Investigation of myofiber composition changes and molecular mechanisms in rabbit meat quality development during growth

Abstract

Background

Rabbit meat is a high-protein, low-fat food with recognized nutritional benefits, often labeled as “healthy meat” and “nootropic meat.” However, its low-fat content leads to relatively poor flavor. Myofibers are the main components of rabbit meat, and their type composition determines the overall metabolic characteristics of the meat, which ultimately affects fresh meat quality. During the growth of rabbits, muscle fibers transform into one another. Nevertheless, the changes in the relative composition of myofiber types as domestic rabbits grow, and the molecular mechanisms behind myofiber transition, remain unclear. This study aimed to analyze the changes in the relative composition of myofiber types in rabbits of different ages and explore the roles of various potential molecules in rabbit myofiber transition at the transcriptional level using whole transcriptome technology.

Results

Significant differences were observed in the relative composition of gluteus (GLU) and gastrocnemius (GAS) muscle fiber types, which impacted rabbit meat color and taste. During growth, the relative composition of GLU muscle fiber types differed significantly between 1-day-old and 14-day-old rabbits. Transcriptome analysis of GLU muscle from these two age groups revealed extensive molecular changes during myofiber transition, including 3,194 differentially expressed mRNAs, 366 circRNAs, 1,394 lncRNAs, 343 miRNAs, 180 differentially expressed transcription factors, and 2,717 genes with significant alternative splicing. These differentially expressed molecules were associated with multiple signaling pathways involved in myofiber transition, such as the AMPK, calcium, PI3K-Akt, MAPK, Hippo, and mTOR pathways. Comprehensive co-expression and protein-protein interaction analyses identified an active interconnected module containing 38 co-expressed proteins related to myofiber transition. Based on the ceRNA (competitive endogenous RNA) theory and these 38 key molecules, a lncRNA/circRNA-miRNA-mRNA network was constructed, involving 9 mRNAs, 10 circRNAs, 18 lncRNAs, and 14 miRNAs.

Conclusions

This study investigated how the relative composition and content of myofiber types change during different growth stages of rabbits, and revealed the complex dynamic biological mechanisms underlying myofiber transition in rabbits through whole transcriptomics. The results of this study can help identify appropriate targets for regulating myofiber transition, thereby facilitating the development of high-quality rabbit meat.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12864-026-12587-0.

Background

As living standards improve, the obesity rate is gradually rising, with approximately 603.7 million adults globally classified as obese [1]. Obesity-related metabolic diseases increasingly impact people’s health and quality of life [2]. Rabbit meat is a healthy option that is high in protein, low in fat, and rich in beneficial nutrients, making it particularly suitable for individuals looking to lose weight or for those with hypertension, hyperglycemia, and hyperlipidemia. However, the low-fat content of rabbit meat can result in a lack of flavor, negatively affecting its market sales.

Myofiber type proportion is a key determinant of rabbit meat quality, particularly affecting flavor and tenderness. The quality of fresh rabbit meat largely depends on the biochemical changes that occur in muscle during its conversion to meat. Skeletal muscle is an extremely heterogeneous tissue composed of different types of fibers with varying contraction and metabolic characteristics. It is classified into slow-twitch (red muscle fibers, type I fibers) and fast-twitch (white muscle fibers, type II fibers) fibers. Based on histochemical methods, muscle fibers can be categorized into type I fibers (with lower ATPase activity) and type II fibers (with higher ATPase activity) [3]. According to myosin heavy chain (MyHC) subtypes, muscle fibers can be further divided into I, IIA, IIX, and IIB fiber types [4]. Additionally, skeletal muscle is a dynamic tissue, and the composition of fiber types exhibits significant variation both between muscles and among different animals [5], influenced by factors such as breed [6], genetics [7], age [8], and nutrition [9]. The relative abundance of each fiber type in muscle tissue is associated with the development of final meat quality. Research indicates that red muscle fibers (oxidative fibers) are positively correlated with higher tenderness, intramuscular fat, water-holding capacity, brightness, and color, while white muscle fibers (glycolytic fibers) are associated with higher shear force and brightness [10–12]. Moreover, the high myoglobin content and flavor-enhancing amino acids in red muscle fibers particularly enhance the color and flavor of meat [13, 14]. Therefore, gradually improving fresh meat quality through the regulation of the relative abundance of muscle fiber types is an effective approach to enhancing meat quality.

In recent years, advancements in transcriptomics, genomics, metabolomics, and other methodologies have gradually elucidated the molecular and cellular mechanisms that control the interconversion of muscle fibers [15, 16]. However, these mechanisms remain largely unknown and quite complex. This study focuses on rabbits of different ages, aiming to investigate how the relative composition of muscle fiber types varies with growth by comparing fiber type compositions at different growth stages. Utilizing whole transcriptome technology, we further explore the potential molecular characteristics involved in the interconversion of rabbit muscle fibers. This research aims to reveal the RNA transcriptional regulatory mechanisms during the conversion process of rabbit muscle fibers at the transcriptional level and clarify the key activation network mechanisms involved in muscle fiber transformation. Ultimately, this study seeks to identify target genes for fiber type regulation to produce high-quality fresh rabbit meat.

Materials and methods

Animal and sample collection

In this study, we used three healthy male Hyla rabbits at each of the following ages: 1, 7, 14, 21, 28, 35, and 75 days. All rabbits were fed standard commercial rabbit feed and allowed to eat freely. They were housed in individual cages (80 cm × 60 cm × 40 cm) under controlled conditions (temperature: 22 ± 2℃, humidity: 55 ± 5%, light cycle: 12 h light / 12 h dark). Prior to sample collection, the rabbits were fasted for 12 h and deprived of water. Rabbits were euthanized by intravenous injection of sodium pentobarbital (50 mg/kg body weight) via the ear vein, and death was confirmed by the absence of corneal reflex and respiratory arrest. Then, we collected samples of the gluteus muscle (GLU) and gastrocnemius muscle (GAS). Some samples were designated for electronic tongue analysis, while others were fixed in formaldehyde for immunofluorescence analysis. Additional samples were rapidly cooled in liquid nitrogen and stored at -80 °C for RNA extraction and ATPase staining. All animal studies adhered to the guidelines established by the Ethics Committee of Northwest A&F University (XN2024-0335).

ATPase stain

Samples were embedded in OCT compound (Sakura, Japan) and sectioned into 10-µm-thick slices using a POLAR DM frozen microtome (Sakura, Japan) at -20℃. Staining of the frozen sections was performed according to the instructions provided in the ATPase staining kit (Solarbio, Beijing, China). Briefly, the sections were covered with pre-warmed acidic pre-incubation solution (0.1 mol/L acetate-acetic acid buffer, pH 4.3) and incubated for 15 min. After discarding the liquid, the pre-warmed ATPase staining working solution was added, and the sections were incubated at 37 °C for 45 min. The sections were then washed three times with 1% calcium chloride solution for 1 min each, and excess liquid was removed. A suitable amount of Co solution was added to cover the sections for 3 min, followed by four washes with distilled water for 3 min each. The sections were subsequently incubated in the color development working solution for 2 min until a uniform dark brown color was achieved. Finally, the sections were washed four times with distilled water for 3 min each, and after removing excess moisture, they were mounted with glycerol gelatin. Observations were made using an optical microscope (SMZ25, Nikon, Japan).

