Improvement of organoleptic properties of Polygonatum sibiricum-rice wine by Co-fermentation of Saccharomyces cerevisiae, Lacticaseibacillus casei and Limosilactobacillus fermentum

Abstract

The growing consumer interest and expanding market for rice wine have intensified the focus on developing high-quality products with the enhanced health-promoting properties. In the present study, the organoleptic and functional properties of Polygonatum sibiricum–rice wine was enhanced by developing a novel fermentation strategy using pre-hydrolyzed P. sibiricum extract as substrate. Results showed that the sequential inoculation of S. cerevisiae followed by L. casei (S—C) increased acidity, enhanced antioxidant capacity, promoted the release and transformation of phenolic acids and free amino acids, and improved the volatile aroma profile by increasing esters and acids (e.g., γ-nonalactone and ethyl lactate) with the reduced higher alcohol contents. Multivariate and sensory analyses confirmed that S—C group was the optimal strategy since S. cerevisiae provided the alcoholic and ester precursors, while subsequent L. casei contributed to balance the acidity and liberate antioxidant phenolic compounds. The outputs of the present study demonstrated that sequential co-fermentation with S. cerevisiae and L. casei was a promising approach to developing a high-quality, functional P. sibiricum-rice wine with enhanced organoleptic and nutritional properties.

Highlights

• •A novel fermentation strategy was developed to produce P. sibiricum-rice wine.
• •Sequential inoculation of S. cerevisiae and L. casei was optimal for quality of P. sibiricum-rice wine.
• •Co-fermentation boosts the contents of phenolic acids and umami amino acids.
• •Co-fermented P. sibiricum-rice wine had an increased acidity and volatile aroma profile.
• •Unique volatiles γ-nonalactone and ethyl lactate enhanced aroma of P. sibiricum-rice wine.

1. Introduction

Polygonatum sibiricum (Siberian Solomon's seal, or Huang Jing 黄精), a plant of the genus Polygonatum (Liliaceae), typically refers to the dried roots of species like P. sibiricum, P. kingianum, and P. cyrtonema (Sun et al., 2020). It is primarily distributed in humid, northern temperate regions at altitudes of 500–3600 m (Deng et al., 2024). As a traditional food-medicine homologous substance, P. sibiricum has a long history of food and medicine and significant health care effects. The description of P. sibiricum first appeared in the Book of Famous Doctors, and later it was recorded in Compendium of Materia Medica and Dietary Materia Medica (Xu et al., 2021). It is considered a life-prolonging tonic with properties that nourish Yin and qi, lowering blood sugar and lipids (Cui et al., 2018; Li et al., 2018), anti-oxidation, anti-aging, immune regulation (Li et al., 2018; Liu et al., 2021), antibacterial and anti-inflammatory, anti-tumor, anti-atherosclerosis, protecting myocardium and promoting sleep (Cui et al., 2018; Liu et al., 2021). Therefore, it is widely used in the product development of food and medicines (Cui et al., 2018; Li et al., 2018; Liu et al., 2021).

The growing consumer interest and expanding market for rice wine have intensified the focus on developing high-quality products with enhanced health-promoting properties, where innovations in brewing technology and flavor profiles present significant opportunities (Mu et al., 2023). For example, Mu et al. applied a potential probiotic Saccharomyces cerevisiae strain BR14 to produce functional Chinese rice wine (Huangjiu) with an increased ethanol content and total acidity, and improved total volatile flavor compounds such as 1-propanol, 3-methyl-1-butanol and phenethyl alcohol, and nutrients and functionality of glutamic acid, ornithine and antiradical activity. P. sibiricum is also a potential raw material for development of nutritious low-alcohol fermented rice wine. (Zhang & Zhou, 2003) Wang using P. sibiricum revealed that the P. sibiricum fermented by co-culture of S. cerevisiae and fermentation starters (Qu) exhibits superior nutritional values with polysaccharides, saponins, and flavonoids compared to those of traditionally processed samples (Wang, 2020).

Production of traditional Chinese rice wine typically employs solid-state or semi-solid-state fermentation, such as the tanfan (spread-cooling) or weifan (fed-batch) processes, which rely on natural microbial communities and involve minimal raw material pretreatment (Lin, 2014). The raw or simply processed P. sibiricum and glutinous rice are mixed and fermented using traditional fermentation starters (Qu), which consist of complex, undefined microbial communities. This approach often results in limited and inconsistent release of bioactive compounds from the raw materials, uncontrolled acidification, and variable flavor profiles (Lin, 2014; Wang et al., 2020). However, the raw or simply processed P. sibiricum without prior hydrolysis pretreatment usually leads to the limited release of bioactive compounds such as polysaccharide and saponin inconsistent flavor profiles, and occasionally excessive acidity (Wang et al., 2020). To overcome these limitations, the present study aimed to: (1) introducing a liquid-state fermentation method coupled with enzymatic pretreatment of P. sibiricum to produce Polygonatum sibiricum-rice wine, (2) developing a co-fermentation strategy using Saccharomyces cerevisiae along with lactic acid bacteria (Lacticaseibacillus casei and Limosilactobacillus fermentum), and (3) comparing the sensory and functional properties of P. sibiricum-rice wine including in vitro antioxidant capacity, targeted bioactive compounds (phenolic acids, amino acids), volatile profiling (via E-nose and GC–MS), and sensory analysis, with the different inoculation methods of L. casei, Limosilactobacillus fermentum, and S. cerevisiae. The outputs of the present study would provide a scientific and reproducible framework for developing premium functional rice wines with the traditional sensory attributes and enhanced nutritional value for addressing a key market and technological need.

