Effects of potassium polyaspartate on phenolic compounds, volatile compounds, color, and sensory properties of cabernet sauvignon wine Authors: Zeyu Zhou, Yibin Lan, Meng Zhou, Wenzhe Tong, Yixuan Ge, Cong Yu, Xuerong Duan, Changqing Duan, Xuemei Fan, Nan Meng, Guangfeng Wu Published: 2026 ABSTRACT AbstractResearch on potassium polyaspartate (KPA) in wines has focused primarily on tartrate stabilization. This first systematic study on the impact of KPA on flavor offers novel production insights. KPA was applied at commercially relevant dosage levels, either alone or following 48 h of pre-freezing, together with an untreated control and an extended cold-treatment reference (−4 °C for 30 d). Phenolic and volatile compounds, along with sensory properties, were analyzed before and after bottle aging for 6 months. Although the untreated wine already exhibited high potassium bitartrate (KHT) stability (Δχ < 30 μS/cm), KPA-containing treatments further decreased Δχ, indicating a lower residual KHT crystallization tendency under the test conditions. Individual application of 45 mL/hL KPA maintained a deeper, redder hue at 6 M, associated with a higher vitisin concentration. Concurrently, it contained more norisoprenoids, carbonyl compounds, and fatty acids. The combined treatment of 48 h pre-freezing with 35–45 mL/hL KPA preserved phenolic substances, including monomeric anthocyanins, anthocyanin derivatives, and flavonols, and showed no significant effect on aroma compounds or sensory properties at 6 M. Overall, under the conditions of this study, KPA treatments produced more pronounced changes in the color-related characteristics of the Cabernet Sauvignon wine than in its sensory aroma profile.Keywords: Potassium polyaspartate, Potassium bitartrate, Phenolic compounds, Volatile compounds, Sensory properties Introduction During wine storage, tartrate precipitates composed of potassium bitartrate (KHT) and calcium tartrate dihydrate (CaT) gradually precipitate out. This natural physicochemical phenomenon is frequently misinterpreted by consumers as a quality defect, resulting in significant economic losses for wineries (Cui et al., 2024). There are currently two distinct technological approaches for tartaric stabilization. The first approach is the “subtraction strategy”, which involves removing either existing bitartrate or components that may potentially form it in wine. Applicable methods include cold stabilization, ion exchangers (IE), and electrodialysis (ED). The cold stabilization method is based on the principle that the solubility of tartrates decreases as the temperature decreases. By cooling the wine from ambient conditions (20 °C) to near its freezing point, tartrates are precipitated, followed by filtration to remove crystallized precipitates. This method is currently widely adopted and proven effective in winery applications. However, it is time-consuming, economically costly, and generates environmental pollution (Dabare et al., 2023; Reilly et al., 2023). This process may lead to the loss of wine color and specific aromatic compounds, with partial oxidation due to the continuous stirring required for heat exchange and the high solubility of oxygen at low temperatures (Cosme et al., 2024; Martínez-Pérez et al., 2020). In the IE method, the cations (H+) displace potassium ions (K+) and calcium ions (Ca2+) in wine. This process reduces the content of both polyphenolic and aromatic substances (Benı´tez et al., 2002; Lasanta et al., 2013), with the most pronounced effect on monomeric anthocyanins (Sarapulova et al., 2018). ED utilizes an external electric field to induce unidirectional migration of K+, Ca2+, and hydrogen tartrate (TH−) from wine into a nitric acid buffer solution. This method demonstrates eight times lower energy consumption than cold stabilization (Bdiri et al., 2020). However, it must be noted that the International Organization of Vine and Wine (OIV) limits acidification to a maximum pH reduction of 0.3 units, which implies that certain wines cannot achieve complete stabilization (Canals & Zamora, 2017; Pasechnaya et al., 2023). The second approach adopts an “additive strategy” by employing appropriate stabilizers to prevent the aggregation or formation of bitartrate. Available additive options include metatartaric acid (MTA), carboxymethylcellulose (CMC), and yeast mannoproteins (YMP). MTA undergoes hydrolysis after 12–18 months of storage at 10–18 °C, making it suitable only for immediate consumption or short-term marketing wine (Sprenger et al., 2015). CMC can adversely affect the chromatic characteristics and turbidity of red wines. Consequently, the OIV has restricted its oenological application exclusively to white and sparkling wine productions (Sommer et al., 2016). The interaction between YMP and tannins decreases tannin concentration and astringency intensity in wine (Mekoue Nguela et al., 2016). Overall, traditional tartrate stabilization techniques each have their own limitations. Balancing effective tartrate stabilization with maintaining wine sensory and qualities remains a challenge, highlighting the need for more sustainable and versatile stabilization strategies. Furthermore, while stabilizers effectively inhibit tartrate crystallization, they perform ineffectively against cold-labile compounds in wine, such as pigments, proteins, polysaccharides, and tannins. These cold-labile compounds may aggregate to form cold colloidal haze, potentially interfering with stabilizer efficacy. Thus, hybrid techniques combining two or more methods are gradually emerging as a developmental trend. For instance, integrating cold treatment with stabilizer addition to mitigate the limitations of the “additive strategy”. Nevertheless, these novel technologies still require progressive validation and optimization. In recent years, the novel tartrate-stabilizing additive potassium polyaspartate (KPA) has gained attention. The European Commission evaluated the safety of potassium polyaspartate as a food additive. It concluded that its application in wine at maximum levels of 300 mg/L, typical usage range 100–200 mg/L, posed no safety concerns (Galbusera et al., 2017). The National Health Commission of China approved potassium polyaspartate as a novel food additive in 2023. As KPA has been widely approved for use, researchers have begun to focus on its tartrate stabilization effects. Studies have found that compared to traditional additives such as MTA and CMC, KPA offers advantages in stabilization efficacy at higher temperatures and during long-term storage (Bosso, Motta, Panero, Lucini, et al., 2020; Canuti et al., 2019; Eder et al., 2019). In fact, KPA is a highly effective tool for KHT stabilization, preventing precipitation in Austrian white and red wines over a 12-month period (Philipp, Strauss, Willach, et al., 2022). However, for wines with elevated calcium ion concentrations, KPA does not eliminate the risk of CaT precipitation (Philipp, Strauss, Schlögl, et al., 2022). Overall, existing research has primarily focused on the tartrate stabilization effects and applicable scope of KPA. The effects of varying KPA concentrations and combining with pre-freezing treatment remain to be explored. Considering all the above context, experiments were conducted involving the addition of KPA at varying concentrations and with or without pre-freezing treatment. The primary objective of this study was to systematically evaluate the effects of these treatments on wine phenolic compounds, volatile compounds, color, and sensory quality under commercially relevant application conditions. Conductivity was additionally measured to characterize treatment-related differences in residual KHT crystallization tendency. The findings were expected to provide reference information regarding the compositional and sensory changes associated with KPA application in Cabernet Sauvignon wine. Materials and methods Materials The potassium polyaspartate sample was the commercial Enartis stabilizer ZENITH UNO (marketed as “Zhenli No.1” in China, potassium polyaspartate A-5D K/SD*, containing 0.3 ± 0.1% SO2) chosen from ESSECO Co., Ltd. (Italy). Its formulation is a high-purity KPA solution, free of other tartrate stabilizers, which allows for a scientific evaluation of the singular impact of KPA on the flavor characteristics of the wine. Standards of phenolic and aroma compounds were all purchased from Sigma-Aldrich Chemical Co. (St Louis, MO, USA). Chromatographic grade solvents, including methanol (≥99.9%) and acetonitrile (≥99.9%), were purchased from Honeywell Burdick & Jackson (Morris Plains, NJ, USA), while formic acid (≥99.9%) was purchased from Roe Scientific Inc. (Newark, NJ, USA). Analytical grade chemicals, including gallic acid, hydrochloric acid, sulfuric acid, and absolute ethanol, were purchased from Beijing Chemical Works (Beijing, China), with tartaric acid, sodium hydroxide, and sodium chloride purchased from Macklin (Shanghai, China). Deionized water was prepared using a Milli-Q water purification system (Millipore, Bedford, MA, USA). Sample treatments After the completion of malolactic fermentation in Cabernet Sauvignon (blended dry red wine, provided by a commercial winery), the sample treatments were divided as follows. (1) Control group (CK): No treatment; (2) extended cold-treatment reference group (Cold-T): the wine was maintained at −4 °C for 30 d; (3) Combined