Skip to main content
NCBI home page
Food Chemistry: X logoLink to Food Chemistry: X
. 2026 Jun 24;37:104137. doi: 10.1016/j.fochx.2026.104137

Metabolomics analysis reveals how feed intake-regulated nutritional compensation strategies promote flavor precursor accumulation and aroma formation in sheep meat

Yining Xie a,b,1, Xiaokang Jing a,1, Zhaohan Zhan a, Suo Langda c, Deqing Zhuoga c, Ruqing Zhong a, Bao Yi a, Liang Chen a,⁎, Hongfu Zhang a,⁎
PMCID: PMC13320271  PMID: 42389340

Abstract

Feed intake–regulated nutritional compensation is widely applied in ruminant production, yet its metabolic mechanisms underlying meat flavor formation remain unclear. This study used an integrated metabolomics approach to evaluate how graded feed restriction followed by refeeding regulates flavor precursor metabolism in Gangba sheep. Feed restriction at 85% of ad libitum intake increased taste-active free amino acids and monounsaturated fatty acids and enhanced aroma intensity detected by an electronic nose. Untargeted metabolomics revealed coordinated activation of carbohydrate and lipid metabolism, redox homeostasis, and bile acid–related pathways, establishing an anabolic state that promoted precursor accumulation and conversion. In contrast, moderate and high restriction shifted muscle metabolism toward oxidative and energy-compensatory pathways, limiting flavor improvement. These findings suggest that restriction at 85% of ad libitum intake followed by refeeding enhances meat flavor via coordinated metabolic regulation, providing a mechanistic basis for precision feeding strategies under feed-limited plateau systems.

Keywords: Feed intake restriction, Nutritional compensation, Sheep meat, Meat flavor, Aroma formation

Graphical abstract

Unlabelled Image

Open in a new tab

Highlights

  • •

    Metabolomics analysis linked muscle metabolism with aroma formation in Gangba sheep meat.

  • •

    Feed restriction at 85% of ad libitum intake increased flavor-related amino acids and fatty acids.

  • •

    Electronic nose analysis indicated enhanced aroma complexity and intensity under moderate restriction.

  • •

    Metabolic pathway analysis suggested coordinated regulation of energy metabolism and redox balance.

1. Introduction

Meat quality, nutritional value, and flavor characteristics are key determinants of consumer acceptance and market competitiveness of sheep products (Bryant & Barnett, 2018; Furnols & Guerrero, 2014). Among these attributes, flavor is the primary driver of consumer preference and repeat purchase, reflecting the integrated effects of muscle biochemistry and postmortem metabolism (Li et al., 2022; Yu et al., 2020). Flavor development is closely associated with the composition and metabolism of amino acids, fatty acids, and other small-molecule metabolites in muscle tissue (Fu et al., 2022; Sun et al., 2022). These compounds determine intrinsic taste properties and serve as precursors for volatile aroma compounds generated during cooking (Khan et al., 2015; Rizzi, 2008). Thus, nutritional strategies that regulate muscle metabolic composition represent a practical approach to improving meat quality and flavor.

Feed restriction followed by refeeding, also known as compensatory feeding, is widely applied in ruminant production to improve feed efficiency and regulate body composition (Addah et al., 2017; Hu et al., 2019). Previous studies have demonstrated that this strategy affects carcass traits, intramuscular fat deposition, and protein accumulation, with responses depending on restriction intensity (Deng et al., 2024; Hansen et al., 2006). In addition to growth performance, increasing evidence indicates that feed intake regulation can alter muscle amino acid and fatty acid profiles, which are key precursors of meat flavor (Wang et al., 2022). However, the metabolic mechanisms linking feed intake–regulated nutritional compensation to flavor precursor accumulation and aroma formation remain poorly understood.

Recent advances integrating metabolomics with electronic nose and electronic tongue technologies provide powerful tools to characterize flavor-related metabolic changes (Zhang et al., 2024; Zhou et al., 2024). However, such integrative approaches have rarely been applied to evaluate the effects of feed intake–regulated nutritional compensation on flavor formation in sheep. Gangba sheep, a representative Tibetan breed raised under fluctuating forage availability, provide a relevant model to investigate how nutritional regulation influences muscle metabolism and flavor development (Liu et al., 2025; Zhang et al., 2021). Therefore, this study aimed to evaluate the effects of graded feed restriction followed by refeeding on meat quality, flavor characteristics, and muscle metabolite profiles in Gangba sheep, and to elucidate the associated metabolic pathways underlying flavor precursor formation.

2. Materials and methods

2.1. Experimental design and feeding management

The experiment was conducted from May to October 2024 at the Gangba Sheep Breeding Farm, Gangba County, Shigatse City, Tibet Autonomous Region, China (28.262° N, 88.418° E). All animal procedures were approved by the Animal Ethics Committee of the Institute of Animal Science, Chinese Academy of Agricultural Sciences (Approval No. IAS2024–126). All procedures were conducted in accordance with relevant institutional guidelines and internationally recognized standards for the care and use of laboratory animals, equivalent to Directive 2010/63/EU on the protection of animals used for scientific purposes and the Guide for the Care and Use of Laboratory Animals published by the National Research Council.

Twenty-four healthy male Gangba sheep (initial body weight 16.22 ± 0.35 kg) were individually housed and randomly assigned to four treatments: ad libitum feeding (ALF), feed restriction at 85% (low-level restriction, LRF), 70% (moderate-level restriction, MRF), and 55% (high-level restriction, HRF) of the ALF intake. Each treatment included six replicates (one sheep per pen), and the individual animal served as the experimental unit. The trial consisted of a 14-day adaptation, 30-day restriction, and 60-day compensatory refeeding period. Sheep were fed twice daily (09:00 and 17:00). During restriction, ALF sheep were fed ad libitum, with refusals maintained at 5–10%. Feed allowance was adjusted daily based on the previous day's feed intake to maintain this level of refusals. Feed refusals were collected and weighed daily to determine intake and calculate average daily feed intake (ADFI). The restricted groups were fed based on the daily feed intake of the ALF group recorded on the previous day, and feed allowance was adjusted according to the predefined restriction levels (85%, 70%, and 55%). During refeeding, all sheep were fed ad libitum. During the entire experimental period, no animals received antibiotics, veterinary drugs, or other clinical therapeutic interventions. All animals received the same pelleted complete diet formulated to meet the nutrient requirements for growing sheep recommended by the National Research Council (2007). Diet composition and nutrient levels are provided in Table S1.

2.2. Slaughter procedures and sample collection

At the end of the experiment, all sheep were transported to a local commercial abattoir and fasted for 24 h prior to slaughter. Slaughter was conducted according to national regulations (MARA, 2019). Fasting body weight and body measurements were recorded before slaughter. After exsanguination, visceral organs were excised and weighed to calculate organ indices. Hot carcass weight (30–40 min postmortem), carcass length and width, backfat thickness, and eye muscle area were measured (Honikel, 1998). Longissimus dorsi samples were collected between the 9th and 13th ribs, trimmed of connective tissue and fat, and divided into three portions for meat quality (4 °C), proximate composition (−20 °C), and metabolomic analyses (−80 °C), respectively.