2. Materials and methods

2.1. Strains and culture conditions

S. cerevisiae was provided by Danyang Yihe Food Co., Ltd. and kept at the China Center for Type Culture Collection (CCTCC) under accession number CCTCC NO: M 2022914. L. fermentum strain JDYJZ-3 has been deposited under accession number M 2026138. L. casei strain YH80JZ-3 is currently held in the laboratory. The lactic acid bacteria (LAB) strains L. casei YH80JZ-3 and L. fermentum JDYJZ-3 were isolated from natural fruits, and identified based on the 16S rDNA sequence. The 16S rRNA gene sequences of L. casei YH80JZ-3 and L. fermentum JDYJZ-3 fruits andeposited in the GenBank database with the accession numbers of PP795338 and PP795337, respectively. 2-Octanol (GC ≥ 99.5%) was purchased from Shanghai Macklin Biochemical Technology Co., Ltd. MRS media were obtained from Qingdao Hope Bio-Technology Co., Ltd. Liquefying enzyme and saccharifying enzyme were supplied by Xiasheng Enzyme Bio-Technology Co., Ltd. All other chemical reagents were sourced from Sinopharm Chemical Reagent Co., Ltd.

2.2. Fermentation process of P. sibiricum-rice wine

To overcome the limited and inconsistent release of bioactive compounds from the raw materials, uncontrolled acidification, and variable flavor profiles during the conventional production of P. sibiricum-rice wine, a reproducible liquid-state fermentation platform with defined microorganisms herein was developed including (1) intensive enzymatic pretreatment of raw materials to enhance precursor availability; (2) the use of a liquid fermentation media; and (3) controlled inoculation with selected pure cultures. Briefly, 5-year-old P. sibiricum powder (Jiuhua Mountain, Chizhou City, China) was mixed with distilled water at a solid-to-liquid ratio of 1:25 (m/v). The amylase with the activity of ≥100,000 U/g (Shanghai Aladdin Biochemical Technology Co., Ltd. China) was used for pretreatment of P. sibiricum powder with dosage of 3.3% (w/w), pH 5.0, ultrasonic power of 290 W at 44 °C for 25 min. The supernatant was collected by centrifuging at 4000 rpm for 10 min as the Polygonatum extract. Glutinous rice flour provided by Yihe Food Co., Ltd. (DanYang, China) was blended with water at a ratio of 1:2 (m/v), gelatinized at 100 °C, and then treated with 0.1% (w/w) liquefying enzyme (Xiasheng Enzyme Biotechnology Co., Ltd.) at 60 °C for 1 h. Subsequently, 0.1% (w/w) saccharifying enzyme (Xiasheng Enzyme Biotechnology Co., Ltd.) was added and reacted at 60 °C for 2 h to yield rice slurry. The P. sibiricum-rice slurry was prepared by mixing Polygonatum extract with the glutinous rice slurry at a ratio of 1:1 (v/v) with the final pH of 6.0, heat-treating at 100 °C for 10 min and cooling to room temperature.

The activated strains of S. cerevisiae, L. casei YH80JZ-3 and/or L. fermentum JDYJZ-3 were inoculated into the P. sibiricum-rice slurry at inoculum size of 3.5% (v/v), with a Lactobacillus-to-S. cerevisiae ratio of 1:6. The detailed inoculation methods were shown as Table 1. Fermentation was conducted at 37 °C for 4 d when only LAB were present and adjusted to 30 °C after S. cerevisiae inoculation. The fermented samples were aged at 4 °C for 15 d, bottled, sterilized at 85 °C for 15 min and cooled to room temperature to obtain the P. sibiricum-rice wine.

Table 1.

Inoculation protocols for P. sibiricum-rice wine co-fermentation.

Group	Co-fermentation Protocol
	Sequence of inoculation	Inoculation Timing
S	S. cerevisiae	Single inoculation (control)
S-C	S. cerevisiae → L. casei	24-h interval
S-F	S. cerevisiae → L. fermentum
S-CF	S. cerevisiae → L. casei + L. fermentum
SC	S. cerevisiae + L. casei	Simultaneous inoculation
SF	S. cerevisiae + L. fermentum
SCF	S. cerevisiae + L. casei + L. fermentum
C-S	L. casei → S. cerevisiae	24-h interval
F-S	L. fermentum → S. cerevisiae
CF-S	L. casei + L. fermentum → S. cerevisiae

The sequences and timing of inoculating S. cerevisiae (Y), L. casei (P), and L. fermentum (F) are detailed to define the co-fermentation strategies compared in this study.

2.3. Physical and chemical characteristics of P. sibiricum-rice wine produced with different inoculation and fermentation methods

The physico/chemical analyses of final fermented P. sibiricum-rice wine samples were performed after cold aging, sterilization, and cooling. The pH values of fermented samples were measured using a pH meter. Total acidity (expressed as lactic acid) and amino acid nitrogen content in the samples were determined with a pH meter, followed by using the protocols listed in National Standard GB 12456–2021 “Determination of Total Acidity in Foods” and GB 5009.235–2016 “Determination of Amino Acid Nitrogen in Foods”.

Total phenolic content (TPC) in the P. sibiricum-rice wine samples was assayed using the Folin-Ciocalteu colorimetric method. Briefly, a standard curve was prepared by aliquoting 0.2, 0.4, 0.6, 0.8, 1.0 and 1.2 mL of 0.10 mg/mL gallic acid standard solution into 10 mL stoppered colorimetric tubes. Each aliquot was diluted to 5.0 mL with distilled water, mixed with 1.0 mL Folin-Ciocalteu reagent, and allowed to stand for 0.5–8 min. Subsequently, 1.0 mL of 20% sodium carbonate solution was added, and the mixture was brought to volume with deionized water. After incubation at 75 °C for 10 min, absorbance was measured at 760 nm. A blank control was prepared by replacing the gallic acid solution with 5.0 mL distilled water. The standard curve was plotted with absorbance (Y-axis) against gallic acid concentration (X-axis). TPC in the P. sibiricum-rice wine samples was calculated using a gallic acid standard curve and expressed in micrograms of gallic acid equivalent per milliliter of wine (μg/mL).