application of 48 h pre-freezing and KPA addition groups (PF-A1, PF-A2, PF-A3): Cooled the wine to −4 °C for 48 h, filtered (0.45 μm pore size) and added KPA at different concentrations (35, 45, 55 mL/hL); (4) Individual KPA addition groups (NF-A1, NF-A2, NF-A3): skipped the pre-freezing step, directly added KPA at different concentrations (35, 45, 55 mL/hL). The KPA dosages were selected to represent commercially relevant application levels for practical winemaking. Each treatment was repeated three times. Samples were taken for analysis at the initial stage (0 M) and after bottle storage for 6 months. Conductivity test The potassium bitartrate (KHT) crystallization tendency of the wines was assessed using a TartarCheck conductometer (Exacta Labcenter SpA, Italy) (Bosso et al., 2016). Aseptically filtered and degassed wine sample (25 mL) was placed in a cold bath maintained at 0 °C (±1.0 °C) with continuous agitation. After calibrating the conductivity meter, the probe was inserted into the sample, and the initial conductivity reading (C1) was recorded when the sample reached the test temperature. Subsequently, the precipitating agent KHT (16 g/L) was added, and measured again (C2) after 30 min of reaction. Phenolic compounds analysis Analysis of monomeric anthocyanins, non-anthocyanin phenolics, and anthocyanin derivatives was performed using an Agilent 1200-6410B high-performance liquid chromatography triple quadrupole tandem mass spectrometry (HPLC-QqQ-MS/MS, Agilent Technologies, Santa Clara, CA, USA). The chromatographic column was an Agilent Poroshell 120 EC-C18 column (150 mm × 2.1 mm, 2.7 μm). Mobile phase A: 0.1% formic acid in water; Mobile phase B: 0.1% formic acid in a 1:1 (v/v) mixture of methanol and acetonitrile. Wine samples were filtered through a 0.22 μm polyethersulfone membrane filter (Jinteng Experimental Equipment Co., Ltd., China) prior to analysis. Monomeric anthocyanins Analysis of monomeric anthocyanins was conducted according to our previously reported method (Li et al., 2016). The chromatographic conditions were gradient elution with an injection volume of 1 μL. Mobile phase B was from 10% to 46% for 28 min, from 46% to 10% for 1 min, and re-equilibration at 10% for 5 min. The flow rate was 0.4 mL/min, and the column temperature was set at 55 °C. Mass spectrometry was operated in positive electrospray ionization mode using multiple reaction monitoring (MRM). Capillary voltage was 4 kV. Drying gas temperature was 350 °C with a flow rate of 12 L/min. Nebulizer pressure was 35 psi. Identification was achieved by comparing retention times and qualitative transition ions with reference standards and compounds in our in-home library. Quantification was performed using calibration curves established for the corresponding primary anthocyanins in the simulated wine solution. The simulated solution was prepared as an aqueous solution containing 14% (v/v) ethanol, 7 g/L tartaric acid, and 4 g/L glucose, with the pH adjusted to 3.3. The phenolic standard was dissolved and subjected to a two-fold serial dilution, resulting in a total of 10 concentration gradients. This dilution series could cover the concentration range of phenolic compounds in the wine samples. Non-anthocyanin phenolics MS was operated in negative electrospray ionization mode. Other conditions were identical to 2.4.1. Anthocyanin derivatives Analysis of anthocyanin derivatives was conducted according to our previously reported method (X.-K. Zhang et al., 2021). The chromatographic conditions were gradient elution with an injection volume of 10 μL. Mobile phase B was 0% for 1 min, from 0% to 25% for 2 min, from 25% to 30% for 12 min, and 100% for 5 min. The concentration of anthocyanin derivatives was calculated and expressed as malvidin-3-O-glucoside equivalents. Other conditions were identical to 2.4.1. Volatile compounds analysis According to our previously reported method (Tong et al., 2023), headspace solid-phase microextraction combined with gas chromatography–mass spectrometry (HS-SPME-GC–MS) was used to determine volatile compounds in wine. The HS-SPME was performed using a CTC CombiPAL autosampler (CTC Analytics, Zwingen, Switzerland) equipped with a 2 cm DVB/CAR/PDMS (Supelco, Bellefonte, PA., USA; 50/30 μm). A 5 mL wine sample was placed in a 20 mL vial with 1.5 g NaCl and 10 μL 4-methyl-2-pentanol (internal standard, 1.0086 g/L). The vial was capped with a PTFE‑silicone septum. The sample was equilibrated at 40 °C with 500 rpm agitation for 30 min, followed by extraction for 30 min. Subsequently, the fiber was desorbed in the GC injector for 8 min. GC–MS (7890 A-5975C, Agilent Technologies, Santa Clara, CA, USA) was equipped with an HP-INNOWAX column (60 m × 0.25 mm, 0.25 μm, J&W Scientific, Folsom, CA, USA). A 5:1 split injection was used with the injector temperature at 250 °C. The carrier gas was helium at a flow rate of 1 mL/min. The temperature program: 50 °C held for 1 min, increased to 220 °C at 3 °C/min, and held at 220 °C for 5 min. Mass spectrometry used an electron impact ionization mode. The ion source temperature was 230 °C, the quadrupole temperature was 150 °C, and the acquisition range was m/z 29–350. Identification was verified by matching mass spectra with standards and the NIST 2011 library, supplemented by calculating retention indices. Quantification was performed using calibration curves of volatile compound standards in simulated wine, as described in 2.4.1. Color measurements CIELab color parameters (lightness L*, red/green a*, yellow/blue b*) were analyzed using a UV–Vis spectrophotometer (UV-2450, Shimadzu Corporation, Kyoto, Japan) (Ayala et al., 1997). Wine samples were filtered through 0.22 μm polyethersulfone membranes (Jinteng Experimental Equipment Co., Ltd., Tian, China). Using distilled water as the blank reference, absorbance values at 440 nm, 530 nm, and 600 nm were measured in a 2 mm optical path length glass cuvette. Sensory evaluation Sensory evaluation was conducted using the quantitative descriptive analysis (QDA) method (Stone et al., 1974). For this study, 20 individuals with wine tasting experience (15 females and 5 males, aged 22–32 years) were recruited from the Center for Viticulture and Enology (CFVE) at China Agricultural University to participate in sensory training. All assessors, including the experts, provided written informed consent prior to participation in accordance with the principles of the 1975 Declaration of Helsinki. Ethical approval for this study involving human subjects was granted by the Research Ethics Committee of China Agricultural University (Approval No. CAUHR-20231103). Prior to the formal tasting, the recruited individuals underwent a total of 32 h (16 sessions) of systematic sensory training based on the GB/T 46555–2025 standard. The training content covered learning descriptor development, familiarization with reference standards, intensity calibration, and pre-experimental evaluations. Six experts conducted a preliminary evaluation to determine descriptor for 6 aroma attributes (red berry, black berry, flowery, herbaceous, cooked fruit, and overall aroma) and 4 palate attributes (astringency, acidity, bitterness, and smoothness), and selected representative samples as training references. The members underwent descriptor discrimination and screening training using reference standard solutions (Supplementary Table 1), followed by standard intensity training using reference standard solutions with dilution gradients and untreated Cabernet Sauvignon wine samples from the control group. An 11-point scale was utilized (0–10, where 0 = “absent”, 1 = “barely perceptible”, 3 = “perceptible”, 5 = “directly identifiable”, 7 = “relatively prominent”, and 9 = “extremely intense”), and consensus was reached through open discussion. Finally, the members performed pre-evaluations on representative training samples over three consecutive days. A three-way ANOVA was conducted using PanelCheck v1.4.2 software (https://www.panelcheck.com) to determine if significant differences existed among the members and across rounds (p > 0.05). The sensitivity, stability, and scoring consistency of the members were also evaluated. Ultimately, 16 panelists (12 females and 4 males, aged 22–32 years) were selected to participate in the formal experiments. Before each formal tasting session, two significantly different wine samples from the training set were selected as references for scoring and discussion to ensure a unified evaluation scale. The formal tasting procedure was identical to the training. Specifically, 30 mL of each wine sample was poured into transparent ISO tasting glasses in a standardized tasting booth and presented to the assessors in a randomized order to eliminate systematic errors caused by the sample sequence. Assessors used water and unflavored soda crackers as palate cleansers to maintain their oral condition. Finally, PanelCheck software was used again to evaluate and calibrate the data of each assessor, excluding data from panelists demonstrating poor reproducibility for the same sample. Statistical analysis One-way ANOVA and Duncan's multiple comparison test (p < 0.05) were performed using SPSS 24.0 (Chicago, IL, USA). The graphs were drawn using GraphPad Prism 9 (GraphPad Software Inc). Results and discussion Conductivity response The conductivity change (Δχ) of all wine samples, including the untreated