2.3. Meat quality and nutritional composition analysis

At 45-min postmortem, the cut surface of the Longissimus dorsi muscle was exposed to air for blooming. Meat color parameters (L*, a*, and b*) were then measured using a CM-600d spectrophotometer (Konica Minolta, Tokyo, Japan) after calibration with a standard white plate according to the manufacturer's instructions. Measurements were performed under standard conditions (D65 illuminant, 10° observer angle, specular component excluded, SCE). The measuring aperture was placed in full contact with the muscle surface. Six measurements were taken at different locations for each sample, and the average value was used for statistical analysis. Muscle pH at 45 min (pH45min) was measured using a calibrated portable pH meter (Testo 205, Testo, Lenzkirch, Germany). The electrode was inserted approximately 3 cm into the caudal end of the Longissimus dorsi muscle, and the value was recorded after stabilization. Measurements were taken at three different positions and averaged. After storage at 4 °C for 24 h, muscle pH at 24-h postmortem (pH24h) was determined using the same procedure. Cooking loss was determined using approximately 50 g of meat samples (three subsamples per animal). Samples were wrapped in aluminum foil and cooked in a water bath set at 90 °C. The internal temperature was continuously monitored using a probe thermometer (Beijing Hongkong Instrument Co., China) in combination with a thermocouple thermometer (Sinomeasure, Hangzhou, China). Heating was terminated when the internal temperature reached 70 °C, and samples were immediately removed without additional holding time. This procedure followed standard meat cooking methods (Honikel, 1998). After cooling to room temperature (20 ± 2 °C) under ambient conditions, samples were reweighed. Cooking loss was calculated as the percentage of weight loss relative to the initial weight. Shear force was determined using six cylindrical cores randomly cut parallel to the muscle fiber direction from each cooked sample using a core sampler (MAasic200, BDS Maschinen GmbH, Düsseldorf, Germany). The cores were cut perpendicular to the fiber direction using a Warner–Bratzler shear device (GR Manufacturing, Manhattan, KS, USA) equipped with a V-shaped blade. The crosshead speed was set at 200 mm/min, and the peak shear force was recorded as the maximum force required to cut through each core. Six measurements were performed for each sample, and the average value was used for statistical analysis, following standard procedures (Destefanis et al., 2008; Honikel, 1998). Proximate composition (moisture, ash, crude protein, and crude fat) was determined according to AOAC methods (AOAC, 2020).

2.4. Electronic tongue analysis

Electronic tongue analysis was performed as previously described by Chen et al. (2022), with minor modifications. Approximately 5 g of minced meat was placed in a 50 mL centrifuge tube and extracted with 20 mL of ultrapure water preheated to 37 °C. The mixture was continuously shaken for 30 min, followed by the addition of another 20 mL of ultrapure water at the same temperature. The samples were then subjected to ultrasonic extraction at 37 °C for 30 min and centrifuged at 5000 rpm (approximately 4000 ×g) for 10 min. The supernatant was filtered through a sintered glass funnel and diluted (1:1, v/v) before analysis. Taste attributes, including sourness, bitterness, astringency, saltiness, umami, richness, aftertaste-A, and aftertaste-B, were analyzed using an electronic tongue system (TS-5000Z, Insent Inc., Tokyo, Japan) equipped with lipid membrane sensors simulating human taste receptors. Each sample was measured five times; the first two measurements were used for sensor stabilization and calibration, and the last three replicates were used for statistical analysis.

2.5. Electronic nose analysis

Electronic nose analysis was conducted as previously described by Wang et al. (2019) and Zhou et al. (2024). Volatile profiles of the Longissimus dorsi were analyzed using an electronic nose system (PEN3, Airsense Analytics, Schwerin, Germany). Approximately 1 g of minced sample was placed in a headspace vial and equilibrated prior to measurement. The system was equipped with 10 metal oxide semiconductor (MOS) sensors, including W1C (Aromatic), W5S (Broad range), W3C (Aromatic), W6S (Hydrogen), W5C (Aromatic aliphatic), W1S (Broad methane), W1W (Sulfur organic), W2S (Broad alcohol), W2W (Sulfur chlorine), and W3S (Methane aliphatic). Each sample was measured in triplicate.

2.6. Fatty acid profile analysis

Fatty acid composition was determined by gas chromatography–mass spectrometry (GC–MS) according to a previously described method (Liu et al., 2026). Approximately 1 g of meat was ground under liquid nitrogen and mixed with 4 mL of 0.5 mol/L methanolic sodium hydroxide, followed by saponification at 55 °C for 30 min. Subsequently, 5 mL of 15% (w/v) boron trifluoride–methanol solution was added for methylation at 55 °C for 3 min. After cooling, fatty acid methyl esters (FAMEs) were extracted with n-hexane, dried over anhydrous sodium sulfate, and the organic phase was collected for analysis. FAMEs were separated and detected using a GC–MS system (7890 A GC coupled with a 5975C MSD, Agilent Technologies, USA) equipped with a DB-WAX capillary column (30 m × 0.25 mm × 0.20 μm). Nitrogen was used as the carrier gas (0.80 mL/min), with an injector temperature of 250 °C. The oven program was 180 °C for 1 min, increased to 230 °C at 5 °C/min, and held for 8 min. The mass spectrometer operated in electron impact ionization mode (70 eV; m/z 35–550). Fatty acids were identified using FAME standards and the NIST library and quantified by peak area normalization. The proportions of polyunsaturated fatty acids (PUFA), MUFA, saturated fatty acids (SFA), and total fatty acids were calculated.

2.7. Amino acid profile analysis

The amino acid composition of the Longissimus dorsi muscle was determined using a method previously described by Liu et al. (2024). Approximately 0.5 g of meat sample was ground into a fine powder in liquid nitrogen and transferred into a hydrolysis tube. Then, 10 mL of 6 mol/L hydrochloric acid and about 0.2 mL of phenol as an antioxidant were added. The mixture was vortexed for 2 min and then frozen on ice for 3 min. The hydrolysis tube was sealed under a nitrogen atmosphere and incubated in a thermostatic drying oven (LBAO-150H, STIK, Guangzhou, China) at 110 ± 1 °C for 22 h. After hydrolysis, the samples were cooled to room temperature, filtered, and the hydrolysis tubes were rinsed several times with deionized water. The filtrates were combined and evaporated to dryness, and the residue was re-dissolved in 6 mol/L hydrochloric acid and made up to a final volume of 50 mL. The solution was filtered through a 0.22 μm membrane filter and transferred into autosampler vials. Individual amino acids were then identified and quantified using an automatic amino acid analyzer (L-8900, Hitachi, Tokyo, Japan). A total of 16 amino acids were identified and quantified, and the calibration curves were prepared using an amino acid mixed standard solution (Sigma-Aldrich, St. Louis, MO, USA), based on an external standard method.

2.8. Untargeted metabolomics analysis

Untargeted metabolomics analysis was performed using a liquid chromatography–tandem mass spectrometry (LC–MS/MS) platform based on methods previously described by Wang et al. (2023). Approximately 50 mg of frozen muscle was homogenized in pre-chilled organic solvent containing internal standards, vortexed, ultrasonicated in an ice–water bath, and centrifuged (14,000 rpm, 15 min, 4 °C). The supernatants were collected for analysis. Quality control (QC) samples were prepared by pooling equal aliquots from all samples and injected periodically throughout the sequence. Chromatographic separation was conducted on an ultra-high-performance liquid chromatography system coupled with a TripleTOF mass spectrometer (SCIEX, Framingham, MA, USA) using a C18 column (HSS T3, 2.1 × 100 mm, 1.8 μm). The column temperature was 40 °C, flow rate 0.40 mL/min, and injection volume 2 μL. The mobile phase consisted of water containing 0.1% formic acid or 25 mM ammonium acetate (A) and acetonitrile (B) with gradient elution. Mass spectrometry was performed in both positive and negative electrospray ionization modes (ESI+/ESI−) over m/z 50–1000. Data were acquired in information-dependent acquisition mode. Raw data were processed using Progenesis QI for baseline correction, peak detection, alignment, and normalization. Data were log-transformed and scaled prior to multivariate analysis. Metabolites were annotated by matching accurate mass, retention time, and MS/MS spectra against the Human Metabolome Database (HMDB), METLIN, and in-house libraries. Multivariate analyses were conducted in R (version 4.2.2) using the ropls package (version 1.6.2). Principal component analysis (PCA) and orthogonal partial least squares discriminant analysis (OPLS-DA) were applied. Differential metabolites were selected based on variable importance in projection (VIP > 1.0) and P < 0.05. To reduce potential false-positive findings, P-values were adjusted using the false discovery rate (FDR), and FDR-adjusted P-values were used to identify differential metabolites. Pathway enrichment analysis was performed using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database.

2.9. Correlation analysis

Correlation analysis was conducted among differential amino acids, fatty acids, odor sensor signals, and metabolites using Spearman's rank correlation in R (version 4.2.2). Hierarchical clustering heatmaps were generated with the pheatmap package, and correlation networks were visualized using Cytoscape (version 3.5.1). Significance was set at P < 0.05, P < 0.01, and P < 0.001.

