Metabolomic and bioactivity analysis of plum pulp fermented with Lactiplantibacillus, Lacticaseibacillus, and Bifidobacterium strains

Abstract

Probiotic fermentation is a promising method for enhancing fruit functionality. This study investigated the effects of Lactiplantibacillus plantarum, Lacticaseibacillus paracasei, and Bifidobacterium infantis fermentation on plum pulp quality and bioactive properties. Mixed probiotics maintained viable counts above 7 log CFU/mL during fermentation. Compared to unfermented samples, total sugar content decreased by 24 % on day 5, while phenolic and flavonoid contents increased by 17.5 % and 31.4 %, respectively, on day 3. Non-targeted metabolomics detected 2113 metabolites, revealing pathways related to flavonoid biosynthesis and organic acid metabolism. Quercetin and proanthocyanidin B2 were key metabolites significantly upregulated by fermentation. These biochemical changes resulted in optimal antioxidant capacity observed in day 3 fermentation samples across both chemical and cellular antioxidant assays. Correlation analysis identified 30 metabolites, including 4-hydroxybenzoic acid, niacinamide, and resolvin D1, that were strongly associated with antioxidant capacity. This study demonstrates that mixed probiotic fermentation transforms plum pulp into a functional product with enhanced bioactive properties, providing valuable insights for developing health-focused fermented fruit products.

Highlights

• •A co-culture of L. plantarum, L. paracasei and B. infantis fermented plum pulp.
• •Fermentation reduced total sugars by 24 % and enhanced functional oligosaccharides, including trehalose.
• •Fermentation significantly elevated key organic acids like succinic acid and lactic acid.
• •Key phenolics, notably catechin and proanthocyanidin B2, increased substantially.
• •Antioxidant activity was markedly increased post-fermentation.

1. Introduction

Crisp plum (Prunus salicina), a member of the genus Prunus in the Rosaceae family, is an important and unique fruit tree resource in China, widely distributed throughout the Yangtze River Basin and southwestern regions (Cioni et al., 2024; Xiao et al., 2024). Wushan crisp plum, a distinctive variety native to Wushan County, Chongqing, is renowned for its unique flavor and rich nutritional profile. These plums are rich in vitamins, minerals, dietary fiber, and various bioactive compounds, including phenolic compounds, organic acids, and sugars (Ayub et al., 2023; Xiao et al., 2024). Such compounds possess antioxidant, anti-inflammatory, and hypoglycemic properties, playing a vital role in preventing chronic diseases such as cardiovascular diseases, cancer, and diabetes (Ayub et al., 2023; Johnson et al., 2022). However, Wushan crisp plums are highly perishable, with a short shelf life, making them unsuitable for storage or long-distance transportation (Tian et al., 2023). This perishability results in notable seasonal limitations and significant postharvest losses, severely restricting their commercialization and economic value. Therefore, the exploration of novel processing methods is critical for the broader commercialization of fresh crisp plums.

Fermentation, a time-honored and economically valuable biotechnological method, offers significant advantages in enhancing the nutritional value, sensory qualities, safety, and shelf-life of products (Saud et al., 2024). Lactic acid bacteria, generally recognized as safe microorganisms, utilize nutrients in fruit juice, such as sugars, amino acids, and nucleotides, to produce a variety of flavor compounds and bioactive metabolites, including organic acids, esters, alcohols, and phenolic compounds. These processes not only improve the sensory quality of the product but also enhance the absorption potential of bioactive compounds (Wang, He, et al., 2024; Yuan, Mu, et al., 2024). Probiotic strains such as L. plantarum, L. paracasei, and B. infantis have been widely used in the fermentation of various fruit and vegetable juices, demonstrating excellent metabolic conversion capabilities and proliferation characteristics (Yuan, Wang, et al., 2024). For instance, An et al. (2024) reported that goji berry juice fermented with L. rhamnosus for 32 h achieved viable cell counts exceeding 9.0 log CFU/mL while enhancing its flavor. Fermentation with this single strain produced novel volatile compounds such as 4-methylpentanol and 2-butanol, imparting a distinctive aroma to the juice. Similarly, Marnpae et al. (2022) exhibited that fermented gac fruit juice showed elevated levels of β-carotene, DPPH radical scavenging activity, ferric reducing antioxidant power (FRAP), and lipid peroxidation inhibition capacity compared to unfermented juice.

Recent studies have underscored the advantages of co-cultivation fermentation with probiotics. For instance, a co-fermentation system involving Saccharomyces cerevisiae and L. plantarum was shown to modify the odor and taste profiles of mature coconut water by increasing the concentrations of key aromatic compounds, such as ethyl caproate, ethyl decanoate, and isopentanol, compared to single-strain fermentation (Xu et al., 2025). Similarly, the co-cultured fermentation of L. plantarum and Limosilactobacillus fermentum with rambutan juice enhanced both antioxidant activity and flavor complexity (Zhang et al., 2025). Compared to single-strain fermentation, mixed-strain fermentation offers substantial benefits. The synergistic interactions among strains result in a richer metabolite profile, more efficient substrate utilization, and superior biofunctional properties (Yuan, Mu, et al., 2024). L. plantarum exhibits remarkable adaptability across a wide range of fruit and vegetable substrates and is capable of producing diverse volatile compounds that enhance flavor profiles (An et al., 2024; Fonseca et al., 2022). L. paracasei is particularly effective in the biotransformation of flavonoid compounds, notably through deglycosylation reactions that improve the bioavailability of phenolic compounds (Tang et al., 2025; Wang, Wang, et al., 2024). Previous studies have exhibited the L. paracasei can successfully co-ferment with strains such as L. plantarum, as exhibited in passion fruit juice mixed fermentation, where both strains synergistically optimize the sensory and flavor characteristics of the product (Fonseca et al., 2021; Fonseca et al., 2022). As a well-established probiotic, B. infantis is recognized for its safety and multifunctional probiotic properties. Accordingly, we hypothesize that its incorporation into a mixed fermentation system with L. plantarum and L. paracasei could yield synergistic benefits through strain interactions and cooperative metabolic activity.

Recent years have seen a growing interest in probiotic fermentation of fruit juices. For example, Duan et al. (2023) exhibited that lactic acid bacteria fermentation of goji berry juice produced 23 unique metabolites, significantly enhancing its antioxidant activity. Xia et al. (2025) investigated the effects of plant-derived probiotic mixtures on the flavor characteristics, metabolic profile, bioactive components, and potential health benefits of fermented apple (Malus prunifolia) juice. Their findings showed that probiotic fermentation significantly increased the content of aromatic compounds such as naringin and benzaldehyde, while producing 44 differential metabolites, including maltitol and isonicotinamide, and bioactive components like caffeic acid and hydroxycinnamic acid. These metabolites led to pronounced health benefits, including antioxidant, antiviral, and anti-alcoholic fatty liver activities. Similarly, Han et al. (2025) found that fermentation of mulberry juice significantly improved its sensory qualities and promoted the biotransformation of bioactive compounds, enhancing their absorption potential. After fermentation, the total anthocyanin, total flavonoid, and total polyphenol contents of mulberry juice increased by 13.34 %, 43.75 %, and 9.40 %, respectively.

However, probiotics exhibit strain specificity and possess varying biotransformation abilities in different food matrices (Yuan et al., 2024). Research on the fermentation of Prunus fruits, particularly Wushan crisp plums, remains limited, with most existing studies focused on single-strain fermentation or other juice matrices. To date, there is a lack of comprehensive studies combining mixed-strain fermentation with metabolomic profiling of Wushan crisp plums.