control (CK), was below 30 μS/cm. The lowest Δχ values were observed for PF-A2, PF-A3, NF-A2, and NF-A3, followed by PF-A1 and NF-A1 (Fig. 1). Previous research indicated that a conductivity change of less than 50 μS/cm before and after stabilizer addition indicates stability, while a change below 30 μS/cm signifies a very stable wine (Bosso et al., 2016). As dedicated CaT stability testing was not included in this study, the conductivity results should be interpreted strictly in relation to KHT stability. Accordingly,the untreated control wine already exhibited a low residual potassium bitartrate (KHT) crystallization tendency. Within this already stable wine matrix, the lower Δχ values observed at the higher KPA addition rates (45 and 55 mL/hL) indicate relatively lower residual KHT crystallization tendencies. This finding aligns with previous research, which observed a progressive decrease in Δχ with KPA additions of 10, 20, 50, and 100 mL/hL. Only at the excessive dosage of 100 mL/L, a negative effect on the colloidal balance of the wine was observed, resulting in a significant increase in turbidity (Bosso, Motta, Panero, Petrozziello, et al., 2020). Changes in electrical conductivity of different wine samples. CK: Control, no pre-freezing treatment or addition of KPA; Cold-T: Extended cold-treatment reference at −4 °C for 30 d; PF-A1, PF-A2, PF-A3: Pre-freezing for 48 h with KPA addition at concentrations of 35, 45, 55 mL/hL; NF-A1, NF-A2, NF-A3: KPA addition without pre-freezing at concentrations of 35, 45, 55 mL/hL. Different letters represent significant differences in data between different treatments (Duncan, p < 0.05). The NF-A and PF-A treatments exhibited lower Δχ values than the extended cold-treatment reference. However, because the untreated control wine already showed a low Δχ value, these differences should be interpreted only as indicating that KPA treatment resulted in a relatively lower residual KHT crystallization tendency than the extended cold treatment within an already stable wine matrix. Previous studies have also shown that KPA has certain advantages (Bosso, Motta, Panero, Lucini, et al., 2020; Canuti et al., 2019; Eder et al., 2019). Studies have shown that MTA, CMC, and KPA can all effectively inhibit KHT crystallization, although they differ in their duration of action, sensitivity to storage temperature, and applicability to different wine matrices. In particular, KPA has demonstrated persistent efficacy under elevated temperatures and prolonged storage conditions (Bosso, Motta, Panero, Lucini, et al., 2020). Only KPA consistently prevented tartrate precipitation across all tested wine varieties, even at half the recommended dosage (Eder et al., 2019), or under demanding storage conditions such as elevated storage temperatures and prolonged aging (40 °C for 45 days) (Canuti et al., 2019). The precise mechanism by which KPA inhibits KHT crystallization in wine has not yet been fully elucidated. It may be associated with the previously reported metal-chelating capacity of polyaspartate, whereby adjacent carboxyl groups and the lone pair electrons from the nitrogen (CONH) in the polymer can bind metal ions (Adelnia et al., 2023). One possible explanation is that such metal chelation may interfere with KHT nucleation or crystal growth. The strong KHT-stabilizing effect of KPA should not be extrapolated as being equally effective for CaT stability. Although the specific calcium concentration of the studied Cabernet Sauvignon wine was not directly measured in this work, Cabernet Sauvignon red wines generally exhibit relatively low calcium levels (75–80 mg/L) compared with certain white wines (Laurie et al., 2026). From a practical standpoint, CaT instability is particularly difficult to control due to its prolonged and unpredictable induction period, where precipitation may only emerge months after bottling or even after reaching the market (Mckinnon et al., 1995). Furthermore, CaT instability is governed not only by calcium concentration but also by factors such as pH, tartaric acid and its ionic species, ethanol content, organic acid composition, and the wine colloidal matrix (Cataldo et al., 2026). Previous studies have demonstrated that KPA cannot reliably eliminate the risk of CaT precipitation in wines with elevated calcium levels. Moreover, the precipitation of CaT can act as a physical nucleus, disrupting the thermodynamic equilibrium of the colloidal system and triggering the co-precipitation of previously stabilized KHT (Philipp, Strauss, Schlögl, et al., 2022). Therefore, for wines with elevated calcium concentrations or a potential risk of exogenous calcium contamination, CaT risk should be independently evaluated, and necessary enological measures should be implemented. Such measures include delayed bottling to allow natural CaT to partially precipitate before adding KPA, the staggered addition of KPA (Philipp, Strauss, Schlögl, et al., 2022), or alternative strategies like cation exchange resins, electrodialysis, or sodium alginate (Laurie et al., 2026). Although the combination of cold stabilization and stabilizer addition represents a novel direction in tartrate stabilization techniques (Cui et al., 2024), the group PF-A and the group NF-A with equivalent KPA concentration showed no significant difference in Δχ values (Fig. 1), indicating that 48 h of pre-freezing did not produce additional improvement in KHT stability based on conductivity evaluation. Furthermore, because CaT exhibits slow crystallization kinetics, typically requiring weeks or even months to begin precipitating (Philipp, Strauss, Schlögl, et al., 2022), the 48 h pre-freezing treatment was fundamentally insufficient to induce CaT stabilization. This further confirms that the performance of KPA in the combined pre-freezing process was primarily attributed to its inhibitory effect on KHT. In fact, the pre-freezing and filtration processes might alter certain wine matrix components that contribute less to total conductivity, such as macromolecular cold colloids (formed through the aggregation of pigments, proteins, polysaccharides, tannins, and so on) (Prusisz et al., 2008). Therefore, in the NF-A treatment, KPA acts directly on the intact original colloidal matrix; whereas in the PF-A treatment, KPA is added to a partially altered wine matrix. This matrix difference caused by the processing methods may provide a physicochemical basis for the subsequent differences observed between the two treatments regarding phenolic and volatile compounds. Phenolic composition To comprehensively investigate the impact of KPA on phenolic compounds in wine, monomeric anthocyanins and their derivatives, as well as non-anthocyanic phenolics, were detected. Before and after bottle aging, a total of 22 anthocyanic phenolics were quantified (Table 1), including 5 primary anthocyanins, 5 corresponding acetylated anthocyanins, 5 corresponding coumaroylated anthocyanins, and 7 anthocyanin derivatives. Additionally, 34 non-anthocyanic phenolics were quantified (Table 2), including 9 flavan-3-ols, 16 flavonols, 6 hydroxybenzoic acids, and 3 hydroxycinnamic acids. Table 1Content of anthocyanin-related phenolic compounds in different wine samples before and after 6 months of bottle storage.Anthocyanin-related phenolics (mg/L)CKCold-TPF-A1PF-A2PF-A3NF-A1NF-A2NF-A30 MMonomeric anthocyaninsDelphinidin-3-O-glucoside6.014 d4.765 f6.722 a6.679 ab6.465 abc5.500 e6.198 cd6.352 bcdCyanidin-3-O-glucoside1.795 cd1.560 ef1.984 a1.946 a1.921 ab1.769 cd1.851 bc1.916 abPetunidin-3-O-glucoside7.527 c5.913 e8.473 a8.199 ab8.078 ab6.877 d7.512 c7.855 bcPeonidin-3-O-glucoside13.077 de10.988 f14.980 a14.477 ab13.849 c13.202 d14.112 bc14.573 abMalvidin-3-O-glucoside57.021 de48.282 f64.639 a63.029 ab59.826 cd57.067 de60.776 bc62.786 abDelphinidin-3-O-acetylglucoside1.827 cd1.532 f2.068 a2.009 a1.884 bc1.736 de1.834 cd1.968 abCyanidin-3-O-acetylglucoside0.853 bc0.770 f0.901 a0.905 a0.878 ab0.816 de0.870 ab0.891 aPetunidin-3-O-acetylglucoside2.808 bcd2.233 f3.055 a2.926 ab2.880 abcd2.547 e2.696 cde2.910 abcPeonidin-3-O-acetylglucoside4.505 cd3.723 f4.951 a4.961 a4.739 b4.216 e4.682 bc4.852 abMalvidin-3-O-acetylglucoside24.139 cd19.992 e26.609 a26.805 a25.119 bc23.036 d25.285 bc26.190 abDelphinidin-3-O-coumaroylglucoside (cis + trans)0.670 ab0.621 cde0.698 a0.696 a0.682 a0.617 de0.667 abc0.680 aCyanidin-3-O-coumaroylglucoside (cis + trans)0.563 abcd0.548 bcde0.588 a0.589 a0.569 abc0.541 bcde0.562 abcd0.574 abPetunidin-3-O-coumaroylglucoside (cis + trans)0.748 a0.685 bc0.759 a0.767 a0.752 a0.671 c0.741 a0.756 aPeonidin-3-O-coumaroylglucoside (cis + trans)1.450 ab1.234 cde1.538 a1.492 ab1.476 ab1.214 def1.355 bcd1.484 abMalvidin-3-O-coumaroylglucoside (cis + trans)4.694 a3.589 cd4.955 a4.975 a4.807 a3.800 bc4.550 a4.974 aTotal monomeric anthocyanins127.690 cd106.388 e142.919 a140.456 ab133.927 bc123.608 d133.689 bc138.762 abAnthocyanin derivativesF-A2.354 bc2.168 d2.502 a2.550 a2.501 a2.557 a2.461 ab2.507 aA-e-F0.366 b0.797 a0.349 b0.365 b0.362 b0.346 b0.380 b0.414 bA-v-F0.312 abcde0.302 bcde0.342 abc0.348 ab0.333 abc0.296 cde0.325 abcd0.352 abVitisin3.750 c5.444 a4.387 b4.326 b4.324 b4.281 b4.373 b4.563 bPinotin7.318 abcd6.789 cde8.025 ab8.172 a7.970 abc6.582 de7.710 abcd8.112 aTotal anthocyanin derivatives14.101 def15.499 bcde15.605 bcd15.761 bcd15.490 bcde14.062 def15.250 bcde15.949 abc6 MMonomeric anthocyaninsDelphinidin-3-O-glucoside2.472 