Age-dependent dynamics of cecal caproic acid in free-range Lueyang black-boned chickens and its effects on primary myoblasts

Abstract

The age-dependent dynamics of cecal caproic acid and its effects on chicken myoblasts remain poorly characterized. A total of 560 female Lueyang black-boned chickens were allocated to cage-reared (CR) or free-range (FR) systems, with seven replicate units per system and 40 birds per unit. Cecal samples were collected at 120, 150, 180, and 360 d of age. Age-dependent microbial and metabolic profiles were characterized within FR chickens using shotgun metagenomics and untargeted metabolomics, whereas caproic acid abundance was compared between CR and FR chickens. Primary chicken myoblasts were used to assess cellular responses to caproic acid. Cecal caproic acid abundance was higher in FR than in CR chickens (P < 0.05). Within FR chickens, its abundance increased at 150 d, declined at 180 d, and increased again at 360 d. These dynamics coincided with age-related differences in microbial community composition and predicted carbohydrate-processing functions. The metabolite cluster containing caproic acid was positively associated with breast muscle yield and negatively associated with lipid-related traits. Caproic acid reduced myoblast viability at 500 μmol/L after 24 and 48 h and at 100-500 μmol/L after 72 h. Treatment with 300 or 400 μmol/L for 48 h reduced EdU incorporation and PCNA and CCND1 mRNA expression and their corresponding protein abundances. It also increased the proportion of cells in G0/G1 and increased apoptosis, without detectable between-group differences in culture-medium pH. The p-AMPK/AMPK and p-p38/p38 ratios increased, whereas the p-ERK/ERK ratio decreased (P < 0.05). MYOD1 mRNA expression also increased following caproic acid treatment (P < 0.05). Together, these findings identify caproic acid as an age-responsive cecal metabolite with direct effects on myoblast proliferation and survival in vitro. They provide evidence for a potential link between cecal metabolic variation and muscle-cell responses in chickens.

Graphical abstract

Introduction

Poultry meat is an important source of high-quality animal protein, and consumer demand increasingly includes desirable sensory characteristics, intramuscular fat deposition, and overall meat quality (Cui et al., 2024; Song et al., 2025). Rearing conditions can influence poultry behavior, environmental exposure, intestinal microbial composition, metabolism, and selected carcass or meat-quality-related traits (Chen et al., 2019; Song et al., 2022). Free-range (FR) systems provide birds with access to outdoor environments, opportunities for physical activity, and exposure to plant-derived materials during foraging. These features distinguish FR production from conventional cage-reared (CR) systems and may contribute to differences in host metabolism and meat characteristics. The Lueyang Black-boned (LYB) chicken is an indigenous breed with distinctive carcass and meat-quality traits. Our previous study identified rearing-system-associated differences in intestinal metabolism and microbial homeostasis in LYB chickens and found that caproic acid was enriched in the FR group (Li et al., 2025). However, the temporal variation of caproic acid within the FR system and its potential relationship with muscle-related cellular processes remain unclear.

The chicken cecum is an important site of microbial fermentation and contains a complex microbial community involved in the transformation of dietary and host-derived substrates (Oakley et al., 2014). Cecal microbial activity may therefore contribute to intestinal and systemic physiology, including through the production of short-chain and medium-chain fatty acids (Ramírez et al., 2022; Ali et al., 2022). These fatty acids can participate in energy metabolism, lipid metabolism, and cell signaling (Sun et al., 2021). Caproic acid, also known as hexanoic acid, is a medium-chain fatty acid that can be generated through microbial chain-elongation pathways, in which shorter-chain fatty acids and suitable electron donors serve as metabolic substrates (Spirito et al., 2014). Studies of the gut–muscle axis have suggested that gut-derived metabolites may be related to skeletal muscle metabolism and muscle-cell physiology (Lahiri et al., 2019; Frampton et al., 2020). Nevertheless, most evidence has been obtained from non-poultry models, and the biological relevance of caproic acid for chicken muscle cells remains insufficiently characterized. Thus, the temporal behavior of cecal caproic acid and its possible association with host muscle-related traits warrant further investigation.

The chicken cecal microbiota undergoes age-related succession, including changes in microbial composition, functional potential, and metabolite production (Videnska et al., 2014). However, studies conducted at a limited number of developmental stages provide incomplete information about the continuity and direction of these changes. In particular, it remains unclear whether age-related microbial succession is accompanied by coordinated variation in caproic acid abundance and carbohydrate-processing functions within the cecum. This question is relevant to LYB chickens because the FR system exposes birds to a more complex and variable environment than the CR system. At the host-cell level, myoblast proliferation, cell-cycle progression, apoptosis, and subsequent fate decisions are important components of skeletal muscle development (Velleman, 2007; Yin et al., 2013). Whether a gut-associated metabolite such as caproic acid affects these cellular processes in primary chicken myoblasts has not been established. Addressing this question may help connect microbial metabolic variation with muscle-related cellular responses, while distinguishing correlation from direct cellular effects.

The present study was designed as a longitudinal follow-up of our previous CR-versus-FR study rather than a repetition of the complete rearing-system comparison. We characterized age-dependent changes in cecal metabolites, microbial composition, and predicted microbial functions in FR LYB chickens at 120, 150, 180, and 360 days of age. The CR-versus-FR comparison of caproic acid abundance was retained to provide the biological context for the subsequent FR-focused analyses. We then used primary chicken myoblasts to examine the effects of caproic acid exposure on cell viability, proliferation, cell-cycle distribution, apoptosis-related markers, and AMPK/MAPK-associated signaling. The in vivo analyses were intended to identify associations between cecal caproic acid dynamics and host phenotypic traits, whereas the in vitro experiments were used to characterize cellular responses under controlled conditions. Together, this study evaluates the potential relationship between age-related cecal metabolic succession and muscle-related cellular responses in LYB chickens.

Materials and Methods

Ethical Approval

All animal care, handling, and experimental procedures were conducted in accordance with the animal welfare guidelines and were approved by the Institutional Animal Care and Use Committee of Northwest A&F University (Protocol number: WAFAC1008).

Animal Husbandry and Sample Collection

A total of 560 one-day-old female Lueyang Black-boned (LYB) chicks from the same hatch were obtained from Shaanxi Long Jia Agricultural Science and Technology Development Co., Ltd. All birds underwent a standardized brooding period until 70 d of age. Birds with comparable body weights were then randomly allocated to a cage-reared (CR, n = 280) or free-range (FR, n = 280) system. The first sampling was conducted at 120 d of age, following 50 d of exposure to the assigned rearing systems after the brooding period. Each rearing system comprised seven biological replicate units, with 40 birds per replicate.

In the CR system, each bird was housed individually in a three-tier breeding cage (35 × 45 × 50 cm) equipped with a linear feeder and an automatic nipple drinker. Each cage provided approximately 0.15 m² of floor space per bird, corresponding to an approximate stocking density of 6.3 birds/m². Forty individual cages constituted one CR replicate unit.