In this study, Wushan crisp plums were used as the raw material and were inoculated with a mixed probiotic culture consisting of L. plantarum, L. paracasei, and B. infantis for liquid fermentation. The changes in physicochemical properties and biological activities of the fermented plum pulp were investigated at different fermentation stages. Additionally, non-targeted metabolomics was employed to analyze the key metabolites in fermented plum pulp across various fermentation stages, providing theoretical insights into the biotransformation pathways of Wushan crisp plums fermented with mixed probiotic strains.

2. Methods

2.1. Chemicals

Tartaric acid (≥98 %), Malic acid (≥98 %), Citric acid (≥98 %), Fructose (≥98 %), Glucose (≥98 %) and Sucrose (≥98 %) were purchased from Macklin Biochemical Technology Co., Ltd. (Shanghai, China). Standards included rutin and gallic acid were sourced from Yuanye Bio-Technology Co., Ltd. (Shanghai, China). Sigma Aldrich (St. Louis, MO, USA) supplied Folin-Ciocalteu reagent, ABTS, DPPH and TPTZ. Methanol, acetonitrile, dihydrogen phosphate (KH₂PO₄), aluminum nitrate (Al(NO₃)₃) and potassium persulfate (K₂O₈S₂) were purchased from Chongqing Yuexiang Chemical Co., Ltd. (Chongqing, China). Pectinase (30,000 U/g), cellulase (50,000 U/g) and hemicellulase (50,000 U/g) were purchased from Guangdong Kangda Biotechnology Co.,Ltd. (Guangdong, China). MRS medium was purchased from Beijing Solarbio Science & Technology Co., Ltd. (Beijing, China).

2.2. Materials

Mature, healthy Wushan crisp plums free of disease were harvested from QuChi Township (Wushan, Chongqing, China).

The bacterial strains Lactobacillus plantarum, Bifidobacterium infantis, and Lactobacillus paracasei were obtained from Shaanxi Jushengyuan Biotechnology Co., Ltd. (Shaanxi, China).

RAW264.7 mouse macrophages and culture medium were purchased from Wuhan ProCells Biomedical Technology Co., Ltd. (Wuhan, China). Malondialdehyde kit, glutathione kit, superoxide dismutase kit, catalase kit, and glutathione reductase kit were purchased from Beijing Solarbio Science & Technology Co., Ltd. (Beijing, China).

2.3. Enzymatic digestion and fermentation of fruit pulp

Mature plum fruits were selected, thoroughly washed, and deseeded. The fruits were then homogenized using a pulverizer for 5 min. Enzymatic digestion was performed by adding a composite enzyme mixture (4 mg/g) consisting of pectinase, cellulase, and hemicellulase in a 1:1:2 ratio. The mixture was incubated at pH 3.25 and a temperature of 41.73 °C for 2.58 h.

The resulting enzymatically digested pulp was transferred to sterile conical flasks. A mixed bacterial inoculum containing L. plantarum, B. infantis, and L. paracasei (in a 1:1:1 ratio) was added at an initial concentration of 9 log CFU/mL with a 2 % inoculation volume. Anaerobic fermentation was then conducted at 37 °C for 5 days.

2.4. Probiotic survival and pH changes

Aseptic sampling was performed throughout the fermentation cycle. The specific steps were: after thorough mixing of the pulp in each experimental group, 1 mL of sample was aseptically collected and immediately subjected to tenfold gradient dilution. An aliquot of 0.05 mL from the appropriate dilution was inoculated onto MRS agar plates and incubated at constant temperature in an anaerobic incubator at 37 °C for 48 h. At the end of the incubation period, plates containing 30–300 colonies were selected for enumeration. Results were expressed as log CFU/mL (logarithmic colony-forming units per milliliter). Each value represents the mean ± standard deviation of three independent experiments. Real-time changes in the fermentation system's acidity were monitored using a pH meter.

2.5. Determination of soluble sugars and organic acids

Based on a previously described method with slight modifications (Xiao et al., 2024). Briefly, the samples were ground into powder in liquid nitrogen. Subsequently, 8 mL of 80 % methanol (extraction solvent) was added to 2 g of the sample, and ultrasonic extraction was performed (KQ-500DE, Kunshan Ultrasonic Technology Co., Kunshan, China) at 50 °C for 30 min. The mixture was centrifuged at 25 °C and 11,100 ×g for 5 min. The residue was subjected to two additional extractions with 8 mL of 80 % methanol. The supernatants from the three extractions were combined and diluted with 80 % methanol to a final volume of 25 mL. Before measuring soluble sugars, organic acids, Total Phenol Content (TPC), Total Flavonoid Content (TFC), and chemical antioxidant capacity, store the extract in a dark bottle at 4 °C.

Soluble sugars (glucose, fructose, and sucrose) and organic acids (tartaric acid, citric acid, and malic acid) were analyzed using an HPLC system (Watres e2695, Waters, Massachusetts, USA) as described by Tao et al. (2022). Specifically, 1 mL of extract was centrifuged at 5000 ×g for 2 h, evaporated at 45 °C, and resuspended in 1 mL of distilled water. The aqueous solution was filtered into HPLC sample vials using a 0.22 μm hydrophilic polytetrafluoroethylene (PTFE) membrane filter (Anpel Laboratory Technologies Inc., Shanghai, China). Soluble sugars were separated using a SunFire C18 column (Waters, Massachusetts, USA). The mobile phase consisted of 85 % (v/v) acetonitrile (phase A) and 15 % (v/v) water (phase B), with a flow rate of 1.0 mL/min. The column temperature was maintained at 40 °C. Organic acids were separated using a SunFire C18 column (5 μm particle size, 150 × 4.6 mm; Waters, Massachusetts, USA). The mobile phase was a 40 mM KH₂PO₄-H₃PO₄ buffer (pH 2.4). The flow rate was 0.8 mL/min, the column temperature was 30 °C, and the detection wavelength was 210 nm. The concentrations of soluble sugars and organic acids were calculated using standard curves and expressed as mg/g fresh weight (FW) with standard deviations.

2.6. Determination of Total phenol and total flavonoid contents

2.6.1. Determination of total phenol content

Total phenols were determined using the Folin-Ciocalteu method as described by Tao et al. (2022). Briefly, 1 mL Folin-Ciocalteu reagent was added to 0.3 mL of the sample, then 5 mL of the Na₂CO₃ solution (5 %, w/v) was added and reacted in the dark for 5 min. The mixture was made up to 25 mL with distilled water and kept in the dark for 60 min. Absorbance was measured at 765 nm using a spectrophotometer (DU730, Beckman, Germany), and values were expressed as gallic acid equivalents (GAE) based on a gallic acid standard curve.

2.6.2. Determination of total flavonoid content

Total flavonoids were determined using the Al(NO₃)₃ colorimetric method (Tao et al., 2022). Briefly, 0.2 mL of NaNO₂ (5 %, w/v) and 0.7 mL distilled water were added to 0.5 mL of the samples, and the mixture was incubated for 6 min. Then, 0.2 mL of Al(NO3)₃ (10 %, w/v) was added, vortexed, and incubated for another 6 min. Finally, 2 mL of NaOH (1 M) and 1.4 mL of distilled water were added to the mixture to raise the volume to 5 mL. The mixture was then left in the dark for 15 min. Aabsorbance was measured at 500 nm using a spectrophotometer (DU730, Beckman, Germany), and values were expressed as rutin equivalents (RE) based on a gallic acid standard curve.