bc2.114 f2.506 a2.497 ab2.442 cd2.353 e2.330 e2.416 dCyanidin-3-O-glucoside1.820 a1.737 e1.823 a1.817 ab1.810 bc1.791 d1.789 d1.804 cPetunidin-3-O-glucoside2.836 ab2.286 e2.861 a2.845 ab2.787 b2.615 d2.603 d2.725 cPeonidin-3-O-glucoside4.596 a3.343 e4.603 a4.586 a4.504 a4.121 c4.098 c4.346 bMalvidin-3-O-glucoside13.344 bc8.678 e13.857 ab13.941 a13.159 c11.950 d11.565 d12.924 cDelphinidin-3-O-acetylglucoside1.828 a1.729 d1.829 a1.825 ab1.816 ab1.809 b1.791 c1.819 abCyanidin-3-O-acetylglucoside1.706 a1.670 e1.705 a1.702 ab1.701 abc1.695 d1.696 cd1.699 bcdPetunidin-3-O-acetylglucoside2.177 a1.896 f2.185 a2.181 a2.151 b2.082 d2.057 e2.124 cPeonidin-3-O-acetylglucoside2.414 bc2.070 d2.526 a2.524 a2.496 ab2.379 c2.350 c2.428 abcMalvidin-3-O-acetylglucoside6.759 ab4.304 e6.847 a6.839 a6.620 b5.951 d5.724 e6.259 cDelphinidin-3-O-coumaroylglucoside (cis + trans)1.669 c1.648 g1.674 a1.671 b1.666 d1.665 e1.661 f1.666 dCyanidin-3-O-coumaroylglucoside (cis + trans)1.651 ab1.638 e1.653 a1.651 ab1.650 bc1.648 cd1.646 d1.649 bcPetunidin-3-O-coumaroylglucoside (cis + trans)1.702 ab1.662 f1.705 a1.702 b1.694 c1.687 d1.682 e1.692 cPeonidin-3-O-coumaroylglucoside (cis + trans)1.835 b1.720 j1.849 a1.838 b1.819 c1.801 e1.784 f1.810 dMalvidin-3-O-coumaroylglucoside (cis + trans)2.511 c2.016 g2.570 a2.536 b2.446 d2.383 f2.282 g2.417 eTotal monomeric anthocyanins49.317 ab38.511 e50.194 a50.156 a48.762 bc45.929 d45.058 d47.778 cAnthocyanin derivativesF-A1.212 a1.152 e1.206 ab1.208 a1.211 a1.192 c1.184 d1.200 bA-F0.291 abc0.287 bc0.292 ab0.292 abc0.293 a0.279 d0.287 c0.293 aA-v-F1.509 de1.534 a1.520 bc1.510 ef1.503 f1.483 g1.513 d1.514 cdA-e-F1.026 e1.350 a1.028 de1.032 de1.041 d1.034 de1.147 b1.067 cVitisin3.453 e5.832 a3.615 d3.585 d3.547 de3.493 de4.499 b3.819 cPinotin6.502 c6.203 e6.949 a6.641 b6.244 de6.289 d6.587 b6.605 bPortisin0.555 bc0.556 ab0.556 abc0.556 abc0.555 abc0.555 abc0.554 c0.555 abcTotal anthocyanin derivatives14.256 e16.626 a14.875 c14.531 d14.101 ef14.044 f15.485 b14.761 cCK: Control, no pre-freezing treatment or addition of KPA; Cold-T: Extended cold-treatment reference at −4°C for 30 d; PF-A1, PF-A2, PF-A3: Pre-freezing for 48 h with KPA addition at concentrations of 35, 45, 55 mL/hL; NF-A1, NF-A2, NF-A3: KPA addition without pre-freezing at concentrations of 35, 45, 55 mL/hL. F-A: Direct flavanol-anthocyanin condensation product (B type); A-F: Direct anthocyanin-flavanol condensation products (A type); A-v-F: Flavanyl-pyranoanthocyanins; A-e-F: Anthocyanin ethyl-linked flavanols/Acetaldehyde-bridged flavanol-anthocyanins condensation product. Different letters in a row represent significant differences in data between different treatments (Duncan, p < 0.05). Content of anthocyanin-related phenolic compounds in different wine samples before and after 6 months of bottle storage. CK: Control, no pre-freezing treatment or addition of KPA; Cold-T: Extended cold-treatment reference at −4°C for 30 d; PF-A1, PF-A2, PF-A3: Pre-freezing for 48 h with KPA addition at concentrations of 35, 45, 55 mL/hL; NF-A1, NF-A2, NF-A3: KPA addition without pre-freezing at concentrations of 35, 45, 55 mL/hL. F-A: Direct flavanol-anthocyanin condensation product (B type); A-F: Direct anthocyanin-flavanol condensation products (A type); A-v-F: Flavanyl-pyranoanthocyanins; A-e-F: Anthocyanin ethyl-linked flavanols/Acetaldehyde-bridged flavanol-anthocyanins condensation product. Different letters in a row represent significant differences in data between different treatments (Duncan, p < 0.05). Table 2Content of non-anthocyanin phenolic compounds in different wine samples before and after 6 months of bottle storage.Non-anthocyanin phenolics (mg/L)CKCold-TPF-A1PF-A2PF-A3NF-A1NF-A2NF-A30 MFlavan-3-olsCatechin24.931 cde25.047 cde28.706 abcde31.099 abc33.081 a22.443 e24.329 de29.141 abcdGallocatechin8.309 bc7.810 def8.429 b8.918 a8.887 a7.583 f7.875 cdef7.783 efEpicatechin33.334 abc31.491 cd31.515 cd35.115 ab35.723 a30.159 d31.209 cd31.954 cdEpigallocatechin3.347 def3.035 g3.533 bcd3.715 ab3.827 a3.017 g3.146 fg3.478 cdProcyanidin B159.477 a56.661 ab53.536 bc56.325 ab59.960 a55.239 ab53.870 bc55.323 abProcyanidin B233.770 a30.692 de30.990 cde32.723 abc33.062 ab30.758 de29.713 def31.248 cdeProcyanidin C111.851 a10.775 b9.318 d10.754 b10.457 bc9.267 d9.657 cd10.436 bcEpicatechin-3-gallate0.397 ef0.395 f0.409 abcdef0.412 abcde0.421 ab0.398 ef0.406 bcdef0.419 abcEpigallocatechin 3-gallate0.247 a0.246 a0.250 a0.246 a0.247 a0.244 a0.245 a0.244 aTotal flavan-3-ols175.664 bc169.539 bcde166.686 cdef179.306 ab185.664 a159.108 f160.450 ef170.026 bcdeFlavonolsMyricetin3.862 ab3.227 bcde3.407 abcd3.603 abc3.810 ab2.485 e2.822 cde3.370 abcdMyricetin-3-O-glucoside14.730 ab14.753 ab13.244 d14.126 abc14.014 bc11.884 f12.350 ef12.890 deMyricetin-3-O-galactoside1.954 bc2.294 abc2.101 bc2.356 abc2.351 abc1.726 c1.906 c1.893 cQuercetin7.787 ab7.035 ab7.259 ab7.598 ab8.021 a4.462 c6.320 abc7.597 abDihydroquercetin1.176 a1.055 abcd0.966 cd1.106 abc1.095 abc0.942 d0.978 bcd1.038 abcdQuercetin-3-O-glucoside5.027 a5.178 a4.498 cd4.946 ab4.729 bc4.087 e4.279 de4.431 dQuercetin-3-O-glucuronide24.966 ab24.382 abc22.915 cd24.183 bc24.021 bc19.643 g21.057 ef21.942 deQuercetin-3-O-galactoside2.810 a2.781 a1.990 cd2.478 abc2.724 ab2.340 abcd2.225 abcd1.784 dQuercetin-3-O-rhamnoside0.427 a0.410 ab0.375 c0.416 ab0.413 ab0.367 c0.365 c0.369 cDihydrokeampferol0.921 a0.839 ab0.844 ab0.882 ab0.934 a0.786 b0.767 b0.827 abKaempferol-3-O-glucoside0.376 a0.381 a0.355 abc0.369 a0.364 ab0.326 c0.324 c0.328 cKaempferol-3-O-galactoside0.339 ab0.361 a0.291 cd0.309 abcd0.328 abc0.305 bcd0.281 d0.289 cdSyringetin0.201 b0.201 b0.215 a0.201 b0.201 b0.206 ab0.207 ab0.200 bSyringetin-3-O-glucoside7.332 a7.161 ab6.605 cde7.123 abc6.983 abc6.097 e6.349 de6.695 bcdLaricitrin0.625 a0.595 a0.570 a0.546 a0.535 a0.563 a0.658 a0.619 aIsorhamnetin-3-O-galactoside2.241 a2.192 a2.011 bc2.167 a2.148 a1.788 d1.877 d1.879 dTotal flavonols74.678 a73.627 a67.629 b72.354 a72.546 a55.970 e62.668 cd66.138 bcHydroxybenzoic acidsGallic acid58.424 a58.930 a40.968 c47.322 b49.223 b35.312 de33.691 e36.856 d4-Hydroxybenzoic acid1.096 a1.092 a0.725 d0.845 c0.925 b0.576 e0.572 e0.621 eProtocatechuic acid4.378 a4.182 b3.129 e3.509 d3.782 c2.519 g2.486 g2.760 fGentisic acid1.875 a1.779 ab1.352 e1.521 d1.593 cd1.150 f1.157 f1.208 fVanillic acid3.899 ab3.979 ab3.112 c2.944 cd3.720 b2.569 de2.313 e2.693 cdeTotal hydroxybenzoic acids69.673 ab70.268 a49.286 d56.141 c59.243 c42.126 e43.364 e44.138 eHydroxycinnamic acidsChlorogenic acid0.175 ab0.172 b0.183 ab0.171 b0.176 ab0.187 a0.179 ab0.172 bCaffeic acid11.265 a11.099 ab7.219 e8.193 d9.009 c5.618 fg5.320 g6.111 fFerulic acid0.391 a0.369 a0.244 cd0.278 bc0.329 abc0.188 de0.108 e0.171 deTotal hydroxycinnamic acids11.830 a11.750 a7.645 de8.642 cd9.514 bc5.993 f3.916 g6.454 efTotal non-anthocyanin phenolics331.942 a323.180 abc291.262 e316.499 bcd327.093 ab265.235 f269.038 f286.769 e6 MFlavan-3-olsCatechin21.544 a19.723 c20.345 bc21.566 a21.696 a19.803 bc19.947 bc20.519 bGallocatechin20.582 a16.785 ab16.787 ab20.402 a18.714 ab16.840 ab16.281 b16.761 abEpicatechin40.989 a36.970 de39.381 bc39.714 b41.098 a38.609 c37.602 d38.772 bcEpigallocatechin0.840 bc0.793 de0.867 ab0.865 ab0.870 a0.877 a0.821 cd0.872 aProcyanidin B126.376 d23.085 e27.765 ab26.830 cd27.288 bc28.017 a26.858 cd27.402 abcProcyanidin B228.812 b25.271 f27.977 c29.393 a29.516 a27.429 d26.898 e27.771 cdProcyanidin C14.655 ab3.783 cd4.288 abcd4.728 a4.818 a4.402 abc3.954 bcd4.410 abcTotal flavan-3-ols143.798 a126.410 d137.409 b143.498 a144.000 a135.977 b132.361 c136.507 bFlavonolsMyricetin9.174 b8.660 c8.652 c10.141 a9.175 b7.850 d7.710 d8.595 cMyricetin-3-O-glucoside20.012 bcd20.694 abcd19.546 bcd21.351 ab21.116 ab20.461 abcd18.767 cd18.705 dMyricetin-3-O-galactoside0.141 bcde0.161 bc0.055 bcde0.183 ab0.125 bcde0.013 de0.029 cde0.002 eQuercetin19.179 b17.073 d20.049 a20.345 a18.894 b17.870 c18.285 c20.245 aQuercetin-3-O-glucoside3.147 c4.962 a4.242 b4.591 ab4.523 ab2.533 d2.652 cd2.665 cdQuercetin-3-O-glucuronide42.381 a41.054 cd40.450 d42.523 a41.546 bc38.876 f39.789 e38.596 fQuercetin-3-O-galactoside1.418 bc1.424 bc1.328 cd1.596 a1.561 a1.054 f1.090 f1.183 eDihydrokeampferol0.997 a0.811 b0.996 a1.133 a1.070 a0.723 bc0.806 b1.107 aSyringetin0.740 d0.714 d0.954 a0.964 a0.846 bc0.712 d0.542 e0.928 abSyringetin-3-O-glucoside17.661 ab17.140 bc16.894 cd17.784 a17.793 a16.082 e16.115 e16.276 eIsorhamnetin3.678 bc2.816 e3.836 b3.759 b3.516 c3.234 d3.318 d4.205 aIsorhamnetin-3-O-glucoside1.102 cdef1.233 abcde1.335 abc1.476 a1.328 abcd0.934 f1.071 def1.096 cdefTotal flavonols119.630 cd116.742 ef118.339 de125.845 a121.492 bc110.341 h110.175 h113.602 gHydroxybenzoic acidsGallic acid86.848 b91.362 a83.983 de83.045 e85.081 cd82.659 ef83.068 e85.263 cd4-Hydroxybenzoic acid2.635 a2.681 a2.623 ab2.458 de2.522 bcd2.449 de2.455 de2.579 abcProtocatechuic acid2.284 a2.312 a2.022 c2.177 b2.212 b1.995 c1.995 c2.026 cGentisic acid2.440 b2.549 a2.560 a2.374 b2.420 b2.557 a2.575 a2.584 aVanillic acid3.382 a3.318 a3.333 a3.462 a3.425 a3.355 a3.380 a3.240 aSyringic acid0.152 f0.233 bcde0.328 a0.163 ef0.199 cde0.251 bcd0.215 cdef0.304 abTotal hydroxybenzoic acids97.740 b102.454 a94.851 de93.680 ef95.859 cd93.265 f93.688 ef95.996 cdHydroxycinnamic acidsCaffeic acid14.373 b14.746 a13.725 d13.845 cd14.014 c13.195 e13.646 d13.878 cd4-Hydroxycinnamic acid18.033 ab18.202 a17.024 e17.544 d17.719 