In the FR system, 280 birds were randomly allocated to seven replicate units, with 40 birds per replicate. The total FR facility consisted of an indoor floor-rearing area measuring 20 × 20 m and an adjacent outdoor pasture measuring 20 × 50 m. These areas were partitioned into seven corresponding replicate units using fishing nets, with 40 birds assigned to each unit. Based on the total number of birds and the total available area, the calculated stocking densities were approximately 0.7 birds/m² for the indoor area and 0.3 birds/m² for the outdoor pasture. The indoor and outdoor areas of each replicate were directly connected to allow birds unrestricted access between them during the daytime. The indoor area had concrete floors, natural ventilation, and 80-cm-high plastic mesh barriers, whereas the outdoor pasture was enclosed with wire mesh. Both areas were equipped with perches, feeders, and plastic water tanks. Birds had unrestricted access to the outdoor pasture from 06:00 to 18:00 daily and were confined to the indoor area from 18:00 to 06:00.

From March onward, both rearing systems were maintained under the same indoor temperature (20–25°C), relative humidity (65–70%), and photoperiod (16 h light:8 h dark). All birds received the same basal diet, formulated according to the nutrient requirements for meat-type chickens specified in the Chinese National Standard (NY/T 33−2004), and had ad libitum access to feed and water.

At 120, 150, 180, and 360 d of age, one bird was randomly selected from each replicate unit in both the CR and FR groups, yielding seven biological replicates per group at each time point. Following a 12-h fast, birds were euthanized by electrical stunning and exsanguination. Cecal contents were collected aseptically, snap-frozen in liquid nitrogen, and stored at −80°C until multi-omics analysis. Samples from the CR and FR systems were analyzed separately and were not pooled before metabolomic or metagenomic analysis. The age-dependent longitudinal analyses of the cecal metabolome and metagenome were conducted using FR samples from 120, 150, 180, and 360 d of age, with seven biological replicates per age group. The CR-versus-FR comparison was retained specifically for caproic acid abundance at each sampling age.

Metagenomic Sequencing and Functional Annotation

Total genomic DNA was extracted from cecal contents using the Fecal Genome DNA Extraction Kit (AU46111-96, BioTeke, China). DNA integrity and concentration were assessed before library construction. Sequencing libraries were prepared using the TruSeq Nano DNA Library Preparation Kit (Illumina, USA). Paired-end sequencing (150 bp) was performed on the Illumina NovaSeq 6000 platform. Raw sequencing reads were quality-filtered, and host DNA contamination was removed by aligning high-quality reads against the Gallus gallus reference genome using Bowtie2 (v2.2). Clean microbial reads were de novo assembled into contigs using MEGAHIT (v1.2.9). Putative coding sequences were predicted, and the resulting sequences were clustered into a nonredundant unigene catalog using CD-HIT (v4.6.1).

Taxonomic annotation was performed by aligning predicted unigene protein sequences against the NR_meta database, a microbial subset of the NCBI nonredundant protein database containing 52,375,954 sequences, using DIAMOND (v0.9.14) in BLASTP mode with an E-value threshold of ≤ 1 × 10⁻⁵. The best-scoring hit for each unigene was used for taxonomic assignment. Taxonomic information was mapped using the NCBI Taxonomy database to obtain annotations at the superkingdom, phylum, class, order, family, genus, and species levels. Taxonomic abundance tables were summarized at the phylum, genus, and species levels and converted to relative abundances for downstream community analyses.

Functional profiles were annotated by aligning unigenes against the Kyoto Encyclopedia of Genes and Genomes (KEGG, v87.1) and Carbohydrate-Active enZymes (CAZy) databases using DIAMOND (v0.9.14).

Microbiome analyses were conducted in R (v4.5.1) using phyloseq (McMurdie and Holmes, 2013) and vegan (v2.7-2; Oksanen et al., 2025). Alpha-diversity differences were assessed using the Kruskal–Wallis test. Beta diversity was calculated using Bray–Curtis dissimilarities with the vegdist function, and community differences among age groups were evaluated by permutational multivariate analysis of variance (PERMANOVA) using the adonis2 function with 999 permutations. Principal coordinate analysis and nonmetric multidimensional scaling were used for ordination.

LEfSe analysis was performed using microbiomeMarker (Cao et al., 2022), with Kruskal–Wallis and Wilcoxon cutoffs of 0.05 and an LDA cutoff of 2.0. Species-level taxa with LDA scores ≥ 2.5 were retained for subsequent temporal clustering and correlation analyses. Random Forest classification was performed using the randomForest package (Liaw and Wiener, 2002) with 1,000 trees. Ten-fold cross-validation was used to select the feature subset, and taxa were ranked according to Mean Decrease Accuracy. For temporal clustering, the mean relative abundance of each retained species was calculated for each age group and standardized as a Z score. K-means clustering was performed with 25 random starts. The number of clusters was selected according to the mean silhouette width across candidate cluster numbers using the cluster package (v2.1.8.1; Maechler et al., 2025), resulting in five age-related abundance patterns.

Microbial Functional and Integrative Analyses

Level 2 CAZyme abundances were converted to relative abundances within each sample. Families with a maximum mean relative abundance > 0.0005 across age groups were retained. The 25 families with the highest variance across the four age groups were selected, standardized by row-wise Z scores, and visualized in a heatmap.

KEGG functional shifts were assessed using the ReporterScore package (v0.1.9; Peng et al., 2024). KO abundances were normalized to counts per million, and each age group (150, 180, or 360 d) was compared with the 120-d group using the directed Reporter Score mode. Pathways with an absolute Reporter Score > 1.96 in at least one comparison were retained. When more than 25 pathways met this criterion, the 25 pathways with the greatest variance across comparisons were displayed.

Spearman correlation analysis between Cluster 3 metabolites and host phenotypic and serum biochemical variables was performed using the psych package (v2.5.6; Revelle, 2025). Correlations with an absolute Spearman correlation coefficient > 0.60 and an unadjusted P value < 0.05 were displayed; caproic acid was retained in the visualization irrespective of this filtering criterion. For taxon–caproic acid correlations, all species-level LEfSe taxa with LDA scores ≥ 2.5 were tested using Spearman correlation analysis. The 30 taxa with the largest absolute correlation coefficients were displayed, and unadjusted P values were used for significance annotation. For pathway–caproic acid correlations, KEGG pathways with relative abundances > 0.0001 in at least 20% of samples were retained. Positively correlated pathways with unadjusted P < 0.05 were ranked by Spearman correlation coefficient, and up to 25 pathways were displayed.

Untargeted Metabolomics Analysis

Metabolites from the cecal contents were extracted using an 80% (v/v) pre-cooled methanol aqueous solution. The homogenates were incubated at −20°C for 30 min, centrifuged (20,000 × g, 15 min), and the supernatants were vacuum-dried before being reconstituted for liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis. A pooled quality control (QC) sample was utilized to monitor system stability.

Chromatographic separation was performed on an UltiMate 3000 UPLC System coupled with an ACQUITY UPLC T3 column (Waters, USA), followed by mass spectrometry detection using a TripleTOF 6600 system (SCIEX, USA) operating in both positive and negative ion modes. Raw MS data were processed using XCMS (v3.22.0) in R for peak picking, retention-time alignment, and feature quantification (Smith et al., 2006; Louail et al., 2025). Metabolite annotation was conducted by matching the exact mass-to-charge ratio (m/z) and fragmentation patterns against the Human Metabolome Database (HMDB) and KEGG databases.

