Effects of supercooling, sub−freezing, and deep−freezing pre−cooling on color, pH, shear force, sensory quality, and myofibrillar protein properties of beef during expanded polystyrene−based distribution Authors: Yuqing Tang, Depeng Sun, Qian Li, Jinghe Zhang, Baide Mu, Hongmei Li, Tingyu Li, Chunxiang Piao, Jian Feng, Guanhao Li, Juan Wang Published: 2026 ABSTRACT This study evaluated the effects of different pre−cooling phase states on beef quality during e−commerce distribution. Beef steaks were subjected to four treatments: 4 °C (CK), − 0.5 °C (supercooling), − 5 °C (sub−freezing), and − 18 °C (deep−freezing), followed by 48 h of EPS−based passive distribution at 25 °C and subsequent storage at 4 °C. Changes in color, pH, shear force, sensory quality, myoglobin oxidation, myofibril fragmentation index (MFI), and the biochemical and conformational properties of myofibrillar proteins (MPs) were analyzed. The results showed that pre−cooling temperature significantly affected beef quality during distribution and storage. Although deep−freezing (− 18 °C) provided better initial color protection, the associated ice crystal damage likely intensified moisture exudation, myoglobin oxidation, and protein oxidation during the subsequent temperature rise and refrigerated storage, which together may have contributed to structural disruption and faster quality deterioration. As a result, the − 18 °C and − 5 °C groups showed faster deterioration in color, texture, and sensory quality, accompanied by higher MetMb content, carbonyl content, surface hydrophobicity, and β − sheet proportion, as well as lower protein solubility. In contrast, the − 0.5 °C supercooling treatment showed the best overall performance by maintaining higher redness, better texture stability, and a longer period of sensory acceptability. At the molecular level, supercooling delayed MP oxidation, reduced hydrophobic group exposure and structural rearrangement, and better preserved the ordered secondary structure of proteins. Overall, − 0.5 °C supercooling was more effective than sub−freezing and deep−freezing in maintaining beef quality during EPS−based distribution. Introduction In recent years, with the rise of large e−commerce platforms and the continuous development of cold−chain logistics, the online consumption market for fresh beef has grown rapidly. To balance distribution cost and timeliness, the EPS box + ice pack distribution model has been widely used in fresh e−commerce logistics for the “last mile”, usually with a delivery time of 24–48 h. However, due to the limited insulation performance of the EPS box and the finite cooling capacity of the ice packs, meat products are highly susceptible to temperature fluctuations during complex transportation and distribution processes (Guo, 2025; Navaranjan, 2013; Xu, 2023). During summer, ambient temperatures in e-commerce cold-chain distribution may approach or exceed 25 °C, and such conditions are relevant to practical ambient environments encountered during short-distance delivery. Previous studies have reported that, under passive cooling systems, parcel temperatures commonly rise within the range of 20–30 °C during 24–48 h of transportation, and this range has also been widely adopted to evaluate the performance of cold-chain packaging systems (Waldhans et al., 2024). Therefore, 25 °C was selected in the present study as a representative ambient temperature for simulating summer e-commerce distribution, and a 48 h duration was used to reflect the typical delivery period of EPS box + ice pack systems. Previous studies have shown that such thermal shock during logistics is one of the key factors accelerating oxidative deterioration, drip loss, and microbial proliferation in fresh meat (Zhang, 2020). To alleviate heat intrusion during logistics, low‑temperature pre‑cooling treatment is usually carried out before shipment. Depending on the target temperature, the water in meat can exist in different phase states, including supercooling (approximately − 0.5 °C, with no or minimal ice formation), sub‑freezing (approximately − 5 °C, with partial ice formation), and deep‑freezing (approximately − 18 °C, with complete freezing) (Li, 2018; Stone, 2021). In theory, lower pre−cooling temperatures provide a greater cold reserve. However, during the freezing process and the subsequent temperature fluctuations during distribution, meat inevitably undergoes ice crystal formation, melting, and recrystallization—that is, repeated freeze–thaw cycles, which are known to induce physicochemical instability in muscle tissues (Kiani, 2011; Leygonie et al., 2012; Xu et al., 2021). Previous studies have confirmed that heterogeneous and coarse ice crystals can damage cell membranes and organelles, resulting in the release of endogenous pro‑oxidants and proteolytic enzymes, thereby causing irreversible quality deterioration in meat (Lund, 2011; Zhang, 2021). Myofibrillar proteins (MPs) are the core structural components that maintain the three−dimensional muscle network and are closely related to the water−holding capacity (WHC) and tenderness of meat. These proteins are particularly sensitive to oxidative stress induced by temperature fluctuations and ice crystal damage (Hughes, 2014; Koohmaraie and Geesink, 2006). Under the attack of reactive oxygen species (ROS), the amino acid side chains of MPs are prone to carbonylation, which leads to the unfolding of the native α − helix structure and the exposure of hydrophobic groups originally buried inside the molecule. As a result, MPs further undergo cross−linking and aggregation through disulfide bonds or hydrophobic interactions (Bao & Ertbjerg, 2019; Estévez, 2011; Li, 2025). However, most existing studies have concentrated on packaging optimization or macroscopic shelf‑life assessment, whereas comprehensive investigations covering the full pathway from logistics‑induced temperature fluctuations to ice‑crystal recrystallization, protein oxidation and structural alteration, and ultimately quality deterioration, are still scarce. In particular, the differential effects of different initial pre−cooling states, namely different extents of phase transition, on the molecular conformation of MPs and the endogenous enzymatic system under the same thermal shock during logistics have not yet been fully clarified. Distinct from conventional storage research, this study establishes a critical link between beef quality and myofibrillar protein (MP) molecular stability under passive logistics thermal shock, elucidating how the initial pre−cooling stage dictates subsequent quality deterioration. An e−commerce logistics process—simulated as 25 °C passive distribution for 48 h followed by refrigeration—was evaluated by systematically comparing three pre−cooling strategies (− 0.5 °C, − 5 °C, and − 18 °C) against a 4 °C control. By integrating protein biochemical indicators, including carbonylation, solubility, surface hydrophobicity, and secondary structure, with macroscopic physicochemical traits (pH, color, shear force, and myofibril