The reducing sugar content in the P. sibiricum-rice wine samples was determined using the anthrone‑sulfuric acid method with glucose as the standard. A standard curve was constructed by aliquoting 0.1, 0.2, 0.3, 0.4, 0.5 and 0.6 mL of 0.33 mg/mL glucose standard solution into 10 mL stoppered colorimetric tubes. Each aliquot was diluted to 2.0 mL with distilled water, mixed thoroughly, and chilled in an ice-water bath. Subsequently, 0.2% anthrone‑sulfuric acid solution was slowly added to the mark under continuous cooling. After equilibration, the mixtures were heated in a boiling water bath for 10 min, immediately cooled in an ice bath for 10 min, and their absorbance measured at 625 nm. A blank control was prepared by replacing the glucose solution with 2.0 mL distilled water. The standard curve was plotted with absorbance (Y-axis) against glucose concentration (X-axis). The reducing sugar content in the P. sibiricum-rice wine samples was calculated using a standard curve, expressed in micrograms per milliliter (μg/mL).

The alcohol content was determined using the potassium dichromate colorimetric method, in accordance with National Standard GB 5009.225–2016 “Determination of Ethanol Concentration in Foods”.

2.4. Determination of antioxidant capacity in vitro of P. sibiricum-rice wine under different inoculation and fermentation methods

The 2,2-Diphenyl-1-picrylhydrazyl (DPPH) radical scavenging capacity was determined following the method of Cai et al. with minor modifications. Briefly, 2 mL of 0.2 mmol/L DPPH solution was mixed with 2 mL of P. sibiricum-rice wine at equivalent dilution (Cai, Tang, et al., 2019). The mixture was incubated in the dark at room temperature for 30 min, and absorbance was measured at 517 nm. The scavenging activity was calculated using the formula:

$$\text{DPPH free radical scavenging ability}=\left(1-\frac{A_1-A_2}{A_0}\right)\times 100\%$$ (1)

A₁: Absorbance of sample (P. sibiricum-rice wine + DPPH).

A₂: Absorbance of sample control (P. sibiricum-rice wine + absolute ethanol).

A₀: Absorbance of blank control (absolute ethanol + DPPH).

The 2,2′-Azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS) radical scavenging capacity was analyzed according to Cai et al. with minor modifications(Cai, Zhang, et al., 2019). Briefly, 5 mL of 7.4 mmol/L ABTS solution was mixed with 88 μL of 180 mmol/L potassium persulfate (K₂S₂O₈) solution in equal volumes. The resulting suspension was stored in the dark for 12–16 h to generate ABTS radicals, then diluted with deionized water until the absorbance at 734 nm reached 0.70 ± 0.05, yielding the ABTS radical stock solution.

For testing ABTS free radical scavenging ability, 0.2 mL of P. sibiricum-rice wine (at equivalent dilution) was mixed with 4.8 mL of ABTS radical stock solution. The mixture was incubated in the dark at room temperature for 30 min, and absorbance was measured at 734 nm. The scavenging activity was calculated using the formula:

$$\text{ABTS free radical scavenging ability}=\left(1-\frac{A_1-A_2}{A_0}\right)\times 100\%$$ (2)

A₁: Absorbance of sample (P. sibiricum-rice wine + ABTS stock).

A₂: Absorbance of sample control (P. sibiricum-rice wine + PBS buffer).

A₀: Absorbance of blank control (PBS buffer + ABTS stock).

Hydroxyl radical scavenging capacity was evaluated according to Zhang et al. with minor modifications (Zhang et al., 2022). In brief, 1 mL of equivalently diluted P. sibiricum-rice wine was added to a test tube, followed by sequential addition of 1 mL ferrous sulfate (FeSO₄), 1 mL hydrogen peroxide (H₂O₂), and 1 mL salicylic acid solution. The mixture was vortexed and incubated in a 37 °C water bath for 1 h. Absorbance was measured at 510 nm, and the scavenging activity was calculated as follows:

$$\text{Hydroxyl radical scavenging ability}=\left(1-\frac{{\mathrm{A}}_1-{\mathrm{A}}_2}{{\mathrm{A}}_0}\right)\times 100\%$$ (3)

A₁: Absorbance of sample (P. sibiricum-rice wine + FeSO₄ + H₂O₂ + salicylic acid).

A₂: Absorbance of sample control (P. sibiricum-rice wine + deionized water + FeSO₄ + salicylic acid).

A₀: Absorbance of blank control (deionized water + FeSO₄ + H₂O₂ + salicylic acid).

2.5. Determination of bioactive components of P. sibiricum-rice wine produced with different inoculation and fermentation methods

Phenolic acids in the P. sibiricum-rice wine samples were analyzed by High performance liquid chromatography (HPLC) method with modifications based on the method of Xiao (Xiaoli, 2021). Briefly, the sample was acidified to pH = 2.0 using 10% (v/v) hydrochloric acid and then extracted twice with an equal volume of ethyl acetate The ethyl acetate extract was collected by centrifugation at 8000 rpm for 15 min, and removed by rotary evaporation. The residue was re-dissolved in anhydrous ethanol and filtered through a 0.45 μm membrane. Standards of gallic acid, neochlorogenic acid, protocatechuic acid, chlorogenic acid, p-hydroxybenzoic acid, vanillic acid, butyric acid, p-coumaric acid and ferulic acid was weighted to 5.0 mg, dissolved in methanol, and then fixed into a 25 mL volumetric flask at a concentration of 200 μg/mL, respectively. The each standard phenolic acid was diluted to the concentrations of 1.0, 2.0, 4.0, 8.0, 12.0, 16.0 and 20.0 μg/mL for preparing the the standard curve, respectively. The chromatographic column was Agilent HC-C18 with inner diameter of 4.6 mm and column length of 250 mm. The gradient elution procedure employed a binary mobile phase system, with 0.1% (v/v) acetic acid aqueous solution as Mobile Phase A and methanol as Mobile Phase B. The elution profile proceeded as follows: From 0 to 10 min, Mobile Phase A linearly decreased from 90% to 80% while Mobile Phase B increased from 10% to 20%. Between 10 and 15 min, Mobile Phase A rapidly decreased from 80% to 62% with Mobile Phase B rising from 20% to 38%. From 15 to 25 min, near-isocratic elution was maintained (Mobile Phase A: 62% → 60%; Mobile Phase B: 38% → 40%). At 25–26 min, a steep gradient change occurred (Mobile Phase A: 60% → 0%; Mobile Phase B: 40% → 100%) for column flushing. Between 26 and 30 min, 100% Mobile Phase B was held. Finally, from 30 to 35 min, the column was re-equilibrated (Mobile Phase A: 0% → 90%; Mobile Phase B: 100% → 10%). The total run time was 35 min. The flow rate was 0.8 mL/min, the UV detection wavelength was 260 nm, and the injection volume was 10 μL. The concentration of each phenolic acid in the P. sibiricum-rice wine samples was calculated using a standard curve, expressed in μg/mL.