Metabolomic data were analyzed in R (v4.5.1). Orthogonal partial least squares-discriminant analysis (OPLS-DA) was performed using the ropls package (v1.42.0; Thévenot et al., 2015). Models were constructed with one predictive and one orthogonal component after Pareto scaling. Model performance was assessed using seven-fold cross-validation and 200 permutation tests. For each pairwise age comparison, univariate differences were assessed using two-sided Welch’s t-tests. Metabolites with a variable importance in projection (VIP) value > 1 and an unadjusted P value < 0.05 were defined as differentially abundant metabolites (DAMs). P values were adjusted using the Benjamini–Hochberg procedure. The resulting q values for the selected DAMs are reported in Supplementary Table 3. Temporal metabolite profiles were analyzed using Mfuzz (v2.70.0; Kumar and Futschik, 2007). Log2-transformed metabolite abundances were averaged within each age group and standardized before clustering. Fuzzy c-means clustering was performed using six clusters and a fuzzification parameter of 1.5. Heatmaps were generated using ComplexHeatmap (Gu et al., 2016). DAMs shared across the three consecutive pairwise comparisons were visualized by hierarchical clustering after row-wise Z-score standardization. HMDB superclass annotations and KEGG pathway annotations of DAMs were summarized according to the number of annotated metabolites in each category or pathway.

Primary Myoblast Isolation and Culture

To examine cellular responses to caproic acid exposure in vitro, primary myoblasts were isolated from the pectoral muscles of 12-day-old LYB chicken embryos. Briefly, the excised muscle tissues were minced and subjected to enzymatic digestion using 0.25% trypsin (BC—CE-006, Biochannel, China) and 0.1% collagenase type I (1904MG100, BioFroxx, Germany) at 37°C. The cell suspension was filtered through a 40 μm cell strainer, centrifuged, and resuspended in a growth medium consisting of DMEM/F12 (BC-M-002, Biochannel, China) supplemented with 10% fetal bovine serum (BC-SE-FBS01, Biochannel, China) and 1% penicillin-streptomycin (BC—CE-022, Biochannel, China). To eliminate rapidly adhering fibroblasts, the suspension was subjected to differential adhesion purification in uncoated culture flasks for 2 h. The unattached, purified myoblasts were collected, seeded into culture plates, and maintained in a humidified incubator at 39°C with 5% CO₂. The myogenic purity of the isolated myoblasts was verified via PAX7 immunofluorescence staining, and the cells were confirmed to be free of mycoplasma contamination prior to downstream assays.

Cell Viability Assay (CCK-8)

Primary myoblasts were seeded in 96-well plates at a density of 8,000 cells per well. After overnight culture, when the cells reached approximately 70% confluence, the culture medium was replaced with fresh growth medium containing caproic acid at final concentrations of 0, 100, 200, 300, 400, or 500 μmol/L. Control wells received an equal volume of vehicle. Cells were exposed to the treatments for 24, 48, or 72 h. Each treatment included six technical replicate wells, and the experiment was independently repeated three times. At each time point, 10 μL of CCK-8 reagent was added to each well, followed by incubation for 2 h at 37°C in the dark. Absorbance was measured at 450 nm using a microplate reader. Based on the CCK-8 results, 300 and 400 μmol/L caproic acid were selected for subsequent functional and molecular assays.

Assessment of Culture-Medium pH

To determine whether caproic acid treatment altered culture-medium pH, DMEM/F12 supplemented with 10% fetal bovine serum was prepared with 0, 100, 200, 300, 400, or 500 μmol/L caproic acid. The media were maintained under the same experimental conditions used for cell culture (39°C and 5% CO₂). The pH was measured immediately after treatment preparation and after 24, 48, and 72 h using a calibrated pH meter. Measurements were obtained from 3 independent experiments.

Caproic Acid Treatment for Downstream Assays

For downstream functional and molecular assays, including EdU incorporation, flow cytometry, RT-qPCR, and Western blotting, primary myoblasts were seeded in the appropriate culture plates. When the cells reached approximately 70% confluence, the growth medium was replaced with fresh medium containing 300 or 400 μmol/L caproic acid. Control cells received an equal volume of vehicle. After 48 h of treatment, cells were harvested for the respective assays.

EdU Proliferation Assay

Cell proliferation was evaluated using the 5-ethynyl-2′-deoxyuridine (EdU) Cell Proliferation Kit with Alexa Fluor YF 594 (UElandy, China). Following the 48-h caproic acid treatment, myoblasts in 96-well plates were incubated with 50 μM EdU labeling medium for 2 h at 37°C. The cells were then fixed with 4% paraformaldehyde, permeabilized with 0.5% Triton X-100, and reacted with the Apollo staining solution in the dark. Cell nuclei were counterstained with Hoechst 33342. Images were captured using an inverted fluorescence microscope (OLYMPUS CKX53, Japan), and the EdU-positive cells were quantified using ImageJ software (v1.5.3, National Institutes of Health, USA).

Cell Cycle Analysis by Flow Cytometry

To analyze the cell cycle distribution, the treated myoblasts were harvested, washed twice with cold PBS, and fixed overnight in 70% pre-cooled ethanol at 4°C. Following fixation and ethanol removal, the cells were washed and stained using a commercial Cell Cycle and Apoptosis Analysis Kit (UElandy, China) according to the manufacturer’s instructions. Briefly, the cell pellets were resuspended and incubated with the provided propidium iodide (PI) and RNase A staining solutions for 30 min at 37°C in the dark. The cellular DNA content was subsequently analyzed utilizing a BD FACSAria™ III flow cytometer (Becton Dickinson, USA), and the percentages of cells in the G0/G1, S, and G2/M phases were quantified using ModFit LT software (v5.0, Verity Software House, USA).

Cell Apoptosis Analysis by Flow Cytometry

The apoptosis rate of the treated myoblasts was determined using an Annexin V-FITC/PI Apoptosis Detection Kit (UElandy, China) according to the manufacturer’s instructions. Briefly, cells were collected, washed twice with cold PBS, and resuspended in 100 μL of 1× binding buffer. Subsequently, 5 μL of Annexin V-FITC and 5 μL of PI were added to the cell suspension, followed by an incubation for 15 min at room temperature in the dark. Finally, 400 μL of 1× binding buffer was added to each tube, and the apoptotic cells were immediately detected using a BD FACSAria™ III flow cytometer (Becton Dickinson, USA). Data analysis was performed using FlowJo software (v10.8.1).

Immunofluorescence (IF) Staining

Myoblasts cultured on coverslips were fixed with 4% paraformaldehyde for 15 min and permeabilized with 0.5% Triton X-100 for 10 min. After blocking with 5% bovine serum albumin (BSA) for 1 h at room temperature, the cells were incubated overnight at 4°C with specific primary antibodies (PAX7, Affinity, AF7584, 1:200). The following day, the cells were washed and incubated with fluorophore-conjugated secondary antibodies (Proteintech, 1:100) for 1 h in the dark. Nuclei were stained with DAPI (Solarbio, China). Fluorescence images were captured using an inverted fluorescence microscope (CKX53, Olympus, Japan).