Immunofluorescence

The immunofluorescence method was based on previous laboratory protocols [17]. Briefly, samples were embedded and sectioned. After sectioning, a hydrophobic barrier pen was used to draw a circle around the tissue. The sections were then blocked with BSA for 30 min. Following the removal of the blocking solution, the prepared primary antibody (GB112130, 1:5000, Servicebio, Wuhan, China) was added, and the sections were incubated overnight at 4 °C. The sections were washed with PBS and incubated with the corresponding HRP-conjugated secondary antibody (GB23303, 1:500, Servicebio, Wuhan, China) at room temperature for 50 min. TSA dye was then added, and the sections were incubated for 10 min. Subsequently, the sections were placed in antigen retrieval buffer and microwaved for 7 min. The second primary antibody (GB112131, 1:500, Servicebio, Wuhan, China) was added, and the sections were incubated overnight at 4 °C. The corresponding secondary antibody (GB27303, 1:400, Servicebio, Wuhan, China) was then applied and incubated in the dark at room temperature for 50 min. Afterward, the sections were stained with DAPI and mounted with a drop of antifade mounting medium. Finally, the sections were observed under a fluorescence microscope (Eclipse C1, Nikon, Japan).

Electronic tongue analysis

Eighty grams of samples from both the 75-day-old GLU and GAS muscles were weighed and placed in a water bath at 40 °C until the internal temperature of the meat reached 40 °C. The samples were then transferred to a blender, and pre-warmed distilled water at 40 °C was added at a mass-to-volume ratio of 1:5, followed by homogenization for 1 min. The mixture was filtered through a funnel equipped with a 0.45 μm water-based filter membrane to obtain the supernatant. Subsequently, the taste characteristics were analyzed using the French α-ASTREE electronic tongue system (α-ASTREE, AlphaMOS, France), which is equipped with seven sensors (ZZ, JE, BB, CA, GA, HA, JB).

RNA extraction, library construction, and sequencing

Samples of GLU muscle tissue from 1-day and 14-day-old Hyla rabbits were used for total RNA extraction with Trizol reagent (Invitrogen, Carlsbad, CA, USA), following the manufacturer’s protocol. The quantity and purity of the total RNA were analyzed using a Bioanalyzer 2100 and RNA 6000 NanoLabChip Kit (Agilent, CA, USA, 5067 − 1511), ensuring a RIN number greater than 7.0. Approximately 5 µg of total RNA was treated with the Ribo-Zero Gold rRNA Removal Kit (Illumina, cat. MRZG12324, San Diego, USA) to remove ribosomal RNA. The remaining RNA was processed into short fragments at high temperature using the NEBNext^® Magnesium RNA Fragmentation Module (NEB, cat. E6150S, USA) and subsequently reverse transcribed to generate six libraries for mRNA, lncRNA, and circRNA. Paired-end sequencing (PE150) was performed on the Illumina Novaseq™ 6000 platform (LC-Bio Technology CO., Ltd., Hangzhou, China). The miRNA libraries were prepared using the TruSeq Small RNA Sample Prep Kits (Illumina, San Diego, USA). Following the preparation of the six miRNA libraries, sequencing was conducted using the Illumina HiSeq 2000.

Identification and analysis of mRNAs, LncRNAs and circrnas

Reads were first filtered to remove poor-quality sequences using Cutadapt [18] (https://cutadapt.readthedocs.io/en/stable/,version:cutadapt-1.9). Sequence quality was then verified using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/,version0.11.9), which included assessments of Q20, Q30, and the GC content of the clean data. The reads were aligned to the rabbit (Oryctolagus cuniculus) reference genome (GCF_964237555.1) and assembled into transcripts using Hisat2 [19]. Transcript assembly and definition were performed with StringTie [20], comparing against annotation files to identify known mRNAs and lncRNAs. The expression levels of mRNAs were measured using FPKM (Fragments Per Kilobase Million) [21]. Differentially expressed genes between the two groups were analyzed using DESeq2, with transcripts meeting the criteria of |log₂FC| ≥ 1 and P < 0.05 classified as significantly different genes.

To identify we first removed, transcripts overlapping with known mRNAs, those shorter than 200 bp, and those with exon counts ≤ 1. The remaining transcripts were then assessed for coding potential using CPC (Coding Potential Calculator) [22] and CNCI (Coding-Non-Coding Index) [23]. LncRNAs were identified by combining CPC < -1, CNCI < 0, and FPKM > 0.5 (expressed in at least 3 samples) to exclude low-expression transcripts without biological significance. Differentially expressed lncRNAs (DElncRNAs) were analyzed using DESeq2 [24], applying the criteria |log₂FC| ≥ 1 and P < 0.05. The subcellular localization of lncRNA was predicted using three software tools: RNAlight, iLoc_LncRNA, and DeepLncLoc [25–27].

CircRNAs were identified by intersecting results from CIRCexplorer2 and CIRI tools to reduce false positives [28, 29]. CircRNA expression levels were analyzed using srpbm (spliced reads per billion mapping) [30]. Differential expression analysis of circRNAs was conducted using edgeR, with |log₂FoldChange| ≥ 1 and P-value < 0.05 defining differentially expressed circRNAs.

Identification and analysis of MiRNAs

Raw reads were processed using the internal program ACGT101-miR (v4.2) [31] to remove adapter dimers, contaminants, low-complexity sequences, common RNA families (including rRNA, tRNA, snRNA, and snoRNA), and repetitive sequences. Unique sequences ranging from 18 to 26 nucleotides were then mapped to specific precursor sequences in miRbase 22.1 through BLAST searches to identify known miRNAs, as well as novel 3p- and 5p-miRNAs. A P-value calculation model based on a normal distribution was employed to compute P-values, and a t-test was used to analyze differences in miRNA expression between the two sample groups. miRNAs with |log₂FC| > 1 and P < 0.05 were considered differentially expressed miRNAs [32]. The target genes of the significantly differentially expressed miRNAs were predicted using TargetScan (score ≥ 50) and miRanda (energy < -10) [33, 34], and the intersection of results from both tools was used to reduce false positives.

Transcription factor and alternative splicing analysis

Transcription factors are a diverse family of proteins that typically act in multisubunit complexes to activate or repress gene transcription. We annotated transcription factors for genes using AnimalTFDB database, based on gene ID and gene name, and counted the family distribution of differentially expressed transcription factors. rMATS (version 4.1.1) (http://rnaseq-mats.sourceforge.net) was employed to identify alternative splicing events and analyze differential splicing events between samples [35]. We considered alternative splicing (AS) events with a false discovery rate (FDR) < 0.05 in a comparison as significant AS events.

DElncRNA and DEmRNA co-expression analysis

Based on the expression levels of DElncRNA and DEmRNA, we filtered out all protein-coding genes and lncRNAs with expression levels below 0.5 across all samples, including those without expression in one-third of the samples. The Pearson correlation between lncRNAs and protein-coding genes was calculated using the Hmisc package, with screening thresholds set at |cor| ≥ 0.95 and P < 0.05. Subsequently, functional enrichment analysis was performed for mRNAs co-expressed with lncRNAs, allowing for an examination of the relationships among lncRNAs, co-expressed mRNAs, and functional annotations.