2.10. Statistical analysis

All data were analyzed using the general linear model (GLM) procedure of SAS 9.4 (SAS Institute Inc., Cary, NC, USA) to perform analysis of variance (ANOVA). Results are presented as means ± standard error of the mean. For measurements conducted in subsamples or repeated determinations, the average value per experimental unit (animal) was used for statistical analysis. Differences were considered statistically significant at P < 0.05, and Tukey–Kramer's multiple comparison test was applied for post hoc comparisons. Values of 0.05 ≤ P < 0.10 were considered indicative of a trend. Orthogonal polynomial contrasts were further used to evaluate whether the responses of each parameter to different levels of feed restriction followed linear or quadratic patterns. All figures were generated using GraphPad Prism 9.5 (GraphPad Software Inc., La Jolla, CA, USA).

3. Results

3.1. Slaughter performance and visceral organ indices

Compared with the ALF group, fasting body weight, carcass weight, carcass length, and carcass width were decreased in the MRF and HRF groups, with a significant linear effect (P < 0.05; Table 1). In contrast, eye muscle area was significantly increased in the MRF and HRF groups and also exhibited a significant linear effect. Moreover, no significant differences were observed in visceral organ indices among different treatments (Table 2).

Table 1.

Effects of feed intake-regulated nutritional compensation strategies on slaughter performance and carcass traits of Gangba sheep.

 Average daily feed intake, ALF basis1
P-values2a – c
ALF LRF MRF HRF SEM Treatment Linear Quadratic
Fasting body weight, kg 26.65a 27.82a 24.32b 24.15b 1.381 0.039 0.021 0.503
Body height, cm 57.50 57.50 60.50 57.83 1.502 0.443 0.559 0.386
Chest circumference, cm 78.92 76.83 77.33 76.00 2.114 0.797 0.392 0.861
Carcass weight, kg 13.07a 13.73a 11.30b 11.35b 0.671 0.042 0.042 0.243
Carcass length, cm 59.67a 57.00ab 54.00bc 52.50c 1.224 0.003 0.000 0.640
Carcass width, cm 25.17a 23.67ab 20.83b 20.83b 0.881 0.007 0.001 0.173
Backfat thickness, mm 0.97 1.49 1.76 1.43 0.202 0.066 0.740 0.603
Eye muscle area, cm2 16.05b 16.52ab 17.56a 18.62a 0.643 0.048 0.007 0.529
Slaughter percentage, % 49.04 49.37 46.47 46.84 1.103 0.109 0.725 0.143
Open in a new tab
1

ALF, ad libitum feeding; LRF, low restricted feeding; MRF, medium restricted feeding; HRF, high restricted feeding.

2

Linear, linear effect of feed intake-regulated nutritional compensation strategies; Quadratic, quadratic effect of feed intake-regulated nutritional compensation strategies.

a – c

Values within a row with no common superscripts differ significantly (P < 0.05).

Table 2.

Effects of feed intake-regulated nutritional compensation strategies on the visceral organs of Gangba sheep.

 Average daily feed intake, ALF basis1
P-values2
ALF LRF MRF HRF SEM Treatment Linear Quadratic
Heart, % 0.65 0.66 0.68 0.70 0.043 0.790 0.336 0.850
Liver, % 1.59 1.54 1.52 1.66 0.102 0.794 0.682 0.377
Spleen, % 0.20 0.28 0.17 0.16 0.041 0.124 0.157 0.225
Lungs, % 1.18 1.37 1.24 1.33 0.105 0.552 0.498 0.633
Kidneys, % 0.57 0.62 0.56 0.73 0.042 0.326 0.179 0.427
Rumen, % 2.47 2.12 2.25 2.10 0.175 0.425 0.217 0.561
Reticulum, % 0.59 0.42 0.40 0.45 0.054 0.056 0.624 0.444
Omasum, % 0.36 0.39 0.32 0.42 0.043 0.354 0.499 0.375
Abomasum, % 0.57 0.57 0.56 0.59 0.031 0.416 0.654 0.658
Small intestine, cm 2630.58 2679.17 2570.83 2446.83 76.402 0.194 0.048 0.272
Duodenum, cm 96.67 97.17 94.83 93.67 3.751 0.904 0.507 0.826
Jejunum, cm 2301.67 2380.00 2289.33 2189.33 68.381 0.299 0.177 0.207
Ileum, cm 232.25 202.00 186.67 163.83 11.152 0.172 0.195 0.874
Large intestine, cm 173.08 226.17 186.00 187.00 13.614 0.064 0.980 0.702
Colon, cm 66.58 114.83 85.67 83.67 13.712 0.128 0.723 0.082
Cecum, cm 65.50 65.83 63.83 63.00 4.452 0.964 0.638 0.897
Rectum, cm 41.00 45.50 36.50 40.33 4.552 0.589 0.595 0.942
Open in a new tab
1

ALF, ad libitum feeding; LRF, low restricted feeding; MRF, medium restricted feeding; HRF, high restricted feeding.

2

Linear, linear effect of feed intake-regulated nutritional compensation strategies; Quadratic, quadratic effect of feed intake-regulated nutritional compensation strategies.

3.2. Meat quality and nutritional composition

Compared with the ALF group, feed intake–regulated nutritional compensation strategies had no significant effects on overall meat quality parameters (Table 3). Moisture and fat contents tended to differ among treatments (P = 0.06 and 0.07, respectively), with a significant linear effect across restriction levels (P < 0.05). Moreover, crude protein content of the Longissimus dorsi was increased in the MRF and HRF groups, showing a significant linear trend (P < 0.05).

Table 3.

Effects of feed intake-regulated nutritional compensation strategies on the meat quality and proximate composition of Gangba sheep.

 Average daily feed intake, ALF basis1
P-values2a – b
ALF LRF MRF HRF SEM Treatment Linear Quadratic
L* 38.71 39.82 37.46 40.97 1.072 0.150 0.366 0.275
a* 13.59 11.82 13.31 13.30 1.024 0.611 0.892 0.401
b* 15.36 14.58 14.75 15.36 0.923 0.898 0.966 0.458
pH45mn 6.30 6.24 6.10 6.13 0.094 0.356 0.115 0.601
pH24h 5.55 5.55 5.56 5.54 0.022 0.866 0.906 0.561
Cooking loss (%) 68.84 69.22 68.88 69.12 0.674 0.972 0.870 0.918
Shear force (N) 27.95 29.57 29.98 29.88 1.863 0.855 0.161 0.834
Moisture (%) 68.27 67.69 69.33 70.65 0.754 0.055 0.017 0.222
Crude protein (%) 19.62b 20.54ab 22.39a 22.82a 0.704 0.009 0.001 0.923
Fat (%) 9.71 9.87 7.35 6.80 0.948 0.070 0.016 0.714
Open in a new tab
1

ALF, ad libitum feeding; LRF, low restricted feeding; MRF, medium restricted feeding; HRF, high restricted feeding.

2

Linear, linear effect of feed intake-regulated nutritional compensation strategies; Quadratic, quadratic effect of feed intake-regulated nutritional compensation strategies.

a – b

Values within a row with no common superscripts differ significantly (P < 0.05).

3.3. Electronic tongue

Electronic tongue analysis of the Longissimus dorsi muscle indicated that feed intake–regulated nutritional compensation strategies had no significant effect on the overall taste profile of Gangba sheep meat (Fig. 1A–F). Because the response values of sourness, astringency, aftertaste-A, and saltiness were negative, these values indicate that the taste intensities were lower than those of the reference solution rather than the absence of these taste attributes. Accordingly, they were not subjected to further statistical comparison. In contrast, bitterness, umami, richness, and aftertaste-B showed no significant differences among the treatment groups.

Fig. 1.

Fig. 1

Open in a new tab

Evaluation of the taste characteristics of Longissimus dorsi muscle in Gangba sheep using an electronic tongue (n = 6). (A) Principal component analysis (PCA); (B) Radar chart; (C) Bitterness; (D) Umami; (E) Richness; (F) Aftertaste B.