RNA Extraction and Quantitative Real-Time PCR (qPCR)

Total RNA was extracted from the caproic acid-treated myoblasts using TRIzol reagent (Agbio, China) in accordance with the manufacturer’s instructions. The concentration and purity of the isolated RNA were spectrophotometrically determined using a NanoDrop 2000 instrument (Thermo Fisher Scientific, USA), ensuring an A260/A280 ratio between 1.8 and 2.0. Subsequently, 1 μg of total RNA was reverse-transcribed into complementary DNA (cDNA) using a First-Strand cDNA Synthesis Kit with gDNA Remover (Cat. No. ZR108; Zhuangmeng, China) to eliminate potential genomic DNA contamination.

Quantitative real-time PCR was performed on a CFX96 Real-Time PCR Detection System (Bio-Rad Laboratories, Hercules, CA, USA) utilizing a SYBR Green Real-Time PCR Kit (Zhuangmeng, China). The primer sequences used in this study were listed in Supplementary Table 1. The thermal cycling conditions were set as follows: an initial denaturation at 95°C for 30 s, followed by 40 cycles of amplification at 95°C for 10 s and 60°C for 30 s. A melt curve analysis was routinely performed at the end of each run to verify primer specificity and the absence of primer-dimer formation. The relative mRNA expression levels of the target genes were calculated using the standard 2^−ΔΔCt method and normalized to β-actin as the reference gene.

Western Blotting

Total cellular proteins were extracted from treated myoblasts using RIPA lysis buffer containing 1% phenylmethanesulfonyl fluoride (PMSF) and phosphatase inhibitors (Agbio, China). Protein concentrations were determined using a BCA Protein Assay Kit (Solarbio, China). Equal quantities of protein (30 μg per lane) were separated on 12% SDS-polyacrylamide gels and transferred onto 0.22-μm PVDF membranes (Millipore, USA). The membranes were blocked with 5% skim milk or 5% BSA for phosphorylated proteins for 2 h at room temperature and then incubated overnight at 4°C with the following primary antibodies: anti-PCNA (Immunoway, 1:500), anti-CCND1 (HUABIO, 1:5,000), anti-BCL-2 (Bioss, 1:1,000), anti-BAX (Proteintech, 1:2,000), anti-AMPK (HUABIO, 1:2,000), anti-p-AMPK (Immunoway, 1:5,000), anti-ERK1/2 (Proteintech, 1:5,000), anti-p-ERK1/2 (HUABIO, 1:5,000), anti-p38 (HUABIO, 1:2,000), anti-p-p38 (HUABIO, 1:1,000), anti-GAPDH (HUABIO, 1:6,000), and anti-β-actin (Proteintech, 1:20,000). After washing, the membranes were incubated with HRP-conjugated secondary antibodies (Proteintech, 1:5,000) for 1 h at room temperature. Protein bands were visualized using an enhanced chemiluminescence (ECL) kit (Seven, China) and an imaging system (ChampChemi 610 Plus, China). Band optical densities were quantified using ImageJ software. β-actin was used as the loading control for PCNA, CCND1, BCL-2, BAX, ERK1/2, and p-ERK1/2, whereas GAPDH was used as the loading control for AMPK, p-AMPK, p38, and p-p38. Phosphorylation levels were expressed as the ratios of phosphorylated to corresponding total proteins. Original, uncropped Western blot images are provided in Supplementary Data 6.

Statistical Analyses

Statistical analyses of phenotypic and cell-based data were performed using GraphPad Prism 9.5. Data are presented as means ± standard deviation (SD). Differences between two independent groups were analyzed using Student’s t-test. Multiple-group comparisons were performed using one-way analysis of variance followed by Tukey’s multiple-comparisons test. For the dose-response CCK-8 assay, Dunnett’s multiple-comparisons test was used to compare each caproic acid concentration with the untreated control. Multiplicity-adjusted P values were reported, and P < 0.05 was considered statistically significant.

Results

Age-Related Changes in the Cecal Metabolome and Caproic Acid Abundance

To characterize age-associated variation in the cecal metabolome, untargeted metabolomic analyses were conducted using FR samples collected at 120, 150, 180, and 360 d of age, with seven biological replicates per age group. OPLS-DA score plots showed separation between consecutive age groups (Fig. 1A). The corresponding models had R²Y values of 0.972, 0.993, and 0.991 and Q² values of 0.798, 0.892, and 0.924 for the 120 vs. 150 d, 150 vs. 180 d, and 180 vs. 360 d comparisons, respectively. Permutation tests returned Q² intercepts below zero for all three models (Fig. 1B). Together, the cross-validated Q² values and permutation-test results indicated satisfactory performance of the pairwise OPLS-DA models. VIP scores from the OPLS-DA models were subsequently combined with unadjusted P values to prioritize DAMs for downstream analyses.

Fig. 1.

Temporal dynamics of the cecal metabolome in free-range (FR) Lueyang Black-boned (LYB) chickens. All age-dependent analyses in this figure were conducted using FR samples collected at 120, 150, 180, and 360 d of age. (A) Orthogonal partial least squares discriminant analysis (OPLS-DA) score plots showing the separation of cecal metabolite profiles among sampling ages. (B) Permutation tests for the OPLS-DA models. (C) Numbers of DAMs showing increased or decreased abundance in pairwise age comparisons, identified using VIP > 1 and unadjusted P < 0.05. (D) Venn diagram showing DAMs shared among the three consecutive pairwise age comparisons. (E) Hierarchical clustering heatmap showing age-related abundance patterns of the shared DAMs.

Pairwise age comparisons identified substantial numbers of DAMs. At 150 d relative to 120 d, 139 DAMs showed higher abundance and 99 showed lower abundance. Between 150 and 180 d, 169 and 154 DAMs showed higher and lower abundance, respectively. Between 180 and 360 d, 62 DAMs showed higher and 168 showed lower abundance (Fig. 1C). Thirty-eight DAMs were shared across the three consecutive pairwise comparisons (Fig. 1D) and showed distinct abundance profiles across the four sampling ages (Fig. 1E).

Functional annotation provided a biochemical overview of the DAMs. KEGG annotation most frequently assigned these DAMs to pathways related to biosynthesis of secondary metabolites, glycerophospholipid metabolism, microbial metabolism in diverse environments, biosynthesis of plant secondary metabolites, biosynthesis of amino acids, protein digestion and absorption, choline metabolism, central carbon metabolism, and biosynthesis of cofactors (Fig. 2A). HMDB superclass annotation classified the DAMs as lipids and lipid-like molecules (n = 121, 30%), organic acids and derivatives (n = 107, 26%), organoheterocyclic compounds (n = 69, 17%), benzenoids (n = 44, 11%), phenylpropanoids and polyketides (n = 24, 6%), alkaloids and derivatives (n = 9, 2%), and nucleosides, nucleotides, and analogues (n = 9, 2%) (Fig. 2B).

Fig. 2.

Functional annotation of the cecal metabolome and age-related variation in caproic acid abundance. (A) KEGG pathway annotation summary of differentially abundant metabolites (DAMs). (B) Classification of DAMs according to Human Metabolome Database (HMDB) superclass categories. (C) Caproic acid abundance in cage-reared (CR) and free-range (FR) chickens at each sampling age. Asterisks indicate significant differences between the CR and FR groups at the same age (*P < 0.05, **P < 0.01, and ***P < 0.001). (D) Mfuzz clustering of metabolite abundance profiles, with caproic acid assigned to Cluster 3. (E) Spearman correlation analysis between Cluster 3 metabolites and host phenotypic and serum biochemical parameters. Metabolites with an absolute Spearman correlation coefficient > 0.60 and unadjusted P < 0.05 were displayed, and caproic acid was retained irrespective of this display criterion. Asterisks indicate significant correlations based on unadjusted P values (*P < 0.05, **P < 0.01, and ***P < 0.001).