Function enrichment analysis

We conducted enrichment analysis of GO terms and KEGG pathways based on the differentially expressed mRNAs (DEmRNAs) using the ClusterProfiler 4.0 R package, considering a threshold of P < 0.05 as significant. The enriched GO terms and KEGG pathways were then visualized using the ggplot2 package.

Gene set enrichment analysis (GSEA)

Gene set enrichment analysis was performed using GSEA software (v4.1.0) [36]. Briefly, we input the gene expression matrix and ranked genes using the Signal2Noise normalization method. Enrichment scores and P-values were calculated using default parameters. GO terms and KEGG pathways that met the criteria of |NES|>1, NOM P-value < 0.05, and FDR P-value < 0.25 were considered significantly different between the two groups.

PPI analysis and CeRNA network construction

To identify key genes involved in muscle fiber transformation, we conducted Pearson correlation analysis based on the expression levels of differentially upregulated genes, selecting pairs with |r| > 0.8 and P-value < 0.05. Interaction predictions for the differentially upregulated genes were performed using the STRING database (https://cn.string-db.org/). We obtained the intersection of the identified relationships and extracted genes within these pairs as key genes. Based on the selected key genes from the PPI analysis, we predicted target binding relationships between miRNAs and mRNAs, lncRNAs, and circRNAs using TargetScan (v5.0) and miRanda (v3.3a). Correlation analyses were then conducted on the expression levels of DEmRNAs, DEmiRNAs, DElncRNAs, and DEcircRNAs based on common miRNA binding ceRNA relationships. Pairs of miRNA-mRNA and miRNA-lncRNA with a correlation coefficient (r) < -0.9, as well as mRNA-lncRNA pairs with r > 0.9, were intersected to construct lncRNA-miRNA-mRNA networks. Similarly, circRNA-miRNA-mRNA networks were constructed. Utilizing Cytoscape software, we created network diagrams to visualize the regulatory relationships.

RT-qPCR

RT-qPCR was performed to validate the accuracy and reliability of the RNA sequencing data, using the same RNA samples employed in RNA-Seq for cDNA synthesis. From the identified differentially expressed transcripts, seven mRNAs, three lncRNAs, and four miRNAs were randomly selected for primer design (Supplementary Table S1). RT-qPCR was conducted using the 2×ChamQ SYBR qPCR Master (Takara, Kyoto, Japan), and the quantitative results were analyzed to determine the relative expression levels of the selected differentially expressed genes using the 2^−ΔΔCt method.

Statistical analysis

All statistical analyses and figures generated in this study were performed using GraphPad Prism (v8.0.2), SPSS (v15.0), and OriginPro (2019b). Data are presented as mean ± SEM, and results were compared using paired t-tests and one-way analysis of variance (ANOVA). A significance threshold of P < 0.05 was defined (*P < 0.05, **P < 0.01, ***P < 0.001). Different lowercase letters (a, b, c) indicate significant differences among groups (P < 0.05).

Result

Analysis of relative muscle fiber composition at different growth stages in rabbits

The relative composition of muscle fibers is a key factor influencing the quality of fresh meat. To examine whether the relative composition of muscle fibers affects rabbit meat quality, we collected samples from the GLU and GAS muscles. We first assessed the proportion of type I and type II muscle fibers in these samples using immunofluorescence. As shown in Fig. 1A and B, the GLU muscle contains 99.23% type II fibers and 0.77% type I fibers, while the GAS muscle has 83.61% type II fibers and 16.39% type I fibers. The a* value (redness) of GAS muscle was 8.2 ± 0.9, significantly higher than that of GLU muscle (0.9 ± 0.1, P < 0.01) (Fig. 1C). PCA results from an electronic tongue analysis showed that the first principal component accounted for 81.8% of the variance, while the second principal component accounted for 16%, with the combined contribution of the first two components exceeding 90%. Notable differences in flavor characteristics were observed between the GLU and GAS muscles (Fig. 1D). As illustrated in Fig. 1E, the flavor profiles of the GLU and GAS muscles differ significantly, likely due to the varying proportions of muscle fibers. The distinct compositions of muscle fibers in the GLU and GAS muscles result in significant variations in overall flavor compounds, consistent with the PCA results. Previous laboratory findings have indicated that differences in the content of type I and type II muscle fibers significantly impact the flavor quality of rabbit meat, with a higher red fiber content correlating with better flavor quality [17]. However, there is a lack of research on how rabbit muscle fibers change during growth and how myofibers transform into one another. To explore these changes, we collected frozen sections of the GLU muscle from Hyla Rabbits at different ages and employed ATPase staining to identify muscle fiber types. The study indicated that as age increased, the proportion of type I muscle fibers gradually decreased while type II fibers increased. A significant reduction in type I fibers was observed at 14 days of age, decreasing from 97.17% to 27.0%, while type II fibers significantly increased from 2.83% to 73.0% (Fig. 1F and G). To investigate the molecular mechanisms underlying muscle fiber transformation, we selected GLU muscle samples from rabbits at 1 and 14 days of age for comprehensive transcriptomic analysis, aiming to explore changes in gene expression and regulatory networks of total RNA during muscle fiber transformation.

Fig. 1.

Analysis of Muscle Fiber Composition during Growth in Rabbits. A Immunofluorescence of fast and slow fibers in the rabbit GLU and GAS muscles. DAPI (blue) indicates cell nuclei; Slow Fibers (red) indicate oxidative muscle fibers; Fast Fibers (green) indicate glycolytic muscle fibers. Scale bar = 50 μm. B Quantification of the proportion of fast and slow muscle fibers in the rabbit GLU and GAS muscles (**P<0.01). C Measurement of muscle color in the rabbit GLU and GAS muscles, The A value represents the redness of the meat. D Principal component analysis (PCA) of electronic tongue data for the GLU and GAS muscles; Black dots represent GLU muscles; red dots represent GAS muscles. E Radar chart of sensor response values from the electronic tongue for the GLU and GAS. F ATP staining of GLU fibers in rabbits of different ages. In the ATP staining results, the red arrows indicate type I muscle fibers, while the white arrows indicate type II muscle fibers. The scale is 100 μm. G Quantitative statistical plots of ATP-stained muscle fibers of both type I and type II, different letters (a, b, c) indicate significant differences among groups (P < 0.05)

Identification of differential mRNAs during myofiber transition in rabbits

We conducted RNA-seq analysis on GLU muscle samples from rabbits at 1 and 14 days of age, constructing six cDNA libraries to represent each age group. Each library yielded between 74,115,540 and 86,493,570 raw reads. After filtering out low-quality sequences using Cutadapt (Q20 > 98.84%, Q30 > 95.43%), the average number of clean reads was 70,141,308 for the 1-day group and 77,396,419 for the 14-day group (Supplementary Table S2). The clean reads were aligned to the reference genome using HISAT2, with the percentage of high-quality sequence reads ranging from 90.70% to 92.94%. A total of 85.17% to 89.20% of the tags across all libraries were uniquely mapped to the reference genome (Supplementary Table S3). Transcript reconstruction performed with StringTie identified a total of 33,433 transcripts.

