2-Ethylfuran, methyl butanoate, 2-methylthiophene, methyl hexanoate, 3-octen-2-one, 2-pentylthiophene, dimethyl sulfone, methanethiol, ethyl acetate, hexanal, acetoin, nonanal, (E)-2-nonenal, (Z)-5-octen-1-ol, (E,E)-2,4-decadienal, dimethyl trisulfide, 1-octen-3-one, 1-octen-3-ol, 2-pentylpyridine, 2-acetyl-2-thiazoline, butanoic acid, methyl ester, and 2-methylthiophene were supplied by Aladdin Biochemical Technology (Shanghai, China) as analytical-grade standards. The analytical standards n-alkane mixture (C₇–C₄₀), 4-heptanone, 2-heptanone, 3-octanone, 2-octanone, γ-nonalactone, heptanal, octanal and pentanal were from Sigma-Aldrich (Shanghai, China). The gas chromatography standards 2-butylfuran, 2-heptylfuran and 2-pentylfuran were supplied by Alfa Aesar (Shanghai, China). Seventeen amino acids reference standards including glycine, L-alanine, L-serine, L-proline, L-valine, L-threonine, L-leucine, isoleucine, L-aspartic acid, L-lysine, L-glutamic acid, L-methionine, L-histidine, L-phenylalanine, L-arginine, L-tyrosine and L-cystine, each with purity ≥98%, were purchased from Alta Scientific Ltd. (Tianjin, China).

All animal procedures were conducted according to the regulations of the Animal Care and Use Committee of the Institute of Animal Sciences of the Chinese Academy of Agricultural Sciences (No. IAS 2020-69). In brief, flocks of 176 Hu male sheep and 76 Tan male sheep, with similar weights, were reared in the same environment, in individually ventilated pens (0.8 × 1.0 m², one sheep per pen). After weaning at 56 days old, all animals were fed ad libitum on total mixed ration pellets (Supplementary Table S1) and given free access to water until 6 months old. After fasting for 12 h, all animals were electrically stunned for 3 s (SQ05A stunner, Wujiang Aneng Electronic Technology, Suzhou, China), exsanguinated, eviscerated, and split. Psoas major muscles between the right 1st and 4th lumbar vertebrae were collected at 45 min postmortem for IMF, fatty acid, and amino acid analysis. The remaining psoas major muscles of each left and right carcass were vacuum-packed, and chilled for 24 h (2–4°C) for later determination of meat quality characteristics and volatile analysis. Following analysis of the IMF content, 40 individuals, 20 replicates per group, were selected according to the IMF content in order to reduce the impact of IMF variation within breed on characteristics differences between breeds.

Muscle pH was recorded with a pH meter (HI99163, Hanna Instruments Inc., Washington, DC, United States). Calibration of the pH probe was performed using pH 4.0 and 7.0 standard buffers. Color coordinates (lightness, L*; redness, a*; yellowness, b*; hue angle; chroma) were determined at three random locations using a Minolta CR-400 colorimeter, equipped with a D65 illuminator, 8 mm aperture and 2° viewing angle (Konica Minolta Sensing Inc., Osaka, Japan). Chroma ((a∗)2+(b∗)2) and hue angle (tan−1a∗b∗) values were calculated according to Honikel (23). Pressing loss was calculated as the difference in weight before and after pressing, divided by initial weight, as reported previously (24). Briefly, a muscle core sample (25 mm diameter, 10 mm thickness) was weighed, pressed for 5 min under a force of 343 N using a dilatometer, then reweighed. Drip loss and cooking loss of psoas major muscles were measured as reported previously (25). For cooking loss determination, samples were weighed, sealed in bags, heated to center temperature of 70°C in a water bath (80°C), cooled, refrigerated, wiped dry, and reweighed. After determining cooking loss, the Warner-Bratzler shear force was conducted, on the same samples, with 10 strips (10 mm × 10 mm × 20 mm) using an HDP/BSW V-shaped blade attached to a texture analyzer (TA.XT. Plus, Stable Micro Systems, Godalming, United Kingdom).

Intramuscular fat was extracted using the Soxhlet extraction method reported previously (25) and indicated as g/100 g wet muscle sample. After saponification, fatty acids from muscle tissues were reacted with sodium methoxide to produce fatty acid methyl esters and methyl undecanoate was employed as internal standards. The obtained methyl esters were quantitated using an Agilent 6890 gas chromatograph with DB-23 capillary column (60 m × 0.25 mm i.d., 0.25 μm film thickness) coupled with a flame-ionization detector (Agilent Technologies, Santa Clara, CA, United States). Helium as carrier gas was set at a flow rate of 3.0 ml/min with a split ratio of 1 to 20. The temperature gradient was as follows: 100–220°C with increasing rate at 4°C/min, followed by 5°C/min to 250°C, which was maintained for another 5 min. The temperature of inlet was set at 250°C while that of flame-ionization temperature was shown at 280°C. Fatty acid methyl esters were identified using retention time, confirmed by authentic standards, and quantitated using internal standard based method (methyl undecanoate).