The free amino acid content in the P. sibiricum-rice wine samples was determined using an automatic amino acid analyzer (Saikem Scientific Instrument Co., Ltd.), following the methodology outlined in National Standard GB/T 5009.124–2016 “Determination of Amino Acids in Foods”.

2.6. Flavor profiles of P. sibiricum-rice wine produced with different inoculation and fermentation methods

Aliquots (5 mL) of P. sibiricum-rice wine samples fermented with different inoculation methods were transferred to 20 mL headspace vials. Volatile profiles were analyzed using a PEN3 electronic nose (Air sense Analytics GmbH, Germany) under the following parameters: Sensor purge time: 210 s; Sample preparation time: 5 s; Acquisition time: 180 s; Internal flow rate: 400 mL/min; Injection flow rate: 200 mL/min.

The aroma components of P. sibiricum-rice wine were extracted by solid-phase microextraction (SPME) method using a 50/30 μm DVB/CAR/PDMS coated fiber (Supelco, Bellefonte, PA, USA). A 5 mL aliquot of P. sibiricum-rice wine fermented with different inoculation methods was transferred to a 20 mL headspace vial, mixed with 1.5 g NaCl, and spiked with 10 μL of 2-octanol (1600 μg/mL) as an internal standard. The vial was sealed and equilibrated at 50 °C for 10 min for SPME. Gas chromatography-mass spectrometry (GC–MS) analysis was performed using a triple quadrupole gas chromatograph-mass spectrometer (GC–MS/MS, Shimadzu GCMS-TQ8040 NX, Kyoto, Japan). Separation was achieved on an Rtx-WAX column (30 m × 0.25 mm × 0.25 μm) conditioned at 250 °C for 1 h, with an injector temperature of 250 °C and helium carrier gas flow rate of 1.0 mL/min in spitless mode. The oven temperature program initiated at 50 °C (10 min hold), ramped to 150 °C at 6 °C/min (5 min hold), then to 240 °C at 8 °C/min (10 min hold). Mass spectrometric detection was operated in the full-scan mode under the following conditions: electron impact (EI) ionization at 70 eV, interface temperature of 250 °C, ion source temperature of 230 °C, quadrupole temperature of 150 °C, and mass scan range of 45–450 m/z. Detected compounds were tentatively identified by matching mass spectra against the NIST17 library, with verification based on chromatographic retention times and sample-specific metadata. For semi-quantitative comparison of volatile compounds across samples, 2-octanol was employed as an internal standard. The relative content of each compound was expressed as the peak area ratio relative to 2-octanol (μg/mL, equivalent).

2.7. Sensory evaluation of P. sibiricum-rice wine with different inoculation and fermentation methods

All sensory panelists have been informed the written consent and adherence to ethical guidelines. A panel of 10 trained evaluators (5 males and 5 females, aged 18–30 years) conducted sensory assessments of the P. sibiricum-rice wine. The evaluation was based on a quantitative descriptive analysis method, using a weighted scoring system specifically developed for rice wine quality assessment (total score = 100 points). The product was evaluated across four key attributes with their respective weights determined based on their contribution to the overall quality of rice wine: appearance (10 points), aroma (30 points), taste (45 points) and typicality (characteristic style, 15 points). For each attribute, assessors scored the intensity and quality on a continuous scale from 0 to the maximum points, with higher scores indicating better performance in that attribute. The total score represents the overall sensory quality. Prior to formal evaluation, the panel was trained on the use of the scorecard and the definition of each attribute using reference samples.

2.8. Data analysis

All experiments were conducted in triplicate and the data presented in the tables and figures are the average values with the standard deviations of triplicate experiments using Microsoft Excel. Graphical representations were generated using Origin 2022, while correlation and significance analyses were performed with IBM SPSS Statistics 22.0 with statistical significance defined at p < 0.05. Multivariate analyses included principal component analysis (PCA) using SIMCA 14.1 and cluster heatmap analysis via TB tools. For the hierarchical cluster analysis, Z-score normalized data of all measured physicochemical and bioactive parameters were used, with Euclidean distance as the metric and Ward's method for linkage.

3. Results & discussion

3.1. Effects of different inoculation and fermentation methods on the physicochemical characteristics of P. sibiricum-rice wine

The physicochemical indices are shown in Fig. 1. For clarity, the following codes are used: S: S. cerevisiae; C: L. casei YH80JZ-3; F: L. fermentum JDYJZ-3. A hyphen (−) denotes sequential inoculation with a 24-h interval, while its absence indicates simultaneous inoculation. For example, S—C represents S. cerevisiae inoculated first, followed by L. casei after 24 h; SCF denotes all three strains inoculated simultaneously.Samples F—S and CF-S of P. sibiricum-rice wine exhibited the significantly higher contents of total phenolics, reducing sugars, and AAN (p < 0.05), while having a significantly lower ethanol content (p < 0.05), compared to those in other samples (Fig. 1A, C, D and E). In contrast, the sample S—C showed significantly lower reducing sugar contents and pH values, and higher total acidity than those in S sample (p < 0.05) with the comparable levels of total phenolics and ethanol (p > 0.05).

Fig. 1.

Physicochemical properties and in vitro antioxidant capacity of P. sibiricum-rice wine under different fermentation protocols. (A) Total phenols and Amino nitrogen;(B) pH value;(C) Reducing sugar;(D) Total acidity;(E) Alcohol. Different lowercase letters above the bars indicate significant differences (p < 0.05). (E) Alcohol content; (F) DPPH, ABTS, and hydroxyl radical scavenging activities. Data are presented as mean ± standard deviation (n = 3). Different lowercase letters above bars within the same indicator denote significant differences (p < 0.05).

