes by assigning open chromatin regions to the nearest DEGs (Fig. 4b, Additional file 2: Fig. S2f).

Fig. 4
figure 4

Integration of RNA-seq and ATAC-seq to identify candidate genes. a Venn diagrams showed the DEGs carrying DARs. b The relationship between DEGs and DARs. c The 16 RBGs at D21 and the 65 D21BGs of RR were combined for GO enrichment analysis. d Frequency analysis of DEGs of top 20 terms. e The network analysis and functional enrichment analysis of the DEGs was performed using GeneMANIA. f Differential expression results among the four groups. ** and * indicate P value less than 0.01 and 0.05, respectively

Full size image

Insights into muscle development at D21 in broilers

Based on the data showed in Fig. 4a, it was determined that there were four distinct classes of DEGs exhibiting higher count level, including CC-biased genes at D42, RR-biased genes at D21, D42-biased genes of CC, and D21-biased genes of RR, within the eight identified clusters. Considering the differential expression patterns, we have merged a set of 16 RR-biased genes at D21 with 65 D21-biased genes of CC. This combination of DEGs, resulting in a total of 81 DEGs that were highly expressed in RR at D21, for subsequent analysis. Enrichment analysis indicated that these genes were mainly involved in biological processes such as cholesterol metabolic process and actin filament-based movement (Fig. 4c). For further analysis, the DEGs of top 20 terms were chosen for frequency analysis, and the 14 genes were ranked (Fig. 4d). It was observed that FDPS and CHRM4 had the highest frequency, appearing 6 times, followed by MTNR1A, CETP, PRPS2, VTG1, HBBA and HBA1, indicating their crucial role in the development of RR at D21. The subsequent analysis was conducted using the GeneMANIA, which confirmed the strong association of ACTC1, NRG1, and MYC with mesenchyme development, and FDPS and CETP with sterol metabolic process (Fig. 4e).

The gene expression profiles in broiler chickens at both D21 and D42 are depicted in Fig. 4f. The findings indicated a significant upregulation of ACTC1, HBBA, HBA1, FDPS, and NRG1 in broiler chickens at D21 compared to D42. It is known that the rapid muscle development occurs in broiler chickens during the D21 period. Based on these observations, it can be hypothesized that ACTC1 and NRG1 play crucial roles in muscle development during this period, while FDPS, HBBA and HBA1 may exert a coordinated regulatory influence on muscular development.

Muscle regulation dynamics and regeneration mechanisms at D42

On the flip side, we noted that the 12 CBGs at D42 and 69 D42BGs of CC could be classified into a singular category that were highly expressed in CC at D42. This classification facilitated an in-depth exploration of the regulatory patterns governing muscle development in broiler chickens at D42, resulting in the identification of 81 DEGs. Subsequent functional enrichment analysis of these 81 DEGs revealed their significant involvement in organ developmental process, cell differentiation, and SMAD protein signal transduction (Fig. 5a). Using the method described above, we selected the top 10 terms for frequency analysis of the DEGs, and the resulting 9 genes were ranked accordingly (Fig. 5b). Notably, TGFB3 exhibited the highest frequency, appearing 8 times, followed by FOS, WNT9A, HDAC9, and MUSTN1, underscoring their pivotal role in the development of CC at D42.

Fig. 5
figure 5

Intense muscle damage with rapid muscle growth at D42. a The 12 CBGs at D42 and the 69 D42BGs of CC were combined for GO enrichment analysis. b Frequency analysis of 9 DEGs of top 10 terms. c Network analysis and functional enrichment analysis using GeneMANIA on the 9 DEGs that were present in at least 2 terms among the 20 DEGs. d The network analysis of DEGs was performed using STRING. e Differential expression results among the four groups. ** and * indicate P value less than 0.01 and 0.05, respectively. f Gene expression and chromatin accessibility of FOS and MUSTN1 displayed similar dynamic changes in CC at D42

Full size image

Utilizing GeneMANIA, we conducted the network analysis and functional enrichment analysis on the 9 DEGs in Fig. 5c. The results revealed that LMOD2 was associated with myofiber contraction and cytoskeletal organization. Protein–protein interaction analysis revealed that LMOD2 interacts with MUSTN1 and MB (Fig. 5d). This finding suggested a potential functional relationship between these DEGs, which could be significant in the context of muscle development and function. Notably, we observed that MUSTN1, FOS, TGFB3, and WNT9A were highly expressed in CC at D42 among the four groups (Fig. 5e). Furthermore, the gene expression and chromatin accessibility of FOS and MUSTN1, which are related to muscle regeneration, exhibited similar dynamic changes in CC at D42 (Fig. 5f).

Discussion

As a crucial tissue in meat-producing animals, skeletal muscle plays a vital role in initiating movements, supporting respiration, and maintaining homeostasis [51, 52]. Chickens serve as significant scientific models for studying skeletal muscle development in animals, given that the progression of chicken anatomy closely mirrors that of mammals [53, 54]. Through prolonged selective breeding practices, broiler chickens have developed an exceptionally rapid growth rate, primarily attributed to the prioritization of the pectoral muscle’s weight as a breeding goal, which rapidly increases in size after hatching [55, 86, 87]. It is noteworthy that broiler chickens exhibit sustained accelerated muscle growth at D42, which may be closely related to the elevated expression of TGFB3. Importantly, the study revealed the presence of several genes associated with muscle regeneration, such as FOS and MUSTN1, in broilers at D42. Research has shown that FOS plays a crucial role in initiating key stem cell functions including cell cycle entry, proliferative expansion, and muscle regeneration [88]. The function of MUSTN1 in skeletal muscle physiology is multifaceted, with the knockdown of MUSTN1 has been shown to induce changes in myofiber types [89] and to exert influences on both myoblast differentiation and muscle regeneration processes [90]. Myoblast differentiation is impaired and the fusion of myoblasts fails to occur with the absence of MUSTN1. In-depth vivo experiments have demonstrated that MUSTN1 functions as an early marker of myoblast fusion, playing a role in skeletal muscle development and regeneration [90]. Furthermore, inhibition of MUSTN1 expression impairs myoblast differentiation by targeting MYOD and MYOG [91]. Interestingly, our H&E staining results of muscle results showed that numerous myonuclei within the muscle fibers of broilers during the growth period were located centrally rather than at the periphery, suggesting a state of continuous muscle repair [92, 93]. This further indicates that intense muscle damage may coincide with rapid muscle growth until D42. To ensure continuous muscle mass growth, the organism responds by upregulating FOS and MUSTN1, facilitating repair and achieving a stable state of muscle differentiation in the body.

Conclusion

This research provides a high-resolution exploration of the transcriptome and chromatin accessibility landscapes of pectoral muscle in broiler chickens. Our study explores the developmental dynamics of the pectoral muscle, focusing on various dimensions and the results underscore that temporal differences outweigh those observed in the breed dimension. Moreover, our findings illuminate that the rapid growth observed in broiler chickens induces continuous muscle damage followed by subsequent regeneration. Key candidate genes, including ACTC1, FDPS, NRG1, TGFB3, MUSTN1 and FOS, were identified to play crucial roles in the rapid post-hatching development of broiler chickens. Overall, our research contributes to a deeper understanding of the regulatory mechanisms governing late-stage muscle development in broiler chickens. Additionally, the identification of key candidate genes lays a foundation for potential future studies delving into the functional aspects of muscle development.