Gene expression levels were analyzed for distribution and correlation. The gene expression profiles of GLU muscle samples between the 1-day and 14-day age groups showed significant divergence, while the biological replicates within each group exhibited high consistency (Fig. 2A, Supplementary Figure S1). Differential gene expression analysis was performed using DESeq2, applying thresholds of |log₂ FC| ≥ 1 and P < 0.05, which revealed a total of 3,194 differentially expressed genes (DEGs) between the two age groups. Among these, 517 genes were upregulated and 2,677 were downregulated, with a strong correlation observed in the expression patterns of the DEGs (Fig. 2B and D).

Fig. 2.

Identification of differentially expressed mRNAs (DEmRNAs) during muscle fiber transformation in rabbits. A Pearson correlation coefficient plot. Both the x-axis and y-axis represent various samples, with color intensity indicating the magnitude of the correlation coefficient between samples. B Heatmap of differentially expressed genes. The x-axis represents the samples, while the y-axis represents the genes; Different colors represent different levels of gene expression, with the color ranging from blue through white to red indicating expression levels from low to high; red represents highly expressed genes, and blue represents lowly expressed genes; scale range: -1 to 1.5. C Bar chart of overall statistics for differentially expressed genes. Red indicates upregulated genes, while blue indicates downregulated genes. D Volcano plot of differentially expressed genes. The x-axis represents the fold change in gene expression between different samples, while the y-axis represents the statistical significance of changes in gene expression levels; red dots indicate significantly upregulated genes, blue dots indicate significantly downregulated genes, and gray dots represent genes with non-significant differential expression. E Bar chart of GO enrichment for upregulated and downregulated genes. The x-axis indicates the number of differentially expressed genes enriched in GO terms, with upregulated and downregulated genes represented in red and blue, respectively, while the y-axis represents the GO terms. F Bar chart of KEGG pathway enrichment analysis. The y-axis lists the pathway names, while the x-axis represents the -log₁₀ P-values for the enrichment analysis of the KEGG pathways

Functional enrichment analysis of the DEGs was conducted using Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses. The GO terms for the DEGs showed significant enrichment in biological process (BP) categories, particularly in “regulation of DNA-templated transcription” and “DNA-templated transcription” (Fig. 2E). In the cellular component (CC) category, significant functions included “cytoplasm” and “nucleoplasm,” while the molecular function (MF) category exhibited enrichment in “protein binding,” “metal ion binding,” “zinc ion binding,” and “ATP binding.” KEGG pathway analysis identified the top 20 enriched pathways related to muscle fiber transformation, including the calcium signaling pathway, Notch signaling pathway, p53 signaling pathway, and Hippo signaling pathway (Fig. 2F).

Changes in transcription factors and mRNA alternative splicing analysis during rabbit myofiber transition

To further elucidate potential upstream regulatory molecules involved in genes associated with biological processes, we annotated transcription factors using the AnimalTFDB database, identifying a total of 828 transcription factors, including 180 differentially expressed transcription factors (as shown in Fig. 3A and Supplementary Table S4). The transcription factor families primarily included zf-C2H2, Homeobox, and ZBTB, among others (Fig. 3B).

Fig. 3.

Analysis of transcription factors and mRNA alternative splicing during muscle fiber transformation in rabbits. A Heatmap of transcription factors. The x-axis represents the samples while the y-axis represents the genes. Different colors indicate varying levels of gene expression, with red signifying high-expression genes and blue indicating low-expression genes. B Bar chart of transcription factor family distribution. The x-axis shows the number of differentially expressed transcription factors, and the y-axis represents the families to which these transcription factors belong. C Pie chart of alternative splicing (AS) types. This chart illustrates the proportion of various alternative splicing events. D Bar chart of differential AS statistics. The x-axis represents the types of AS, while the y-axis indicates the number of significantly different AS events. E Visualization of alternative splicing. At the bottom of the figure, a schematic diagram of the alternative splicing isoforms is presented. RPKM (Reads Per Kilobase Million) values quantify the distribution of sequencing depth in the samples, with different sample groups represented in distinct colors. F Bubble chart of GO enrichment for differentially spliced genes. The x-axis (Rich factor) represents the proportion of differentially expressed genes within the GO term to the total number of genes in that GO term. The y-axis lists the GO functional annotations, with bubble size representing the number and bubble color indicating the P-value of the enrichment analysis. G Bar chart of KEGG pathway enrichment for differentially spliced genes. The x-axis indicates the number of differentially expressed genes included in each pathway, while the y-axis lists the pathway names

To investigate alternative splicing (AS), we utilized rMATS to detect five types of splicing events: skipped exons (SE), alternative 5’ splicing sites (A5SS), alternative 3’ splicing sites (A3SS), mutually exclusive exons (MXE), and retained introns (RI). A total of 34,235 genes exhibited alternative splicing, with the distribution of events as follows: A3SS (9.15%), A5SS (6.89%), MXE (9.60%), RI (8.42%), and SE (65.94%) (Fig. 3C). Using a threshold of false discovery rate (FDR) < 0.05 and |IncLevelDifference| > 0.1, we identified 2,717 significantly differentially alternative splicing genes (Fig. 3D), with distinct splicing patterns observed in the GLU muscle between 14-day-old and 1-day-old samples. For instance, the YBX3 gene exhibited different splicing forms at 14 days compared to 1 day of age (Fig. 3E).

To further investigate the functional implications of these differentially spliced genes on muscle activity, we performed GO and KEGG enrichment analyses on genes with differential splicing events (based on JC). The top 20 enriched GO terms were predominantly associated with nucleoplasm, protein binding, nucleus, cytoplasm, nucleotide binding, and ATP binding (Fig. 3F). KEGG pathway analysis revealed significant enrichment in signaling pathways such as the Wnt signaling pathway, PI3K-Akt signaling pathway, and MAPK signaling pathway, all of which are involved in muscle fiber transformation (Fig. 3G). These results indicate that genes modify their functions through alternative splicing to facilitate muscle fiber transformation.

Screening of key differential genes involved in the myofiber transition process

To elucidate the key activating factor network involved in the transformation of rabbit muscle fibers, we employed a network-based approach (Fig. 4A). A Pearson correlation analysis was conducted on 517 differentially upregulated genes, identifying co-expression relationships with |r| > 0.8 and P-value < 0.05, resulting in a total of 145,118 absolute correlation pairs. Concurrently, protein-protein interaction (PPI) predictions for the 517 genes were performed using the STRING database, leading to the identification of 742 interaction pairs. Based on the intersection of co-expression relationships and PPI pairs, we selected 223 activation protein interaction pairs relevant to muscle fiber transformation.

Fig. 4.