The physicochemical properties of P. sibiricum-rice wine reflect the fundamental quality attributes and the results were shown in Fig. 1. Generally, starch is converted into dextrins and polysaccharides via liquefaction and saccharification during production of rice wine, followed by microbial hydrolysis to glucose, which is further converted to ethanol and organic acids. Additionally, residual sugars during post-fermentation also contribute to sweet taste (Lai et al., 2019). Amino acid nitrogen (AAN) serves as an indicator of protein hydrolysis efficiency, nitrogen utilization, and product quality (Wang et al., 2019). The balance between ethanol and organic acid contents is governed by the interaction of S. cerevisiae and lactic acid bacteria (LAB). In the present study, LAB showed an inhibitory effect on the ethanol production by S. cerevisiae. Similarly, Kin and Huang observed that the inhibitory effect of lactic acid bacteria on Saccharomyces cerevisiae during wine fermentation primarily occurs before the completion of alcoholic fermentation. This manifests as a suppression of the alcoholic fermentation rate, leading to acetic acid production and reduced wine quality (King & Beelman, 1986; Huang et al., 1996). This directly reflects competitive interactions among microorganisms. Lactic acid bacteria rapidly acidify the environment through homolactic fermentation, thereby inhibiting yeast metabolic activity and redirecting carbon flux (Liao et al., 2023). From this point, the final profile of sugars, acids, and ethanol in the P. sibiricum-rice wine were the synergistic effect of substrate-converted product and the direct outcome of the dynamic and order-dependence between the S. cerevisiae and L. casei YH80JZ-3 and/or L. fermentum JDYJZ-3.

3.2. Effects of different inoculation and fermentation methods on the antioxidant capacity of P. sibiricum-rice wine in vitro

The in vitro antioxidant capacity is a critical quality indicator for functional fermented beverages. Three putative mechanisms could explain the enhancement of antioxidant activity via fermentation process, including release of the bound bioactive compounds from raw materials, the promotion of their synthesis, and the direct production of new antioxidants by microorganisms (Elfahri et al., 2016; Zhang et al., 2021). In this study, DPPH, ABTS, and hydroxyl radical scavenging were used to evaluate the antioxidant profiles of the P. sibiricum-rice wines. As shown in Fig. 1 F, the distinct and significant changes between the P. sibiricum-rice wine samples were observed (Fig. 1F), indicating that the inoculation strategy significantly shaped the specific antioxidant pathways activated by the complex interplay of microbial types, substrate composition, and fermentation conditions (Sarıtaş et al., 2024). The DPPH radical scavenging capacity ranged from 21.49% to 79.48% within all groups. Group SCF, produced by simultaneous inoculation of S. cerevisiae, L. casei YH80JZ-3 and L. fermentum JDYJZ-3, exhibited the highest activity (79.48 ± 1.25%), while first inoculation of L. fermentum (F—S and CF-S) seemed to result in the lowest DPPH scavenging abilities (37.12% and 21.49%, respectively). The DPPH scavenging is closely related to the hydrogen-donating ability of antioxidants. The superior performance of groups SCF, SF, and S-CF is strongly correlated with their significantly higher total phenolic acid contents (Section 3.3), which could be attributed to the secretion of β-galactosidases, proteases, and/or peptidases by lactic acid bacteria (LAB) to degrade and hydrolyze the raw materials, thereby facilitating the release of bound phenolic compounds (Escobar-Beiza et al., 2023). Phenolic acids like gallic acid and chlorogenic acid, which possess multiple hydroxyl groups ideal for hydrogen atom transfer, are likely key contributors liberated through this enzymatic activity (Hur et al., 2014). A different pattern emerged for ABTS radical cation scavenging activity. Group S presented the highest activity of 70.70 ± 1.85%, and SCF showed the highest value of 65.00 ± 0.98% among the co-fermentation groups. The ABTS assay measures both electron and hydrogen transfer capabilities over a broader pH range. The high ABTS scavenging activity of S. cerevisiae -solely fermented sample might be explained by the dual role of phenolic compounds in microbial metabolism. While LAB could release phenolics, high concentrations of phenols also inhibit LAB activity and metabolism (Escobar-Beiza et al., 2023). Therefore, it is deduced that in the S group, the absence of LAB prevented such inhibition, allowing for the preservation or exclusive microbial synthesis by S. cerevisiae of a distinct set of potent electron-donating antioxidants. These may include specific phenolic structures or other metabolites like certain peptides, which are particularly efficient in the ABTS assay. Hydroxyl radical scavenging capacity varied from 13.68% to 28.56%. Groups F—S, CF-S, C—S, SCF with first inoculation of LABs or simultaneous inoculation with S. cerevisiae generally exhibited higher activities, with F—S showing the highest value of 28.56 ± 1.40%. Notably, S group had the lowest hydroxyl radical scavenging capacity of 13.68 ± 1.59%. The hydroxyl radical is the most reactive oxygen species, and its scavenging often involves complex mechanisms like metal ion chelation to inhibit the Fenton reaction. Enhanced activity in LAB-prominent groups might be attributed to the increased concentration of metal-chelating phenolic acids like gallic acid released via fermentation, and/or production of specific metal-chelating peptides or organic acids during LAB fermentation (Spyropoulos et al., 2011). The significantly-altered metabolic environment in these groups, potentially characterized by lower pH and different organic acid profiles, could favor the formation or stability of the metal-chelating agents (Hur et al., 2014). Methods for measuring antioxidant capacity include DPPH, ABTS, and ORAC assays, each possessing distinct advantages (such as ease of operation and sensitivity to specific compounds) but also inherent limitations. Consequently, no single assay can comprehensively reflect antioxidant capacity, highlighting the specificity of antioxidant capacity testing methods and the multifaceted influences of the fermentation process (Hamed et al., 2025). Group SCF presented a robust and balanced profile, ranking highest in DPPH, second in ABTS, and above average in hydroxyl radical scavenging. This comprehensive performance aligns with the synergistic metabolic activities of its diverse microbial consortium, which likely promotes both the release and production of a wide spectrum of antioxidants. In contrast, while S excelled in ABTS scavenging, it was less effective in neutralizing DPPH and hydroxyl radicals, highlighting a more specialized antioxidant profile shaped solely by S. cerevisiae metabolism. Groups F—S and CF-S, despite their high acidity, showed a weak overall antioxidant profile except for moderate hydroxyl radical scavenging. The results suggested that first-inculcation of L. fermentum potentially promoted to produce the chelating agents and create the environments (e.g., rapid acidification) inhibiting the synthesis or release of the hydrogen/electron-donating antioxidants (Escobar-Beiza et al., 2023). These divergent profiles conclusively demonstrated that microbial consortia and inoculation sequences tailor the antioxidant repertoire of the final product through distinct biochemical pathways, effectively programming the functional output of the fermentation process.