fragmentation index [MFI]) and sensory metrics, this work reveals the “pre−cooling → MP stability → quality” causal chain, providing a theoretical basis for optimizing e−commerce cold chain strategies from molecular to macroscopic levels. Materials and methods Experimental design Rump meat samples (Yanbian Yellow Cattle, male, 36 months old, m. biceps femoris) were obtained from cattle collected from Mule Co., Ltd. (Yanji, China) within 24 h post−slaughter. Visible fat and connective tissue were removed, and the samples were cut into 250 ± 20 g, 3 − cm−thick steaks. The steaks were vacuum−packaged and divided into four groups from the same carcass. The CK group was stored at 4 °C. Steaks from the supercooling, sub−freezing, and deep−freezing groups were rapidly frozen in a − 30 °C freezer (BC/BD−122WDG, Aucma, China). Core temperature was monitored in real time using a K−type thermocouple (TA612A, TASI, China) inserted into the geometric center of a reference sample. When the core temperature reached − 0.5 °C, − 5 °C, and − 18 °C, eight steaks from the same treatment group and six standardized gel ice packs (500 ± 10 g each, pre−frozen at − 20 °C for more than 48 h) were arranged in alternating layers to ensure the interior space of the package was completely filled. The assembly was then wrapped with an insulating thermal bag, placed into an EPS box (external dimensions: 30 cm × 25 cm × 20 cm), and finally sealed inside a corrugated cardboard carton. The boxes were placed in a constant-temperature and humidity incubator set at 25.0 ± 0.5 °C for 48 h to simulate short-term e-commerce distribution under representative summer ambient conditions. This 25 °C/48 h simulation represents a representative warm-season distribution condition for an EPS box + ice pack system. This temperature and duration are consistent with field-monitored package temperatures during summer e‑commerce transportation, which commonly rise to 20–30 °C within 24–48 h even when ice packs are used. The core temperature profiles of beef during the 48 h simulated distribution are shown in Fig. 1A. During the initial ice‑melting phase (approximately 0–15 h), the core temperature of all pre‑cooling groups remained below 0 °C. Thereafter, the temperature gradually increased and reached − 1.7 to 3.3 °C at the delivery stage (DA), depending on the pre‑cooling treatment. These profiles confirm that the EPS box + ice pack system generated an initial cooling phase followed by a moderate temperature rise, which is characteristic of distribution. Fig. 1(A) Temperature ranges of beef core temperature during the 48‑h EPS‑based distribution after different pre-cooling treatments. (B) TVB-N changes during distribution and refrigerated storage. Cross-shaded areas indicate stages where sampling was discontinued because TVB-N exceeded 20 mg/100 g (A) Temperature ranges of beef core temperature during the 48‑h EPS‑based distribution after different pre-cooling treatments. (B) TVB-N changes during distribution and refrigerated storage. Cross-shaded areas indicate stages where sampling was discontinued because TVB-N exceeded 20 mg/100 g Sampling intervals were determined based on spoilage kinetics. Samples (n = 3) were collected at the freezing stage (FS), at delivery (DA), and every 24 h during refrigerated storage (DA–24 h to DA–120 h). TVB-N was used as the key indicator to define the end of acceptable quality. A TVB-N value of 20.0 mg/100 g was strictly adopted as the spoilage threshold (Fig. 1B). To ensure the relevance of quality analysis, sampling and further measurements were discontinued for any treatment group once its TVB-N value exceeded this limit, which explains the “ND” (Not Determined) entries at later storage stages. The key temperature ranges during the 48‑h EPS-based distribution and the changes in TVB‑N used to determine the spoilage threshold are summarized in Fig. 1B. Experimental method Color measurement After opening the package, the beef samples were removed and the residual liquid on the surface was blotted off. The samples were then left at room temperature for 10 min. The surface color of the beef samples was measured using a portable colorimeter (CR − 10 Plus, Konica Minolta, Japan), and the L* (lightness), a* (redness), and b* (yellowness) values were recorded. Determination of myoglobin oxidation state The myoglobin oxidation state was determined using a multi−wavelength spectrophotometric method (Bai et al., 2024). Briefly, a 5 g meat sample was homogenized with 20 mL of ice−cold 40 mM PBS, pH 6.8. The mixture was homogenized (16,000 rpm and 15 s) (IKA homogenizer, T10 basic ULTRATURRAX), and the absorbances at 525, 545, 565, and 572 nm of the filtered supernatant were measured using a UV − 2600 spectrophotometer (Shimadzu, Japan). The Mb content and metmyoglobin (MetMb) percentage were calculated separately according to the following formulas.\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}{\mathrm{Mb}}\left( {{\mathrm{mg}}/{\mathrm{g}}} \right) &= - 0.{\mathrm{166}}A{\mathrm{572}} + 0.0{\mathrm{86}}A{\mathrm{565}} + 0.0{\mathrm{88}}A{\mathrm{545}} \\&\qquad + 0.0{\mathrm{99}}A{\mathrm{525}} \end{aligned}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}{\mathrm{MetMb}}\left( \% \right) &= ( - {\mathrm{2}}.{\mathrm{514}}A{\mathrm{525}}/A{\text{572 }} + 0.{\mathrm{777}}A{\mathrm{525}}/A{\mathrm{565}} \\&\qquad + 0.{\mathrm{8}}00A{\mathrm{525}}/A{\mathrm{545}} + {\mathrm{1}}.0{\mathrm{98}} ) \times {\mathrm{1}}00 \end{aligned}$$\end{document} Sensory evaluation Sensory evaluation was conducted by 10 trained panelists (5 males and 5 females, aged 22 − 30 years) who had received 30 h of training in accordance with ISO 8586 standards (ISO, 2023). The beef samples were served at 20 ± 2 °C. They were evaluated for marbling, meat color, fat color, wetness, and firmness, and scores were recorded on a 5 − point hedonic scale (5 = excellent, 1 = unacceptable). pH determination A 5.00 g meat sample was weighed and mixed with 45 mL of distilled water, and then homogenized at 6,000 rpm for 60 s using a homogenizer (F6/10, Jingxin, China). The homogenate was filtered through filter paper, and the pH of the filtrate was measured using a pH meter (PHS − 25, INESA, China). Shear force determination Beef samples were heated in a 75 °C water bath for 30 min and then cooled at 0–4 °C for 24 h. Six rectangular pieces (1 × 1 × 3 cm) were cut from each sample, and the samples were sheared three times perpendicular to the muscle fiber direction using a texture analyzer (TMS − PRO, FTC, USA). The peak shear force value (N) was recorded. Myofibril fragmentation index (MFI) A 0.5 g meat sample was weighed and mixed with 10 volumes of MFI buffer (0.1 mol/L KCl, 0.02 mol/L Na₂HPO₄, 1 mmol/L EDTA, 1 mmol/L MgCl₂, pH 7.1). The sample was homogenized in an ice bath, followed by centrifugation and washing. The protein concentration was adjusted to 0.5 mg/mL, and the absorbance at 540 nm was measured. The MFI was calculated as follows:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm{MFI}= \mathrm{OD}_{540} \times 