bcd16.970 e17.592 cd17.700 cdTotal hydroxycinnamic acids32.406 b32.948 a30.749 f31.388 de31.733 cd30.165 f31.239 def31.578 cdeTotal non-anthocyanin phenolics393.573 a378.554 b381.348 cd394.411 a393.084 a369.748 e367.463 ef377.683 bCK: Control, no pre-freezing treatment or addition of KPA; Cold-T: Extended cold-treatment reference at −4°C for 30 d; PF-A1, PF-A2, PF-A3: Pre-freezing for 48 h with KPA addition at concentrations of 35, 45, 55 mL/hL; NF-A1, NF-A2, NF-A3: KPA addition without pre-freezing at concentrations of 35, 45, 55 mL/hL. Different letters in a row represent significant differences in data between different treatments (Duncan, p < 0.05). Content of non-anthocyanin phenolic compounds in different wine samples before and after 6 months of bottle storage. CK: Control, no pre-freezing treatment or addition of KPA; Cold-T: Extended cold-treatment reference at −4°C for 30 d; PF-A1, PF-A2, PF-A3: Pre-freezing for 48 h with KPA addition at concentrations of 35, 45, 55 mL/hL; NF-A1, NF-A2, NF-A3: KPA addition without pre-freezing at concentrations of 35, 45, 55 mL/hL. Different letters in a row represent significant differences in data between different treatments (Duncan, p < 0.05). Principal component analysis At 0 M of bottle storage, the first two principal components explained 81.4% of the variance among the sample set, with specific clustering observed according to the different experimental treatments. The KPA supplementation groups (group NF-A and PF-A) were clustered on the right side of the plot. Group PF-A was concentrated in the upper-right quadrant, while group NF-A was distributed in the lower-right quadrant. In contrast, the control and extended cold-treatment reference groups (CK, Cold-T) occupied the left side (Fig. 2A). Meanwhile, direct flavanol-anthocyanin condensation product (F-A) was grouped on the right side of the plot, while nonacylated anthocyanins, acetylated anthocyanins, coumaroylated anthocyanins, flavanyl-pyranoanthocyanins (A-v-F), and pinotin clustered in the upper-right quadrant. In contrast, anthocyanin ethyl-linked flavanols (A-e-F) were located on the left side of the plot (Fig. 2C). This indicates that KPA addition was associated with higher direct flavanol-anthocyanin condensation product (F-A) levels and lower anthocyanin ethyl-linked flavanols (A-e-F) levels. Furthermore, combined pre-freezing and KPA showed strong correlations with higher flavanyl-pyranoanthocyanins (A-v-F), pinotin, and various monomeric anthocyanins. A possible mechanism underlying these treatment-related associations might result from interactions between the carboxylate-rich KPA chain and multivalent metal ions, which may modify the oxidative environment of wine pigments (Vernhet et al., 1999). This hypothesis still needs to be verified in the future by measuring metal ion concentrations and their binding behavior. Principal component analysis (PCA) score and loading plots of phenolic compounds in different wine samples before and after 6 months of bottle aging. A, C: Score and loading plots at 0 M of bottle aging; B, D: Score and loading plots at 6 M of bottle aging. CK: Control, no pre-freezing treatment or addition of KPA; Cold-T: Extended cold-treatment reference at −4 °C for 30 d; PF-A1, PF-A2, PF-A3: Pre-freezing for 48 h with KPA addition at concentrations of 35, 45, 55 mL/hL; NF-A1, NF-A2, NF-A3: KPA addition without pre-freezing at concentrations of 35, 45, 55 mL/hL. F-A: Direct flavanol-anthocyanin condensation product (B type); A-F: Direct anthocyanin-flavanol condensation products (A type); A-v-F: Flavanyl-pyranoanthocyanins; A-e-F: Anthocyanin ethyl-linked flavanols/Acetaldehyde-bridged flavanol-anthocyanins condensation product. During 6 months of bottle storage, 71.5% of the variance among the sample set could be explained by the first two PCs. Clustering according to the different experimental treatments was also observed. Cold-T wines were distributed on the left side of the plot. The NF-A group clustered toward the center, while the PF-A group aggregated near the center-right with overlapping regions (Fig. 2B). This indicates that wines with or without pre-freezing treatment prior to KPA addition exhibited similarities in phenolic composition. Pre-freezing treatment did not produce significant effects, which may be related to the duration of freezing treatment (Cui et al., 2024). Nonacylated anthocyanins, acetylated anthocyanins, coumaroylated anthocyanins, direct flavanol-anthocyanin condensation product (F-A), pinotins, and flavanols were grouped on the right side of the plot, while flavanyl-pyranoanthocyanins (A-v-F), anthocyanin ethyl-linked flavanols (A-e-F), and hydroxycinnamic acids were located on the left (Fig. 2D). These loading patterns indicate treatment-related differences in phenolic composition. The extended cold-treatment reference (Cold-T) was more closely associated with flavanyl-pyranoanthocyanins (A-v-F), anthocyanin ethyl-linked flavanols (A-e-F), and hydroxycinnamic acids, whereas the combined pre-freezing and KPA treatments were more closely associated with monomeric anthocyanins, direct flavanol-anthocyanin condensation products (F-A), pinotins, and flavanols. Anthocyanin-related phenolic compounds Table 1 shows the composition and content of anthocyanin phenolic compounds in wines subjected to various treatments before and after bottle storage for 6 months. The monomeric anthocyanins and anthocyanin derivatives provide insights into the color characteristics of wines. In Cold-T, all types of monomeric anthocyanins were consistently significantly lower than the control group (CK), both before and after bottle storage. The lower monomeric anthocyanin concentrations in Cold-T may be related to anthocyanin adsorption onto and coprecipitation with tartrate crystals during the extended cold treatment. As previous research detected in tartrate crystals after cold stabilization, anthocyanins accounted for 0.2%–0.3% of their composition (Vernhet et al., 1999). The total content of anthocyanin derivatives in Cold-T remained slightly higher than that of the control group both at 0 M and 6 M. Previous studies have reported that low-temperature stabilization may increase dissolved oxygen levels in wine (Cant, 1960). Accordingly, the slightly higher anthocyanin derivative content observed in Cold-T may be associated with differences in oxidative exposure during the extended cold treatment (X.-K. Zhang et al., 2021). Within the NF-A group, only the NF-A3 exhibited a significantly higher total monomeric anthocyanin content than the control group (CK) at 0 M, and the contents of nonacylated anthocyanins, acetylated anthocyanins, and coumaroylated anthocyanins were all relatively high. After 6 months, the total monomeric anthocyanin content in all NF-A groups was lower than that of the control group. Previous studies similarly reported unchanged (Bosso, Motta, Panero, Lucini, et al., 2020) or reduced (Martínez-Pérez et al., 2020) total anthocyanins after KPA addition for 6 months. Regarding anthocyanin derivatives, NF-A3 was significantly higher than the control at 0 M, while both NF-A2 and NF-A3 were significantly higher than the control after bottle storage for 6 months. This is primarily attributed to vitisin. In summary, within the individual KPA addition treatment, the treatment concentration of 45–55 mL/hL (NF-A2 and NF-A3) showed the optimal performance for wine anthocyanin phenolics. Although it had a slightly negative impact on the retention of monomeric anthocyanins post-bottling, it enhanced the content of anthocyanin derivatives. A previous study also observed that a KPA-type tartrate stabilizer (containing potassium polyaspartate, gum arabic, and approximately 0.4% sulfur dioxide) had a significantly negative effect on anthocyanins but a less pronounced effect on their derivatives (Wang et al., 2025). Notably, there is some similarity between our results using high-purity KPA and the findings of Wang et al. (2025), as both observed lower monomeric anthocyanin contents in the treatment groups compared to the control group. This indicates that this phenomenon is primarily driven by KPA itself or the stabilization process, rather than being a side effect induced by the gum arabic additive. In fact, during wine aging and storage, monomeric anthocyanins can react covalently with other substances, particularly phenolics, to form more stable anthocyanin derivatives with unique colors (X.-K. Zhang et al., 2021). One possible mechanism is that the polyaspartic acid chains, rich in carboxylate groups, can partially chelate multivalent metal ions involved in wine oxidation reactions (Adelnia et al., 2023). Consequently, the oxidative degradation of monomeric anthocyanins might be somewhat impeded, allowing them to continue reacting with substances like phenolics to form anthocyanin derivatives. Therefore, the decrease in monomeric anthocyanin content alongside the increase in derivatives such as vitisin likely reflects a redistribution of pigments from monomeric anthocyanin forms to anthocyanin derivative forms. Nevertheless, this hypothesis requires further verification through future measurements of metal binding, oxidative status, and pigment-conversion kinetics. In the PF-A group, at 0 M, the content of all types of monomeric anthocyanins was significantly higher than or equal to that in the control group (CK), with PF-A1 and PF-A2 exhibiting the highest total content. After 6 months, only some monomeric anthocyanins in PF-A3 were lower than those of the control group. Regarding anthocyanin derivatives, at 0 M, all the PF-A groups were not lower than the control group. 