Identification of key Networks Associated with Muscle Fiber Transformation. A Flowchart detailing the process used to construct the key active subnet. The Pearson correlation coefficients of all pairs among 517 annotated differentially expressed mRNAs (DE mRNAs) were calculated. Protein-protein interaction pairs were filtered from the STRING database (https://cn.string-db.org/). B KEGG pathway analysis of the mRNAs constituting the key active subnet. The x-axis (Rich factor) represents the proportion of differentially expressed genes within each pathway relative to the total number of genes in that pathway; the y-axis lists the KEGG pathways. Bubble size corresponds to the number of genes, and bubble color represents the P-value of the enrichment analysis. C PPI Interaction Network Diagram of 38 Key Genes. D Bar chart of KEGG enrichment analysis. The x-axis represents the normalized enrichment scores (NES) of the gene sets, while the y-axis lists the gene set names from the KEGG database, with colors representing P-values or false discovery rate (FDR) values. E Line graph of KEGG enrichment scores (ES). The x-axis represents the ranked list of all genes; The y-axis denotes the enrichment score (ES); The peak ES value corresponds to the core gene set driving pathway enrichment. F Gene Expression Profiles of Glycolysis/Gluconeogenesis, Fructose and Mannose Metabolism, and Glycine, Serine, and Threonine Metabolism Pathways

Subsequently, KEGG pathway analysis of these 223 genes revealed significant enrichment in pathways related to glycolysis/gluconeogenesis, fructose and mannose metabolism, glycine, serine, and threonine metabolism, calcium signaling, and AMPK signaling pathway (Fig. 4B). From the key genes enriched in pathways associated with muscle fiber transformation (calcium signaling, and AMPK signaling), we identified a total of 38 critical genes. PPI analysis of these 38 key genes demonstrated strong interactions among them (Fig. 4C).

Additionally, through gene set enrichment analysis (GSEA) of mRNA, we found that the glycolysis/gluconeogenesis, fructose and mannose metabolism, and glycine, serine, and threonine metabolism pathways were significantly activated in 14-day-old rabbits (Fig. 4D and E), with a notable increase in the associated genes (Fig. 4F). In our previous research on the metabolic differences between red and white muscle, we found that pathways related to glycolysis/gluconeogenesis, fructose and mannose metabolism, and glycine, serine, and threonine metabolism were significantly enriched in white muscle [17]. This further indicates that the GLU muscle of rabbits gradually transforms from red muscle to white muscle during growth.

Identification and characterizing LncRNAs in rabbit muscle fiber transformation process

First, known mRNAs and transcripts shorter than 200 bp were removed. The remaining novel transcripts were then subjected to coding potential prediction using CPC and CNCI software, resulting in the identification of 6,512 novel lncRNAs (Fig. 5A). Subsequently, subcellular localization of lncRNAs was predicted using three software programs: RNAlight, iLoc_LncRNA, and DeepLncLoc. Venn diagram analysis revealed the overlap between cytoplasmic and nuclear localization: 4,224 lncRNAs were found to be located in the cytoplasm, while 500 were located in the nucleus (Fig. 5B and C).

Fig. 5.

Analysis of DELncRNA during muscle fiber transformation in rabbits. A Venn diagram illustrating the predictions of CPC and CNCI for lncRNA. B Venn diagram analyzing the intersections of predicted lncRNAs in the cytoplasm using RNAlight, iLoc_LncRNA, and DeepLncLoc. C Venn diagram analyzing the intersections of predicted lncRNAs in the nucleus using the same three software tools. D Bar chart summarizing the overall statistics of differentially expressed lncRNAs. Red bars indicate upregulated lncRNAs, while blue bars indicate downregulated lncRNAs. E Clustering heatmap of differentially expressed lncRNAs. The x-axis represents the samples, and the y-axis represents lncRNA names. Red indicates high-expression lncRNAs, while blue indicates low-expression lncRNAs; scale range: -1.5 to 1.5. F Volcano plot of differentially expressed lncRNAs. The x-axis represents the fold change of lncRNA expression across different samples, while the y-axis indicates the statistical significance of gene expression changes. Red points represent significantly upregulated lncRNAs, blue points represent significantly downregulated lncRNAs, and gray points represent lncRNAs with non-significant differential expression. G Co-expression network diagram of lncRNAs and protein-coding genes. “Positive: Pearson correlation coefficient r > 0.95 and P < 0.05, indicated by blue lines; Negative: r < -0.95 and P < 0.05”, indicated by orange lines. H Bar chart of GO enrichment analysis for co-expressed differentially expressed mRNAs (DEmRNAs). The x-axis represents the GO terms, while the y-axis indicates the number of differentially expressed genes enriched in each GO term. I Bubble chart of KEGG enrichment analysis for co-expressed DEmRNAs. The x-axis (Rich factor) represents the proportion of differentially expressed genes within each pathway relative to the total number of genes in that pathway; the y-axis lists the KEGG pathways. Bubble size corresponds to the number of genes, and bubble color represents the P-value of the enrichment analysis

Expression levels of lncRNAs were quantified using FPKM to create a lncRNA expression profile, followed by differential analysis. Using thresholds of |log₂FC| ≥ 1 and P < 0.05, a total of 1,394 lncRNAs were found to be significantly differentially expressed (Fig. 5F), including 317 lncRNAs that were upregulated and 1,077 lncRNAs that were downregulated (Fig. 5D). The differential expression trends of lncRNAs among groups were similar (Fig. 5E).

Next, co-expression analysis was performed between differentially expressed lncRNAs and differentially expressed protein-coding genes, using |cor| ≥ 0.95 and P < 0.05 criteria. This resulted in the identification of 1,048,576 strong co-expression relationships. The three lncRNAs with the highest connectivity (MSTRG.12175.1, MSTRG.16233.1, MSTRG.753.1) were selected, and their co-expression relationships with 2,059 differentially expressed mRNAs were visualized in a network diagram (Fig. 5G).

Furthermore, Gene Ontology (GO) and KEGG enrichment analyses were conducted on the genes highly correlated with lncRNAs to explore the biological processes and pathways potentially involved in their co-expression with protein-coding genes. The GO enrichment analysis revealed that differentially expressed lncRNAs were significantly enriched in the canonical Wnt signaling pathway and glycolytic process (biological process category), as well as in the nucleus and cytoplasm (cellular component category), and in protein binding and ATP binding (molecular function category) (Fig. 5H). The KEGG enrichment analysis indicated that differentially expressed lncRNAs primarily participate in several signaling pathways, including fructose and mannose metabolism, glycolysis/gluconeogenesis, the p53 signaling pathway, the Hippo signaling pathway, and the Notch signaling pathway (Fig. 5I). These findings suggest that lncRNAs may regulate muscle fiber transformation through pathways associated with muscle differentiation, such as the p53, Hippo, and Notch signaling pathways.

Identification and characterization of miRNAs during muscle fiber transformation in rabbits

In the GLU muscle of rabbits at 1 and 14 days of age, we identified a total of 897 miRNAs (Supplementary Table S5). Correlation analysis was then conducted based on the expression levels of miRNAs from the samples. The Pearson correlation coefficients and Principal Component Analysis (PCA) results indicated significant differentiation in the miRNA expression profiles between different age groups of rabbits (Fig. 6A and B). The samples demonstrated a high correlation coefficient, and the expression levels clustered well.

Fig. 6.