200$$\end{document} Determination of biochemical properties and conformation of myofibrillar proteins MP was extracted following the method of Zheng (Zheng et al., 2025) with minor modifications. Trimmed the beef samples to remove visible fat and connective tissue, and recorded the weight. The samples were then homogenized with ten volumes (w/v) of Buffer 1 (0.1 M KCl, 2 mM MgCl₂, 2 mM EGTA, 20 mM K₂HPO₄, pH 6.8) and centrifuged at 2000 × g for 15 min. The resulting pellet was washed twice with eight volumes of Buffer 2 (Buffer 1 containing 10% Triton X‑100, pH 6.8), followed by two additional washes with the same volume of 0.1 M KCl. The final pellet obtained after washing was collected as the myofibrillar protein (MP) fraction. Protein solubility Protein solubility was determined according to the method of Dai (Dai et al., 2021). MPs were dispersed in phosphate buffer and the protein concentration was adjusted to 5 mg/mL. After centrifugation at 4 °C (4100×g, 15 min), the protein content in the supernatant was determined. Protein solubility was expressed as the percentage of protein content in the supernatant relative to the initial protein content. Carbonyl content Carbonyl content was determined by the DNPH method described by Wang (Wang et al., 2025) with slight modifications. The MPs solution was adjusted to 5 mg/mL, and then mixed separately with DNPH solution and HCl blank solution. After reaction at room temperature in the dark, trichloroacetic acid was added to precipitate the protein. The precipitate was collected by centrifugation and washed with an ethanol−ethyl acetate mixture to remove unreacted reagents. The final precipitate was dissolved in phosphate buffer containing guanidine hydrochloride, and the absorbance at 370 nm was measured. Carbonyl content was calculated as follows: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\text{Carbonyl content}}\left( {{\mathrm{nmol}}/{\mathrm{mg}}} \right) = {\mathrm{1}}0^{6} \times {\mathrm{A37}}0/\left( {{\mathrm{22}},000 \times {\mathrm{C}}} \right)$$\end{document} where A370 is the absorbance at 370 nm, and C is the protein concentration (mg/mL). Surface hydrophobicity Surface hydrophobicity was determined according to the bromophenol blue (BPB) binding method of Chelh (Chelh et al., 2006). The concentration of the MP solution was adjusted to 5 mg/mL. For the sample group, 200 µL of BPB solution (1 mg/mL) was added to 1 mL of MP solution. For the blank group, 1 mL of MP solution was replaced with 1 mL of phosphate buffer (20 mM, pH 6.0). After thorough mixing, the samples were allowed to react at room temperature for 10 min and then centrifuged at 2000 × g for 10 min at 25 °C. The supernatant was diluted 10 − fold, and the absorbance at 595 nm was measured as Asample and Ablank. The amount of bound BPB was calculated using the following formula: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\text{BPB bound }}\left( {\mu {\mathrm{g}}} \right) = {\mathrm{2}}00\mu {\mathrm{g}} \times ({\mathrm{A}}_{{{\mathrm{blank}}}} - {\mathrm{A}}_{{{\mathrm{sample}}}} )/{\mathrm{A}}_{{{\mathrm{sample}}}}$$\end{document} Secondary structure analysis The secondary structure of MPs was analyzed by Fourier transform infrared spectroscopy (FTIR). An appropriate amount of the MPs sample was placed on a Fourier transform infrared spectrometer (IRSpirit − T, Shimadzu, Japan) for scanning. The scanning range was 650–4,000 cm⁻¹, with a resolution of 4 cm⁻¹ and 32 scans. After spectral collection, ATR correction and baseline processing were performed. The amide I band (1600–1700 cm⁻¹) was deconvoluted and curve−fitted using PeakFit v4.12 software to calculate the relative contents of α − helix, β − sheet, β − turn, and random coil. Data analysis All experiments were conducted with at least three independent replicates, and the results are expressed as mean ± standard deviation (SD). One − way analysis of variance (ANOVA) followed by Tukey’s honestly significant difference (HSD) test was used to determine significant differences between means, with the significance level set at p < 0.05. Statistical analyses were performed using SPSS 26.0 (IBM, USA), and figures were generated using Origin Pro 2021 (OriginLab, USA). All measurements were performed on three independent biological replicates (n = 3) per treatment and time point. This sample size is commonly used in meat quality studies and was sufficient to detect the major treatment effects observed here; however, it may limit the statistical power to resolve more subtle time-dependent interactions, which is acknowledged as a limitation of the present work. Results and discussion Color characteristics and myoglobin oxidation Surface color is one of the most important sensory attributes affecting consumers’ purchase decisions. In this study, the L*, a*, and b* values of beef samples at different storage time points were analyzed to evaluate the effects of different pre−cooling treatments on color stability (Table 1). For L*, all groups showed clear differences after pre−cooling and a general upward trend during the subsequent storage period. At the FS stage, the PC − T−18 °C group (samples freezing to − 18 °C) showed the highest L* value (31.46), which was significantly higher than that of the CK group (26.32) and the other treatment groups (p < 0.05). This may be attributed to the formation of ice crystals during deep−freezing, which compressed muscle fibers and increased the amount of free water exuded to the surface after thawing, thereby enhancing light reflection and scattering (Leygonie et al., 2012). After simulated distribution and during the subsequent refrigerated storage, the L* values of all groups gradually increased, which may be related to the decrease in water−holding capacity caused by myofibrillar protein denaturation and the continuous migration of internal water to the meat surface. Among the treatment groups, the PC − T−18 °C group remained at a relatively high level throughout storage, whereas the PC − T−0.5 °C group showed the smallest increase, indicating better maintenance of microstructure and lower moisture exudation. Table 1Effects of different pre-cooling treatments on the color parameters (L*, a*, and b*) of beef during simulated distribution and subsequent refrigerated storageSampling TimeTreatment GroupL*a*b*FSCK26.32 ± 0.54Dd12.3 ± 2.13Dd11.12 ± 1.32DdPC-T-0.5℃28.42 ± 0.76CDc16.8 ± 1.94BCc14.3 ± 1.21CcPC-T-5℃29.58 ± 0.97BCb17.76 ± 1.61ABb16.38 ± 0.74ABbPC-T-18℃31.46 ± 1.68Ba18.54 ± 1.09Aa17.14 ± 0.61AaDACK30.5 ± 0.85BCa18.42 ± 0.75Aa15.62 ± 0.34BbPC-T-0.5℃30.06 ± 1.09Ca16.92 ± 1.03BCb14.74 ± 0.78CcPC-T-5℃30.6 ± 1.21Ba17.54 ± 0.86Bb15.64 ± 0.4BbPC-T-18℃30.52 ± 0.86Ba15.1 ± 1.59Cc13.34 ± 0.83DdDA-24CK33.96 ± 1.27Aa19.74 ± 1.79Aa17.36 ± 1.22AaPC-T-0.5℃34.38 ± 1.74Aa18.98 ± 1.48Aa17.2 ± 3.16AaPC-T-5℃32.94 ± 1.79Ab17.86 ± 2.04ABb16.06 ± 1.11BbPC-T-18℃29.4 ± 0.99Cc16.92 ± 0.98Bc14.64 ± 0.87DcDA-48CK32.92 ± 2.67ABb17.7 ± 0.94Ba15.98 ± 0.28BbPC-T-0.5℃31.02 ± 1.26Bc16.46 ± 1.25Cb14.72 ± 0.9CdPC-T-5℃33.28 ± 1.76Aa16.04 ± 1.61Cb15.64 ± 1.1BcPC-T-18℃29.38 ± 1.18Cd15.86 ± 1.23Cb14.32 ± 0.99DdDA-72CK31.82 ± 2.9Ba16.6 ± 3.64Ca16.44 ± 1.07ABaPC-T-0.5℃29.26 ± 1.21Db16.64 ± 1.84Ca14.48 ± 1.13CcPC-T-5℃29.2 ± 0.37Cb15.86 ± 0.59Ca14.28 ± 0.38CcPC-T-18℃30.88 ± 1.2Ba15.8 ± 1.99Ca14.62 ± 1.02DcDA-96CK30.54 ± 0.68BCb20.6 ± 2.17Aa16.52 ± 0.31ABPC-T-0.5℃32.4 ± 0.83ABa18.32 ± 1.99ABb16.04 ± 1.18ABDA-120CK17.48 ± 0.85E17.48 ± 0.85B16.12 ± 0.39BValues are expressed as mean ± SD (n = 3). Different uppercase letters (A-G) within the same treatment indicate significant differences among different sampling time points (p < 0.05), and different lowercase letters (a-g) within the same sampling time point indicate significant differences among different treatments (p < 0.05) Table 1Effects of different pre-cooling treatments on the color parameters (L*, a*, and b*) of beef during simulated distribution and subsequent refrigerated storageSampling TimeTreatment GroupL*a*b*FSCK26.32 ± 0.54Dd12.3 ± 2.13Dd11.12 ± 1.32DdPC-T-0.5℃28.42 ± 0.76CDc16.8 ± 1.94BCc14.3 ± 1.21CcPC-T-5℃29.58 ± 0.97BCb17.76 ± 1.61ABb16.38 ± 0.74ABbPC-T-18℃31.46 ± 1.68Ba18.54 ± 1.09Aa17.14 ± 0.61AaDACK30.5 ± 0.85BCa18.42 ± 0.75Aa15.62 ± 0.34BbPC-T-0.5℃30.06 ± 1.09Ca16.92 ± 1.03BCb14.74 ± 0.78CcPC-T-5℃30.6 ± 1.21Ba17.54 ± 0.86Bb15.64 ± 0.4BbPC-T-18℃30.52 ± 0.86Ba15.1 ± 1.59Cc13.34 ± 0.83DdDA-24CK33.96 ± 1.27Aa19.74 ± 1.79Aa17.36 ± 1.22AaPC-T-0.5℃34.38 ± 1.74Aa18.98 ± 1.48Aa17.2 ± 3.16AaPC-T-5℃32.94 ± 1.79Ab17.86 ± 2.04ABb16.06 ± 1.11BbPC-T-18℃29.4 ± 0.99Cc16.92 ± 0.98Bc14.64 ± 0.87DcDA-48CK32.92 ± 2.67ABb17.7 ± 0.94Ba15.98 ± 0.28BbPC-T-0.5℃31.02 ± 1.26Bc16.46 ± 1.25Cb14.72 ± 0.9CdPC-T-5℃33.28 ± 1.76Aa16.04 ± 1.61Cb15.64 ± 1.1BcPC-T-18℃29.38 ± 1.18Cd15.86 ± 1.23Cb14.32 ± 0.99DdDA-72CK31.82 ± 2.9Ba16.6 ± 3.64Ca16.44 ± 1.07ABaPC-T-0.5℃29.26 ± 1.21Db16.64 ± 1.84Ca14.48 ± 1.13CcPC-T-5℃29.2 ± 0.37Cb15.86 ± 0.59Ca14.28 ± 0.38CcPC-T-18℃30.88 ± 1.2Ba15.8 ± 1.99Ca14.62 ± 1.02DcDA-96CK30.54 ± 0.68BCb20.6 ± 2.17Aa16.52 ± 0.31ABPC-T-0.5℃32.4 ± 0.83ABa18.32 ± 1.99ABb16.04 ± 1.18ABDA-120CK17.48 ± 0.85E17.48 ± 0.85B16.12 ± 0.39BValues are expressed as mean ± SD (n = 3). Different uppercase letters (A-G) within the same treatment indicate significant differences among different sampling time points (p < 0.05), and different lowercase letters (a-g) within the same sampling time point indicate significant differences among different treatments (p < 0.05) Effects of different pre-cooling treatments on the color parameters (L*, a*, and b*) of beef during simulated distribution and subsequent refrigerated storage Values are expressed as mean ± SD (n = 3). Different uppercase letters (A-G) within the same treatment indicate significant differences among different sampling time points (p < 0.05), and different lowercase letters (a-g) within the same sampling time point indicate significant differences among different treatments (p < 0.05) In contrast, a* value showed a continuous downward trend throughout storage. At the FS stage, all pre−cooling treatment groups showed significantly higher a* values (16.8 − 18.54) than the CK group (12.3) (p < 0.05), with the PC − T−18 °C group showing the highest value (18.54), indicating that rapid pre−cooling delayed the initial oxidation of myoglobin. However, after the temperature rise during logistics at the DA stage, the decrease in a* value became more pronounced, especially in the PC − T−18 °C group, which decreased to 16.92 at DA − 24 and 15.86 at DA − 48, and remained at the lowest level during the subsequent storage period. The CK and PC − T−5 °C groups showed intermediate rates of decline, whereas the PC − T−0.5 °C group showed the slowest decrease and maintained the highest redness throughout the storage period. Consistent with the change in a* value, the b* value showed an increasing trend during storage. At the FS stage, the b* values of all pre−cooling treatment groups (14.3 − 17.14) were significantly higher than that of the CK group (11.12) (p < 0.05). From the DA stage, the b* values of the PC − T−18 °C and PC − T−5 °C groups increased markedly and reached relatively high levels at the end of storage, whereas the increase in the PC − T−0.5 °C group was smaller. During the late storage period (DA − 72 to DA − 120), only the CK and PC − T−0.5 °C groups were measured. In the CK group, the abnormal increase in the a* value at DA − 96, followed by a decrease at DA − 120, along with the sharp decrease in the L* value, suggested that the samples had entered the spoilage stage. To further explain the above color changes, the relative contents of myoglobin (Mb) and metmyoglobin (MetMb) were analyzed (Fig. 2). The decrease in a* value is mainly associated with the conversion of oxygenated myoglobin to brown metmyoglobin (Mancini & Hunt, 2005), and the changes in Mb and MetMb provided direct evidence for this process. Overall, Mb content decreased and MetMb content increased with storage time in all groups, indicating continuous oxidation of pigment proteins. From the FS to DA stage, the pre−cooling groups subjected to the 25 °C logistics condition, especially the − 5 °C and − 18 °C groups, showed a rapid decrease in Mb content, indicating that temperature rise during distribution strongly affected pigment stability. This was consistent with the concurrent decrease in a* value. During the subsequent refrigerated storage, the oxidation behavior differed among groups. The CK group, which was stored continuously at 4 °C, maintained a relatively high Mb content before DA − 72, and the increase in MetMb was relatively slow. In contrast, the − 18 °C group showed the most severe oxidation, with MetMb increasing rapidly after DA − 72 and remaining at the highest level among all groups (p < 0.05), which corresponded well to its lowest a* value. This may be associated with structural damage caused by ice crystal melting and recrystallization, which could increase the susceptibility of Mb to oxidation to MetMb and may be accompanied by changes in the activity of endogenous metmyoglobin reductase. Among the pre−cooling groups, the − 0.5 °C treatment showed the best pigment protection effect. Its MetMb content remained at a relatively low level throughout storage, and the decrease in Mb was significantly slower than that in the − 5 °C and − 18 °C groups. This indicates that − 0.5 °C supercooling not only provided a certain low−temperature reserve against external thermal shock during logistics, but also avoided the physical damage caused by freezing, thereby maintaining a more stable endogenous antioxidant environment. This result was also consistent with the color results, in which the − 0.5 °C group showed the highest a* value, the smallest increase in L* value, and the lowest increase in b* value. Fig. 2Effects of different pre-cooling treatments on (A) myoglobin (Mb) content and (B) metmyoglobin (MetMb) proportion of beef during simulated distribution and subsequent refrigerated storage. Values are expressed as mean ± SD (n = 3). Different uppercase letters (A-G) indicate significant differences among different sampling time points within the same treatment (p < 0.05), and different lowercase letters (a-g) indicate significant differences among different treatments at the same sampling time point (p < 0.05) Effects of different pre-cooling treatments on (A) myoglobin (Mb) content and (B) metmyoglobin (MetMb) proportion of beef during simulated distribution and subsequent refrigerated storage. Values are expressed as mean ± SD (n = 3). Different uppercase letters (A-G) indicate significant differences among different sampling time points within the same treatment (p < 0.05), and different lowercase letters (a-g) indicate significant differences among different treatments at the same sampling time point (p < 0.05) In summary, although the PC − T−18 °C treatment showed better color protection at the initial stage, the formation of ice crystals under deep−freezing damaged cell membrane integrity and subsequently accelerated pigment oxidation and browning during storage. In contrast, the PC − T−0.5 °C treatment effectively avoided obvious freezing damage and showed the best color stability throughout storage, which is consistent with a better preservation of structural integrity and a slower oxidation process. pH changes The pH value is a fundamental indicator reflecting postmortem glycolysis and quality deterioration during storage. As shown in Fig. 3, the pH values of all groups showed a fluctuating pattern of being relatively stable at first, then increasing, and finally decreasing throughout the simulated distribution and subsequent refrigerated storage period. At the FS and DA stages, the pH values of all groups remained at a relatively low level of 5.8 − 6.1, which is consistent with the normal postmortem physiological characteristics of meat. This was primarily attributed to the anaerobic glycolysis of residual glycogen in muscle, which led to the production and accumulation of lactic acid and resulted in a weakly acidic system (Huff Lonergan et al., 2010). With the extension of storage time, the pH values of all groups increased markedly at DA − 48, reaching the highest level during the whole monitoring period (close to or slightly higher than 6.8). This general increase in pH may be related to the activation of endogenous proteases after the temperature rise during logistics. Macromolecules such as myofibrillar proteins were extensively hydrolyzed into peptides and free amino acids, followed by deamination reactions that generated alkaline nitrogen−containing metabolites, such as ammonia and free amines including TVB − N (Bekhit et al., 2021). The accumulation of these alkaline compounds neutralized the lactic acid formed at the early stage, resulting in a significant increase in pH (p < 0.05). From DA − 72 to DA − 120, the pH values of all groups began to decrease, and clear differences in the rate of decline were observed among treatments. This secondary decline in pH during late storage has been attributed in previous studies to the proliferation of psychrotrophic spoilage bacteria that produce organic acids (Pothakos et al., 2015), although microbial counts were not determined in the present study. Fig. 3Effects of different pre-cooling treatments on pH changes of beef during simulated distribution and subsequent refrigerated storage. Values are expressed as mean ± SD (n = 3). Different uppercase letters (A-G) indicate significant differences among different sampling time points within the same treatment (p < 0.05), and different lowercase letters (a-g) indicate significant differences among different treatments at the same sampling time point (p < 0.05) Effects of different pre-cooling treatments on pH changes of beef during simulated distribution and subsequent refrigerated storage. Values are expressed as mean ± SD (n = 3). Different uppercase letters (A-G) indicate significant differences among different sampling time points within the same treatment (p < 0.05), and different lowercase letters (a-g) indicate significant differences among different treatments at the same sampling time point (p < 0.05) Shear force and MFI The tenderness of meat is jointly regulated by physical damage, water migration, and the endogenous proteolytic system. In this study, shear force and myofibril fragmentation index (MFI) were determined to evaluate the effects of different pre−cooling treatments and temperature rise during logistics on the texture stability of beef. As shown in Fig. 4A, shear force showed clear differences among groups at the FS stage. The CK group showed the highest shear force (123.56 ± 2.16 N), whereas the PC − T−18 °C group showed the lowest value (82.65 ± 7.24 N) (p < 0.05). This initial tenderization may be attributed with the physical disruption of muscle fibers caused by ice crystal formation during deep−freezing (Lagerstedt et al., 2008). However, at the DA stage, the shear force of the PC − T−18 °C and PC − T−5 °C groups increased sharply to 142.37 ± 13.84 N and 132.36 ± 12.02 N, respectively. This hardening phenomenon may be related to the alternating freezing–thawing process, in which ice crystal melting can reduce the water‑holding capacity of damaged muscle fibers and promote drip loss, thereby contributing to muscle fiber shrinkage and protein denaturation (Kim et al., 2015). During the subsequent refrigerated storage, the shear force of all groups generally decreased. Among them, the PC − T−18 °C group showed the largest decline and decreased to about 53.02 ± 4.59 N at DA − 72, indicating excessive softening, whereas the PC − T−0.5 °C group showed a relatively stable downward trend. Fig. 4Effects of different pre-cooling treatments on (A) shear force and (B) myofibril fragmentation index (MFI) of beef during simulated distribution and subsequent refrigerated storage. Values are expressed as mean ± SD (n = 3). Different uppercase letters (A-G) indicate significant differences among different sampling time points within the same treatment (p < 0.05), and different lowercase letters (a-g) indicate significant differences among different treatments at the same sampling time point (p < 0.05) Effects of different pre-cooling treatments on (A) shear force and (B) myofibril fragmentation index (MFI) of beef during simulated distribution and subsequent refrigerated storage. Values are expressed as mean ± SD (n = 3). Different uppercase letters (A-G) indicate significant differences among different sampling time points within the same treatment (p < 0.05), and different lowercase letters (a-g) indicate significant differences among different treatments at the same sampling time point (p < 0.05) The changes in MFI further explained the above texture changes at the microstructural level (Fig. 4B). At the FS stage, the MFI value of the PC − T−18 °C group (24.73 ± 3.59) and the PC − T−5 °C group (23.27 ± 2.20) was significantly lower than that of the CK group (40.40 ± 1.20) and the PC − T−0.5 °C group (54.60 ± 4.12) (p < 0.05). According to Koohmaraie and Geesink (Koohmaraie and Geesink, 2006), inhibition of endogenous calpains can interrupt normal postmortem ageing and limit myofibril fragmentation. In