2.7. Determination of antioxidant capacities

2.7.1. DPPH radical scavenging activity assay

DPPH radical scavenging activity was determined based on the method reported by Barreca et al. (2011). In brief, 200 μL of the sample was added to 2.5 mL of the DPPH solution (75 μmol/L). After 30 min of incubation in the dark at 25 °C, the absorbance of the mixture was measured using a UV–VIS spectrophotometer at 517 nm. The standard sample was a Trolox solution dissolved in 80 % methanol, and the antioxidant capacity was expressed as Trolox equivalents (TE).

2.7.2. ABTS radical scavenging activity assay

ABTS radical scavenging activity was determined as the previously reported method (Tao et al., 2022).The ABTS radical was prepared by mixing 5 mL ABTS (7 mol/L) stock solution with 88 μL potassium persulfate (140 mmol/L) and keeping the mixture for 12–16 h until the absorbance was stable. The ABTS· + solution was dissolved with ethanol to an absorbance of 0.70 ± 0.02 at 734 nm. Then, 40 μL of the sample was added to 3.9 mL of the ABTS· + solution and kept in the dark for 10 min at 25 °C. The absorbance was measured using a UV-VIS spectrophotometer at 734 nm. The standard sample was a Trolox solution dissolved in 80 % methanol, and the antioxidant capacity was expressed as Trolox equivalents (TE).

2.7.3. Reducing power activity assay

The ferric reducing antioxidant power (FRAP) was determined based on the previously described method (Tao et al., 2022). Briefly, 200 mL of acetate buffer solution (0.3 mol/L), 20 mL of TPTZ (10 mmol/L), and 20 mL of FeCl₃ (20 mmol/L) were mixed to obtain the reagents. Afterward, 200 μL of the sample was mixed with 3.8 mL of the reagents and allowed to react for 30 min at 25 °C. Absorbance was measured using a UV–VIS spectrophotometer at 593 nm. The standard sample was a Trolox solution dissolved in 80 % methanol, and the antioxidant capacity was expressed as Trolox equivalents (TE).

The antioxidant potency composite index (APCI) is used to evaluate overall antioxidant capacity. APCI = (value measured by this method / maximum value measured by this method) × 100 %. The final APCI used is the average of the APCI scores of the three antioxidant methods.

2.7.4. Cellular antioxidant assay

The cellular antioxidant assay was performed based on a previously described method (Tan et al., 2025). Briefly, samples obtained from different fermentation time points were dissolved in DMSO and subsequently diluted with complete medium to a fixed concentration. RAW264.7 cells were seeded at a density of 1 × 10⁴ cells per well in 96-well plates. Cytotoxicity was evaluated using the CCK-8 assay, following the manufacturer's instructions, to determine the optimal dosage. An oxidative stress model was then established by treating the cells with 0.004 % H₂O₂ for 2 h. After treatment, the medium was removed, and the cells were washed twice with PBS. The model group received fresh medium, while the treatment groups were exposed to samples from different fermentation stages. The normal control group was treated with complete medium only. After 24 h of incubation, the cells were harvested to assess oxidative stress markers, including malondialdehyde (MDA), glutathione (GSH), superoxide dismutase (SOD), catalase (CAT), and glutathione reductase (GR). All assays were conducted in strict accordance with the respective manufacturer's protocols.

2.8. Untargeted metabolomics analysis

High-resolution metabolite detection was performed using LC-MS/MS for comprehensive metabolomic profiling. The liquid chromatography conditions were as follows: a Thermo Vanquish UHPLC system was used, with an ACQUITY UPLC® HSS T3 column (2.1 × 100 mm, 1.8 μm; Waters, Milford, MA, USA). The column flow rate was set to 0.3 mL/min, the column temperature was maintained at 40 °C, and the injection volume was 2 μL per sample. In positive ion mode, mobile phase A2 consisted of 0.1 % formic acid in water, and mobile phase B2 consisted of 0.1 % formic acid in acetonitrile. The gradient elution program was as follows: 0–1 min, 8 % B2; 1–8 min, 8–98 % B2; 8–10 min, 98 % B2; 10–10.1 min, 98–8 % B2; 10.1–12 min, 8 % B2. In negative ion mode, mobile phase A3 consisted of 5 mM ammonium formate in water, and mobile phase B3 consisted of acetonitrile. The gradient elution program was as follows: 0–1 min, 8 % B3; 1–8 min, 8–98 % B3; 8–10 min, 98 % B3; 10–10.1 min, 98–8 % B3; 10.1–11.2 min, 8 % B3.

Mass spectrometric detection of metabolites was performed on Orbitrap Exploris 120 (Thermo Fisher Scientific, USA) with ESI ion source. Simultaneous MS1 and MS/MS (Full MS-ddMS2 mode, data-dependent MS/MS) acquisition was used. The parameters were as follows: sheath gas pressure, 40 arb; aux gas flow, 10 arb; spray voltage, 3.50 kV and − 2.50 kV for ESI(+) and ESI(−), respectively; capillary temperature, 325 °C; MS1 range, m/z 100–1000; MS1 resolving power, 60000 FWHM; number of data dependant scans per cycle, 4; MS/MS resolving power, 15000 FWHM; normalized collision energy, 30 %; dynamic exclusion time, automatic (Want et al., 2013).

To detect and quantify metabolites, we used primary mass spectrometry (MS1) and secondary mass spectrometry (MS2) data. Metabolite annotation was performed using our self-built BioDeep database (BioDeepDB) and public databases (HMDB, mzVault, METLIN, etc.) to ensure high reliability of metabolite characterization. Metabolomic data analysis was conducted using the R language. Based on metabolite abundance data, the OPLS-DA model was applied to calculate variable importance in projection (VIP) scores. KEGG pathway enrichment analysis was performed using MetaboAnalyst.

2.9. Statistical analysis

Data were processed using Microsoft Excel 2021, visualized using GraphPad Prism 8.0, and the use of analysis of variance with Duncan's post-hoc test using IBM SPSS Statistics V23.0 software. Principal Component Analysis (PCA) was performed using the method described by Braşoveanu et al. (2023). Statistical significance was defined as p < 0.05. All experiments were conducted in triplicate, and results are presented as mean ± standard deviation (SD).

3. Results and discussion

3.1. Physicochemical properties of fermented plum pulp

3.1.1. Viable bacterial counts and pH changes

Viable bacterial counts are a key indicator of fermentation efficiency, as they directly reflect the growth and metabolic activity of microorganisms in the fermentation system. Plum pulp, rich in fermentable sugars such as glucose and fructose, serves as an abundant carbon source for the synergistic growth of L. plantarum, L. paracasei, and B. infantis (Xiao et al., 2024). As shown in Fig. 1A, the mixed probiotic strains exhibited favorable growth characteristics in the plum pulp. During fermentation, viable bacterial counts initially increased, reaching 7.43 ± 0.52 log CFU/mL on the first day, peaking at 8.22 ± 0.35 log CFU/mL on the third day, and then slightly declining to 7.15 ± 0.24 log CFU/mL on the fifth day. Fruit pulp serves as an ideal fermentation substrate, offering abundant reducing sugars and other nutrients that support the rapid proliferation of probiotic strains. Previous studies have exhibited that certain juices, such as goji berry and passion fruit juices, provide excellent growth media for probiotics (Duan et al., 2023). As fermentation progresses, these nutrients are gradually depleted, while the accumulation of organic acids, such as lactic acid, leads to a decrease in pH (Duan et al., 2023). This increasingly acidic environment imposes stress on bacterial survival, resulting in a decline in viable cell counts during the later stages of fermentation. Importantly, viable counts remained above 7 log CFU/mL throughout the fermentation process, indicating that plum pulp provided a conducive environment for the growth of the selected mixed strains. This observation aligns with the dominant growth of mixed probiotic strains reported by Lan et al. (2024) in kiwi juice fermentation.