6 months later, the content of anthocyanin derivatives in PF-A1 and PF-A2, particularly vitisin, was significantly higher than that of the control group. Vitisin in wine primarily originates from monomeric anthocyanins reacting with pyruvic acid to form vitisin A or with acetaldehyde to form vitisin B (De Freitas & Mateus, 2011). This may be attributed to either low-temperature processing conditions or the metal chelation between potassium ions (K+) in KPA and anthocyanin derivatives, effectively stabilizing the compound (Enaru et al., 2021). In summary, within the process, combined pre-freezing and KPA addition, the optimal treatment concentrations were 35 mL/hL (PF-A1) and 45 mL/hL (PF-A2). They promoted the retention of monomeric anthocyanins and increased the content of anthocyanin derivatives after bottle storage. Non-anthocyanin phenolic compounds This study further analyzed non-anthocyanin phenolic compounds (Table 2), including flavanols, flavonols, and phenolic acids, which can also act as co-pigments contributing to wine color formation. The Cold-T group showed no significant change at 0 M. However, after bottle storage, the content of non-anthocyanin phenolics, particularly flavanols and flavonols, was significantly lower than that of the control group (CK). Previous research similarly found that after cold stabilization treatment exceeding 4 days, the concentration of flavonols was only about 75% of the initial level (Wang et al., 2025). In the NF-A group, the contents of flavanols, flavonols, hydroxybenzoic acids, and hydroxycinnamic acids were significantly lower than those in the control group (CK) both before and after bottle storage, with NF-A3 showing the smallest loss. At 0 M, the non-anthocyanin phenolic content in the NF-A group was approximately 80% of the control group, while at 6 M it was about 95%. Previous research similarly found that after short-term treatment (6 d) with a KPA mixed stabilizer, hydroxybenzoic acids, hydroxycinnamic acids, and flavanols in wine suffered substantial losses, whereas flavonols were relatively unaffected (Wang et al., 2025). Our treatment with pure KPA also resulted in a reduction of non-anthocyanin phenolic compounds, indicating that the interaction with the specific non-anthocyanin phenolics is an inherent characteristic of KPA itself. In the PF-A group, only PF-A3 showed no significant difference in total non-anthocyanin phenolic content compared to the control group (CK) at 0 M. In PF-A1, the contents of almost all types of non-anthocyanin phenolics were significantly lower than those in the control group. PF-A2 exhibited lower levels of hydroxybenzoic acids and hydroxycinnamic acids. During aging, hydroxycinnamic acids can serve as substrates for derivative formation reactions. For example, pinotins are generated from anthocyanins and hydroxycinnamic acids, while Portisins (B type) are formed from anthocyanins, pyruvic acid, and hydroxycinnamic acids (Quaglieri et al., 2017). This mechanism might concurrently explain the decrease in phenolic acid content and the increase in certain anthocyanin derivatives. After bottle storage for 6 months, the flavonol content in PF-A2 and PF-A3 was significantly higher than that of the control group. This increase was the contribution of myricetin-3-O-glucoside, quercetin-3-O-galactoside, quercetin-3-O-glucoside, isorhamnetin-3-O-glucoside, and syringetin. It demonstrated that the combined application of pre-freezing and KPA facilitated the retention of flavonoids, aligning with findings from Bosso, Motta, Panero, Lucini, and Guaita (2020). Volatile compounds To comprehensively analyze the impact of KPA on wine aroma, a total of 68 volatile compounds were detected before and after bottle storage (Supplementary Table 2). These included 28 esters (10 ethyl esters of straight-chain fatty acids, 4 higher alcohol acetates, 4 ethyl esters of branched acids, and 10 other esters), 19 higher alcohols, 4 terpenes and norisoprenoids, 7 medium- and short-chain fatty acids, 6 aldehydes and ketones, and 4 other volatile compounds. Among them, some esters (including straight-chain fatty acid ethyl esters, higher alcohol acetates, and ethyl esters of branched acids), terpenes and norisoprenoids, medium- and short-chain fatty acids, as well as aldehydes and ketones among the different experimental treatments showed significant differences compared to the control group. Principal component analysis Prior to bottle aging, 68% of the variance among the sample sets could be explained by the first two PCs, with observable clustering according to the different experimental treatments. CK was located on the left side of the plot, Cold-T was positioned in the upper middle region, while the PF-A and NF-A groups were predominantly clustered on the right side (Fig. 3A). Terpenes and lactones were grouped on the right side of the plot (Fig. 3C). This indicates that, compared to the control group, the addition of KPA, particularly when combined with the pre-freezing process, resulted in wine samples exhibiting stronger correlations with volatile compounds such as terpenes and lactones. In red wine, there is limited direct research on the influence of KPA addition on the content of volatile compounds. In white wine, KPA could significantly reduce unstable proteins in wine, thereby indirectly supporting aroma stability by preventing protein-related haze and instability that could otherwise affect the aromatic profile (Quaglieri et al., 2017). Principal component analysis (PCA) score and loading plots of aroma compounds in different wine samples before and after 6 months of bottle aging. A, C: Score and loading plots at 0 M of bottle aging; B, D: Score and loading plots at 6 M of bottle aging. CK: Control, no pre-freezing treatment or addition of KPA; Cold-T: Extended cold-treatment reference at −4 °C for 30 d; PF-A1, PF-A2, PF-A3: Pre-freezing for 48 h with KPA addition at concentrations of 35, 45, 55 mL/hL; NF-A1, NF-A2, NF-A3: KPA addition without pre-freezing at concentrations of 35, 45, 55 mL/hL. After bottle storage for 6 months, 69.4% of the variance among the sample sets was explained by the first two PCs, accompanied by clustering according to different experimental treatments. The PF-A group was distributed on the left side of the plot, the NF-A group on the lower middle side, and CK and Cold-T were located in the upper right side (Fig. 3B). Lactones were positioned in the lower-left quadrant, and volatile phenols were located in the upper-left quadrant (Fig. 3D). These results suggest that, compared to the control group, the wine samples treated with pre-freezing combined with KPA exhibited stronger correlations with lactones and volatile phenols after bottle aging for 6 months. Esters Esters are key contributors to the freshly fermented fruit aroma of wine, typically categorized into ethyl esters of straight-chain fatty acids, higher alcohol acetates, ethyl esters of branched acids, and other esters. At 0 M, the total content of ethyl esters of straight-chain fatty acids in the extended cold-treatment reference (Cold-T) and KPA addition treatment (PF-A and NF-A groups) showed no significant difference compared to the control group (CK) (Fig. 4A). This is because ethyl acetate, which accounts for over 99% of the total ethyl esters of straight-chain fatty acids, did not show a significant difference in its content across the various experimental groups. Although the contents of other ethyl esters of straight-chain fatty acids were very low, they still contribute to the aroma. Because even sub- or perithreshold esters can significantly affect the overall odour intensity of red wine (de-la-Fuente-Blanco et al., 2020). At 0 M, the contents of ethyl butanoate, ethyl hexanoate, and ethyl octanoate in the KPA-added groups (PF-A and NF-A groups) were significantly lower than those in the control group (CK) (Supplementary Table 2). After 6 M of bottle storage, the contents of ethyl esters of straight-chain fatty acids in PF-A2, PF-A3, NF-A1, and NF-A2 were lower than those in the control group (Fig. 4A). Similarly, Natolino et al. (2023) also observed a reduction in esters due to KPA treatment. Changes in volatile compound content under different treatments before and after 6 months of bottle aging. A: Ethyl esters of straight-chain fatty acids; B: Higher alcohol acetates; C: Ethyl esters of branched acids; D: Terpenoids and norisoprenoids; E: Medium- and short-chain fatty acids; F: Aldehydes and ketones. CK: Control, no pre-freezing treatment or addition of KPA; Cold-T: Extended cold-treatment reference at −4 °C for 30 d; PF-A1, PF-A2, PF-A3: Pre-freezing for 48 h with KPA addition at concentrations of 35, 45, 55 mL/hL; NF-A1, NF-A2, NF-A3: KPA addition without pre-freezing at concentrations of 35, 45, 55 mL/hL. Different uppercase or lowercase letters in the figure indicate significant differences among treatment groups (Duncan, p < 0.05). Before and after bottle storage, the extended cold-treatment reference (Cold-T) and KPA treatment alone (NF-A group) did not significantly affect the content of higher alcohol acetates. In contrast, the combined treatment of