For total amino acid analysis, psoas major muscles (~1 g) were hydrolyzed with HCl (10 ml, 6 mol/l) at 110°C for 22 h. Hydrolysates were transferred, filtered, and diluted with deionized water to volume. The filters were dried under nitrogen and the residue redissolved in HCl aqueous solution (2 ml, 0.02 mol/l) for further analysis. Amino acid concentrations were determined on an L-8900 amino acid analyzer (Hitachi, Tokyo, Japan). Identification of amino acids was performed by comparison with authentic standards.

For free amnio acid analysis, psoas major muscles (~100 mg) and 1 ml of sulphosalicylic acid (30 g/l) were mixed, homogenized, and centrifuged for 15 min at 14000 rpm at 4°C. The supernatant was vortex-mixed with hexane, centrifuged, and filtered through a 0.22 μm membrane. The filtrate was carried out an ACQUITY ultra-performance liquid chromatography (UPLC, Waters Corp., Milford, MA, United States) coupled with triple-quadrupole mass spectrometer (SCIEX QTRAP 6500, Framingham, MA, United States) in the electron spray ionization (ESI) mode. Two microliters of the samples were injected into LC–MS/MS system using an autosampler kept at 10°C. An ACQUITY UPLC HSS T3 column (2.1 × 100 mm, particle size 1.8 μm, pore size 100 Å) was applied to separate target analysts in meat samples. Mobile phase A was deionized water while mobile phase B was methanol. The 10-min gradient conditions were set as follows: 0–3 min, 2% B; 3–7 min, 2–95% B; 7–8 min, 95% B; 8–8.1 min, 95–2% B; 8.1–10 min, 2% B. The MS parameters was an ion-spray voltage of 4500 V, turbo gun source temperature of 400°C and curtain gas of 300 psi. Free amino acids were quantitated by external standard curves containing varying concentrations of reference standards.

Short loins were cut into small pieces (30 mm × 25 mm × 15 mm), sealed in bags, vacuum packed, heated for 60 min in an 80°C-water bath, cooled, snap-frozen, and ground. Cooked meat samples (~4.0 g) were weighed into a 20 ml vial and 10 μl of 2-methyl-3-heptanone (0.05 μg/l in methanol) were added as an internal standard. The vial was incubated for 20 min at 55°C. A solid-phase microextraction fiber coated with a 50/30 divinylbenzene/Carboxen^®/polydimethylbenzene fiber (Supelco, Bellefonte, PA, United States) was exposed to the headspace for 40 min to extract the volatile compounds. After extraction, the fiber was injected to desorb for 3 min at 250°C in split mode, and the split injection were set with a split ratio of 5:1.

Volatile compounds were separated with a VF-WAXms capillary column (60 m × 0.25 mm i.d., 0.25 μm film thickness, Agilent Technologies, Santa Clara, CA, United States) attached to Q Exactive GC/Orbitrap-MS (Thermo Fisher Scientific, Waltham, MA, United States). Chromatographic conditions and identification of volatile compounds were carried out as previously reported (17). Volatile compounds were first semi-quantified by an internal standard method. Then compounds having odor activity values (OAVs) greater than 1 were accurately quantified using external standard curves, constructed with various concentrations of standards (10 ‍mg/L–10 ‍g/L) in an odorless matrix. After mixing with the odorless matrix, the reference compounds were extracted and determined under the same analysis conditions. The odorless matrix was prepared as described previously (26), and was mainly composed of benzene, ethylbenzene, o-xylene, styrene, and linear alkane hydrocarbons such as octane, heptane, and nonane. The OAVs were calculated as dividing each compound concentration by its reported odor threshold in water (27).

Data were expressed as the mean ± standard error of the mean. Statistical analyses were performed using the SPSS software (Version 22.0, SPSS, Inc., Chicago, IL, United States). The significance of the differences in meat quality parameters, fatty acid profiles in IMF, amino acid profiles, and odor-active volatile compounds were determined by multivariate analysis of variance using breed as a fixed factor and IMF as a covariate. Pearson’s correlation coefficients were conducted with meat quality characteristics and the IMF, and a two-sided test for significance was performed. The significant differences were identified as p < 0.05.