Among the age-responsive metabolites, caproic acid abundance was examined in the CR and FR systems at each sampling age. Caproic acid abundance was higher in FR chickens than in CR chickens at 120, 150, 180, and 360 d (P < 0.05; Fig. 2C). Within the CR system, caproic acid abundance did not differ among sampling ages (P > 0.05). Within the FR system, caproic acid abundance increased at 150 d, decreased at 180 d, and increased again at 360 d, with the decrease from 150 to 180 d reaching significance (P < 0.05). Mfuzz analysis assigned caproic acid to Cluster 3 (Fig. 2D). Metabolites in Cluster 3, including caproic acid, were negatively correlated with intramuscular fat content, abdominal fat percentage, and serum TG and TC concentrations, and positively correlated with breast muscle yield, meat lightness (L*), and whiteness index (unadjusted P < 0.05; Fig. 2E). The host phenotypic and serum biochemical data used for this analysis are provided in Supplementary Data 5.

Age-Dependent Dynamics and Maturation of the Cecal Microbiota

Age-related metabolite variation was accompanied by changes in the cecal microbial community. All age-dependent metagenomic analyses were conducted using FR samples, with seven biological replicates per age group. Taxonomic richness indices including Observed, Chao1, and ACE did not differ significantly among ages (P > 0.05), whereas Shannon, Simpson, and InvSimpson indices differed among ages (P < 0.01; Fig. 3A). PCoA and NMDS ordinations based on Bray-Curtis dissimilarities suggested separation among the 120-, 150-, and 180-d communities, whereas the 150- and 360-d communities showed greater overlap. PERMANOVA indicated overall compositional differences among sampling ages (R² = 0.495, P = 0.001), and the NMDS stress value was 0.128 (Fig. 3B).

Fig. 3.

Age-related changes in cecal microbiota diversity in FR LYB chickens. All age-dependent metagenomic analyses in this figure were conducted using FR samples. (A) Temporal variation in α-diversity indices, including the Observed, Chao1, and ACE richness indices and the Shannon, Simpson, and InvSimpson diversity indices. (B) β-diversity analyses based on Bray–Curtis dissimilarities and visualized using principal coordinate analysis (PCoA) and nonmetric multidimensional scaling (NMDS).

Based on relative abundance, the dominant phyla included Bacteroidota, Bacillota, Uroviricota, Actinomycetota, Spirochaetota, Thermodesulfobacteriota, Pseudomonadota, Deferribacteriota, Campylobacterota, and Synergistota. The dominant genera included Bacteroides, Phocaeicola, Alistipes, Parabacteroides, Muribaculum, Clostridium, Prevotella, Lachnoclostridium, and Mediterranea (Fig. 4A).

Fig. 4.

Taxonomic composition and age-associated patterns of the cecal microbiota in FR LYB chickens. All age-dependent metagenomic analyses in this figure were conducted using FR samples. (A) Relative abundance of dominant taxa at the phylum, genus, and species levels. (B) Linear discriminant analysis effect size (LEfSe) analysis identifying species-level taxa that discriminated among age groups using an LDA score > 2.0. (C) Random Forest analysis at the genus level, with genera ranked according to mean decrease accuracy. (D) Comparison of taxa prioritized by LEfSe and Random Forest after mapping species-level taxa to their corresponding genera. (E) K-means clustering of species-level taxa with LDA scores ≥ 2.5 into five age-related abundance patterns.

LEfSe analysis identified species-level taxa that discriminated among sampling ages at an LDA score > 2.0 (Fig. 4B). Taxa enriched at 120 d included Mucispirillum schaedleri, Lachnoclostridium sp. An76, and Mediterraneibacter glycyrrhizinilyticus. The 150 d group included Bacteroides caecigallinarum and Gallibacter intestinalis, the 180 d group included Bacteroides sp. An322 and Paraprevotella clara, and the 360 d group included Treponema brennaborense and Muribaculum sp. An289.

The genus-level Random Forest classifier provided a complementary view of age-associated taxa (Fig. 4C). Leadbetterella, Candidatus Argoarchaeum, Methanomicrobium, Baoshanvirus, and Methanocorpusculum ranked highly according to Mean Decrease Accuracy. After species-level LEfSe taxa were mapped to their corresponding genera, Muribaculum, Ruminococcus, and Methanocorpusculum were represented in both analyses. These genera corresponded to Muribaculum sp. An289, Ruminococcus sp. cLA-AA-H200, and Methanocorpusculum vombati and M. petauri, respectively (Fig. 4D and Supplementary Table 2). K-means clustering grouped the selected species-level taxa into five age-related abundance patterns (Fig. 4E).

Functional Shifts in Microbial Polysaccharide Degradation and Metabolism

Age-related variation was also evident in predicted microbial functional profiles. The 25 selected Level 2 CAZyme families showed distinct abundance patterns across ages (Fig. 5A). GH0, GT5, GH1, CE4, CBM48, GH13_9, GT35, GH3, GH23, and GT4 showed the highest relative abundances at 120 d and decreased thereafter. GH13_31, CBM50, GT51, CBM32, GT2, GH20, GH13, and GT28 decreased from 120 to 180 d and increased at 360 d. GH2, GH78, GH35, and GH97 increased from 120 to 180 d and decreased at 360 d. GH28, GH92, and CBM20 increased across the sampling ages.

Fig. 5.

Age-related changes in microbial functional profiles and associations with cecal caproic acid abundance in free-range (FR) Lueyang Black-boned (LYB) chickens. All age-dependent metagenomic analyses in this figure were conducted using FR samples. (A) Heatmap showing the 25 most variable Level 2 carbohydrate-active enzyme (CAZyme) families. (B) Spearman correlation network showing associations between selected species-level taxa and cecal caproic acid abundance. The displayed taxa were selected from species-level LEfSe taxa with LDA scores ≥ 2.5 and ranked according to the absolute Spearman correlation coefficient. Red and blue nodes and edges indicate positive and negative associations, respectively. (C) Heatmap showing age-related shifts in predicted KEGG pathway abundance identified using the Reporter Score (RS) algorithm, with |RS| > 1.96. (D) Spearman correlation analysis showing associations between KEGG pathways and cecal caproic acid abundance. Correlations in panels B and D were evaluated using unadjusted P values. Asterisks indicate statistically significant correlations (*P < 0.05, **P < 0.01, and ***P < 0.001).