pre-freezing and KPA (PF-A group) resulted in a significantly lower content of higher alcohol acetates compared to the control group (CK) (Fig. 4B), which was primarily attributed to the lower contents of isoamyl acetate and isobutyl acetate (Supplementary Table 2). The differences among the various treatment groups may be attributed to whether cold treatment was applied and the duration of treatment. For example, Xia et al. (2022) observed that short-term cold stabilization (10 d) led to a significantly lower isoamyl acetate content in wine before bottle storage compared to the control, with no significant difference after bottle storage for 6 months. In contrast, long-term cold stabilization (15 d) had no effect on isoamyl acetate either before or after bottle storage for 6 months. For ethyl esters of branched acids, the KPA-added groups (PF-A and NF-A groups) were significantly lower than the control group (CK) both before and after bottle storage, and the negative impact was more pronounced in the KPA treatment combined with pre-freezing (PF-A group) (Fig. 4C). This was primarily due to the lower contents of ethyl 3-methylpentanoate, ethyl 3-methylbutanoate, and ethyl 2-methylbutanoate (Supplementary Table 2). Only the extended cold-treatment reference (Cold-T) showed no significant difference in the ethyl esters of branched acids content compared to the control group. Terpenes and norisoprenoids Before and after bottle storage, 2 norisoprenoids were detected, including β-damascenone and β-ionone. These compounds are characterized by extremely low thresholds and intense floral and fruity aromas, representing one of the core varietal aromas of Cabernet Sauvignon (Falcão et al., 2008), the wine sample used in this study. Additionally, 2 terpenes were detected, including linalool and α-terpineol, which also contribute potentially to the aroma profile (Mele et al., 2021). At 0 M, the contents of terpenes and norisoprenoids in nearly all treatment groups were significantly higher than those in the control group (CK). After 6 M, only the individual KPA treatment (NF-A group) maintained this trend, with the highest levels observed at a concentration of 35 mL/hL (NF-A1) (Fig. 4D). This was mainly attributed to the higher content of β-damascenone (Supplementary Table 2). Proteins and polysaccharides can retain volatile compounds such as norisoprenoids and fatty acids within the wine matrix through non-covalent interactions like hydrophobic interactions, van der Waals forces, and hydrogen bonding (Jones-Moore et al., 2022). Furthermore, studies on protein-containing food and beverage systems indicate that protein addition generally reduces aroma release intensity and prolongs the retention of volatile compounds in the matrix (Barallat-Pérez et al., 2024). Conversely, under wine pH conditions, KPA contains multiple negatively charged carboxylate groups that can interact with unstable proteins and other colloidal components, promoting their aggregation or precipitation (Natolino et al., 2023). One possible explanation is that this could reduce the macromolecular binding sites for aroma substances, thereby increasing the gas-liquid partition coefficients of volatile compounds (Terta et al., 2006). Therefore, the elevated measured concentrations of certain norisoprenoids and fatty acids likely stem from their release from the wine matrix and improved analytical availability, rather than KPA inducing the formation of new aroma compounds. Medium- and short-chain fatty acids Medium- and short-chain fatty acids also contribute significantly to the aroma profile of wine. Among them, branched-chain fatty acids (e.g., isobutanoic acid) imparted odors described as rancid or fatty, while medium-chain fatty acids (e.g., hexanoic, octanoic, and decanoic acids) exhibited cheesy or sweaty notes (Francis & Newton, 2005). At 0 M, 3 medium- and short-chain fatty acids were detected in the wine samples, including isobutanoic acid, hexanoic acid, and 1-decanoic acid. The content of medium- and short-chain fatty acids in the Cold-T group was significantly lower than that in the control group (CK), whereas both the PF-A and NF-A groups showed higher levels than the control. After bottle storage for 6 months, the variety of medium- and short-chain fatty acids increased, with a total of 6 types detected, including acetic acid, butyric acid, 2-methylbutyric acid, hexanoic acid, octanoic acid, and 1-decanoic acid were detected. Only the NF-A group maintained a higher content of medium- and short-chain fatty acids compared to the control, with NF-A1 showing the highest level (Fig. 4E). This indicates that the addition of KPA alone, particularly at 35 mL/hL, can result in a higher content of medium- and short-chain fatty acids. However, this finding is not entirely consistent with previous research. Natolino et al. (2023) observed no significant fatty acid changes in Cortese and Lugana wines after 100 mg/L KPA treatment at 90 d. This discrepancy may be due to differences in the initial chemical composition of the wine samples and the storage duration. Aldehydes and ketones Aldehyde and ketone compounds, including furfural and furfuryl alcohol, are considered important contributors to the roasty aroma notes in wine, imparting characteristics such as toasted, baked, caramel, and nutty scents (Dumitriu (Gabur et al., 2019). At 0 M, 3 aldehyde and ketone compounds were detected in the wine, namely difurfuryl ether, furfural, and furfuryl alcohol. The furfural content in all treatment groups was significantly higher than that in the control group (Supplementary Table 2). Because furfural has been described as an oxidation marker in wines not aged in oak barrels (Escudero et al., 2002), its higher concentration could indicate differences in oxidative exposure among the treatments. One possible mechanism is that wine handling during the extended cold treatment and KPA dispersion might introduce dissolved oxygen and might result in oxidative reactions. Fenton-type reactions represent one possible pathway involved in wine oxidation (Han, 2015), and it will be possible to determine whether the mechanism in this study involves this pathway by measuring relevant parameters in the future. From a practical winemaking perspective, inert-gas protection, such as CO2 or a CO2/Ar gas mixture, may be considered to limit oxygen exposure during stabilizer addition (Del Barrio-Galán et al., 2023). After 6 M, 6 aldehyde and ketone compounds were detected, including 3 newly emerged compounds (3-furaldehyde, octanal, and 2-methylcyclopentanone). Only the treatment with 45 mL/hL KPA (NF-A2) showed a higher content of aldehyde and ketone compounds compared to the control group (CK) (Fig. 4F), primarily due to its higher level of furfuryl alcohol. Color parameters Compared to the control group (CK), the Cold-T group exhibited a significantly lower L* value and a significantly higher a* value at 0 M. This trend persisted after bottle storage for 6 months, with the b* value of the Cold-T group also being significantly lower than that of the control group (Fig. 5). This indicates that the extended cold-treatment reference was associated with lower brightness and higher red hue, and slowed down the yellowing of wine during bottle storage. This variation in apparent color can be reasonably explained by the content of phenolic compounds, which is associated with the higher levels of anthocyanin derivatives in the Cold-T group compared to the control group. Because anthocyanin derivatives can contribute up to 50% of wine color after prolonged aging (X.-K. Zhang et al., 2021). Changes in color parameters of wine under different treatments before and after 6 months of bottle aging. The figure represents the difference in content between the treatment group and CK. A: Brightness (L*); B: Red hue (a*); C: Yellow hue (b*). CK: Control, no pre-freezing treatment or addition of KPA; Cold-T: Extended cold-treatment reference at −4 °C for 30 d; PF-A1, PF-A2, PF-A3: Pre-freezing for 48 h with KPA addition at concentrations of 35, 45, 55 mL/hL; NF-A1, NF-A2, NF-A3: KPA addition without pre-freezing at concentrations of 35, 45, 55 mL/hL. Different uppercase or lowercase letters in the figure indicate significant differences among treatment groups (Duncan, p < 0.05). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) Compared with the control group (CK), the NF-A group showed significantly lower L* and b* values and a significantly higher a* value at 0 M. After 6 M of bottle storage, the same trends for L* and a* values were maintained, while the b* value in the NF-A group were higher than that in the control group (Fig. 5). A lower L* value indicates higher color intensity (CI). This suggests that KPA treatment had a positive effect on wine color, resulting in stronger color intensity and higher redness. Martínez-Pérez et al. (2020) similarly observed that the combination of potassium polyaspartate and gum arabic increased the CI of red wine after both 6 and 12 months of storage. Notably, our study demonstrated that the use of KPA alone can also significantly enhance the CI and red hue of red wine during aging. This comparison confirms that KPA itself plays a key role in color preservation and stabilization, independent of the presence of gum arabic. Bosso et al. (2015) also reported a decrease in the color hue (A420/A520) in some red wine samples after KPA addition, which indicated a shift toward redder tone. In the NF-A group, the addition of KPA at 45 mL/hL (NF-A2) yielded the lowest brightness (L*) and yellow hue (b*) and the highest red hue (a*) after bottle storage. The superior color performance of NF-A2 may be related to the higher content of anthocyanin derivatives, especially vitisin, as mentioned earlier. Compared with its precursor malvidin-3-O-glucoside, vitisin exhibited a shift of the maximum absorption peak in the visible spectrum toward shorter wavelengths, presenting an orange-red hue in the acidic medium of wine (Sun et al., 2020). Moreover, vitisin A has a higher molar extinction coefficient than malvidin-3-O-glucoside, thus contributing to a deeper color (He et al., 2010). Both before and after bottle storage, the PF-A group exhibited significantly higher brightness (L* value) and yellow hue (b* value) than the control group (CK), while its red hue (a* value) was lower (Fig. 5). This indicates that the combined use of pre-freezing and KPA performed poorly in terms of color. The mechanism underlying this phenomenon remains unclear. There is scarcely any literature investigating the effect of combining KPA with the pre-freezing process on wine color. Overall, the addition of KPA alone resulted in a wine with outstanding color intensity and red hue, with an optimal dosage of 45 mL/hL. This may be associated with the higher content of anthocyanin derivatives, particularly vitisin, in the wine samples after the addition of KPA alone treatment. Although the combination of pre-freezing and KPA performed well in terms of phenolic composition, specific process parameters still require further investigation to improve its color performance. Sensory evaluation The sensory characteristics of each treatment group were obtained by a trained evaluation panel, which evaluated the aroma (red berry, black berry, flowery, herbaceous, cooked fruit, and overall aroma) and mouthfeel parameters (astringency, acidity, bitterness, and smoothness) (Fig. 6). Radar chart of sensory evaluation for wines under different treatments before and after 6 months of bottle storage. A: 0 M; B: 6 M. CK: Control, no pre-freezing treatment or addition of KPA; Cold-T: Extended cold-treatment reference at −4 °C for 30 d; PF-A1, PF-A2, PF-A3: Pre-freezing for 48 h with KPA addition at concentrations of 35, 45, 55 mL/hL; NF-A1, NF-A2, NF-A3: KPA addition without pre-freezing at concentrations of 35, 45, 55 mL/hL. Different letters represent significant differences in data between different treatments, and ns indicates that there were no significant differences among all groups. (Duncan, p < 0.05). At 0 M, Cold-T and PF-A1 achieved the highest overall aroma scores. Cold-T exhibited the most prominent cooked fruit aroma, followed by KPA-added groups (PF-A and NF-A), while the control group (CK) displayed the weakest cooked fruit aroma (Fig. 6A). Lactones, furanones, furaneol, and certain ketones are the main chemical groups correlated with the intensity of cooked fruit aromas in Cabernet Sauvignon wines (X. Zhang et al., 2023). Therefore, the enhanced cooked fruit aroma in treatment groups (Cold-T, PF-A, and NF-A) might derive from higher contents of whisky lactones, β-damascenone, and furfuryl alcohol. Compared to CK, nearly all treatments (group Cold-T, PF-A, and NF-A) enhanced astringency (Fig. 6A). Astringency primarily results from interactions between wine polyphenols (notably tannins) and salivary proteins, leading to protein precipitation and reduced oral lubrication (X. Zhang et al., 2023). Therefore, the more pronounced astringency in the treated groups (group Cold-T, PF-A, and NF-A) may be attributed to the preservation of substances such as catechin and epicatechin-3-gallate. At 6 M, there were no significant differences among all the treatment groups and the control group (Fig. 6B). Martínez-Pérez et al. (2020) similarly found that KPA addition had no negative impact on sensory characteristics. It is not contradictory that the analytical differences in volatile compounds observed in this study ultimately manifested as only minor sensory differences. The sensory contribution of an individual volatile compounds depends partly on its concentration relative to its odour threshold and can be preliminarily assessed using its odour activity value (OAV). Compounds present below or close to their odour thresholds may have limited sensory effects, even when their concentrations differ significantly among treatments. However, OAVs alone cannot fully predict wine aroma because sensory perception is also influenced by the wine matrix and interactions among volatile compounds (Cosme et al., 2025). Non-volatile wine components, particularly high-molecular-weight polymers, can interact with odorants and alter their headspace release and sensory expression (Gabler et al., 2023). Moreover, synergistic, masking, or suppressive interactions among aroma compounds may further weaken the direct correspondence between analytical concentrations and sensory responses (Li et al., 2025). Conclusions This study systematically evaluated the effects of KPA, applied at commercially relevant dosage levels either alone or in combination with 48 h of pre-freezing, on phenolic compounds, volatile compounds, color, and sensory quality in Cabernet Sauvignon wine. In conclusion, under the conditions of this study, KPA treatments produced significant changes in the color and phenolic composition of the Cabernet Sauvignon wine examined. Treatment-dependent differences were also observed in some volatile compounds, whereas the sensory aroma was only slightly affected. Specifically, the alone addition of KPA at a concentration of 45 mL/hL produced the optimal color effect, yielding a pronounced red hue after bottle storage for 6 months, attributable to its higher vitisin content. The alone addition of KPA (45–55 mL/hL) also exhibited higher contents of norisoprenoids, carbonyl compounds, and medium- and short-chain fatty acids, while showing no significant effect on the sensory aroma profile. The combined use of 35–45 mL/hL KPA and pre-freezing processing preserved phenolic compounds such as monomeric anthocyanins, anthocyanin derivatives, flavanols, and flavanones, without making a significant contribution to volatile compounds or aroma characteristics. Current research on KPA in enology primarily focuses on tartaric stabilization efficacy. This study provides the first comprehensive investigation into the effects of KPA on the flavor compounds and sensory quality of a Cabernet Sauvignon wine. However, the effects of KPA on residual KHT crystallization tendency and its impact on wine quality may vary depending on the specific wine composition, including colloidal characteristics, pH, tartaric acid and potassium levels, as well as concentrations of calcium and other metal ions, among other factors. Future research should include wines from diverse grape varieties, regions, vintages, and matrix compositions with different initial degrees of KHT instability, together with cold-stabilization procedures that more closely reflect industrial practice and extended storage durations, to further validate these findings. Studies in human This study was performed in compliance with relevant laws, regulatory frameworks and guidelines where the research took place. This study was approved by the the Research Ethics Committee of China Agricultural University. (Approval No. CAUHR-20231103) CRediT authorship contribution statement Zeyu Zhou: Writing – original draft, Visualization, Investigation, Formal analysis, Data curation. Yibin Lan: Writing – review & editing, Supervision, Project administration, Funding acquisition, Conceptualization. Meng Zhou: Validation, Investigation, Data curation. Wenzhe Tong: Software, Methodology. Yixuan Ge: Resources. Cong Yu: Resources. Xuerong Duan: Resources. Changqing Duan: Funding acquisition, Conceptualization. Xuemei Fan: Conceptualization. Nan Meng: Writing – review & editing, Project administration, Conceptualization. Guangfeng Wu: Project administration, Conceptualization. Informed consent and patient details Written informed consent to take part in the study and to publish the article has been obtained from all participants or their legal representatives. The privacy rights of participants have been observed. 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. Acknowledgment This research was supported by the Science and Technology Major Project of Xinjiang Uygur Autonomous Region (2024B02021-3), the Open Project of China Food Flavor and Nutrition Health Innovation Center (CFC2023B-022), the earmarked fund for Agriculture Research System of China (CARS-29), and the National Key Research and Development Program of China (2023YFF1104203). We thank the Center for Viticulture and Enology (CFVE) trained panelists for their time and effort with the sensory evaluation. Footnotes Supplementary data to this article can be found online at https://doi.org/10.1016/j.fochx.2026.104422. Contributor Information Xuemei Fan, Email: 565995789@qq.com. Nan Meng, Email: mengn@btbu.edu.cn. Guangfeng Wu, Email: wumaple@cau.edu.cn. Supplementary data Supplementary material Data availability Data will be made available on request. References Associated Data Supplementary Materials Supplementary material Data Availability Statement Data will be made available on request.