Fig. 1.

Physicochemical properties of fermented plum pulp. (A) Viable bacterial counts, (B) pH, (C) Total sugar, (D) Total acid, (E) Total phenols content, (F) Total flavonoid content.

Changes in pH are a crucial parameter for assessing fermentation activity and the acid-production capacity of probiotics (Fig. 1B). Throughout the fermentation process, pH exhibited a continuous decline, decreasing from an initial value of 3.25 ± 0.02 to 3.07 ± 0.01 on the fifth day, representing an overall reduction of 5.5 % (p < 0.05). During the early stage of fermentation (day 1), pH decline was relatively slow, dropping from 3.25 to 3.20. In contrast, a more pronounced decrease was observed during the mid-to-late stage of fermentation (day 5). This pH reduction was driven by the conversion of sugars in the plum pulp into organic acids such as lactic acid and acetic acid through glycolysis, accompanied by the generation of amino acids from protein hydrolysis, which further contributed to the acidification (Alrosan et al., 2023). Acidification of the fermentation environment not only inhibited the growth of spoilage microorganisms but also created favorable conditions for the production of metabolites, such as organic acids and bioactive compounds (Muhialdin et al., 2020).

3.1.2. Changes in sugars and organic acids

Sugars serve as the primary carbon source during probiotic fermentation, and their consumption patterns directly reflect microbial metabolic activity and fermentation progression. As shown in Fig. 1C, total sugar content exhibited a dynamic trend throughout the mixed-strain fermentation of plum pulp. During the early stage of fermentation (day 0–1), total sugar content slightly increased from 72.17 ± 1.59 mg/mL to 78.47 ± 1.23 mg/mL, likely due to the metabolic activity of mixed strains during their early growth phase (An et al., 2024). As fermentation advanced, total sugar content significantly decreased, dropping to 65.28 ± 0.42 mg/mL on day 3 and further to 54.78 ± 0.88 mg/mL on day 5, representing a 24 % reduction compared to the initial value. This decline indicates efficient sugar utilization and biotransformation by the mixed strains (p < 0.05).

Monosaccharide composition analysis elucidated the intricate dynamics of sugar metabolism during fermentation (Table 1). Sucrose, the predominant sugar, steadily declined from an initial concentration of 41.33 ± 0.91 mg/mL to 29.87 ± 0.46 mg/mL by day 5, representing a 27.7 % reduction. Glucose levels rose to 30.37 ± 0.60 mg/mL on day 1 before decreasing to 19.03 ± 0.35 mg/mL, indicating that sucrose was hydrolyzed into glucose and fructose, which were preferentially consumed by the mixed microbial strains. Fructose levels peaked at 8.44 ± 0.18 mg/mL on day 1 before returning to near-baseline values, suggesting a delayed consumption compared to glucose. This sequential sugar metabolism highlights the synergistic interactions and distinct metabolic preferences of L. plantarum, L. paracasei, and B. infantis (Meng et al., 2022).

Table 1.

Sugars and organic acids content of different fermentation times.

	day 0	day 1	day 3	day 5
Sucrose	41.33 ± 0.91a	39.67 ± 0.81a	32.50 ± 0.36b	29.87 ± 0.46b
Glucose	24.80 ± 1.78b	30.37 ± 0.60a	26 ± 0.60ab	19.03 ± 0.35b
Fructose	6.03 ± 0.5c	8.44 ± 0.18a	6.78 ± 0.21b	5.88 ± 0.09c
Malic acid	3.28 ± 0.15a	2.74 ± 0.13b	2.19 ± 0.15c	1.66 ± 0.02c
Tartaric acid	1.85 ± 0.13c	3.14 ± 0.4ab	3.59 ± 0.18a	1.96 ± 0.1bc
Citric acid	0.46 ± 0.02a	0.45 ± 0.04a	0.35 ± 0.03b	0.33 ± 0.03b

Note: Values followed by different superscripts (a–c) are significantly different (p < 0.05).

The dynamics of organic acid profiles during fermentation were closely associated with sugar metabolism. As shown in Fig. 1D, total acid content increased initially from 5.59 ± 0.18 mg/mL to a peak of 6.33 ± 0.35 mg/mL on day 3 before declining to 3.95 ± 0.12 mg/mL by day 5. This pattern suggests that the mid-fermentation phase was pivotal in organic acid accumulation. Analysis of organic acid composition elucidated intricate metabolic transformations during fermentation (Table 1). Malic acid content steadily decreased from 3.28 ± 0.15 mg/mL to 1.66 ± 0.02 mg/mL, representing a 49.4 % reduction. This decline results from the malolactic fermentation pathway, where malic acid is decarboxylated to lactic acid, increasing total acidity and enhancing flavor complexity in the fermented pulp (Wu et al., 2020). Tartaric acid content significantly increased during the early fermentation stage, rising from 1.85 ± 0.13 mg/mL to 3.59 ± 0.18 mg/mL on day 3, before decreasing to 1.96 ± 0.10 mg/mL. This change may be related to the metabolic pathway where tartaric acid is converted to oxaloacetate and subsequently utilized in TCA cycle (Duan et al., 2023). Citric acid content slightly decreased in the late fermentation stage, from 0.46 ± 0.02 mg/mL to 0.33 ± 0.03 mg/mL, indicating its utilization in TCA cycle metabolism.

These dynamics of sugar and organic acid profiles indicate that mixed-strain fermentation markedly altered the biochemical profile of plum pulp. Efficient sugar utilization and organic acid transformation not only improved fermentation efficiency but also laid the biochemical foundation for subsequent metabolite production and bioactivity enhancement. The unique organic acid profile formed during fermentation imparted smoother taste and richer flavor complexity to the plum pulp, consistent with the quality improvement effects observed in fermented fruits and vegetables by lactic acid bacteria and other probiotics (Gao et al., 2019).

3.1.3. Changes in TPC and TFC

Phenolic compounds and flavonoids are vital bioactive components in plant-based foods, offering significant antioxidant, anti-inflammatory, and anti-aging benefits that contribute to human health. Probiotic fermentation, through intricate biochemical transformations, can markedly influence the content and bioavailability of these functional compounds in plum pulp. As shown in Fig. 1E, the fermentation process significantly affected the dynamic changes in TPC of plum pulp. Initially, the TPC was 6.20 ± 0.14 mg GAE/g FW, which increased to 6.74 ± 0.09 mg GAE/g FW on the first day of fermentation and peaked at 7.28 ± 0.22 mg GAE/g FW on the third day, representing a 17.5 % increase compared to the initial value (p < 0.05). However, by the fifth day, TPC had decreased to 5.37 ± 0.26 mg GAE/g FW, falling below the initial level. Correlation analysis between specific metabolites and TPC content showed that 4-hydroxybenzoic acid, fustin, and chlorogenic acid were markedly positively correlated with TPC content (Fig. S1). These substances dynamics may have affected the total change in TPC. This change pattern of TPC suggests that hydrolytic enzymes such as esterases, glycosidases, and polyphenol oxidases produced by mixed strains during the early and mid-fermentation stages released free phenolic compounds from their bound or polymerized forms, thereby enhancing their bioavailability (Parada et al., 2023). The decline in TPC during late fermentation may result from the oxidation and polymerization of phenolic compounds. In addition, probiotics can metabolize and transform phenolic compounds during fermentation, a process often mediated by intracellular enzymes (Duan et al., 2025). For instance, phenolic acid decarboxylase can convert phenolic acids into other bioactive metabolites (Duan et al., 2025). Moreover, pH changes during fermentation can influence the stability of phenolic compounds (Zhang et al., 2023). In our study, the pH of plum pulp decreased to approximately 3.0 after 5 days of fermentation (Fig. 1B). Such pH shifts may alter the structure, solubility, and stability of phenolic compounds in solution.