3.4. Electronic nose

Electronic nose analysis showed that the feed intake–regulated nutritional compensation strategies significantly affected the volatile flavor profile of the Longissimus dorsi muscle. PCA (PC1 = 39.01%, PC2 = 28.65%; Fig. 2A) showed a trend of separation among the treatment groups, although some overlap between groups was observed, indicating differences in aroma-related characteristics. Radar plots and bar charts further illustrated the changes in specific sensor signal intensities (Fig. 2B–L). Specifically, compared with the ALF group, the LRF and MRF groups both exhibited significantly higher intensities of W1C-Aromatic, W3C-Aromatic, Arom-aliph, Broad-methane, Broad-alcohol, and Methane-aliph (P < 0.05).

Fig. 2.

Fig. 2

Open in a new tab

Evaluation of the flavor characteristics of Longissimus dorsi muscle in Gangba sheep using an electronic nose (n = 6). (A) Principal component analysis (PCA); (B) Radar chart; (C) W1C–Aromatic; (D) Broad range; (E) W3C–Aromatic; (F) Hydrogen; (G) Arom–aliph; (H) Broad–methane; (I) Sulfur–organic; (J) Broad–alcohol; (K) Sulph–chlor; (L) Methane–aliph；a – b Values within a row with no common superscripts differ significantly (P < 0.05).

3.5. Fatty acid profile

Fatty acid profile analysis of the Longissimus dorsi muscle showed that feed intake–regulated nutritional compensation affected fatty acid composition (Table 4). The contents of C16:0, SFA, and total fatty acids were increased in the LRF, MRF, and HRF groups, with significant linear and quadratic effects (P < 0.05). Meanwhile, C4:0, C14:1, and C23:0 were decreased in the MRF and HRF groups, showing a linear effect (P < 0.05). In addition, C17:0, C17:1, C18:1n9c, and MUFA were increased in the LRF group, all displaying a quadratic effect (P < 0.05). Furthermore, C6:0 was reduced in the MRF group (P < 0.05).

Table 4.

Effects of feed intake-regulated nutritional compensation strategies on the fatty acid profile of Gangba sheep (mg/100 g fresh muscle).

 Average daily feed intake, ALF basis1
P-values2a – c
ALF LRF MRF HRF SEM Treatment Linear Quadratic
C4:0 8.55a 10.81a 5.42b 4.74b 0.91 0.001 0.001 0.159
C6:0 8.21a 7.11ab 5.71b 7.43a 0.53 0.030 0.131 0.017
C8:0 5.18b 5.56b 4.66b 14.78a 2.47 0.028 0.023 0.060
C10:0 46.64 52.90 40.56 37.54 3.86 0.060 0.038 0.251
C12:0 29.98 31.84 26.24 28.64 3.53 0.719 0.549 0.926
C13:0 6.23ab 9.63a 4.81b 6.37ab 1.07 0.044 0.353 0.429
C14:0 624.06 759.22 641.72 610.42 64.32 0.363 0.588 0.209
C14:1 54.66a 51.05a 31.97b 21.46b 5.91 0.002 0.000 0.583
C15:0 40.33 36.96 39.05 33.98 9.92 0.973 0.731 0.927
C15:1 11.80b 22.29a 4.78c 3.13c 4.41 0.024 0.039 0.188
C16:0 2163.00b 3086.17a 3247.93a 3406.46a 145.41 <0.001 <0.001 0.016
C16:1 57.67 48.48 56.47 40.32 6.08 0.203 0.128 0.596
C17:0 194.94c 327.97a 286.12ab 230.41bc 23.52 0.004 0.548 0.001
C17:1 318.65b 518.09a 258.67bc 197.31c 37.11 <0.001 0.001 0.002
C18:1n9c 277.30b 451.26a 211.54b 148.99b 49.43 0.002 0.011 0.027
C18:2n6t 37.55 41.35 34.02 24.75 6.31 0.310 0.120 0.310
C18:3n3 73.42 87.37 67.86 109.43 20.52 0.504 0.346 0.513
C21:0 9.94 11.29 4.98 16.65 37.42 0.223 0.404 0.198
C22:0 19.30 22.48 17.87 15.40 2.88 0.421 0.226 0.363
C23:0 14.26a 10.91a 1.58b 1.83b 1.94 0.000 <0.001 0.369
C24:1 8.60 10.91 1.58 1.83 8.53 0.558 0.515 0.447
MUFA 728.68b 1097.99 a 572.19bc 420.38c 71.11 <0.001 0.000 0.002
PUFA 110.97 128.72 101.87 134.18 25.48 0.780 0.707 0.772
SFA 3170.63b 4372.87a 4326.66a 4414.65a 184.93 0.000 0.000 0.007
Total fatty acids 4010.28b 5599.58a 5000.73a 4969.20a 242.20 0.002 0.048 0.003
Open in a new tab

PUFA, polyunsaturated fatty acids, PUFA = C18:2n6t + C18:3n3.

MUFA, Monounsaturated fatty acids，MUFA = C14:1 + C15:1 + C16:1 + C17:1 + C18:1n9c + C24:1.

SFA, Saturated fatty acids, SFA = C4:0 + C6:0 + C8:0+ C10:0 + C12:0 + C13:0 + C14:0 + C15:0 + C16:0 + C17:0 + C21:0 + C22:0 + C23:0.

1

ALF, ad libitum feeding; LRF, low restricted feeding; MRF, medium restricted feeding; HRF, high restricted feeding.

2

Linear, linear effect of feed intake-regulated nutritional compensation strategies; Quadratic, quadratic effect of feed intake-regulated nutritional compensation strategies.

a – c

Values within a row with no common superscripts differ significantly (P < 0.05).

3.6. Amino acid profile

Amino acid profile analysis of the Longissimus dorsi muscle showed that feed intake–regulated nutritional compensation affected amino acid composition (Table 5). Compared with the ALF group, Asp, Thr, Ser, Gly, Ala, Leu, Tyr, Phe, Lys, His, and Arg were increased in the LRF group, all showing a significant quadratic effect (P < 0.05). Meanwhile, Glu was increased in both the LRF and MRF groups, also showing a quadratic effect (P < 0.05). In addition, Met and Pro were elevated in the LRF and HRF groups, displaying a significant linear effect (P < 0.05). Notably, Val and Ile showed increasing trends (P = 0.06 and 0.08, respectively), both exhibiting quadratic effects (P < 0.05).

Table 5.

Effects of feed intake-regulated nutritional compensation strategies on the amino acid profile of Gangba meat (g/100 g fresh muscle).

 Average daily feed intake, ALF basis1
P-values2a – c
ALF LRF MRF HRF SEM Treatment Linear Quadratic
Asp 1.99b 2.19a 2.10ab 2.00b 0.052 0.037 0.895 0.008
Thr 1.01b 1.12a 1.08ab 1.04b 0.033 0.039 0.802 0.009
Ser 0.87b 0.96a 0.93ab 0.89b 0.034 0.028 0.707 0.006
Glu 3.16b 3.49a 3.41a 3.31ab 0.063 0.029 0.304 0.008
Gly 0.90b 0.99a 0.95ab 0.89b 0.022 0.025 0.408 0.006
Ala 1.24b 1.37a 1.33ab 1.32ab 0.013 0.042 0.142 0.032
Val 1.05b 1.16a 1.10ab 1.09ab 0.044 0.064 0.584 0.029
Met 0.57b 0.65a 0.61ab 0.64a 0.016 0.032 0.043 0.256
Ile 0.96b 1.06a 1.02ab 1.00ab 0.034 0.077 0.487 0.040
Leu 1.73b 1.91a 1.84ab 1.76b 0.062 0.034 0.944 0.007
Tyr 0.75c 0.82a 0.80ab 0.76bc 0.036 0.031 0.811 0.005
Phe 0.94bc 1.03a 1.00ab 0.89c 0.034 0.001 0.093 0.000
Lys 1.97bc 2.16a 2.09ab 1.94c 0.076 0.011 0.437 0.002
His 0.79b 0.89a 0.84ab 0.78b 0.035 0.011 0.480 0.002
Arg 1.38bc 1.51a 1.46ab 1.36c 0.052 0.011 0.504 0.002
Pro 0.88c 0.97ab 0.93bc 1.03a 0.034 0.001 0.001 0.908
Total amino acids 20.19b 22.26a 21.49ab 20.69b 0.493 0.035 0.735 0.008
Open in a new tab
1

ALF, ad libitum feeding; LRF, low restricted feeding; MRF, medium restricted feeding; HRF, high restricted feeding.