Reporter Score analysis further characterized age-related shifts in predicted KEGG pathway abundance, using 120 d as the reference (|RS| > 1.96; Fig. 5C). Pathways related to nonribosomal peptide structures, ABC transporters, the phosphotransferase system (PTS), flagellar assembly, cardiac muscle contraction, bacterial chemotaxis, 2-oxocarboxylic acid metabolism, pentose and glucuronate interconversions, and galactose metabolism showed negative Reporter Scores at 150, 180, and 360 d. Pathways related to O-antigen nucleotide sugar biosynthesis, biosynthesis of nucleotide sugars, amino sugar and nucleotide sugar metabolism, the proteasome, the ribosome, and DNA replication showed positive Reporter Scores across the three comparisons. Methane metabolism, spinocerebellar ataxia, carbon metabolism, viral carcinogenesis, the HIF-1 signaling pathway, RNA polymerase, aminoacyl-tRNA biosynthesis, carbon fixation in photosynthetic organisms, and basal transcription factors showed positive Reporter Scores at 150 and 360 d and negative Reporter Scores at 180 d.

Multi-Omics Associations Among Cecal Microbes, Functional Pathways, and Caproic Acid

To examine associations between the age-related microbial profiles and caproic acid abundance, Spearman correlation analysis was performed using the selected species-level taxa and KEGG pathways. Paraprevotella clara, Bacteroides sp. 43_108, Paraprevotella xylaniphila, Bacteriophage sp., Prevotella sp. CAG:1320, Prevotella sp. AG:487_50_53, Prevotella lascolaii, and Caudoviricetes sp. were negatively correlated with caproic acid abundance (unadjusted P < 0.05, ρ < 0). Firmicutes bacterium CAG:110, Massilistercora timonensis, Pseudoflavonifractor sp. An184, Flavonifractor sp. An10, Campylobacter avium, and Gallibacter intestinalis were positively correlated with caproic acid abundance (unadjusted P < 0.05, ρ > 0; Fig. 5B). Methanocorpusculum vombati and M. petauri also showed positive Spearman correlation coefficients, although these associations did not reach the unadjusted significance threshold (0.05 < P < 0.10).

At the pathway level, several KEGG pathways were positively correlated with caproic acid abundance at the unadjusted significance threshold (P < 0.05; Fig. 5D). These pathways included terpenoid backbone biosynthesis, biosynthesis of cofactors, nicotinate and nicotinamide metabolism, propanoate metabolism, cysteine and methionine metabolism, porphyrin and chlorophyll metabolism, glycerophospholipid metabolism, d-alanine metabolism, carbon fixation pathways in prokaryotes, d-glutamine and d-glutamate metabolism, pyruvate metabolism, selenocompound metabolism, and peptidoglycan biosynthesis.

Effects of Caproic Acid on Myoblast Proliferation and Cell Cycle Progression

To assess direct cellular responses to caproic acid, primary myoblasts were isolated from embryonic pectoral muscle, and PAX7 immunofluorescence confirmed their myogenic identity (Fig. 6A). Culture-medium pH remained close to 7.2 across all caproic acid concentrations and incubation times. No statistically significant difference in pH was detected among treatment groups at the same time point (P > 0.05; Table 1).

Fig. 6.

Effects of caproic acid exposure on primary chicken myoblast viability, proliferation, and cell-cycle distribution. (A) Immunofluorescence staining of PAX7 (red) and DAPI (blue) for identification and myogenic purity assessment of primary myoblasts. (B) Cell viability after treatment with 0, 100, 200, 300, 400, or 500 μmol/L caproic acid for 24, 48, or 72 h, measured using the CCK-8 assay. (C) Relative mRNA expression of the proliferation-related genes PCNA and CCND1 after treatment with sublethal concentrations of caproic acid, measured using RT-qPCR. (D) Representative Western blot images and densitometric quantification of PCNA and CCND1 proteins. (E) Representative images and quantitative analysis of EdU incorporation in primary myoblasts. (F) Cell-cycle distribution in the G0/G1, S, and G2/M phases determined by propidium iodide staining and flow cytometry. Data are presented as means ± SD from three independent experiments. The CCK-8 assay included six technical replicate wells per treatment in each experiment. Dunnett-adjusted P values were used for comparisons of caproic acid concentrations with the untreated control in panel B; Tukey-adjusted P values were used for multiple-group comparisons. Exact P values are shown in the figure.

Table 1.

Culture-medium pH after caproic acid treatment under the cell-culture conditions used in this study.

Caproic acid concentration	pH at 0 h	pH at 24 h	pH at 48 h	pH at 72 h
0 μmol/L	7.24 ± 0.038	7.22 ± 0.042	7.21 ± 0.032	7.18 ± 0.019
100 μmol/L	7.25 ± 0.022	7.23 ± 0.036	7.20 ± 0.030	7.15 ± 0.025
200 μmol/L	7.24 ± 0.025	7.24 ± 0.025	7.22 ± 0.025	7.14 ± 0.031
300 μmol/L	7.19 ± 0.027	7.20 ± 0.028	7.21 ± 0.024	7.13 ± 0.025
400 μmol/L	7.20 ± 0.033	7.21 ± 0.034	7.22 ± 0.027	7.12± 0.026
500 μmol/L	7.18 ± 0.037	7.20 ± 0.033	7.18 ± 0.035	7.12 ± 0.022

DMEM/F12 supplemented with 10% fetal bovine serum was incubated at 39°C in a humidified atmosphere containing 5% CO₂. Values are presented as mean ± SD from 3 independent experiments. No statistically significant differences in pH were detected among caproic acid concentrations at the same time point (P > 0.05).

The CCK-8 assay showed lower myoblast viability after caproic acid treatment in a concentration- and time-dependent manner. At 24 and 48 h, 500 μmol/L caproic acid reduced cell viability relative to the untreated control (P < 0.01). At 72 h, cell viability was lower in the 100–500 μmol/L treatment groups than in the control group (P < 0.001; Fig. 6B). After 48 h of treatment with 300 or 400 μmol/L caproic acid, PCNA and CCND1 mRNA expression and protein abundance were lower than those in the control group (P < 0.01; Fig. 6C,D).

Consistent with the changes in PCNA and CCND1, treatment with 300 or 400 μmol/L caproic acid reduced the proportion of EdU-positive myoblasts (P < 0.05; Fig. 6E). Flow-cytometric analysis showed an increased proportion of cells in the G0/G1 phase after caproic acid treatment (P < 0.05; Fig. 6F).

Effects of Caproic Acid on Myoblast Apoptosis

Apoptosis-related responses were also altered after caproic acid treatment. The mRNA expression of BID, FAS, CASP3, and BAX increased in primary myoblasts (P < 0.05; Fig. 7A). Western blotting showed lower BCL-2 protein abundance and higher BAX protein abundance in caproic-acid-treated cells than in control cells (P < 0.05; Fig. 7B). Annexin V-FITC/PI staining showed increased early and late apoptosis rates after caproic acid treatment (P < 0.001; Fig. 7C).

Fig. 7.

Effects of caproic acid exposure on apoptosis-related responses in primary chicken myoblasts. (A) Relative mRNA expression of the apoptosis-related genes BID, FAS, CASP3, and BAX measured using RT-qPCR. (B) Representative Western blot images and densitometric quantification of BAX and BCL-2 proteins. (C) Flow-cytometric analysis of apoptosis using Annexin V-FITC/PI double staining, with quantification of early and late apoptosis. Data are presented as mean ± SD from three independent experiments. Tukey-adjusted P values were used for multiple-group comparisons. Exact P values are shown in the figure.