the present study, the relatively low MFI values observed at the FS stage in deep‑frozen samples are consistent with a scenario of reduced proteolytic activity, although calpain activity was not directly measured. Following the temperature rise at the DA stage, the MFI of the PC − T−18 °C group increased sharply to 82.60 ± 7.22. This pattern is in line with previous reports that freeze–thaw‑induced structural damage can facilitate the release of lysosomal proteases such as cathepsins, thereby contributing to enhanced proteolysis (Xu et al., 2021). Although cathepsin activity was not directly assessed here, the sharp increase in MFI after the DA stage is compatible with such a mechanism. During the middle and later storage period, the MFI of this group remained at a relatively high level, with 83.40 ± 0.69 at DA − 24 and 83.27 ± 0.90 at DA − 48, and then increased to 110.33 ± 6.50 at DA − 72, which corresponded well to the excessive decrease in shear force. Similarly, the PC − T−5 °C group showed a marked rise in MFI after the DA stage, increasing to 66.00 ± 2.75 at DA, further rising to 86.33 ± 8.51 at DA − 24 and 105.20 ± 2.83 at DA − 48, and remaining high at 112.33 ± 2.52 at DA − 72. MFI at DA − 96 and DA − 120 was not determined (ND) for the PC − T−5 °C and PC − T−18 °C groups. This was because these samples reached the TVB − N spoilage limit (> 20 mg/100 g) prematurely due to accelerated protein degradation post−distribution, and were thus excluded from further analysis. In contrast, the PC − T−0.5 °C group showed a gradual increase in MFI, reflecting a more orderly endogenous proteolysis process. Overall, shear force and MFI generally showed a negative relationship during normal ageing, which was more evident in the CK group. However, in the PC − T−5 °C and PC − T−18 °C groups, both indicators increased simultaneously at the DA stage. This suggests that texture hardening at the early stage was mainly caused by dehydration and muscle fiber shrinkage, whereas as water loss gradually reached equilibrium, the effect of structural degradation became dominant, ultimately resulting in excessive softening. These results indicate that the PC − T−0.5 °C treatment helped maintain a more orderly endogenous proteolytic process and avoided the chain reaction of cell rupture and dehydration−induced hardening caused by deep−freezing, thereby better preserving the texture stability of beef. Carbonyl content Total carbonyl content is an important indicator for evaluating the oxidation degree of myofibrillar proteins. As shown in Fig. 5A, at the FS stage, the carbonyl contents of the three pre−cooling treatment groups (0.438 ± 0.003, 0.433 ± 0.041, and 0.443 ± 0.027 in the PC − T−0.5 °C, PC − T−5 °C, and PC − T−18 °C groups) were all higher than that of the CK group (0.208 ± 0.002), indicating that low−temperature pre−cooling and phase transition had already caused a certain degree of initial oxidation. Similar early−stage increases in protein carbonyls under freezing and supercooling conditions have been associated with cold−induced oxidative stress during phase transitions in muscle foods (Estévez, 2011; Lund, 2011; Nawaz et al., 2022). Previous studies have shown that protein oxidation can lead to the accumulation of carbonyl compounds and further induce the deterioration of physicochemical and functional properties in meat (Nawaz et al., 2022). Thereafter, the oxidation process gradually diverged among groups after distribution. At DA − 24, the carbonyl content of the PC − T−18 °C group had increased to 1.432 ± 0.026, which was higher than that of the PC − T−0.5 °C group (0.853 ± 0.014) and the PC − T−5 °C group (0.935 ± 0.018). By DA − 72, the PC − T−18 °C and PC − T−5 °C groups further increased to 2.610 ± 0.110 and 2.287 ± 0.037, respectively, whereas the PC − T−0.5 °C group remained at only 1.357 ± 0.044, which was close to that of the CK group (1.216 ± 0.031). These results indicate that under logistics‑related temperature fluctuations, deep‑freezing and sub‑freezing treatments were associated with more severe protein oxidation, whereas supercooling tended to slow the accumulation of oxidative damage to a certain extent. This result is consistent with previous reports showing that supercooling storage helps reduce ice crystal damage and oxidative deterioration (Lin et al., 2022; You et al., 2020). Fig. 5Effects of different pre-cooling treatments on (A) carbonyl content, (B) surface hydrophobicity, (C) protein solubility, and (D) secondary structure of beef myofibrillar proteins during simulated distribution and subsequent refrigerated storage. Values are expressed as mean ± SD (n = 3). Different uppercase letters (A-G) indicate significant differences among different sampling time points within the same treatment (p < 0.05), and different lowercase letters (a-g) indicate significant differences among different treatments at the same sampling time point (p < 0.05) For (A), (B), and (C) Effects of different pre-cooling treatments on (A) carbonyl content, (B) surface hydrophobicity, (C) protein solubility, and (D) secondary structure of beef myofibrillar proteins during simulated distribution and subsequent refrigerated storage. Values are expressed as mean ± SD (n = 3). Different uppercase letters (A-G) indicate significant differences among different sampling time points within the same treatment (p < 0.05), and different lowercase letters (a-g) indicate significant differences among different treatments at the same sampling time point (p < 0.05) For (A), (B), and (C) Surface hydrophobicity Surface hydrophobicity reflects the integrity of the native protein conformation and the exposure degree of internal hydrophobic groups. As shown in Fig. 5B, at the FS stage, the surface hydrophobicity values of the PC − T−18 °C and PC − T−5 °C groups were 336.178 ± 9.673 and 205.623 ± 8.280, respectively, which were markedly higher than those of the CK group (114.754 ± 3.357) and the PC − T−0.5 °C group (104.761 ± 1.698). This indicates that deep−freezing and sub−freezing had already caused stronger disruption to protein conformation at the initial pre−cooling stage. Both protein oxidation and freezing treatment may induce molecular unfolding, thereby exposing hydrophobic groups originally buried inside the protein and resulting in increased surface hydrophobicity (Chen et al., 2022). After distribution, the surface hydrophobicity of all groups increased markedly, whereas the PC − T−0.5 °C group remained at a relatively low level throughout storage. The CK group also showed a pronounced increase after distribution and continued to rise during storage. At the DA stage, the CK value increased to 910.219 ± 16.512, which was higher than that of the PC − T−0.5 °C group (441.801 ± 11.926) but lower than those of the PC − T−5 °C group (629.966 ± 50.135) and the PC − T−18 °C group (788.906 ± 54.381). By DA − 72, the CK value further increased to 1272.232 ± 29.223, remaining higher than the PC − T−0.5 °C group (902.126 ± 21.367) and comparable to the PC − T−18 °C group (1199.436 ± 17.191) while