Large variations in intra-breed IMF content were found in psoas major muscles from Hu (1.38–6.93%) and Tan (1.83–6.39%) sheep, but no inter-breed difference in IMF content was observed (p > 0.05; Table 1). The mean IMF content of the selected samples was 3.27% ± 0.03% for Hu sheep and 3.18% ± 0.05% for Tan sheep, thereby more accurately representing the IMF average of each breed and better reflecting meat quality differences between breeds. The IMF content and meat quality characteristics in selected samples were shown in Table 2. There was no inter-breed difference in IMF, pH, or pressing loss (p > 0.05; Table 2). Drip loss was 32.5% lower for Tan sheep (p < 0.01; Table 2). Cooking loss and shear force were 8.2 and 28.5% higher, respectively for Tan sheep (p < 0.01; Table 2). Higher lightness, redness, yellowness, and chroma value, and a lower hue angle were observed in Tan sheep, indicating that muscles from Tan sheep had a more intense red color (p < 0.01; Table 2). Meat quality characteristics did not have an obvious correlation with the IMF content (Table 2).

Total glycine concentration was lower in Tan sheep compared to Hu sheep (p < 0.05; Figure 1). Free L-proline, L-arginine, L-tyrosine, and L-glutamic acid were 18.5–40.3% lower in Tan sheep against Hu sheep (p < 0.05; Figures 1E,F). Free L-cystine and L-aspartic acid were 22.4% and 13.6% higher, respectively in Tan sheep relative to Hu sheep (p < 0.05; Figure 1H). Tan sheep muscle was 378 mg/100 g lower (p < 0.05) in total saturated fatty acids, reflecting lower concentrations of C18:0, C16:0, and C14:0 than Hu sheep (p < 0.05; Figures 2A,B,F). Concentrations of C16:1, C18:2n6 trans, C18:3n6, and C20:2 were lower in Tan sheep muscle (p < 0.05; Figures 2C–E). No differences were detected for contents of C18:1n9 cis and C20:1n9 (p > 0.05; Figure 2C).

Fifty-three volatile compounds were identified and quantified in Hu and Tan sheep, namely, 14 aldehydes, nine ketones, eight esters, seven S-containing compounds, six alcohols, four furans, three acids, one lactone and one pyridine (Table 3). The total concentrations of nine chemical classes were quantified for inter-breed comparison (Figure 3); the concentration of γ-nonalactone was significantly decreased in psoas major muscles from Tan sheep against Hu sheep (p < 0.05), but no other differences between breeds were observed (p > 0.05; Figure 3A).

To assess volatile compounds’ contribution to the overall aroma, their concentrations and OAVs were compared. Eighteen compounds having OAVs greater than 1 were accurately quantitated using an external standard method. Table 4 lists the odorants, the selected ion fragments, calibration equations, and correlation coefficients. Good linearity of detector response for the odor-active volatile compounds was observed (R² > 0.99; Table 4). The major compounds in Tan sheep were hexanal at 42.89 mg/kg, followed by acetoin (13.14 mg/kg), pentanal (6055.35 μg/kg), nonanal (1102.16 μg/kg), heptanal (2697.63 μg/kg), (E,E)-2,4-decadienal (991.98 μg/kg), and 1-octen-3-ol (1527.90 μg/kg; Figure 3). Similarly, high levels of hexanal (38.52 mg/kg), acetoin (6339.21 μg/kg), 1-octen-3-ol (1333.58 μg/kg), pentanal (4931.36 μg/kg), and heptanal (2656.19 μg/kg) were found in Hu sheep. No differences between breeds were observed for hexanal, pentanal, (E,E)-2,4-decadienal, heptanal, acetoin, 1-octen-3-ol, nonanal, octanal, 2-pentylfuran, (E)-2-nonenal, ethyl acetate, 2-pentylpyridine, dimethyl trisulfide, methanethiol, 1-octen-3-one, (Z)-5-octen-1-ol, (E)-2-octenal, and 2-acetyl-2-thiazoline (p > 0.05; Figures 3B–D).

As presented in Figure 3E, the concentrations of 18 odor-active volatile compounds were greater than their odor thresholds (OAV > 1) in Hu and Tan sheep, indicating that these compounds make important contributions to the aroma profile of lamb meat. The compound with the highest OAV in Tan sheep was 1-octen-3-one, followed by hexanal, octanal, (E)-2-octenal, and (E,E)-2,4-decadienal with OAVs > 1,000, whereas 2-pentylpyridine, 2-pentylfuran, ethyl acetate, and (Z)-5-octen-1-ol had low OAVs between 9 and 19. In addition, no significant differences were observed between Hu and Tan sheep in the 18 odor-active volatile compounds. Taking all together, the odor-active volatile compound profiles of meat from Hu sheep and Tan sheep with the 3–3.3% IMF differed only slightly.