Effects of Caproic Acid on AMPK/MAPK Signaling and MYOD1 Expression

Changes in AMPK/MAPK-associated signaling were subsequently assessed. Western blotting showed that the p-AMPK/AMPK and p-p38/p38 ratios were higher in caproic-acid-treated primary myoblasts than in control cells (P < 0.05; Fig. 8A).

Fig. 8.

Effects of caproic acid exposure on AMPK/MAPK-associated signaling and MYOD1 expression in primary chicken myoblasts. (A) Representative Western blot images and densitometric quantification of AMPK, p38, and ERK1/2 signaling proteins, including the p-AMPK/AMPK, p-p38/p38, and p-ERK/ERK ratios. (B) Relative MYOD1 mRNA expression measured using RT-qPCR. Data are presented as mean ± SD from three independent experiments. Tukey-adjusted P values were used for multiple-group comparisons. Exact P values are shown in the figure.

Caproic acid treatment also increased MYOD1 mRNA expression (P < 0.05; Fig. 8B). In contrast, the p-ERK/ERK ratio was lower in caproic-acid-treated myoblasts than in control cells (P < 0.05; Fig. 8A).

Discussion

Our previous study identified a rearing-system-associated difference in cecal caproic acid abundance, together with broader differences in intestinal metabolism and microbial homeostasis in Lueyang Black-boned (LYB) chickens (Li et al., 2025). This observation provided the starting point for the present study. The cage-reared (CR)-versus-free-range (FR) comparison of caproic acid abundance was retained to make this rationale transparent. The longitudinal analyses then focused on FR birds to characterize age-related changes in caproic acid, the cecal metabolome, and microbial features. Primary-myoblast experiments were used to test whether caproic acid altered muscle-cell responses under controlled conditions. This design connects a rearing-system-associated metabolite with its cecal temporal context and cellular relevance.

Within FR birds, caproic acid followed a non-monotonic trajectory, increasing at 150 d, declining at 180 d, and increasing again at 360 d. Its abundance remained relatively stable across the corresponding ages in CR birds. The transient decline at 180 d may coincide with physiological changes associated with reproductive maturation in female birds. Reproductive endocrine and metabolic changes have been accompanied by gut microbial remodeling in humans (Koren et al., 2012). In FR hens, this physiological transition occurred within a rearing context characterized by outdoor access, greater opportunities for locomotor activity, and potential exposure to plant-derived materials during foraging. These factors can influence the avian microbiome and host physiology (Chen et al., 2019; Song et al., 2022). The temporal pattern was also embedded in a broader cecal metabolic signature. Cluster 3, which contained caproic acid, was negatively correlated with intramuscular fat content, abdominal fat percentage, and serum lipid variables, but positively correlated with breast muscle yield and meat color traits. Medium-chain fatty acids can influence cellular energy metabolism (Schönfeld and Wojtczak, 2016). Collectively, these data position caproic acid as an age-responsive component of a coordinated cecal metabolic state associated with host phenotypes, rather than as the sole mediator of those phenotypes. Because reproductive endocrine status, forage intake, and physical activity were not measured, these factors remain plausible explanations rather than experimentally tested mechanisms.

The metabolic pattern developed alongside age-related restructuring of the cecal community. Observed, Chao1, and ACE indices remained stable, indicating that the number of detected taxa did not change markedly across ages. In contrast, the Shannon, Simpson, and InvSimpson indices, together with the Bray-Curtis ordinations and PERMANOVA, showed age-dependent changes in community evenness and composition. The data therefore indicate redistribution of relative community structure rather than a simple gain or loss of taxa. Age-related succession of the chicken gut microbiota has been described during early development, with community assembly involving successive changes in the relative abundance of bacterial populations (Jurburg et al., 2019). Longitudinal work in laying hens has further shown that cecal bacterial populations continue to change across the life span (Videnska et al., 2014). Recent profiling across different gut sections also indicates that microbial composition varies with both host age and intestinal location (Bajagai et al., 2024). These observations are consistent with the age-related restructuring observed in the present FR chickens. This community reorganization provides a functional context for the metabolite results, because carbohydrate-processing capacity is distributed across microbial populations with different enzyme repertoires. At 120 d, GH3 and GH13_9 were among the families with higher relative abundance. GH3 includes several beta-glycosidase activities, whereas GH13_9 belongs to the broader GH13 group, whose members act on alpha-linked glucans and related glycans but encompass diverse catalytic activities. Several glycosyltransferase, carbohydrate-esterase, and carbohydrate-binding families also characterized the early profile. These CAZyme categories represent different aspects of potential glycan processing. Glycoside hydrolases cleave glycosidic bonds, whereas glycosyltransferases catalyze the transfer of glycosyl groups. Carbohydrate esterases can remove ester-linked substituents that influence substrate accessibility, while carbohydrate-binding modules contribute to enzyme–substrate recognition without directly catalyzing bond cleavage (Drula et al., 2022). The distribution of these CAZyme classes is consistent with the capacity of gut microbial communities to process structurally diverse dietary glycans (Cantarel et al., 2012; Kaoutari et al., 2013).

The later CAZyme profiles indicate further reorganization of the predicted glycan-processing repertoire. GH2, GH35, GH78, and GH97 showed transient enrichment at 180 d and include activities that can act on terminal β-galactosyl, α-l-rhamnosyl, α-glucosyl, or α-galactosyl residues. By contrast, GH28, GH92, and CBM20 increased progressively across the sampling ages. GH28 includes pectin- and galacturonan-related hydrolases, GH92 mainly comprises α-mannosidases, and CBM20 is a noncatalytic module commonly associated with binding to starch or glycogen-like α-glucans (Ndeh et al., 2017; Drula et al., 2022). Reporter Score shifts in the phosphotransferase system, pentose and glucuronate interconversions, and galactose metabolism extend this interpretation to predicted carbohydrate uptake and processing functions. Polysaccharide utilization loci provide one established mechanism through which many Bacteroidota access complex glycans (Grondin et al., 2017). Coordinated glycan processing can release oligosaccharides and fermentation intermediates that become available to other microbial populations (Flint et al., 2012; Grondin et al., 2017). Studies of gut microbial ecosystems have shown that cross-feeding can involve partially degraded carbohydrates as well as fermentation intermediates, including lactate, acetate, succinate, and 1,2-propanediol (Belenguer et al., 2006; Louis and Flint, 2017; Culp and Goodman, 2023). Fermentation of dietary carbohydrates generates short-chain fatty acids, which serve as energy substrates for intestinal epithelial cells and participate in host metabolic signaling (Koh et al., 2016; Liu et al., 2024).