higher than the PC − T−5 °C group (1082.690 ± 38.659). Moreover, the CK group continued to increase at DA − 96 and DA − 120 to 1,386.452 ± 50.228 and 1411.245 ± 19.664, respectively. These results indicate that distribution and storage continuously promoted protein unfolding and the exposure of hydrophobic cores, whereas supercooling was more effective in maintaining the structural stability of myofibrillar proteins. This result is consistent with previous reports showing that changes in beef proteins and muscle structure are relatively slower under supercooling conditions (Lu et al., 2020). Protein solubility Protein solubility reflects the dispersion ability and functional integrity of myofibrillar proteins (Fig. 5C). At the FS stage, the solubility of the CK group was 18.577 ± 0.607, whereas all pre−cooling treatments showed lower values, with the PC − T−0.5 °C group being the lowest (13.059 ± 0.200), followed by the PC − T−18 °C group (14.043 ± 0.453) and the PC − T−5 °C group (14.576 ± 0.661). This indicates that the pre−cooling phase state itself altered MP extractability at the initial stage, and the effect was most pronounced under supercooling at − 0.5 °C. Comparable reductions in extractable myofibrillar proteins under near−freezing or superchilling conditions have been reported previously and were attributed to subtle structural tightening and incipient aggregation rather than extensive denaturation (Lu et al., 2020; Chen et al., 2022). With distribution and storage, protein solubility decreased further in all pre−cooling groups, but the declining rates clearly differed. From FS to DA − 72, the PC − T−18 °C group showed the fastest decrease (14.043 ± 0.453 to 8.791 ± 0.256; −5.252 over 72 h, corresponding to − 1.751 per 24 h), followed by the PC − T−5 °C group (14.576 ± 0.661 to 10.041 ± 0.127; −4.535 over 72 h, corresponding to − 1.512 per 24 h). In contrast, the PC − T−0.5 °C group exhibited the slowest decline (13.059 ± 0.200 to 10.191 ± 0.325; −2.868 over 72 h, corresponding to − 0.956 per 24 h). Consistently, at DA − 48 and DA − 72, the PC − T−18 °C group maintained the lowest solubility (9.074 ± 0.612 and 8.791 ± 0.256, respectively), whereas the PC − T−0.5 °C and PC − T−5 °C groups remained higher at DA − 72 (10.191 ± 0.325 and 10.041 ± 0.127). Overall, the faster loss of solubility in the PC − T−18 °C and PC − T−5 °C groups was consistent with their more rapid accumulation of oxidative and conformational changes after distribution (higher carbonyl content and surface hydrophobicity, and a stronger α − helix − to − β − sheet transition), whereas the PC − T−0.5 °C treatment showed a slower deterioration trend during storage. Secondary structure of myofibrillar proteins To further describe the conformational responses of myofibrillar proteins under different pre−cooling strategies, secondary structure composition was quantified by FTIR curve fitting of the amide I band (Fig. 5D). In meat protein systems, α − helix is usually regarded as a more ordered conformation stabilized by intramolecular hydrogen bonds; random coil reflects increased structural disorder/unfolding; β − sheet enrichment is often associated with conformational rearrangement and aggregation−related intermolecular hydrogen−bonding networks; and β − turn represents flexible motifs connecting secondary−structure segments (Estévez, 2011; Lund, 2011; Bao & Ertbjerg, 2019; Nawaz et al., 2022). At the FS stage, the low−temperature pre−cooling groups retained a relatively higher proportion of α − helix than the CK group. Among them, the PC − T−0.5 °C group showed the highest α − helix (62.69%), followed by the PC − T−18 °C group (56.21%) and the PC − T−5 °C group (53.12%), all higher than CK (45.30%). After distribution (DA), the structural evolution diverged clearly among groups. The PC − T−18 °C and PC − T−5 °C groups exhibited a pronounced decrease in α − helix (to 11.25% and 15.48%, respectively) accompanied by a marked increase in β − sheet (to 50.19% and 43.56%), indicating a rapid shift toward aggregation−related conformations. In contrast, the PC − T−0.5 °C group showed higher α − helix (23.34%) and lower β − sheet (18.55%) at DA, remaining closer to CK at the same stage (22.63% and 21.52%). During subsequent refrigerated storage, the freezing−related groups tended to remain in a relatively stable β − sheet−dominant pattern. Specifically, β − sheet in the PC − T−18 °C group stayed at a high level from DA − 24 to DA − 72 (45.21–46.39% at DA − 24/DA − 48 and 42.72% at DA − 72), and a similar stabilization was observed in the PC − T−5 °C group (46.39–46.76% at DA − 24/DA − 48 and 46.27% at DA − 72), suggesting that the major conformational rearrangement occurred mainly during the temperature rise stage and then changed only slightly thereafter. By comparison, CK and the PC − T−0.5 °C group showed more noticeable fluctuations in secondary structure during storage. CK displayed a decrease in α − helix at DA − 48 (15.56%) followed by an increase at DA − 72 (22.23%), together with corresponding changes in random coil. The PC − T−0.5 °C group showed larger post − DA variability, with α − helix decreasing and random coil increasing markedly at DA − 48, followed by partial adjustment at later time points. Taken together with the oxidation and functionality results (higher carbonyl content and surface hydrophobicity, and lower solubility in the PC − T−18 °C and PC − T−5 °C groups; Fig. 5A–C), the persistent β‑sheet enrichment in these two groups is consistent with a stronger tendency toward oxidation‑associated conformational rearrangement and aggregation, whereas the PC‑T‑0.5 °C treatment appeared to delay such aggregation‑related structural evolution to some extent (Lu et al., 2020). Sensory evaluation Sensory evaluation directly reflects the overall changes in color and texture of beef during distribution and storage. As shown in Table 2, Overall, the samples showed different degrees of sensory deterioration under different pre−cooling treatments, and the − 0.5 °C group maintained relatively better overall color and tissue condition during the middle and later storage period after distribution. Table 2Sensory scores of beef under different pre-cooling treatments during simulated distribution and subsequent refrigerated storageSampling TimeTreatmentGroupMarblingMeat colorFat colorWetnessFirmnessFSCK1.9 ± 0.57Cb4.7 ± 0.48Aa4.3 ± 0.67Aa3.9 ± 1.37Ba2.6 ± 0.7BbPC-T-0.5℃2.1 ± 0.57Cab4.4 ± 0.52Aa3.8 ± 0.92Aab3.8 ± 1.4Aa2.9 ± 0.88 BaPC-T-5℃1.8 ± 0.79Bb3.8 ± 0.79Ab3.5 ± 0.53Ab4.4 ± 0.7Aa2.3 ± 0.67CbPC-T-18℃2.3 ± 0.48Ba3.6 ± 0.52Ab2.5 ± 0.71Bc3.7 ± 0.82Aa2.5 ± 0.85CbDACK2 ± 1.05Cb4.6 ± 0.52Aa3.7 ± 0.48Ba2.8 ± 0.42Cb2.3 ± 0.48BcPC-T-0.5℃2.5 ± 0.53Ba2.7 ± 0.48Cc3.3 ± 0.48Bab3.2 ± 0.42Ba3.5 ± 0.71AaPC-T-5℃1.8 ± 0.92Bb2.5 ± 0.85Bc2.3 ± 0.48Cc3.3 ± 0.67Ba2.4 ± 0.7CbcPC-T-18℃2.1 ± 0.74Bab3.6 ± 0.52Bb2.8 ± 0.63Abc2.7 ± 1.16Bb3.5 ± 0.71AaDA-24CK2.4 ± 0.52BCa3.2 ± 0.79ABb3.3 ± 0.48Ba4.3 ± 0.67Aa3.7 ± 0.67AaPC-T-0.5℃2.7 ± 0.82Ba3.8 ± 0.92Ba3.5 ± 0.71Aa2.9 ± 0.74Bb2.5 ± 0.53BCbPC-T-5℃2.3 ± 0.48ABa3 ± 