2

Linear, linear effect of feed intake-regulated nutritional compensation strategies; Quadratic, quadratic effect of feed intake-regulated nutritional compensation strategies.

a – c

Values within a row with no common superscripts differ significantly (P < 0.05).

3.7. Untargeted metabolomics

In total, 2097 metabolites were identified in the Longissimus dorsi muscle by untargeted LC–MS analysis, with 1193 and 904 metabolites detected in the positive and negative ion modes, respectively. The PCA results (PC1 = 25.10%, PC2 = 12.10%; Fig. 3A) showed a partial separation among samples from different feeding groups, with some overlap between groups, indicating differences in muscle metabolic profiles. The OPLS-DA model exhibited acceptable explanatory and predictive performance (R2X(cum) = 0.47, R2Y(cum) = 0.82, Q2(cum) = 0.41). In the permutation test, the Q2 intercept was 0.085, which was markedly lower than the original model Q2 value, indicating good model validity with no evident overfitting (Fig. 3B).

Fig. 3.

Fig. 3

Open in a new tab

Multivariate analysis of metabolites in the Longissimus dorsi muscle of Gangba sheep under different feed intake–regulated nutritional compensation strategies. (A) Principal component analysis (PCA); (B) Orthogonal partial least squares discriminant analysis (OPLS-DA) score plot; (C) Venn plot; (D) Bar chart of differential metabolites among groups; (E) KEGG pathway topology analysis of LRF vs. ALF; (F) KEGG pathway topology analysis of MRF vs. ALF; (G) KEGG pathway topology analysis of HRF vs. ALF.

A total of 2044, 2047, 2018, and 1988 metabolites were detected in the ALF, LRF, MRF, and HRF groups, respectively (Fig. 3C). Compared with the ALF group, 68 metabolites were up-regulated and 61 were down-regulated in the LRF group; 84 were up-regulated and 102 were down-regulated in the MRF group; and 82 were up-regulated and 220 were down-regulated in the HRF group (Fig. 3D). KEGG pathway topology analysis showed that the differential metabolites in the LRF group were mainly involved in starch and sucrose metabolism, the pentose phosphate pathway, primary bile acid biosynthesis, and cofactor biosynthesis (Fig. 3E). A total of 17 differential metabolites were identified in these pathways (Table S2). Compared with the ALF group, the relative abundances of taurochenodeoxycholic acid, cholic acid, deoxycholic acid, taurocholic acid, nicotinic acid, fructose-6-phosphate, xanthine, D-erythrose-4-phosphate, glucose-6-phosphate, 4-amino-4-deoxychorismate, 3-methyldioxyindole, 3-hydroxybutyric acid, 3-methoxy-4-hydroxyphenylglycol glucuronide, and 2-oxobutanoic acid were significantly increased. However, the relative abundances of cortisol, pantetheine 4′-phosphate, and oxalyl-CoA were significantly decreased. The differential metabolites in the MRF group were mainly associated with alanine, aspartate and glutamate metabolism, the tricarboxylic acid (TCA) cycle, primary bile acid biosynthesis, and phenylalanine metabolism pathways (Fig. 3F), involving 17 differential metabolic features in total (Table S3). Compared with the ALF group, the relative abundances of taurochenodeoxycholic acid, taurocholic acid, cholic acid, glycocholic acid, L-aspartic acid, two putative N-acetylaspartic acid-related features, 3-hydroxybutyric acid, phenaceturic acid, L-xylulose, 2-oxobutanoic acid, fumaric acid, and L-malic acid were significantly increased. However, the relative abundances of 9-oxononanoic acid, 15-hydroxynorandrostene-3,17-dione glucuronide, cis-3-hexen-1-ol, and 2-phenylacetamide were significantly decreased. The differential metabolites in the HRF group were mainly related to the TCA cycle, neomycin/kanamycin/gentamicin biosynthesis, and α-linolenic acid metabolism pathways (Fig. 3G), involving a total of nine metabolites (Table S4). Compared with the ALF group, the relative abundances of ribostamycin, citric acid, isocitric acid, and dihydroorotic acid were significantly increased, whereas those of gentamicin C2, tobramycin, 9-oxononanoic acid, cis-3-hexen-1-ol, and uridine 3′-monophosphate were significantly decreased.

3.8. Spearman correlation analysis

The Spearman correlation heatmap revealed associations between electronic nose signals and fatty acid levels (Fig. 4A). Specifically, Methane-aliph and Broad-methane were negatively correlated with C6:0 (P < 0.05), while Broad-methane was positively associated with C17:1, C17:0, total fatty acids, MUFA, and C18:1n9c. W1C-Aromatic was negatively correlated with C6:0 (P < 0.01) and positively associated with total fatty acids and C17:0. Similarly, W3C-Aromatic and Broad-alcohol showed negative correlations with C6:0 and positive associations with total fatty acids, C17:0, MUFA, and C18:1n9c. In addition, Arom-aliph exhibited a negative correlation with C6:0 and positive associations with total fatty acids, C17:0, MUFA, and C18:1n9c.

Fig. 4.

Fig. 4

Open in a new tab

Spearman correlation heatmaps between electronic nose signals, fatty acids, amino acids, and muscle metabolomic profiles. (A) Correlations between electronic nose signals and fatty acid profiles. (B) Correlations between electronic nose signals and amino acid profiles. (C) Correlations between differential amino acids and fatty acids and differential metabolites identified from the untargeted metabolomic analysis of muscle. Red and blue colors indicate positive and negative correlations, respectively, and color intensity represents the magnitude of the Spearman correlation coefficient. Statistical significance is denoted by *P < 0.05, **P < 0.01, and ***P < 0.001. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

In addition, Spearman correlation heatmaps revealed associations between electronic nose signals and amino acid levels (Fig. 4B). Methane-aliph was positively correlated with multiple amino acids, including Tyr, Gly, His, total amino acids, Val, Glu, and Ala, as well as Ser, Leu, Lys, Arg, Phe, Thr, Asp, and Ile (P < 0.05). Broad-methane showed positive correlations with most amino acids, including Glu, Ala, Tyr, Gly, His, total amino acids, Ser, Leu, Lys, Arg, Phe, Thr, Asp, Val, and Ile. Moreover, W1C-Aromatic, Arom-aliph, Broad-alcohol, and W3C-Aromatic were positively correlated with a wide range of amino acids, including Tyr, Gly, His, total amino acids, Ser, Leu, Lys, Arg, Phe, Thr, Asp, Val, Ile, Glu, and Ala.

To explore the associations between key metabolites and changes in muscle fatty acids and amino acids, representative metabolites identified by KEGG pathway topology analysis were subjected to Spearman correlation analysis (Fig. 4C). Overall, multiple metabolites were associated with both fatty acid composition and amino acid profiles. Regarding fatty acids, nicotinic acid was positively associated with C17:0, C17:1, and total fatty acids, whereas fumaric acid and L-malic acid were negatively correlated with MUFA, C6:0, and C17:1. In addition, 9-oxononanoic acid and cis-3-hexen-1-ol were positively associated with MUFA and C18:1n9c, and also positively associated with C17:1, while fumaric acid and L-malic acid showed consistent negative associations with these fatty acids.

For amino acids, glycocholic acid and L-xylulose were positively correlated with multiple amino acids, including Gly, Lys, Arg, Ser, Glu, Val, Ile, Leu, Tyr, and Ala. Similarly, 3-methyldioxyindole showed positive correlations with Glu, Ala, Val, and Ile. Notably, several metabolites were associated with both fatty acids and amino acids. Taurocholic acid and taurochenodeoxycholic acid were positively associated with total fatty acids, C17:0, and multiple amino acids. In contrast, pantetheine 4′-phosphate, citric acid, isocitric acid, and cortisol were negatively correlated with fatty acids, including MUFA, C18:1n9c, and C17:1, as well as multiple amino acids. Meanwhile, L-aspartic acid showed positive associations with total fatty acids and several amino acids. Additionally, deoxycholic acid and cholic acid exhibited positive correlations with MUFA, C17:0, C17:1, and several amino acids, but negative correlations with C6:0.