3.3. Effects of different inoculation and fermentation methods on contents of phenolic acids in P. sibiricum-rice wine

Plant matrices contain high concentrations of phenolic compounds, which could be released from glycosides or conjugates through fermentation or enzymatic hydrolysis (Cai, Tang, et al., 2019). Phenolic compounds are the potent in vitro antioxidants, exhibiting health-promoting properties such as antioxidant, anti-aging, anti-atherosclerotic, anticancer, and preventive effects against diabetes, cardiovascular diseases, and neurological disorders (Xu et al., 2015). As shown in Fig. 2, phenolic acid contents in P. sibiricum-rice wine samples diversified with the inoculation and fermentation methods, which critically shaped product quality. Gallic acid, neochlorogenic acid, and ferulic acid dominated the phenolic acid profile with peak concentrations of 2.83 ± 0.07 μg/mL, 2.36 ± 0.08 μg/mL, and 2.65 ± 0.05 μg/mL, respectively. The LAB-involved P. sibiricum-rice wine samples showed a significant increase of phenolic acid levels of 10.75 ± 0.3 μg/mL (p < 0.05). Generally, at the early stage of fermentation, hydrolysis of plant cell walls liberates the bound phenolic acids, which subsequently undergo microbial-driven transformations including demethylation, hydroxylation, dehydroxylation, carboxylation, and decarboxylation (Que et al., 2006). Chakraborty et al. reported that Escherichia coli could mediate the demethylation of ferulic acid to caffeic acid (Chakraborty et al., 2016), and Chen et al. also demonstrated that Phomopsis liquidambaris could conduct the coupled oxidation of p-coumaric acid to p-hydroxybenzoic acid, followed by hydroxylation to protocatechuic acid (Chen et al., 2011). As shown in Fig. 2, compared with the S group, the galactic acid content in the lactic acid bacteria-inoculated group significantly increased to 3.80 ± 0.06 μg/mL, ferulic acid content rose to 2.65 ± 0.05 μg/mL, and total phenolic acid content reached 10.75 ± 0.30 μg/mL (p < 0.05).Total phenolic acids in groups S—C and S—F were 1.69- and 1.82-fold higher, respectively, than those in S group. Notably, chlorogenic acid was detected exclusively in LAB-supplemented groups. The neochlorogenic acid content in Group S—C showed no significant difference with that in S group (p > 0.05), while the contents of other phenolic acids were higher than those in S group (p < 0.05).

Fig. 2.

Bioactive components of P. sibiricum-rice wine under different fermentation protocols. (A) Content of individual phenolic acids; (B) Taste activity values (TAV) of umami, sweet, and bitter amino acids; (C) Umami Amino Acids;(D) Sweet Amino Acid;(E) Bitter Amino Acid;(F) Amino Acid Without Taste;(G) Total Acid.

The significant enhancement in antioxidant capacity (DPPH, ABTS, and hydroxyl radical scavenging) observed in LAB-supplemented groups, particularly S—C, could be mechanistically linked to the increased release of specific phenolic acids via microbial enzymatic action. Lactic acid bacteria, including L. fermentum and L. casei, are known to produce enzymes such as feruloyl esterases that hydrolyze ester bonds between cell wall polysaccharides and phenolic acids like ferulic and p-coumaric acid, thereby increasing their free forms (Li et al., 2023; Yao et al., 2013). The increase in gallic acid across all LAB groups suggested the activity of tannase or related esterases (Yao et al., 2013). The potent activity of gallic acid for DPPH and ABTS scavenging is attributed to its ortho-dihydroxyl (catechol) group, which facilitates electron donation and radical stabilization (Moazzen et al., 2022; Ye, 2019). While ferulic acid contributes notably to lipid oxidation inhibition (Sánchez-Moreno et al., 1999), its methoxy substitution might render it less effective in certain radical scavenging assays compared to gallic acid (Moazzen et al., 2022). Therefore, the superior and balanced antioxidant profile of the S—C protocol likely results from the synergistic release of a spectrum of phenolic acids (e.g., gallic, ferulic, chlorogenic).

3.4. Effects of different inoculation and fermentation methods on free amino acid contents in P. sibiricum-rice wine

Proteolysis during fermentation generates amino acids that enhance the nutritional value of rice wine and the impart distinct taste profiles including umami (aspartic acid, glutamic acid), sweetness (threonine, serine, glycine, alanine, proline), bitterness (valine, methionine, isoleucine, leucine, phenylalanine, histidine, arginine), and tastelessness (cystine, tyrosine, lysine) (Xu, 2019; Yang et al., 2021). Fig. 2 presented the contents of free amino acids in Polygonatum rice wine samples produced with different inoculation methods.

The amino acids primarily originated from P. sibiricum and glutinous rice, with taste-active amino acids such as glutamic acid, histidine,valine, Arginine, dominating the profile. S group contained 17 kinds of free amino acids. LAB-supplemented fermentation significantly increased the contents of umami amino acids from 0.849 mg/mL to 2.493 mg/mL while reducing methionine level from 0.033 mg/mL to 0 mg/mL. (p < 0.05). Groups S—C, F—S, and CF-S exhibited higher total amino acid contents by 16.51%, 85.42%, and 81.63% than those of S group, respectively, with a marked increase of glycine, valine, isoleucine, leucine, tyrosine, phenylalanine, histidine, arginine, and umami/bitter amino acids (p < 0.05), indicating that L. fermentum JDYJZ-3could efficiently release the amino acids.