Analysis of Differential miRNAs during Muscle Fiber Transformation in Rabbits. A Principal Component Analysis (PCA) of differentially expressed miRNAs. B Pearson correlation coefficient heatmap. Both the x-axis and y-axis represent the various samples, with color intensity indicating the strength of correlation between pairs of samples. A deeper red color (coefficients closer to 1) indicates a stronger correlation, while lighter colors approaching white signify weaker correlations. C Bar chart summarizing the statistics of upregulated and downregulated miRNAs. Red bars represent upregulated miRNAs, and blue bars represent downregulated miRNAs, with the numbers indicating the count of each category. D Clustering heatmap of differentially expressed miRNAs. The x-axis represents the samples, while the y-axis represents the miRNAs. Different colors indicate varying expression levels, transitioning from blue to white to red, reflecting expression levels from low to high. Red indicates high-expression miRNAs, while deep blue indicates low-expression miRNAs; scale range: -1 to 1.5. E Volcano plot of differentially expressed miRNAs. The x-axis represents the fold change of miRNA expression across different samples, while the y-axis indicates the statistical significance of changes in miRNA expression levels. Red points represent significantly upregulated miRNAs, blue points indicate significantly downregulated miRNAs, and gray points denote miRNAs with non-significant differential expression. F Bar chart of GO enrichment analysis for differentially expressed miRNAs. The x-axis lists the GO terms, while the y-axis shows the number of differentially expressed genes enriched in each GO term. G Bubble chart of KEGG enrichment analysis for differentially expressed miRNAs. The x-axis (Rich factor) represents the proportion of differentially expressed genes within each pathway relative to the total number of genes in that pathway, the y-axis lists the KEGG pathways. Bubble size corresponds to the number of genes, and bubble color indicates the P-value of the enrichment analysis

Differentially expressed miRNAs were filtered using a threshold of P-value < 0.05. We identified a total of 343 miRNAs with significant differential expression (P-value < 0.05), of which 208 miRNAs were upregulated and 135 miRNAs were downregulated (Fig. 6C and E). The expression patterns of the differentially expressed miRNAs were found to be relatively similar (Fig. 6D).

Subsequently, we utilized TargetScan (v5.0) and miRanda (v3.3a) software to predict target genes for the significantly differentially expressed miRNAs, applying the criteria of TargetScan_score ≥ 50 and miranda_Energy < -10. A total of 296,932 target genes were predicted, which were then subjected to GO/KEGG functional annotation. The GO enrichment results revealed that processes such as positive regulation of transcription by RNA polymerase II, DNA-templated transcription, and regulation of DNA-templated transcription were prominently represented in the biological process (BP) category (Fig. 6F). In the cellular component (CC) category, structures such as the cytoplasm, nucleus, and cytosol were highlighted. The molecular function (MF) category mainly included functions related to protein binding, metal ion binding, ATP binding, zinc ion binding, and DNA-binding transcription factor activity.

Additionally, KEGG pathway analysis revealed that these differentially expressed miRNAs were significantly enriched in pathways associated with muscle fiber transformation, including the MAPK signaling pathway, Foxo signaling pathway, Hippo signaling pathway, Wnt signaling pathway, and ErbB signaling pathway (Fig. 6G). These findings suggest that miRNAs play a role in the interconversion of myofibers.

Participation in the identification and characterization of circrna in the transformation of rabbit muscle fibers

Based on the structural features of circRNAs and splicing sequence characteristics, we identified a total of 117,476 circRNAs using CIRCexplorer2 and CIRI software (Supplementary Table S6). Among the identified circRNA types, the results in Fig. 7A show that the majority are exonic circRNAs, followed by intronic circRNAs (ciRNAs) and intergenic circRNAs. CircRNA expression levels were quantified using srpbm, and the distribution of circRNA expression levels across samples is illustrated in Fig. 7B. Overall, the expression levels of circRNAs exhibited low variance within groups, indicating good biological replicates.

Fig. 7.

Analysis of Differential circRNAs during Muscle Fiber Transformation in Rabbits. A Statistical chart of circRNA types; The x-axis represents samples; The y-axis represents the percentage of circRNA types. B Distribution of circRNA expression levels. The boxplot displays the samples on the x-axis and log₁₀(srpbm) on the y-axis. Each boxplot represents five statistical metrics (from top to bottom: maximum, upper quartile, median, lower quartile, and minimum), with scatter points indicating outliers. C Statistics of differentially expressed circRNAs. The red bars indicate the number of upregulated circRNAs, while the blue bars represent the number of downregulated circRNAs. D Clustering analysis of differentially expressed circRNAs. The x-axis represents the samples, while the y-axis represents the circRNAs. Different colors indicate varying expression levels, transitioning from blue to white to red, reflecting expression levels from low to high. Red indicates high-expression circRNAs, while deep blue indicates low-expression circRNAs; scale range: -2 to 1. E Scatter plot of differentially expressed circRNAs. The log₂(srpbm + 1) values for 14-day-old rabbits is plotted on the x-axis, while those of 1-day-old rabbits are plotted on the y-axis. F Bar chart of GO enrichment analysis for differentially expressed circRNAs. The x-axis lists the GO terms, while the y-axis shows the number of differentially expressed genes enriched in each GO term. G Bubble chart of KEGG enrichment analysis for differentially expressed circRNAs. The x-axis (Rich factor) represents the proportion of differentially expressed genes within each pathway relative to the total number of genes in that pathway; the y-axis lists the KEGG pathways. The size of the bubbles corresponds to the number of genes, and the color indicates the P-value of the enrichment analysis

Using thresholds of |log₂FC| ≥ 1 and P-value < 0.05, we filtered out 366 differentially expressed circRNAs, including 81 upregulated and 285 downregulated circRNAs (Fig. 7C and E). The expression patterns of the differentially expressed circRNAs are shown to be relatively similar in Fig. 7D.

GO/KEGG analyses were conducted based on the parent genes of the differentially expressed circRNAs. The GO enrichment analysis results indicate that the biological processes (BP) primarily include skeletal system development, gluconeogenesis, regulation of DNA-templated transcription, and muscle contraction (Fig. 7F). In the cellular component (CC) category, structures such as the nucleus, cytoplasm, membrane, and plasma membrane were highlighted. The molecular function (MF) category mainly encompasses functions related to protein binding, DNA binding, metal ion binding, zinc ion binding, and ATP binding.

KEGG pathway analysis revealed significant enrichment in several signaling pathways, including the insulin signaling pathway, Hippo signaling pathway, MAPK signaling pathway, and ErbB signaling pathway (Fig. 7G).

CeRNA network construction during myofiber transition and qPCR

Based on the detailed ceRNA regulatory mechanisms illustrated in Fig. 8A and C, we constructed an extended regulatory network associated with 38 key genes within the activation-regulatory network. This network focuses on factors that may serve as upstream regulators of the differentially expressed (DE) mRNAs during muscle fiber transformation. A comprehensive approach was employed to identify significantly co-expressed miRNA-mRNA pairs (r < -0.9). By intersecting the results from the TargetScan database with the identified DE miRNAs, we generated 250 putative miRNA-mRNA pairs targeting the 38 DE mRNAs.