4. Discussion

In this study, a feed intake-regulated nutritional compensation model (ALF, LRF, MRF, and HRF) was applied to Gangba sheep and evaluated from a metabolomics perspective by integrating carcass traits, muscle chemical composition, sensory-related instrumental responses, and metabolomic profiles. This integrative approach allowed a comprehensive assessment of how nutritional regulation modulates flavor precursor metabolism and aroma formation at both the biochemical and metabolic network levels. Overall, these strategies altered muscle chemical composition and flavor-related precursor profiles. Among all treatments, LRF achieved the most favorable balance between flavor precursor accumulation and the maintenance of overall nutritional value.

In the present study, LRF did not adversely affect slaughter performance or visceral organ development, consistent with previous reports showing that mild feed restriction enables effective nutritional regulation without compromising production efficiency (Kamalzadeh et al., 1998; Li et al., 2014; Li et al., 2015). In contrast, MRF and HRF reduced fasting body weight and several carcass traits, indicating that stronger restriction may prioritize basal energy metabolism and essential physiological functions. Increasing restriction intensity was accompanied by greater eye muscle area, higher crude protein, and lower fat content, suggesting preferential lean tissue and protein deposition during compensatory refeeding, consistent with compensatory growth theory (Hornick et al., 2000). Enhanced lean accretion likely increases the postmortem pool of amino acids and peptides released through proteolysis, which subsequently participate in Maillard and Strecker reactions to generate flavor precursors (Khan et al., 2015; Sun et al., 2022). Collectively, these carcass responses indicate a linkage between feed intake regulation, muscle nutrient composition, and flavor precursor formation.

Consistent with carcass changes, the amino acid and fatty acid profiles of the Longissimus dorsi were markedly modulated, both serving as fundamental chemical bases for meat flavor formation (Sun et al., 2022). Compared with ALF, LRF increased multiple amino acids and fatty acids, creating a metabolic environment favorable for flavor precursor deposition and volatile formation. Glutamate, aspartate, alanine, glycine, serine, and threonine were elevated, enhancing umami and sweetness while providing substrates for Maillard reactions (Lee et al., 2025; You et al., 2024). Branched-chain and aromatic amino acids, including leucine, isoleucine, valine, phenylalanine, and arginine, were also increased, promoting Strecker degradation and the formation of characteristic meaty and roasted aromas (Bai et al., 2021; Fu et al., 2022). Although electronic tongue responses were not statistically significant, the observed upward trends suggest potential contributions of synergistic taste effects and perception thresholds (Elhadad & Wu, 2025; Lan et al., 2025). LRF additionally promoted MUFA deposition, particularly C17:1 and C18:1n9c, key lipid-oxidation precursors of aldehydes and alcohols that contribute substantially to desirable meat aromas (Domínguez et al., 2019; Han et al., 2023). The coordinated enrichment of taste-active amino acids and MUFAs therefore provides a biochemical basis that may support improved sheep meat flavor.

The effects of feed intake regulation on muscle composition were further reflected in electronic nose responses to volatile profiles. A trend of separation among treatments was observed, indicating that differences in amino acid and fatty acid composition are associated with distinct aroma patterns. LRF exhibited stronger signals in aromatic (W1C, W3C, Arom-aliph) and broad-range volatile sensors, consistent with increased formation of aldehydes, alcohols, ketones, and aromatic compounds. Positive correlations between MUFAs, particularly C17:1 and C18:1n9c, and aromatic sensors suggest that lipid oxidation is an important contributor to aroma enhancement, with hexanal and nonanal imparting characteristic fatty and roasted notes (Song et al., 2022; Zhang et al., 2020). These lipid-derived volatiles act synergistically with Maillard products to shape the cooked meat aroma profile (Insausti et al., 2021). Amino acid-derived reactions also contributed, as elevated phenylalanine, tyrosine, and leucine favored Strecker degradation and the formation of key compounds such as phenylacetaldehyde and 3-methylbutanal (Park & Choi, 2025; Rizzi, 2008). These findings suggest that coordinated activation of lipid oxidation and Strecker pathways under LRF enhances the conversion of precursors into volatile compounds, resulting in a more intense and complex aroma. In contrast, weaker electronic nose responses in MRF and HRF suggest reduced conversion efficiency under stronger restriction, indicating a nonlinear relationship between feed intake regulation and aroma formation, with LRF being the most favorable.

Untargeted metabolomics further clarified the metabolic basis of flavor precursor formation under feed intake regulation. Multivariate analyses revealed a distinct shift in the overall muscle metabolic profile, with LRF enriching pathways related to carbohydrate, lipid, and amino acid metabolism, including starch and sucrose metabolism, the pentose phosphate pathway, bile acid biosynthesis, and cofactor metabolism. This coordinated enrichment suggests an anabolic state integrating energy metabolism and redox balance to favor precursor accumulation. Enhanced carbohydrate metabolism was associated with increased glycolytic intermediates and reducing sugars, supporting amino acid biosynthesis and Maillard reactions (Hosry et al., 2025; Shakoor et al., 2022). Meanwhile, activation of the pentose phosphate pathway elevated NADPH production, providing reducing power for fatty acid synthesis and desaturation, which may promote MUFA accumulation, particularly C18:1n9c (Khan et al., 2015; Park & Choi, 2025; Rimmer & Zumbaugh, 2024). Bile acid metabolism appears to represent an additional regulatory axis and showed positive correlations with fatty acids and amino acids. It supports lipid digestion, transport, and metabolic signaling, thereby facilitating nutrient utilization and flavor precursor deposition (Zhang et al., 2022).

Correlation network analysis further demonstrated that flavor precursor formation is associated with coordinated regulation across multiple metabolic pathways. Nicotinic acid was positively associated with C17:1, C17:0, and total fatty acids, highlighting the role of NAD/NADP-dependent redox systems in lipid synthesis and desaturation (Yang et al., 2016). In contrast, TCA cycle intermediates (citric, isocitric, fumaric, and malic acids) showed negative correlations with MUFAs, C18:1n9c, and several amino acids, indicating that enhanced oxidative metabolism may limit substrate availability for anabolic precursor formation (Fu et al., 2022). Pantetheine-4′-phosphate, a coenzyme A precursor, was similarly negatively correlated with MUFAs and amino acids, consistent with increased β-oxidation and reduced lipid deposition (Czumaj et al., 2020; Zhang et al., 2025). Negative associations between cortisol and flavor-related metabolites further suggest stress-related inhibition of anabolic processes. Collectively, LRF appears to be characterized by a more anabolic-oriented metabolic configuration with enhanced carbohydrate metabolism, redox balance, bile acid signaling, and lipid synthesis. This metabolic state may contribute to amino acid and MUFA accumulation and increase the availability of substrates for Maillard reactions, Strecker degradation, and lipid oxidation. In contrast, MRF displayed an intermediate phenotype, whereas HRF is associated with oxidative and stress-related metabolism, conditions unfavorable for efficient precursor accumulation and stable aroma formation.

Overall, feed intake-regulated nutritional compensation induced a gradient pattern of metabolic adaptation. LRF was associated with the most coordinated anabolic state, characterized by the synergistic accumulation of amino acids and MUFAs and enhanced flavor precursor retention. In contrast, MRF exhibited weaker metabolic coordination, whereas HRF was dominated by oxidative and stress-related metabolism that limited precursor deposition. Increasing restriction intensity therefore may promote a shift from anabolic to energy-compensatory and stress-associated metabolic states. These findings suggest that low-level feed restriction represents a promising precision nutritional strategy for improving sheep meat flavor quality. Although several features were annotated as aminoglycoside antibiotics, including Gentamicin C2, Tobramycin, and Ribostamycin, no antibiotics, veterinary drugs, or clinical therapeutic interventions were administered during the experimental period. These annotations may therefore reflect database matching uncertainty or environmental/background signals inherent to untargeted metabolomics. Furthermore, future studies incorporating larger-scale validation, targeted metabolite verification, and functional approaches are needed to further confirm the proposed metabolic associations. Although the E-nose analysis demonstrated clear divergence in aroma characteristics among groups, sensor-array-based profiling cannot identify or quantify specific volatile compounds. Therefore, targeted validation using gas chromatography–mass spectrometry–olfactometry (GC–MS-O) is required in future studies to structurally confirm the volatile compounds driving the observed aroma profiles. In addition, comprehensive sensory evaluation, including human panels, would facilitate a more robust validation of the relationship between metabolic changes and overall flavor perception.