Against this functional background, the negative associations of several Bacteroides, Paraprevotella, and Prevotella taxa, together with the positive associations of Pseudoflavonifractor, Flavonifractor, and Massilistercora timonensis, are compatible with age-related differences in substrate processing and secondary fermentation. Caproic acid can be produced through anaerobic chain elongation via reverse β-oxidation, in which short-chain carboxylates are elongated to medium-chain products (Spirito et al., 2014). The efficiency of this process depends on substrate availability, electron donors, and environmental conditions in the surrounding microbial community (Angenent et al., 2016). Lactate-based reactor studies have further demonstrated that n-caproate can arise from the conversion of carbon-rich substrates through competing fermentation routes (Kucek et al., 2016). A broader review of anaerobic n-caproic acid production provides additional context for these biochemical routes (Cavalcante et al., 2017). The positive association between M. timonensis and cecal caproic acid is of particular interest because the relative abundance of this species was inversely correlated with serum cholesterol in a human metagenomic association study (Kalnina et al., 2023). This direction is consistent with the lipid-related associations observed for Cluster 3 and identifies M. timonensis as a candidate for targeted validation in chicken cecum. Methanocorpusculum taxa also showed positive correlation coefficients with caproic acid and were prioritized by the age-classification analyses. Hydrogen transfer is an important form of metabolic interaction in anaerobic microbial communities and can affect the redox balance of fermentative processes (Stams and Plugge, 2009). Hydrogen-consuming microorganisms have therefore been considered potential cross-feeders in gastrointestinal ecosystems (Smith et al., 2019). The positive correlations observed for Methanocorpusculum taxa provide an ecological lead for future validation, but do not establish their direct involvement in caproic acid formation. Together, the diversity, CAZyme, pathway, and correlation data are consistent with age-dependent reorganization of the cecal fermentation environment. This reorganization may influence the availability of intermediates relevant to caproic acid formation.

This ecological and metabolic context provided the rationale for direct cellular testing. Experimental work in mice has shown that gut microbial status can influence skeletal muscle mass and function, with microbial metabolites contributing to the phenotype (Lahiri et al., 2019). This cross-species evidence does not identify a cecal-to-muscle route in chickens, but it supports testing a cecal metabolite in a muscle-cell model. Primary myoblasts were used because proliferation, survival, cell-cycle withdrawal, and differentiation are coordinated components of muscle development. Embryonic and post-hatch muscle development in chickens depends on the activity of myogenic precursor cells (Velleman, 2007). Studies of muscle stem-cell biology further indicate that changes in proliferation and survival can influence subsequent cell-state transitions (Sousa-Victor et al., 2022). Caproic acid reduced CCK-8 viability, EdU incorporation, and PCNA and CCND1 expression, while increasing G0/G1 accumulation and apoptosis-related responses. The increase in BAX and decrease in BCL-2 protein abundance were consistent with a shift toward proapoptotic regulation of the mitochondrial pathway, in which BCL-2-family proteins regulate mitochondrial outer-membrane permeabilization (Tait and Green, 2010). BID can connect death-receptor signaling to mitochondrial apoptosis, whereas FAS and CASP3 encode a death receptor and an executioner caspase, respectively. Their increased transcript levels, together with the Annexin V-FITC/PI results, support an apoptosis-associated response to caproic acid without establishing activation of a specific apoptotic pathway. A related coupling of reduced proliferation, G1 accumulation, and apoptosis has also been observed in chicken myoblasts (Jia et al., 2017). Culture-medium pH remained close to 7.2 and did not differ among treatment groups at the same time point. This finding makes a major contribution of treatment-related differences in bulk medium pH to the observed cellular responses less likely. A positive association between Cluster 3 and breast muscle yield does not imply that caproic acid should continuously stimulate precursor-cell proliferation. Instead, the present findings indicate a change in the balance between proliferation and survival. The accompanying apoptosis means that G0/G1 accumulation cannot by itself be interpreted as terminal differentiation.

The cell phenotype was accompanied by coordinated changes in AMPK and MAPK-associated signaling. The p-AMPK/AMPK and p-p38/p38 ratios increased, whereas the p-ERK/ERK ratio decreased and MYOD1 mRNA increased. AMPK functions as a cellular energy sensor and responds to changes in energy availability (Hardie et al., 2012). The ERK1/2 and p38 MAPK pathways are involved in the regulation of proliferation, stress responses, and myogenic progression (Keren et al., 2006; Segalés et al., 2016). MAPK signaling has also been linked to cell proliferation, differentiation, migration, senescence, and apoptosis in different cellular contexts (Sun et al., 2015). p38-dependent transcriptional programs have also been linked to cell-cycle exit and MYOD1-associated gene expression (Lluís et al., 2005; Perdiguero et al., 2007). In chicken skeletal-muscle satellite cells, p38-MAPK signaling has been implicated in the coordinated regulation of proliferation and differentiation (He et al., 2021). The signaling pattern observed here therefore offers a molecular context for the altered myoblast state and identifies AMPK, p38, ERK, and MYOD1 as defined targets for future functional experiments. It does not, by itself, resolve the causal order among these pathways or establish myogenic differentiation.

Overall, this study integrates a rearing-system-associated difference in cecal caproic acid abundance with FR-focused longitudinal multi-omics and primary-myoblast experiments. Within FR birds, age-dependent cecal metabolic and microbial functional-potential changes were associated with caproic acid dynamics, and caproic acid altered myoblast responses under the culture conditions tested. These complementary evidence streams identify caproic acid as one biologically relevant candidate within a broader cecal metabolic signature. The design was not a full-factorial rearing-system trial or an in vivo caproic acid intervention. Direct assignment of caproic acid production to individual taxa, measurement of cecal-to-muscle metabolite transport, and verification of in vivo effects will require targeted intervention, metabolite tracing, and flux-based experiments. Because feed intake and feed conversion were not recorded at the replicate level and stocking density was not matched between systems, the CR-versus-FR comparison reflects the composite rearing context. It cannot isolate the contribution of cecal metabolites from that of other housing-related factors.

Conclusion

In conclusion, this study extends previous CR-versus-FR observations by showing that cecal caproic acid abundance was higher in FR chickens and varied across ages within the FR system. These temporal changes coincided with age-related variation in cecal microbial composition, predicted microbial functional potential, and metabolite profiles. In primary chicken myoblasts, caproic acid exposure altered viability, proliferation-related responses, cell-cycle distribution, apoptosis-related markers, and AMPK/MAPK-associated signaling under conditions in which culture-medium pH did not differ among treatment concentrations. Together, the findings support a potential connection between cecal caproic acid dynamics and muscle-related cellular responses in chickens. Targeted in vivo intervention and functional validation are required to define the causal contribution of caproic acid to gut–muscle communication and muscle phenotypes.

Data Availability

The raw metagenomic sequencing data have been deposited in the NCBI Sequence Read Archive under BioProject accession PRJNA1138003. Supplementary tables and datasets accompanying this article provide metabolite annotations, differential metabolite analysis results, microbial functional abundance profiles, host phenotypic measurements, and Western blot source images, as detailed below.

Author Contributions

Tao Li conceived and designed the study, performed the experiments, curated and analyzed the data, wrote the analysis scripts, and drafted the manuscript. Tong Guo contributed to the methodology and investigation and critically reviewed the manuscript. Peng Wang contributed resources, data curation, and formal analysis and critically reviewed the manuscript. Zhuo Zhi contributed resources and investigation and critically reviewed the manuscript. Yungen Cao contributed to validation and formal analysis and critically reviewed the manuscript. Mengmeng Cui contributed to visualization and investigation and critically reviewed the manuscript. Jianqin Zhang contributed to the conception and design of the study and provided supervision, project administration, funding acquisition, and critical revision of the manuscript. Each author made a substantial contribution to the work, participated in drafting or critically revising the manuscript, approved the final version, and agrees to be accountable for all aspects of the work, including the investigation and resolution of questions concerning its accuracy or integrity.

Disclosures

The authors declare no competing financial interest.