In contrast, TFC displayed a more stable upward trend, as shown in Fig. 1F. TFC steadily increased from an initial value of 7.53 ± 0.22 mg RE/g FW to 8.53 ± 0.15 mg RE/g FW on the first day, peaking at 9.89 ± 0.16 mg RE/g FW on the third day, which represents a 31.4 % increase compared to the initial value. Although TFC experienced a slight decline to 7.66 ± 0.12 mg RE/g FW on the fifth day, it remained close to the initial level. Correlation analysis between specific metabolites and TFC content showed that ent-Gallocatechin 3-gallate, Epicatechin, Afzelechin, Genistein, and Procyanidin B2 were markedly positively correlated with TFC content (Fig. S1). Changes in these substances may have affected the total change in TFC. Notably, compared to the fluctuating changes observed in TPC, TFC exhibited greater stability. This phenomenon may be attributed to the inherent chemical stability of flavonoid compounds and their antioxidant properties, which provide protection against degradation in the fermentation environment (Hasani et al., 2023). In addition, the complex metabolic activities of probiotics during fermentation significantly influence the content and composition of flavonoid compounds. L.acidophilus produces various enzymes, including β-glucosidase and glycosidase, which hydrolyze flavonoid glycosides, thereby releasing the corresponding aglycones (Huang et al., 2024). Moreover, microbial metabolic preferences for specific flavonoid substrates also contribute to changes in flavonoid profiles. For example, L. pentosus has been shown to convert rutin into quercetin, leading to a reduction in rutin content (Wang et al., 2024).

3.2. Data-driven metabolomic analysis

Non-targeted metabolomics analysis detected 1419 metabolites in positive ion mode and 694 metabolites in negative ion mode, resulting in a total of 2113 metabolites. The data were normalized before PCA to ensure comparability across samples. PCA visually illustrated metabolic differences among samples collected at different fermentation time points (Fig. 2A, B). In positive ion mode, first principal component (PC1) and second principal component (PC2) explained 37.8 % and 13.4 % of total variance, respectively, with a cumulative variance contribution of 51.2 % (Fig. 2A). In negative ion mode, PC1 and PC2 contributed 38.4 % and 13.1 % of total variance, with a cumulative variance explanation of 51.5 % (Fig. 2B). To achieve a variance reduction of at least 80 %, it is essential to incorporate additional variables into the analysis (Braşoveanu et al., 2024). The PCA score plots revealed distinct temporal separations along the PC1 axis among samples at different fermentation time points. Day 0 samples formed a separate group from the fermented samples (day 1, day 3, day 5). Each time point contains three biological replicates, and the three biological replicates within the same group are closely clustered, indicating good reproducibility of the experiment. PC1 predominantly represented changes in the metabolic profile driven by fermentation progress, whereas PC2 highlighted nuanced variations among different fermentation stages. These results exhibited that mixed-strain fermentation significantly altered the overall metabolic characteristics of plum pulp, with evident metabolomic differences observed across fermentation time points.

Fig. 2.

Comparison of metabolites before and after fermentation. (A) PCA analysis in positive ion mode, (B) PCA analysis in negative ion mode, (C) heatmap, (D) pie chart.

The functional classification of metabolites further highlighted the dynamic alterations across multiple compound classes during the fermentation process (Fig. 2D). Nine major categories were analyzed, including carbohydrates, organic acids, flavonoids, coumarins and their derivatives, benzoic acids and their derivatives, flavanols, isoflavones, flavonols, and phenolic acids. Carbohydrates consistently accounted for the largest proportion, but their relative content decreased from 57.71 % at day 0 to 49.66 % at day 1, before recovering to 52.26 % and 54.98 % at day 3 and day 5, respectively. This trend corresponded to time-dependent dynamics of sugar consumption and transformation. Organic acids represented the second largest category, with their proportion remaining stable at around 29 % on day 1 and day 3 but significantly decreasing to 20.99 % on day 5, reflecting further metabolic transformation of organic acids during the late fermentation stage.

Flavonoid-related compounds (including flavonoids, flavonoid glycosides, flavanols, flavonols, and isoflavones) exhibited complex variation patterns during fermentation. Their overall proportion fluctuated, decreasing from 9.95 % at day 0 to 8.90 % on day 1 and further to 8.86 % on day 3, before increasing to 9.54 % on day 5. Coumarins and their derivatives exhibited a consistent upward trend, increasing from 3.04 % at day 0 to 6.51 % at day 5, indicating that fermentation promoted the formation of these potentially bioactive compounds. Although phenolic acids accounted for a relatively small proportion, their levels exhibited a marked upward trend, increasing from 0.27 % at day 0 to 2.18 % at day 5—an approximately eight-fold increase. This dramatic rise reflects significant biotransformation and release of phenolic compounds during fermentation, which may enhance the bioactive properties of the final product.

3.3. Differential metabolite analysis

3.3.1. Identification and evaluation of differential metabolites

To comprehensively investigate the dynamic metabolic changes in plum pulp during mixed-strain fermentation, orthogonal partial least squares-discriminant analysis (OPLS-DA) and multivariate statistical methods were employed to systematically identify and evaluate differential metabolites. Using stringent screening criteria (VIP ≥ 1 and p ≤ 0.05) combined with KEGG pathway enrichment analysis, the study elucidated the alterations in key metabolic pathways throughout fermentation. Model validation exhibited high performance for both positive and negative ion mode comparisons, with all models achieving R²Y values close to 1 and Q² values greater than 0.9. These results indicate excellent model fit and strong predictive capability, provide a solid foundation for reliable differential metabolite screening (Supplementary Table S1).

As illustrated in Fig. 3A, volcano plots clearly display the distribution patterns of differential metabolites among samples collected at different fermentation time points. The day0 vs day1 group detected 908 significant differential metabolites, with 614 significantly upregulated and 294 significantly downregulated. The day0 vs day3 group detected 1074 differential metabolites, including 645 upregulated and 429 downregulated ones. The day0 vs day5 group detected 1122 differential metabolites, of which 711 were upregulated and 411 were downregulated (Supplementary Table S2-S4). With the progression of fermentation, the total number of differential metabolites increased steadily, with the proportion of upregulated metabolites consistently exceeding 60 %. This trend indicates that fermentation primarily facilitated the generation and accumulation of new metabolites.

Fig. 3.

Differential metabolites during fermentation. (A) volcano plot, (B) Venn diagram, (C) clustering diagram.