5. Conclusion

Feed intake-regulated nutritional compensation modulated carcass traits, muscle nutrient composition, and flavor formation in Gangba sheep. Among the tested strategies, LRF most effectively promoted the accumulation of amino acids and monounsaturated fatty acids, thereby enhancing meat nutritional value and aroma complexity. Metabolomic analysis further revealed that LRF favored coordinated regulation of energy metabolism and redox balance, supporting efficient nutrient conversion and flavor precursor deposition at the metabolic network level. Therefore, from a metabolomics perspective, restriction followed by low-level refeeding represents an effective and mechanistically supported nutritional strategy to improve meat quality and flavor in Gangba sheep under high-altitude, feed-limited production conditions.

CRediT authorship contribution statement

Yining Xie: Writing – original draft, Validation, Methodology, Investigation, Data curation, Conceptualization. Xiaokang Jing: Visualization, Software, Methodology. Zhaohan Zhan: Methodology, Investigation. Suo Langda: Visualization, Software. Deqing Zhuoga: Writing – review & editing, Formal analysis. Ruqing Zhong: Software, Methodology. Bao Yi: Validation, Data curation. Liang Chen: Writing – review & editing, Supervision, Project administration, Formal analysis. Hongfu Zhang: Resources, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

This research was supported by the National Key Research and Development Program (2022YFD1302102), Agricultural Science and Technology Innovation Program (ASTIPIAS07) and China Scholarship Council (202503250014).

Footnotes

Appendix A

Supplementary data to this article can be found online at https://doi.org/10.1016/j.fochx.2026.104137.

Contributor Information

Liang Chen, Email: chenliang01@caas.cn.

Hongfu Zhang, Email: zhanghongfu@caas.cn.

Appendix A. Supplementary data

Supplementary material

mmc1.docx (37.2KB, docx)

Data availability

Data will be made available on request.