Fig. 8.

Construction of the ceRNA Network and qPCR Validation. A/C Flowchart detailing the construction process of the ceRNA network. The Pearson correlation coefficients for cirRNA /lncRNA-miRNA, mRNA-miRNA, and mRNA- cirRNA /lncRNA pairs were calculated. cirRNA /lncRNA-miRNA interactions were screened using the TargetScan and miRanda databases, while mRNA-miRNA interactions were identified using the TargetScan database. B lncRNA-miRNA-mRNA interaction network diagram (green for mRNA, red for lncRNA, yellow for miRNA). D circRNA-miRNA-mRNA ceRNA interaction network diagram (green for mRNA, red for circRNA, yellow for miRNA). E Correlation analysis of lncRNA-miRNA-mRNA network-associated genes with muscle fiber type composition; * indicates P < 0.05; ** indicates P < 0.01. F Correlation analysis of circRNA-miRNA-mRNA network-associated genes with muscle fiber type composition; * indicates P < 0.05; ** indicates P < 0.01. G The expression pattern of differentially expressed genes (DEGs) was detected by RT-qPCR. The horizontal axis indicates gene names, while the vertical axis represents log2 fold change (14 days/1 day)

Further analysis using the TargetScan and miRanda databases identified lncRNA-miRNA regulatory interactions. This was accomplished by intersecting the datasets for DE lncRNAs and DE miRNAs, resulting in 326 miRNA-lncRNA pairs (r < -0.9). Additionally, within the constructed network, a correlation threshold of r > 0.9 was established between mRNAs and lncRNAs. This strategy yielded a lncRNA-miRNA-mRNA network comprising 18 lncRNAs, 9 miRNAs, and 4 mRNAs (COL4A3, RRAGD, MYLK4, MAP2K6) with 35 interactions, indicating the network’s functional relevance to muscle fiber transformation (Fig. 8B). To further validate the association of this network with myofiber transition, we conducted a correlation analysis between the expression levels of 31 genes from the network and the composition of muscle fiber types. As shown in Figs. 4 and 8E mRNAs and 18 lncRNAs exhibited a significant positive correlation with type II muscle fibers and a significant negative correlation with type I muscle fibers. Conversely, 9 miRNAs demonstrated a significant negative correlation with type II muscle fibers and a significant positive correlation with type I muscle fibers. These findings indicate that the selected network genes play a crucial role in the process of myofiber transition.

Similarly, the circRNA analysis produced a circRNA-miRNA-mRNA network consisting of 10 circRNAs, 10 miRNAs, and 7 mRNAs (COL4A3, RRAGD, MAP2K6, CACNA1S, MYLK4, TFRC, SGK1), with 22 interactions (Fig. 8D). 10 circRNAs and 7 mRNAs showed a significant positive correlation with type II muscle fibers and a significant negative correlation with type I muscle fibers. However, the myosin light chain kinase family member 4 transcript variant X3 (MYLK4, XM_008274288.2) exhibited a significant negative correlation with type II muscle fibers and a significant positive correlation with type I muscle fibers, which is associated with the presence of distinct alternative splicing variants of the MYLK4 gene. Additionally, 10 miRNAs displayed a significant negative correlation with type II muscle fibers and a significant positive correlation with type I muscle fibers (Fig. 8F).

To validate the results from the comprehensive transcriptomic sequencing conducted in this study, we randomly selected 14 differentially expressed transcripts from the differential transcriptome data for RT-qPCR analysis. These findings were consistent with the transcriptomic results (Fig. 8G), underscoring the robustness of the transcriptomic data.

Discussion

The quality of fresh meat is a crucial factor influencing livestock and meat production, and producing high-quality fresh meat is essential to meet the nutritional demands of a growing population. Rabbit meat is a healthy option characterized by high protein content, low fat, and abundant unsaturated fatty acids; however, its low-fat nature can lead to suboptimal flavor. Muscle tissue constitutes the primary component of rabbit meat, and the relative composition of muscle fiber types determines the overall biochemical and functional properties of the muscle, which in turn affects the quality of fresh rabbit meat [37]. In this study, we compared the muscle fiber composition and meat quality between the GLU and GAS muscles, finding that differences in muscle fiber composition significantly impact the color and overall flavor of rabbit meat. Previous research in our laboratory has shown that an increased proportion of red muscle fibers can enhance the flavor quality of rabbit meat [17].

However, skeletal muscle is a dynamic tissue, and influenced by numerous factors that can alter its fiber type composition. For instance, dietary supplementation with resveratrol and alpha-lipoic acid can increase the proportions of MyHC-I and MyHC-IIa fibers while reducing MyHC-IIb fibers, thereby improving the muscle’s antioxidant metabolism and enhancing meat flavor and tenderness [38, 39]. Conversely, diets high in fructose and fat can decrease the quantity of type I oxidative slow fibers, leading to their conversion into type II glycolytic fast fibers [40]. Environmental factors also play a role; exposure to cold environments can promote the conversion of type II fibers to type I fibers [41]. At the molecular and cellular level, PGC-1α facilitates the transformation of glycolytic type II fibers into oxidative type I and IIA fibers by enhancing mitochondrial oxidative respiration and fatty acid metabolism, which contributes to the redness of meat [42]. In dwarf chickens, MYOZ3 promotes the expression of the fast muscle fiber marker MYH1F through PPAR signaling, thereby regulating muscle fiber types [43]. Although the molecular and cellular mechanisms controlling fiber type transformation are gradually being elucidated, they remain largely unknown and complex. Here, we employed a comprehensive transcriptomic approach to characterize the RNA changes during muscle fiber transformation, revealing the regulatory network that governs muscle fiber development.

Using whole transcriptomic sequencing data, we identified 3,194 differentially expressed mRNAs (DEmRNAs), 366 differentially expressed circular RNAs (DEcircRNAs), 1,394 differentially expressed long non-coding RNAs (DElncRNAs), 343 differentially expressed miRNAs (DEmiRNAs), 180 differentially expressed transcription factors, and 2,717 significantly differentially alternatively spliced genes in the GLU muscle at 1 and 14 days of age. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) annotations revealed that the differentially expressed genes were significantly enriched in pathways related to calcium signaling, Notch signaling, p53 signaling, and Hippo signaling. The Hippo signaling pathway is known to regulate muscle hypertrophy, atrophy, muscle fiber formation, and transitions between different fiber types [44]. The Notch signaling pathway is implicated in muscle growth and atrophy [45]. We then conducted co-expression and protein-protein interaction (PPI) analyses on the 517 upregulated genes, constructing a network that included 38 key genes associated with muscle fiber transformation pathways, which were significantly enriched in the calcium signaling pathway and AMPK signaling pathway. Previous reports indicate that these pathways can promote the transition of skeletal muscle fiber types from type II to type I [46, 47].