Venn diagram analysis revealed that 820 differential metabolites were shared among the three comparison groups (Fig. 3B). These core differential metabolites represent the most stable and critical metabolic changes during fermentation. Based on the expression patterns of these shared differential metabolites, clustering analysis categorized them into four distinct expression trend groups (Fig. 3C): the first group comprised 399 metabolites, characterized by high expression at day0 followed by sustained low levels; the second group included 118 metabolites, exhibiting high expression at day1; the third group contained 152 metabolites, showing a fluctuating expression pattern; and the fourth group consisted of 151 metabolites, displaying a continuous increase in content from day0 to day5 (Supplementary Table S5). These distinct expression trends reflect the dynamic metabolic processes occurring during fermentation and provide insights into the role of mixed-strain fermentation in reshaping the metabolic landscape of plum pulp.

3.3.2. KEGG pathway enrichment analysis

To further elucidate the biological functions of the four expression clusters, KEGG pathway enrichment analysis was performed for each cluster (Fig. 4). Cluster 1 was primarily enriched in pathways such as 2-oxocarboxylic acid metabolism (p = 0.012), glucosinolate biosynthesis (p = 0.012), and valine, leucine, and isoleucine biosynthesis (p = 0.021). These metabolic pathways are involved in the degradation of branched-chain amino acids and in the biosynthesis of secondary metabolites, which suggests that microorganisms make extensive use of these compounds as nutrients during the early stages of fermentation. The enrichment of the pyruvate metabolism pathway (p = 0.045) demonstrates the consumption of key compounds in glycolysis, consistent with the homolactic fermentation characteristics of lactic acid bacteria (Liu, 2003).

Fig. 4.

KEGG enrichment pathways of differential metabolites in different comparison groups. (A) Cluster1, (B) Cluster2, (C) Cluster3, (D) Cluster4.

Cluster 2 was significantly enriched in plant hormone signal transduction (p = 0.024) and isoquinoline alkaloid biosynthesis (p = 0.081) pathways. Plant hormone-related compounds (e.g., jasmonic acid) rose briefly. This transient spike likely signals stress responses early in fermentation. Likewise, a rise in 4-hydroxyphenylpyruvate suggests that secondary metabolic pathways are active. Enrichment of galactose metabolism (p = 0.087) and alanine–aspartate–glutamate metabolism (p = 0.087) reveals flexible use of carbohydrates and amino acids.

Cluster 3 metabolites were mainly enriched in pathways such as ABC transporters (p = 0.0005), porphyrin metabolism (p = 0.001), and D-amino acid metabolism (p = 0.002). The activation of the ABC transporter pathway suggests elevated nutrient uptake and metabolite efflux during the mid-stage of fermentation. The presence of amino acids (e.g., threonine and arginine) and nucleosides (e.g., adenosine) highlights the importance of protein and nucleic acid metabolism at this stage. Enrichment of the porphyrin metabolism pathway is associated with the biosynthesis of essential cofactors such as vitamin B12. Furthermore, the upregulation of arginine biosynthesis (p = 0.007) and arginine–proline metabolism (p = 0.012) likely contributes to enhanced nitrogen metabolism and improved microbial stress tolerance.

Cluster 4 was mainly enriched in flavonoid biosynthesis (p = 0.008) and metabolism (p = 0.015) pathways. Flavonoids such as quercetin and leucocyanidin continued to accumulate, showing that late fermentation strongly promotes phenolic conversion and the formation of new flavonoid derivatives. The enrichment of carbon metabolism (p = 0.025) and the pentose phosphate pathway (p = 0.037) reflects a reshaped sugar-metabolism network. Additionally, the participation of arginine biosynthesis (p = 0.048) and nicotinate and nicotinamide metabolism (p = 0.055) indicates ongoing amino-acid metabolism and vitamin synthesis during the late fermentation phase.

KEGG enrichment analysis reveals that mixed-strain fermentation drives multi-level reconstruction of biochemical networks, spanning from primary to secondary metabolism. During the early fermentation phase, available nutrients are rapidly consumed and core metabolic pathways are activated. The mid-fermentation phase is characterized by intensified transport activity and cofactor biosynthesis, whereas the late phase features a pronounced accumulation of flavonoids and other functional metabolites. These time-resolved shifts in metabolic pathways furnish critical metabolomic evidence supporting the enhanced antioxidant activity and improved functional properties of fermented plum pulp. The significant enrichment of flavonoid biosynthesis pathways aligns closely with the observed trends in TPC and TFC, confirming that fermentation activates specific biosynthetic pathways to promote the formation and accumulation of bioactive phenolic compounds.

3.4. Key metabolite analysis

3.4.1. Sugars

Our analysis indicates that mixed-strain fermentation significantly alters the sugar metabolism profile of plum flesh (Fig. 5A). Notably, the levels of trehalose and raffinose increased throughout the fermentation process. Trehalose is well known for its stress-protective properties; when combined with probiotics, it enhances the growth, antioxidant capacity, and immune function of both the host and probiotic strains, with particularly notable effects on Bifidobacterium longum and L. plantarum (Peng et al., 2024). Additionally, trehalose functions as an osmoprotectant, contributing to cellular and protein stability under stress conditions (Kuczyńska-Wiśnik et al., 2024). During the later stages of fermentation, as acidity increases, microorganisms may synthesize trehalose in large quantities as a self-protective mechanism, resulting in its elevated relative abundance.

Fig. 5.

Key metabolites during the fermentation process. (A) Sugars, (B) Organic acids, (C) Phenolic acids, (D) Flavonoids, (E) Polyphenols.

In contrast, Our analysis revealed that mixed-strain fermentation induced significant modifications in the sugar metabolic profile of plum pulp (Fig. 5A). During fermentation, lactic acid bacteria (LAB) metabolize available sugars in the fruit matrix as their primary carbon source to support growth and acid production (Wang et al., 2024). For instance, lactose is hydrolyzed into glucose and galactose, with glucose subsequently entering the glycolytic pathway and being converted to pyruvate, which is then reduced to lactic acid (Feng et al., 2025). This suggests that disaccharides such as sucrose may undergo similar hydrolysis, with the resulting monosaccharides rapidly utilized by the fermenting microorganisms. In our study, sucrose content decreased by more than 99 %, indicating its preferential utilization as a carbon source (Fig. 5A).

3.4.2. Organic acids

Organic acids are the primary contributors to the acidic taste of fermented products and play a crucial role in shaping their flavor profiles (Ruiz Rodríguez et al., 2021). Our analysis revealed that mixed-strain probiotic fermentation significantly altered the content of organic acids in plum pulp (Fig. 5B). Specifically, five organic acids—D-tartaric acid, 3-(3-hydroxyphenyl)propionic acid, succinic acid, 2-hydroxypropionic acid (lactic acid), and citraconic acid—were significantly upregulated, whereas nine organic acids—pyroglutamic acid, malic acid, fumaric acid, α-ketoisovaleric acid, and quinic acid—were notably downregulated (Fig. 5B). These results highlight the complex mechanisms underlying organic acid transformation during mixed-strain fermentation of plum pulp.

The significant increase in succinic acid content is likely due to the activation of TCA cycle, while the observed decrease in malic acid and fumaric acid content suggests that these intermediates were actively utilized for succinic acid synthesis (Lan et al., 2024). The rise in 2-hydroxypropionic acid (lactic acid) content confirms the characteristic metabolic activity of probiotic fermentation. This increase was accompanied by a significant reduction in polyphenolic compounds such as quinic acid, suggesting that the microorganisms may metabolize plant-derived phenolics as substrates for lactic acid production (Filannino et al., 2015). Additionally, the decline in valeric acid content and the corresponding increase in 3-(3-hydroxyphenyl)propionic acid suggest a potential metabolic link, wherein probiotics transform aromatic compounds via side-chain modification and hydroxylation pathways. Notably, the accumulation of succinic acid and lactic acid not only contributes to product preservation and flavor enhancement but may also offer health-promoting benefits, such as improved gut health and reduced inflammation (Wang et al., 2024).