Taste activity value (TAV), calculated as the ratio of amino acid concentration to its taste threshold (Fig. 2B), revealed umami and bitter amino acids as primary taste contributors. LAB inoculation significantly enhanced umami TAV (p < 0.05). Groups F—S and CF-S demonstrated superior TAVs across all taste categories compared to other groups (p < 0.05), while S—C showed higher umami, bitter, and total TAVs than S (p < 0.05). Hence, it could be concluded that LAB supplementation enhanced both umami amino acid content and TAV to offer the flavor complexity.

Hierarchical cluster analysis using average linkage grouped P. sibiricum-rice wines into three distinct clusters based on physicochemical properties, in vitro antioxidant capacity, and bioactive components (Fig. 4A). Cluster I comprised groups F-Sand CF-S, Cluster II included S, S—C, and S—F, while remaining groups formed Cluster III. The proximity of S—C and S—F to S in Cluster II indicated the comparable quality metrics with the S. cerevisiae-only fermented control. In summary, S—C group fermented via sequential inoculation of S. cerevisiae (24 h) followed by L. casei YH80JZ-3 exhibited minimal divergence from S group in overall characteristics while achieving significant enhancements in phenolic acids (e.g., +69% total phenolics) and free amino acids (e.g., +16.5% total amino acids). Based on this, the inoculation combination of S. cerevisiae and L. casei achieves an optimally balances traditional fermentation attributes with improved nutritional and sensory profiles, positioning it as a promising strategy for producing P. sibiricum-rice wine with a better quality.

Fig. 4.

Multivariate statistical analysis for comprehensive quality evaluation of P. sibiricum-rice wine fermented with different inoculation protocols. (A)Hierarchical cluster analysis (HCA) heatmap. Clustering fermentation groups based on standardized values of physicochemical properties, antioxidant capacity, and bioactive components. The color scale represents Z-scores (red: above mean; blue: below mean); (B) PCA biplot of major volatile aroma compounds. PCI vs. PC2 scores plot showing the distribution of fermentation groups based on the relative content of key aroma compounds (see Table 2). Vector arrows indicate the contribution of individual compounds; (C) Heatmap of clustered major aroma compounds. Hierarchical clustering of aroma compounds (rows) and fermentation groups (columns) based on semi-quantitative relative content (μg/mL equivalent). The color intensity represents the relative abundance (log2 transformed); (D) PCA biplot of integrated quality indices. PCI vs. PC2 scores plot based on eighteen key quality indices (see Section 3.7). The cumulative variance explained by the first two principal components is indicated. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

3.5. Electronic nose analysis of P. sibiricum-rice wine with different inoculation and fermentation methods

Electronic nose analysis rapidly characterized aroma profiles of P. sibiricum-rice wine, highlighting key differences across fermentation methods. PCA of sensor data (Fig. 3B) explained 96.22% of total variance, with PC1 (89.80%) and PC2 (6.42%). Samples clustered into four groups: S, S—C, S—F, and SC (PC1-negative/PC2-positive, dominant sensors W1W, W6S, W2S, W1S, W2W, W3S); S-CF, SCF, and C—S (PC1-negative/PC2-negative, W5S); CF-S (PC1-positive/PC2-positive, W3C, W5C, W1C); SF and F—S (PC1-positive/PC2-negative, no distinct features). CF-S exhibited maximal divergence from other groups, while S—C, S—F, and SC showed closer similarity to S but with enhanced aroma signatures, indicating LAB's significant impact on aroma characteristics.

Fig. 3.

Flavor and aroma profile analysis of P. sibiricum-rice wine fermented with different inoculation protocols. (A) Radar chart of electronic nose sensor responses. Normalized response values of 10 metal oxide semiconductor (MOS) sensors (W1W, W5S, W3C, W6S, W5C, W1S, W1C, W3S, W2S, W2W) for each fermentation group; (B) Principal Component Analysis (PCA) biplot of electronic nose data.PCI vs. PC2 scores plot showing sample clustering based on aroma profiles detected by PEN3 electronic nose. The percentage of total variance explained by each principal component is indicated; (C) Linear Discriminant Analysis (LDA) of electronic nose data. LD1 vs. LD2 scores plot providing enhanced separation of sample groups based on aroma profiles; (D) Relative content and categories of major volatile compounds. Semi-quantitative relative contents (μg/mL equivalent, normalized to 2-octanol internal standard) of alcohols, esters, acids, ketones, aldehydes, and other compounds detected by HS-SPME-GC–MS.

Linear Discriminant Analysis (LDA) clarified overlapping clusters from PCA, with LD1 (87.84%) and LD2 (8.91%) explaining 96.75% of total variance (Fig. 3C). Results aligned with PCA, showing marked differences between S-C/SC and S-F/S groups, confirming distinct aroma profiles driven by fermentation protocols. Sensor response magnitudes correlated with volatile compound abundance: higher values indicate greater concentrations, while lower values reflect reduced levels (Karl et al., 2008). Radar plots (Fig. 3A) identified terpenes (W1W), nitrogen oxides (W5S), methyl compounds (W1S), and alcohols (W2S) as dominant aroma contributors. Key findings included S: Peak W5S response (nitrogen oxides), S-F: Highest W2S response (alcohols), S-C: Maximal W1W (terpenes) and W1S (methyl compounds) signals, with W2S levels comparable to S. Aromatic compound sensors (W1C, W3C, W5C, W2W) revealed that low responses overall, except increased W1C/W3C/W5C in F-S/CF-S and moderate W3C/W5C/W2W in S-C/SF. In conclusion, LAB species and inoculation strategies critically shaped aroma profiles and sequential fermentation significantly enhanced terpene, methyl, and aromatic compounds to provide a optimized sensory quality.