The whole transcriptomic analyses performed focus on the differentially expressed non-coding RNAs (ncRNAs) during muscle fiber transformation, including circRNAs, lncRNAs, and miRNAs, to determine how they influence myofiber transition. miRNAs can inhibit protein translation and have been shown to play a significant role in the muscle fiber transformation process [48, 49]. In total, we identified 343 differentially expressed miRNAs and predicted their biological functions through pathways associated with muscle fiber transformation, such as the MAPK signaling pathway. Additionally, we identified 1,394 differentially expressed lncRNAs, which is noteworthy because lncRNAs have previously been described as playing a role in muscle fiber transformation [50]. Circular RNAs (circRNAs) are endogenous, covalently closed non-coding RNAs that participate in the development and transformation of muscle fibers in livestock by sponging miRNAs and regulating the transcription and translation of host genes [51, 52]. In this study, we identified 366 differentially expressed circRNAs, and through GO/KEGG analysis based on their parental genes, we found that they primarily influence muscle fiber transformation via the Hippo signaling pathway, MAPK signaling pathway, and ErbB signaling pathway.

Based on the competitive endogenous RNA (ceRNA) theory, we constructed lncRNA-miRNA-mRNA and circRNA-miRNA-mRNA networks using the transcriptomic data generated in this study, which included 3,194 differentially expressed circRNAs (DEmRNAs, ) 343 differentially expressed miRNAs (DEmiRNAs), 366 differentially expressed circRNAs (DEcircRNAs), and 1,394 differentially expressed lncRNAs (DElncRNAs). By focusing on the 38 key activator subnetworks identified in this study, we analyzed potential interactions among the proteins encoded by the DEmRNAs. This analysis led to the identification of a lncRNA-miRNA-mRNA network comprising 18 lncRNAs, 9 miRNAs, and 4 mRNAs (COL4A3, RRAGD, MYLK4, MAP2K6) with a total of 35 interactions, as well as a circRNA-miRNA-mRNA network consisting of 10 circRNAs, 10 miRNAs, and 7 mRNAs (COL4A3, RRAGD, MAP2K6, CACNA1S, MYLK4, TFRC, SGK1) with a total of 22 interactions. The correlation analysis of network-associated genes with muscle fiber type composition further supports a significant relationship between these genes and myofiber transition.

In summary, this study provides a detailed account of how muscle fibers in rabbits transform during growth, unveiling the molecular mechanisms and regulatory networks involved in this process and characterizing the circRNA/lncRNA-miRNA-mRNA competitive endogenous RNA (ceRNA) network. The network-based analytical approach employed here facilitates the identification of key gene networks activated during muscle fiber transformation, offering unprecedented insights into the complex biological processes underlying this phenomenon. Collectively, these findings may assist in identifying suitable targets for modulating muscle fiber transformation, ultimately contributing to the development of high-quality fresh rabbit meat.

However, this study still has several inherent limitations that need to be objectively acknowledged. First, the sample size of each age group was relatively small (n = 3 per group), which may limit the statistical power of the results, especially for detecting subtle differences in molecular expression or phenotypic traits. Second, in vitro functional verification was not performed to validate the regulatory mechanisms of the identified ceRNA network and key molecules (e.g., MYLK4, lncRNA MSTRG.12175.1, miR-145-5p). The current conclusions regarding the “lncRNA/circRNA-miRNA-mRNA” regulatory axis are based on bioinformatics predictions and expression correlation analysis, lacking direct experimental evidence (e.g., overexpression/knockdown assays, dual-luciferase reporter gene assays) to confirm the causal relationships between these molecules. To address these limitations and extend the current research, future work will focus on the following directions: First, expand the sample size to 5–6 rabbits per age group to enhance the statistical robustness of the results. Second, conduct in vitro functional validation using rabbit myoblasts or satellite cells: Verify the direct binding relationships between key ceRNA network components (e.g., lncRNA MSTRG.12175.1 and miR-145-5p, miR-145-5p and MYLK4) via dual-luciferase reporter gene assays; Perform overexpression or knockdown experiments on core molecules to evaluate their effects on myofiber type transition (e.g., changes in the expression of type I/II fiber marker genes MYHC I and MYHC II).

Conclusion

Our study indicates that the gluteus and gastrocnemius muscles of rabbits have different relative compositions of muscle fibers, which significantly affect the color and flavor of rabbit meat. As rabbits grow, muscle fibers transition from Type I to Type II, with notable differences in composition between 1-day-old and 14-day-old rabbits. Through whole transcriptome sequencing, we identified 3,194 mRNAs, 366 circRNAs, 1,394 lncRNAs, 343 miRNAs, 180 transcription factors, and 2,717 alternatively spliced genes associated with muscle fiber transformation. A comprehensive analysis of co-expression and protein-protein interactions revealed 38 key proteins related to this transformation. Based on these 38 key genes, we constructed a lncRNA/circRNA-miRNA-mRNA network comprising 9 mRNAs, 10 circRNAs, 18 lncRNAs, and 14 miRNAs. In summary, our research clarifies how the relative composition of muscle fibers influences the quality and flavor of rabbit meat, how fiber types transition during rabbit growth, and the molecular mechanisms involved in muscle fiber transformation, laying a foundation for future gene regulation studies aimed at enhancing rabbit meat flavor.

Abbreviations

GLU

Gluteus

GAS

Gastrocnemius

FPKM

Fragments Per Kilobase Million

CPC

Coding Potential Calculator

CNCI

Coding-Non-Coding Index

DElncRNAs

Differentially expressed lncRNAs

DEmRNA

Differentially expressed mRNAs

DEcircRNAs

Differentially expressed circular RNAs

DEmiRNAs

Differentially expressed miRNAs

ncRNAs

Non-coding RNAs

ceRNA

Competitive endogenous RNA

AS

Alternative splicing

GSEA

Gene Set Enrichment Analysis

DEGs

Differentially expressed genes

GO

Gene Ontology

KEGG

Kyoto Encyclopedia of Genes and Genomes

BP

Biological process

CC

Cellular component

MF

Molecular function

SE

Skipped exons

A5SS

Alternative 5' splicing sites

A3SS

Alternative 3' splicing sites

MXE

Mutually exclusive exons

RI

Retained introns

FDR

False discovery rate

PPI

Protein-protein interaction

PCA

Principal Component Analysis

Srpbm

Spliced reads per billion mapping

Authors’ contributions

ZR, GS and TZ conceived and designed the project. GS and TZ performed the experiments and analyzed the data. GS and TZ performed the visualization. GH, TZ, LZ, LC, and WJ collected samples. TZ, LZ, LC, and WJ recruited animal resources. GH, AS, SW, SB, XD, and ZR wrote and revised the manuscript. All authors read and approved the final draft.

Funding

This work was supported by the R&D and demonstration of key technologies for healthy rabbit breeding(K3370219046); Yijun County People’s Government-Northwest A&F University-Meat Rabbit(K4050723310); Xi’an Science and Technology Plan Project (no. 25NJSYB00008).

Data availability

The sequence data have been deposited in NCBI SRA repository under accession number PRJNA1331973. (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1331973).

Declarations

Ethics approval and consent to participate

All animal studies conducted in this research adhered to the guidelines established by the Ethics Committee of Northwest A&F University (Approval No. XN2024-0335). The study was designed to ensure the highest standards of animal welfare and ethical treatment throughout the research process.

Competing interests

The authors declare no competing interests.