1.05Bb2.7 ± 0.67Bb3.1 ± 0.57Bb3.2 ± 0.63BabPC-T-18℃2.6 ± 0.52 Ba4.1 ± 1.1Aa2.1 ± 0.32Cc2.6 ± 0.7Bb2.5 ± 0.85CbDA-48CK3.5 ± 0.71Aa3.5 ± 1.18Ba3.1 ± 0.88Ba3.5 ± 0.53Ba3.2 ± 0.79 AaPC-T-0.5℃3.1 ± 0.74Aa3.8 ± 1.03Ba3.4 ± 0.84Aa3 ± 0.47Ba3 ± 0.82BaPC-T-5℃2.8 ± 0.63Aa2.2 ± 0.63Cc2.7 ± 0.67Ba3.2 ± 0.79Ba3.7 ± 0.48AaPC-T-18℃3.6 ± 0.97Aa2.6 ± 0.97Bb2.9 ± 0.57Aa3.2 ± 0.63Ba3.5 ± 0.71AaDA-72CK3.8 ± 0.63Aa3.2 ± 0.42Ba2.9 ± 0.57Ba3.4 ± 0.7 Ba2.7 ± 0.82 BbPC-T-0.5℃3.1 ± 0.32Ab3.8 ± 0.79Ba3.4 ± 0.52Aa3.2 ± 0.42 Ba3.6 ± 0.7AaPC-T-5℃2.8 ± 0.63Ab3.3 ± 0.67Aa3.5 ± 0.53Aa2.7 ± 0.82Cb3.5 ± 0.53AaPC-T-18℃3.3 ± 0.67Aab2.8 ± 0.79Bb3.1 ± 0.32Aa2.5 ± 0.53 Bb3.1 ± 0.74BabDA-96CK1.7 ± 0.48Cb3.5 ± 0.71Ba2.1 ± 1.1Ca3 ± 0.47 Ca2.2 ± 0.42BaPC-T-0.5℃3.3 ± 0.48Aa2.4 ± 1.17Cb2.4 ± 0.52Ca2.4 ± 0.7 Cb2.1 ± 0.88CaDA-120CK2.8 ± 0.63B1.7 ± 0.67C2.3 ± 0.67C2.8 ± 0.63C2.4 ± 0.7BValues are expressed as mean ± SD (n = 3). Different uppercase letters (A-G) within the same treatment indicate significant differences among different sampling time points (p < 0.05), and different lowercase letters (a-g) within the same sampling time point indicate significant differences among different treatments (p < 0.05) Table 2Sensory scores of beef under different pre-cooling treatments during simulated distribution and subsequent refrigerated storageSampling TimeTreatmentGroupMarblingMeat colorFat colorWetnessFirmnessFSCK1.9 ± 0.57Cb4.7 ± 0.48Aa4.3 ± 0.67Aa3.9 ± 1.37Ba2.6 ± 0.7BbPC-T-0.5℃2.1 ± 0.57Cab4.4 ± 0.52Aa3.8 ± 0.92Aab3.8 ± 1.4Aa2.9 ± 0.88 BaPC-T-5℃1.8 ± 0.79Bb3.8 ± 0.79Ab3.5 ± 0.53Ab4.4 ± 0.7Aa2.3 ± 0.67CbPC-T-18℃2.3 ± 0.48Ba3.6 ± 0.52Ab2.5 ± 0.71Bc3.7 ± 0.82Aa2.5 ± 0.85CbDACK2 ± 1.05Cb4.6 ± 0.52Aa3.7 ± 0.48Ba2.8 ± 0.42Cb2.3 ± 0.48BcPC-T-0.5℃2.5 ± 0.53Ba2.7 ± 0.48Cc3.3 ± 0.48Bab3.2 ± 0.42Ba3.5 ± 0.71AaPC-T-5℃1.8 ± 0.92Bb2.5 ± 0.85Bc2.3 ± 0.48Cc3.3 ± 0.67Ba2.4 ± 0.7CbcPC-T-18℃2.1 ± 0.74Bab3.6 ± 0.52Bb2.8 ± 0.63Abc2.7 ± 1.16Bb3.5 ± 0.71AaDA-24CK2.4 ± 0.52BCa3.2 ± 0.79ABb3.3 ± 0.48Ba4.3 ± 0.67Aa3.7 ± 0.67AaPC-T-0.5℃2.7 ± 0.82Ba3.8 ± 0.92Ba3.5 ± 0.71Aa2.9 ± 0.74Bb2.5 ± 0.53BCbPC-T-5℃2.3 ± 0.48ABa3 ± 1.05Bb2.7 ± 0.67Bb3.1 ± 0.57Bb3.2 ± 0.63BabPC-T-18℃2.6 ± 0.52 Ba4.1 ± 1.1Aa2.1 ± 0.32Cc2.6 ± 0.7Bb2.5 ± 0.85CbDA-48CK3.5 ± 0.71Aa3.5 ± 1.18Ba3.1 ± 0.88Ba3.5 ± 0.53Ba3.2 ± 0.79 AaPC-T-0.5℃3.1 ± 0.74Aa3.8 ± 1.03Ba3.4 ± 0.84Aa3 ± 0.47Ba3 ± 0.82BaPC-T-5℃2.8 ± 0.63Aa2.2 ± 0.63Cc2.7 ± 0.67Ba3.2 ± 0.79Ba3.7 ± 0.48AaPC-T-18℃3.6 ± 0.97Aa2.6 ± 0.97Bb2.9 ± 0.57Aa3.2 ± 0.63Ba3.5 ± 0.71AaDA-72CK3.8 ± 0.63Aa3.2 ± 0.42Ba2.9 ± 0.57Ba3.4 ± 0.7 Ba2.7 ± 0.82 BbPC-T-0.5℃3.1 ± 0.32Ab3.8 ± 0.79Ba3.4 ± 0.52Aa3.2 ± 0.42 Ba3.6 ± 0.7AaPC-T-5℃2.8 ± 0.63Ab3.3 ± 0.67Aa3.5 ± 0.53Aa2.7 ± 0.82Cb3.5 ± 0.53AaPC-T-18℃3.3 ± 0.67Aab2.8 ± 0.79Bb3.1 ± 0.32Aa2.5 ± 0.53 Bb3.1 ± 0.74BabDA-96CK1.7 ± 0.48Cb3.5 ± 0.71Ba2.1 ± 1.1Ca3 ± 0.47 Ca2.2 ± 0.42BaPC-T-0.5℃3.3 ± 0.48Aa2.4 ± 1.17Cb2.4 ± 0.52Ca2.4 ± 0.7 Cb2.1 ± 0.88CaDA-120CK2.8 ± 0.63B1.7 ± 0.67C2.3 ± 0.67C2.8 ± 0.63C2.4 ± 0.7BValues are expressed as mean ± SD (n = 3). Different uppercase letters (A-G) within the same treatment indicate significant differences among different sampling time points (p < 0.05), and different lowercase letters (a-g) within the same sampling time point indicate significant differences among different treatments (p < 0.05) Sensory scores of beef under different pre-cooling treatments during simulated distribution and subsequent refrigerated storage Values are expressed as mean ± SD (n = 3). Different uppercase letters (A-G) within the same treatment indicate significant differences among different sampling time points (p < 0.05), and different lowercase letters (a-g) within the same sampling time point indicate significant differences among different treatments (p < 0.05) For meat color and fat color, the CK group showed relatively higher scores at the FS stage, whereas the − 18 °C group showed relatively lower scores, indicating that deep−freezing had already affected appearance acceptability at the initial stage. With storage time, the scores of the − 18 °C and − 5 °C groups decreased more rapidly, and the scores for meat color and fat color further declined at DA − 48 and DA − 72. In contrast, the − 0.5 °C group still maintained relatively high scores for meat color and fat color at DA − 72, indicating a better ability to maintain appearance quality. For wetness and firmness, the − 18 °C and − 5 °C groups showed a gradual decrease in wetness during storage, characterized by a drier surface and reduced exudation, while firmness also showed certain fluctuations. In contrast, the − 0.5 °C group showed relatively stable wetness and maintained a more stable tissue state. Combined with the above physicochemical results and protein structural changes, the decline in sensory quality in the lower−temperature treatment groups may be mainly attributed to ice crystal damage and the subsequent protein denaturation and aggregation. Overall, the − 18 °C and − 5 °C groups showed limited quality retention after DA − 72, whereas the − 0.5 °C group exhibited a relatively longer period of sensory acceptability. Conclusion This study systematically compared the effects of supercooling (− 0.5 °C), sub−freezing (− 5 °C), and deep−freezing (− 18 °C) pre−cooling treatments on beef quality during EPS−based distribution and subsequent refrigerated storage. The results demonstrated that different pre−cooling phase states significantly affected both the macroscopic quality attributes and the molecular properties of myofibrillar proteins. Although deep−freezing provided a stronger initial cooling effect and temporarily improved redness at the FS stage. However, it also caused severe structural damage to muscle tissue, as ice−crystal formation during deep−freezing was followed by melting and recrystallization during logistics. Such damage is likely to be accompanied by increased moisture migration, pigment oxidation, protein oxidation, conformational unfolding, and aggregation, which together may contribute to faster deterioration in color, texture, and sensory quality during storage. The sub−freezing treatment showed a similar but slightly less severe trend. In contrast, the supercooling (− 0.5 °C) treatment avoided obvious freezing phase transition and therefore reduced ice crystal−related injury. As a result, this treatment better preserved color stability, slowed myoglobin oxidation, maintained a more stable pH and texture profile, and delayed oxidative and conformational deterioration of MPs, as reflected by lower carbonyl content and surface hydrophobicity, higher protein solubility, and less pronounced transition from α − helix to β − sheet. Taken together, supercooling pre‑cooling showed the best overall ability to maintain beef quality under the tested logistics conditions. Its protective effect appears to be related to limiting freezing‑induced microstructural damage and preserving a more stable endogenous biochemical environment during subsequent temperature fluctuations. Therefore, − 0.5 °C supercooling may be a more suitable pre−cooling strategy for fresh beef distributed in EPS box + ice pack systems.