References

  1. Addah W., Ayantunde A., Okine E.K. Effects of restricted feeding and re-alimentation of dietary protein or energy on compensatory growth of sheep. South African Journal of Animal Science. 2017;47(3):389–398. doi: 10.4314/sajas.v47i2.15. [DOI] [Google Scholar]
  2. Bai S., Wang Y., Luo R., Ding D., Bai H., Shen F. Characterization of flavor volatile compounds in industrial stir-frying mutton sao zi by GC-MS, E-nose, and physicochemical analysis. Food Science & Nutrition. 2021;9(1):499–513. doi: 10.1002/fsn3.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bryant C., Barnett J. Consumer acceptance of cultured meat: A systematic review. Meat Science. 2018;143:8–17. doi: 10.1016/j.meatsci.2018.04.008. [DOI] [PubMed] [Google Scholar]
  4. Chen H., Pan D., Du H., Ma J., Kong B., Diao J. Flavor differences of edible parts of grass carp between Jingpo Lake and commercial market. Foods. 2022;11(17) doi: 10.3390/foods11172594. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Czumaj A., Jurga S.S., Hebanowska A., Turyn J., Swierczynski J., Sledzinski T., Stelmanska E. The pathophysiological role of CoA. International Journal of Molecular Sciences. 2020;21(23):9057. doi: 10.3390/ijms21239057. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Deng T., Liang M., Du L., Li K., Li J., Li Q.…Gao H. Transcriptome analysis of compensatory growth and meat quality alteration after varied restricted feeding conditions in beef cattle. International Journal of Molecular Sciences. 2024;25(5) doi: 10.3390/ijms25052704. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Destefanis, G., Brugiapaglia, A., Barge, M. T., & Molin, E. D. (2008). Relationship between beef consumer tenderness perception and Warner–Bratzler shear force. Meat Science, 78(3), 153–156. Doi: 10.1016/j.meatsci.2007.05.031. [DOI] [PubMed]
  8. Domínguez R., Pateiro M., Gagaoua M., Barba F.J., Zhang W., Lorenzo J.M. A comprehensive review on lipid oxidation in meat and meat products. Antioxidants. 2019;8(10) doi: 10.3390/antiox8100429. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Elhadad N., Wu J. Decoding the taste of peptides: Structure, interactions with taste receptors, bioactivities, and applications. Sustainable Food Proteins. 2025;3(2) doi: 10.1002/sfp2.70009. [DOI] [Google Scholar]
  10. Fu Y., Cao S., Yang L., Li Z. Flavor formation based on lipid in meat and meat products: A review. Journal of Food Biochemistry. 2022;46(12) doi: 10.1111/jfbc.14439. [DOI] [PubMed] [Google Scholar]
  11. Furnols M.F.I., Guerrero L. Consumer preference, behavior and perception about meat and meat products: An overview. Meat Science. 2014;98(3):361–371. doi: 10.1016/j.meatsci.2014.06.025. [DOI] [PubMed] [Google Scholar]
  12. Han D., Deng S., Wang H., Huang F., Fauconnier M.L., Li H., Li X. Lipid oxidation and flavor changes in saturated and unsaturated fat fractions from chicken fat during a thermal process. Food & Function. 2023;14(14):6554–6569. doi: 10.1039/d3fo01061a. [DOI] [PubMed] [Google Scholar]
  13. Hansen S., Byrne D.V., Therkildsen M. Effects of a compensatory growth strategy on sensory and physical properties of meat from young bulls. Meat Science. 2006;74(4):628–643. doi: 10.1016/j.meatsci.2006.05.014. [DOI] [PubMed] [Google Scholar]
  14. Honikel K.O. Reference methods for the assessment of physical characteristics of meat. Meat Science. 1998;49(4):447–457. doi: 10.1016/S0309-1740(98)00034-5. [DOI] [PubMed] [Google Scholar]
  15. Hornick J.L., Eenaeme C.V., Gérard O., Dufrasne I., Istasse L. Mechanisms of reduced and compensatory growth. Domestic Animal Endocrinology. 2000;19(2):121–132. doi: 10.1016/S0739-7240(00)00072-2. [DOI] [PubMed] [Google Scholar]
  16. Hosry L.E., Elias V., Chamoun V., Halawi M., Cayot P., Nehme A., Maroun E.B. Maillard reaction: Mechanism, influencing parameters, advantages, disadvantages, and food industrial applications: A review. Foods. 2025;14(11) doi: 10.3390/foods14111881. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Hu R., Zou H., Wang Z., Cao B., Peng Q., Jing X., Kong X. Nutritional interventions improved rumen functions and promoted compensatory growth of growth-retarded yaks as revealed by integrated transcripts and microbiome analyses. Frontiers in Microbiology. 2019;10:318. doi: 10.3389/fmicb.2019.00318. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Insausti K., Arbizu M.T.M., Urrutia O., Mendizabal J.A., Beriain M.J., Colle M.J., Arana A. Volatile compounds, odour and flavour attributes of lamb meat from the Navarra breed as affected by ageing. Foods. 2021;10(3) doi: 10.3390/foods10030493. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Kamalzadeh A., Koops W.J., Bruchem J.V., Tamming S., Zwart D. Feed quality restriction and compensatory growth in growing sheep: Development of body organs. Small Ruminant Research. 1998;29(1):71–82. doi: 10.1016/S0921-4488(97)00111-9. [DOI] [Google Scholar]
  20. Khan, M. I., Jo, C., & Tariq, M. R. (2015). Meat flavor precursors and factors influencing flavor precursors--A systematic review. Meat Science, 110, 278–284. Doi: 10.1016/j.meatsci.2015.08.002. [DOI] [PubMed]
  21. Lan J., Xiong Y., Dang K., Pan D., Du L., Wang Y., Dang Y. Rapid screening and taste mechanism of novel umami peptides from natural tripeptide database. Food Chemistry: X. 2025;28 doi: 10.1016/j.fochx.2025.102565. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Lee D., Lee S., Jo C. Application of animal resources into the Maillard reaction model system to improve meat flavor. Food Science of Animal Resources. 2025;45(1):303–327. doi: 10.5851/kosfa.2024.e133. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Li D.B., Liu X.G., Zhang C.Z., Kao G.L., Hou X.Z. Effects of nutrient restriction followed by realimentation on growth, visceral organ mass, cellularity, and jejunal morphology in lambs. Livestock Science. 2015;173:24–31. doi: 10.1016/j.livsci.2014.12.011. [DOI] [Google Scholar]
  24. Li H., Zheng R., Zuo F., Qian C., Yao Z., Dong R., Li C. Influence of proteolysis on the binding capacity of flavor compounds to myofibrillar proteins. Foods. 2022;11(6) doi: 10.3390/foods11060891. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Li L., Zhu Y., Wang X., He Y., Cao B. Effects of different dietary energy and protein levels and sex on growth performance, carcass characteristics and meat quality of F1 Angus × Chinese Xiangxi yellow cattle. Journal of Animal Science and Biotechnology. 2014;5(1):21. doi: 10.1186/20491891521. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Liu R., Fu J., Li S., Gu M., Li L., Xu L., Chen L. Insight into the meat quality differences of Tibetan sheep from different altitudes based on metabolomics. Food Chemistry. 2026;504 doi: 10.1016/j.foodchem.2026.147974. [DOI] [PubMed] [Google Scholar]
  27. Liu W., Gao H., He J., Yu A., Sun C., Xie Y., Lei Z. Effects of dietary Allium mongolicum regel powder supplementation on the growth performance, meat quality, antioxidant capacity and muscle fiber characteristics of fattening Angus calves under heat stress conditions. Food Chemistry. 2024;453 doi: 10.1016/j.foodchem.2024.139539. [DOI] [PubMed] [Google Scholar]
  28. Liu X., Ding H., Zhang X., Ta N., Zhao J., Zhang Q., Zhang X. Dynamic changes in the gastrointestinal microbial communities of Gangba sheep and analysis of their functions in plant biomass degradation at high altitude. Microbiome. 2025;13(1):17. doi: 10.1186/s40168-024-02022-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Park M.K., Choi Y.S. Effective strategies for understanding meat flavor: A review. Food Science of Animal Resources. 2025;45(1):165–184. doi: 10.5851/kosfa.2024.e124. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Rimmer L.A., Zumbaugh M.D. Skeletal muscle metabolic characteristics and fresh meat quality defects associated with wooden breast. Frontiers in Physiology. 2024;13 doi: 10.3389/fphys.2024.1501362. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Rizzi G.P. The Strecker degradation of amino acids: Newer avenues for flavor formation. Food Reviews International. 2008;24(4):416–435. doi: 10.1080/87559120802306058. [DOI] [Google Scholar]
  32. Shakoor A., Zhang C., Xie J., Yang X. Maillard reaction chemistry in formation of critical intermediates and flavour compounds and their antioxidant properties. Food Chemistry. 2022;393 doi: 10.1016/j.foodchem.2022.133416. [DOI] [PubMed] [Google Scholar]
  33. Song S., Zheng F., Tian X., Feng T., Yao L., Sun M., Shi L. Evolution analysis of free fatty acids and aroma-active compounds during tallow oxidation. Molecules. 2022;27(2) doi: 10.3390/molecules27020352. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Sun A., Wu W., Soladoye O.P., Aluko R.E., Bak K.H., Fu Y., Zhang Y. Maillard reaction of food-derived peptides as a potential route to generate meat flavor compounds: A review. Food Research International. 2022;151 doi: 10.1016/j.foodres.2021.110823. [DOI] [PubMed] [Google Scholar]
  35. Wang B., Zhao X., Zhang B., Cui Y., Nueraihemaiti M., Kou Q., Luo H. Assessment of components related to flavor and taste in Tan-lamb meat under different silage-feeding regimens using integrative metabolomics. Food Chemistry: X. 2022;14 doi: 10.1016/j.fochx.2022.100269. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Wang L., Su S., Zhao J., He X., Fu S., Wang B., Liu Y. Effects of dietary oat supplementation on carcass traits, muscle metabolites, amino acid profiles, and its association with meat quality of small-tail Han sheep. Food Chemistry. 2023;411 doi: 10.1016/j.foodchem.2023.135456. [DOI] [PubMed] [Google Scholar]
  37. Wang Q., Li L., Wu D., Zhang D., Wang J., Reed K., Zhang B. Adulterant identification in mutton by electronic nose and gas chromatography-mass spectrometer. Food Control. 2019;98:431–438. doi: 10.1016/j.foodcont.2018.11.038. [DOI] [Google Scholar]
  38. Yang Z.Q., Bao L.B., Zhao X.H., Wang C.Y., Zhou S., Wen L.H., Qu M.R. Nicotinic acid supplementation in diet favored intramuscular fat deposition and lipid metabolism in finishing steers. Experimental Biology and Medicine. 2016;241(11):1195–1201. doi: 10.1177/1535370216639395. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. You Z., Bai Y., Bo D., Feng Y., Shen J., Wang Y., Bai Y. A review of taste-active compounds in meat: Identification, influencing factors, and taste transduction mechanism. Journal of Food Science. 2024;89(12):8128–8155. doi: 10.1111/1750-3841.17480. [DOI] [PubMed] [Google Scholar]
  40. Yu M., Li Z., Rong T., Wang G., Liu Z., Chen W., Ma X. Different dietary starch sources alter the carcass traits, meat quality, and the profile of muscle amino acid and fatty acid in finishing pigs. Journal of Animal Science and Biotechnology. 2020;11:78. doi: 10.1186/s40104-020-00484-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Zhang B., Sun Z., Yu Z., Li H., Luo H., Wang B. Transcriptome and targeted metabolome analysis provide insights into bile acids' new roles and mechanisms on fat deposition and meat quality in lamb. Food Research International. 2022;162 doi: 10.1016/j.foodres.2022.111941. [DOI] [PubMed] [Google Scholar]
  42. Zhang C., Zhang H., Liu M., Zhao X.G., Luo H. Effect of breed on the volatile compound precursors and odor profile attributes of lamb meat. Foods. 2020;9(9):1178. doi: 10.3390/foods9091178. https://www.mdpi.com/2304-8158/9/9/1178 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Zhang J., Pang S., Yan G., Wang L., Xu Y., Bai Y., Jiang Y. Integrated electronic nose and multi-omics reveal changes in flavour characterization of cashmere goats and tan sheep meat. Food Chemistry: X. 2024;25 doi: 10.1016/j.fochx.2024.102042. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Zhang Q., Que M., Li W., Gao S., Tan X., Bu D. Gangba sheep in the Tibetan plateau: Validating their unique meat quality and grazing factor analysis. Journal of Environmental Sciences. 2021;101:117–122. doi: 10.1016/j.jes.2020.06.024. [DOI] [PubMed] [Google Scholar]
  45. Zhang W.T., Wang M.L., Liu C.Y., Li C., He X.P., Wang H.B., Cheng G. Rumen microbiota–host interactions regulate intramuscular fat deposition in cattle via the alpha-linolenic acid–fatty acid β-oxidation/L-carnitine–MPO axis. Microbiome. 2025;14. Article 49 doi: 10.1186/s40168-025-02252-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Zhou L., Ren Y., Shi Y., Fan S., Zhao L., Dong M., Tang C. Comprehensive foodomics analysis reveals key lipids affect aroma generation in beef. Food Chemistry. 2024;461 doi: 10.1016/j.foodchem.2024.140954. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary material

mmc1.docx (37.2KB, docx)

Data Availability Statement

Data will be made available on request.

Articles from Food Chemistry: X are provided here courtesy of Elsevier

RESOURCES