3.4.3. Phenolic compounds

Phenolic compounds play a crucial role in plant defense and contribute significantly to the antioxidant capacity, bioactivity, and sensory qualities of fermented foods (Feng et al., 2025). In this study, probiotic fermentation markedly altered the phenolic compound profile of plum pulp (Fig. 5C). Specifically, the contents of chlorogenic acid, 4-hydroxybenzoic acid, caffeic acid, and dihydroferulic acid were significantly elevated. These findings are consistent with previous studies; for example, fermentation of wolfberry juice by L. plantarum resulted in increased levels of chlorogenic acid and dihydroferulic acid (Liu et al., 2024). Similarly, L. plantarum-fermented lychee pulp showed a significant increase in 4-hydroxybenzoic acid (Huang et al., 2024). Chlorogenic acid, a hydroxycinnamic acid derivative, is known for its potent antioxidant and anti-inflammatory activities (Xia et al., 2025). It exerts its antioxidant effects by donating electrons and hydrogen atoms to neutralize free radicals, and by chelating metal ions to inhibit peroxide decomposition and free radical generation (Xia et al., 2025). Therefore, the increased levels of chlorogenic acid and dihydroferulic acid observed during fermentation are likely to enhance the health-promoting properties of the final product (Zhang et al., 2018).

Conversely, the content of 4-hydroxycinnamic acid was significantly reduced during fermentation (Fig. 5C). This decrease may be attributed to microbial enzymatic activity. Previous studies have shown that phenylacetic acid decarboxylase can catalyze the decarboxylation of 4-hydroxycinnamic acid, converting it into 4-vinylphenol during lactic acid fermentation (Vanbeneden et al., 2008). Such microbially mediated metabolic transformations likely contribute to the observed reduction in 4-hydroxycinnamic acid content.

3.4.4. Polyphenols and flavonoid compounds

Flavonoids and polyphenols, as the most biologically active secondary metabolites in plum fruits, play a key role in determining the sensory attributes and health benefits of the final product (Ayub et al., 2023). Fermentation resulted in significant alterations to the flavonoid and polyphenol profiles in plum pulp (Fig. 5D&E). Data analysis revealed a substantial increase in the levels of polyphenolic compounds, including catechin, epicatechin, genistein, daidzein, and procyanidin B2. Some flavonoid compounds, particularly flavonoids and isoflavonoids (e.g., genistein and daidzein), are commonly present in plants in glycosylated forms (Tang et al., 2025). LAB can produce hydrolytic enzymes such as β-glucosidase, which cleave glycosidic bonds in phenolic glycosides, thereby releasing more bioactive aglycones (Tang et al., 2025). This enzymatic hydrolysis enhances the levels of free genistein and daidzein, which are more readily absorbed and metabolized by the human body compared to their glycoside counterparts.

In addition, Polyphenols in plants are not only localized in the cell sap but also exist in bound forms, associated with cell wall components such as pectin, polysaccharides, and proteins (Huang et al., 2024). During fermentation, microbial activity and the production of metabolic byproducts—including organic acids and cell wall-degrading enzymes—can modify the plant cell wall structure, weakening the chemical bonds between phenolics and cell wall matrices (Duan et al., 2025; Huang et al., 2024). This facilitates the release of bound phenolics into the fruit pulp, thereby increasing the total extractable and detectable phenolic content.

Additionally, an increase in the content of quercetin was observed (Fig. 5D). Microbial hydrolysis of quercetin-7,4’-O-diglucoside leads to the release of free quercetin, resulting in elevated quercetin levels. This finding aligns with the results reported by Ferreira-Lazarte et al. (2021), who exhibited that L.plantarum facilitates the deglycosylation of naringin and rutin through glycosidase activity, producing isoquercetin and thereby enhancing the flavor profile of the beverage. Notably, probiotic fermentation not only increases the total polyphenol content but also optimizes the composition by enriching the proportion of free, bioavailable forms. This selective enrichment of biologically active polyphenols suggests that fermented plum products may exhibit enhanced health-promoting properties compared to raw plum pulp, particularly in terms of antioxidant capacity, anti-inflammatory activity, and cardiovascular protection.

3.5. Evaluation of antioxidant activity

To comprehensively evaluate the impact of mixed-strain fermentation on the antioxidant performance of plum pulp, this study systematically assessed the antioxidant activity of samples at different fermentation time points using three methods: ABTS radical scavenging, DPPH radical scavenging, and FRAP reducing power. The overall antioxidant activity was further evaluated using the Antioxidant Performance Comprehensive Index (APCI). All three assays exhibited distinct time-dependent changes during fermentation (Table 2).

Table 2.

In vitro antioxidant activity of the extracts of different fermentation times.

Samples	Methods of antioxidant activity evaluation (μmol TE /g FW)			APCI	Rank
	ABTS	DPPH	FRAP
day 0	20.17 ± 1.02c	13.48 ± 0.04c	37.67 ± 1.11d	78.80	4
day 1	21.68 ± 0.79c	14.61 ± 0.11b	42.24 ± 0.77b	86.17	2
day 3	26.94 ± 1.14a	16.48 ± 0.28a	44.21 ± 0.51a	97.77	1
day 5	24.17 ± 0.24b	13.52 ± 0.27c	40.65 ± 0.39c	85.97	3

Note: Values followed by different superscripts (a–c) are significantly different (p < 0.05).

The ABTS radical scavenging activity increased significantly from 20.17 μmol TE/g FW on day 0 to 26.94 μmol TE/g FW on day 3, representing a growth of 33.6 %. The DPPH radical scavenging activity rose from 13.48 μmol TE/g FW to 16.48 μmol TE/g FW, with a growth of 22.3 %. Similarly, FRAP reducing power increased from 37.67 μmol TE/g FW to 44.21 μmol TE/g FW, reflecting a growth of 17.4 %. Antioxidant activity reached its peak on day 3 across all three assays, followed by a slight decrease on day 5, while still maintaining levels significantly above the initial baseline.

Furthermore, an oxidative stress model was established in RAW264.7 cells using H₂O₂. Within the tested concentration range of 12.5–50 μg/mL, all samples exhibited low cytotoxicity, maintaining cell viability above 50 % (Fig. 6A). At a concentration of 50 μg/mL, cellular antioxidant activity was found to be dependent on fermentation time, displaying a trend of initial increase followed by a subsequent decline (Fig. 6B). Among all samples, the fermentation product obtained after 3 days exhibited the most pronounced antioxidant effect, as evidenced by a significant reduction in malondialdehyde (MDA) levels and enhanced activities of superoxide dismutase (SOD), catalase (CAT), glutathione reductase (GR), and glutathione (GSH) (Fig. 6B).

Fig. 6.

Evaluation of antioxidant activity of cell-based and correlation analysis between metabolites and biological activity. (A) Cytotoxicity of different fermentation time samples, (B) Antioxidant activity in cell model, (C) PCA of metabolite-profile, (D) Pearson correlation analysis between metabolites and biological activity. Note: ‘*’, ‘**’, and ‘***’ represent p < 0.05, p < 0.01, and p < 0.001, respectively.

In general, the significant enhancement of antioxidant activity can be primarily attributed to the biotransformation and release of phenolic compounds during the fermentation process. Metabolomics analysis revealed a marked increase in polyphenolic compounds such as quercetin, epicatechin, and procyanidin B2, as well as the accumulation of phenolic acids including caffeic acid and mangostin acid (Fig. 5). These compounds directly contributed to the improved free radical scavenging capacity, thereby substantially enhancing the in vitro antioxidant activity of the fermented plum pulp.

3.6. Correlation of antioxidant potential and metabolites

In order to systematically identify the metabolites most closely related to antioxidant activity, we first performed PCA on the correlation data between all detected metabolites and eight antioxidant indicators (Fig. 6C). The PCA biplot shows the relationship between metabolites (blue points) and eight antioxidant-related phenotypes (red arrows). PC1 and PC2 explain 80 % and 13.5 % of the total variance, respectively (Fig. 6C). The clear separation between MDA and other antioxidant markers along PC1 indicates the primary biological axis from oxidative damage to antioxidant defense. The top 30 metabolites most positively associated with antioxidant capacity were selected for further analysis (Fig. 6C).

To further characterize the antioxidant activity profiles of the 30 candidate metabolites, a metabolite–phenotype correlation heatmap was constructed (Fig. 6D). The results revealed that classical phenolic antioxidants displayed strong antioxidant activity, as expected. Notably, 4-hydroxybenzoic acid showed a strong positive correlation with SOD activity (r = 0.763, p < 0.05), as well as significant correlations with non-enzymatic antioxidant indicators, including ABTS, DPPH, and FRAP (r > 0.6, p < 0.05), indicating its dual role in both enzymatic and non-enzymatic antioxidant systems. Mechanistically, 4-hydroxybenzoic acid exerts antioxidant effects by directly scavenging free radicals via a hydrogen atom transfer mechanism (Farhoosh et al., 2016). Similarly, 3-(3-hydroxyphenyl)propanoic acid, a phenolic acid compound, exhibited a strong positive correlation with GR activity (r = 0.772, p < 0.05), suggesting it may support antioxidant defense by sustaining GR system function. As a known metabolic product of quercetin, 3-(3-hydroxyphenyl)propanoic acid has been reported to exhibit stronger antioxidant activity than its parent compound (Feng et al., 2022). Niacinamide also exhibited strong positive correlations with non-enzymatic antioxidant markers (ABTS, DPPH, FRAP) (r > 0.6, p < 0.05) and a significant negative correlation with MDA (r = −0.817, p < 0.05), underscoring its central role as a precursor to NAD⁺/NADP⁺ in redox metabolism (Boo, 2021). Additionally, Resolvin D1 exhibited balanced antioxidant properties, showing positive associations with multiple antioxidant markers (r > 0.6, p < 0.05), highlighting the role of omega-3 fatty acid derivatives in mitigating oxidative stress (Wang et al., 2022).

Interestingly, these metabolites were more strongly correlated with non-enzymatic antioxidant indicators (ABTS, DPPH, FRAP) (r > 0.6, p < 0.05), which may be attributed to the presence of phenolic hydroxyl groups in their structures, conferring direct free radical-scavenging capacity (Di Meo et al., 2013). Notably, the screening also identified compounds such as styrene, typically regarded as cytotoxic. The observed positive correlation between styrene and antioxidant indicators may reflect a compensatory cellular response to exogenous stress, wherein the antioxidant defense system is upregulated in response to potential toxic insults (Morris et al., 2019). Thus, the correlation likely represents a stress-response coexistence pattern rather than intrinsic antioxidant properties of styrene. This finding underscores the importance of considering the broader biological context when interpreting metabolomics correlation data.

3.7. Impact on the current food industry and future directions

3.7.1. Impact on the current food industry

This study highlights the substantial application potential of mixed-strain fermentation technology in the fruit and vegetable processing industry. Through the synergistic action of three probiotic strains, the highly perishable Wushan crispy plum was successfully transformed into a functional fermented product. This process not only extended the shelf life of the fruit but also significantly enhanced its nutritional value. A 3-day fermentation period led to marked increases in total phenolic and flavonoid contents, along with improved antioxidant activity, offering a promising technological approach for the development of high-value-added functional foods.

3.7.2. Future development directions

Future research should prioritize the large-scale optimization and industrial application of fermentation processes. First, in-depth investigations are needed to elucidate the mechanisms by which varying strain ratios and fermentation conditions (e.g., temperature, pH, oxygen concentration) influence product quality, with the goal of establishing precise control systems. Second, the application of mixed-strain fermentation technology should be extended to a broader range of fruit and vegetable raw materials to develop a diverse array of functional fermented products. Furthermore, more comprehensive in vivo efficacy studies are essential to validate the health benefits of these fermented products, thereby facilitating their integration into the functional food and nutritional supplement markets.

4. Conclusion

In conclusion, mixed-strain fermentation markedly alters the physicochemical properties, metabolic profiles, and bioactive properties of plum pulp. The synergistic interactions of Lactiplantibacillus plantarum, Lacticaseibacillus paracasei, and Bifidobacterium infantis efficiently metabolized plum sugars, generating beneficial metabolites. Fermentation reduced simple sugar levels while increasing oligosaccharides, organic acids, and total phenolic compounds. Total Phenolic Content (TPC) and Total Flavonoid Content (TFC) were markedly increased, with TPC rising by 17.5 % and TFC by 31.4 % during peak fermentation. Non-targeted metabolomics elucidated extensive metabolic changes, with over 2100 metabolites identified and notable enrichment in flavonoid biosynthesis and organic acid metabolism pathways. Key metabolites, including trehalose, succinic acid, and proanthocyanidin B2, exhibited significant increases, contributing to improved functional attributes. Antioxidant capacity was markedly enhanced, with ABTS (2,2′-azinobis-(3-ethylbenzothiazoline-6-sulfonic acid)), DPPH (2,2-diphenyl-1-picrylhydrazyl), and FRAP (Ferric Reducing Antioxidant Power) activities increasing by 33.6 %, 22.3 %, and 17.4 %, respectively. Correlation analysis revealed strong associations between specific metabolites and antioxidant activities, clarifying the molecular mechanisms driving enhanced bioactivity. Key metabolic pathways included carbohydrate metabolism, flavonoid biosynthesis, and organic acid metabolism. These findings provide detailed insight into the biotransformation of plum pulp by mixed-strain probiotic fermentation and establish a theoretical basis for developing functional fermented fruit beverages with enhanced health-promoting properties.

CRediT authorship contribution statement

Zimao Ye: Writing – original draft, Software, Methodology, Data curation. Suhui Sun: Methodology, Data curation. Hong yang Chen: Writing – review & editing, Resources, Methodology. Chen Ji: Writing – review & editing, Resources. Zhiqin Zhou: Writing – review & editing, Validation, Supervision, Funding acquisition.

Funding sources

This work was supported by Chongqing Special Program for Technological Innovation and Application Development (grant number: CSTB2022TIAD-KPX0086). Open Project of Key Laboratory for the Development and Utilization of Authentic Herbs in the Three Gorges Reservoir Area of Chongqing(No. KFKT2022007).

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.

Appendix A. Supplementary data

Figure S1 Correlation between specific phenolic and flavonoid compounds and TPC and TFC.

Table S1 The information of OPLS-DA model.

Table S2 The differential metabolites of day 0 vs day 1.

Table S3 The differential metabolites of day 0 vs day 3.

Table S4 The differential metabolites of day 0 vs day 5.

Table S5 The common differential metabolites.

Data availability

Data will be made available on request.