3.6. SPME-GC/MS analysis of P. sibiricum-rice wine with different inoculation and fermentation methods

A total of 84 volatile compounds were identified within P. sibiricum-rice wine samples. The semi-quantitative relative contents of the major aroma compounds were detailed in Table 2. Among them, S group contained the highest total relative content of major aromatics (50.165 μg/mL equivalent), which was characterized by a prominent profile of higher alcohols, including 3-methyl-1-butanol (19.727 μg/mL) and phenethyl alcohol (18.983 μg/mL). Interestingly, unique compounds such as ethyl benzoate and phenylacetaldehyde were detected, contributing to fruity, cherry, and cocoa notes. For sequential inoculation of S—C led to a distinct shift with higher contents of alcohols, such as 3-methyl-1-butanol of 17.688 μg/mL and phenethyl alcohol of 17.579 μg/mL, and newly-formed esters which were not observed in S group. Most notably, γ-nonalactone (0.116 μg/mL) and ethyl lactate (0.335 μg/mL) were synthesized. The total acid content in S—C was moderate and comparable to S. For fermentations where L. fermentum was inoculated first (F—S, CF-S), the volatile profile was dominated by very high concentrations of organic acids, particularly acetic acid (23.695–26.459 μg/mL), resulting in the highest total acid content among all groups. This was accompanied by a sharp reduction in total alcohol and ester content.

Table 2.

Main aroma compounds and their semi-quantitative relative content in P. sibiricum-rice wine with different inoculation and fermentation methods.

Code	Aroma Compound	CAS	Retention Index	Odor Description	Relative content (μg/mL equivalent)
					S	S-C	S-F	S-CF	SC	SF	SCF	C-S	F-S	CF-S
Alcohol					42.341	36.223	21.63	20.064	38.384	17.968	16.262	31.909	7.029	11.979
O1	butanol	71–36-3	662	Banana, fruity, whiskey.	0.025	0.044	0.01	0.008	0.015	0.007	/	0.012	/	/
O2	3-Methyl-1-butanol	123–51-3	697	Fruity, Brandy	19.727	17.688	10.848	9.135	8.111	8.179	8.438	17.465	/	2.271
O3	3-Methyl-1-pentanol	589–35-5	796	Earthy, cocoa notes	0.03	0.033	/	/	/	/	/	0.057	/	/
O4	1-Octen-3-ol	3391-86-4	969	Grass, Lavender, Mushrooms	/	/	/	/	0.002	/	/	/	1.708	2.016
O5	2,3-Butanediol	513–85-9	743	Fruity, creamy, buttery	3.274	0.611	0.162	0.129	1.925	0.216	0.242	0.601	2.915	4.63
O6	phenethyl alcohol	#######	1136	Floral, Rose, Honey	18.983	17.579	10.286	10.526	19.516	9.353	7.57	13.559	1.006	1.284
O7	hexanol	111–27-3	860	Herbal aroma, raw green flavor	0.161	0.145	0.192	0.18	0.16	0.143	/	0.131	0.198	0.293
O8	heptanol	111–70-6	960	Violet, Peony	0.141	0.113	0.007	/	0.039	/	/	0.034	/	/
O9	3-Ethoxy-1-propanol	111–35-3	837	–	/	0.01	0.021	0.016	0.012	/	/	/	/	/
O10	(S)-(+)-3- Methyl-1-pentanol	42,072–39-9	796	–	/	/	0.019	0.013	0.018	/	0.012	/	/	/
O11	(2R,3R)-(−)-2,3-Butanediol	24,347–58-8	743	–	/	/	0.085	0.057	/	0.07	/	/	0.123	/
O12	2-Methyl-1-butanol	137–32-6	697	Fatty, Leathery, Cocoa	/	/	/	/	8.586	/	/	/	/	/
O13	3-Methyl-3-buten-1-ol	763–32-6	728	Fruity, sweet	/	/	/	/	/	/	/	0.05	0.192	0.377
O14	3-Octanol	589–98-0	979	Herbal, Nutty	/	/	/	/	/	/	/	/	0.253	0.35
O15	trans-2-Octen-1-ol	18,409–17-1	1067	lime flavor	/	/	/	/	/	/	/	/	0.634	0.758
Salts					3.761	3.873	1.355	1.022	3.885	0.583	0.927	3.348	0.46	0.526
E1	isoamyl acetate	123–92-2	820	Fruity, banana, sweet	0.086	0.016	0.012	0.009	0.055	/	0.005	0.005	/	/
E2	Ethyl caproate	123–66-0	984	Fruity, Pineapple, Fatty	0.136	0.076	0.037	0.027	0.126	0.027	0.048	0.079	/	/
E3	Ethyl heptanoate	106–30-9	1083	Pineapple flavor, rum aroma	0.017	0.012	0.013	0.007	0.014	/	0.013	/	/	/
E4	Ethyl octanoate	106–32-1	1183	Sweet fruity, floral, brandy notes	0.901	1.18	/	/	0.557	/	/	1.568	/	/
E5	Ethyl decanoate	110–38-3	1381	Grapey, oily, waxy	0.408	0.56	0.146	0.11	0.63	0.078	0.084	0.402	/	/
E6	Phenethyl acetate	103–45-7	1259	Rose, Fruit Peach, Honey	1.005	0.376	0.225	0.221	0.813	0.131	0.103	0.176	0.023	0.052
E7	Gamma-nonanolactone	104–61-0	1284	Coconut, Creamy, Butter Flavor	/	0.116	0.117	0.101	0.078	0.127	0.144	0.115	0.088	0.096
E8	Ethyl benzoate	93–89-0	1160	Wintergreen, Grape, Cherry Flavors	0.443	/	/	/	/	/	/	/	/	/
E9	Ethyl phenylacetate	101–97-3	1259	Fruity, honeyed	0.649	0.184	/	/	0.245	/	/	/	/	/
E10	Ethyl lactate	97–64-3	848	Butter Flavor, Creamy Candy Flavor	/	0.335	0.162	0.187	0.083	0.143	0.229	0.208	0.021	0.019
E11	Diethyl succinate	123–25-1	1151	Floral, Apple, Chocolate	/	0.319	/	0.074	0.517	/	0.026	0.136	/	/
E12	Monoethyl succinate	1070-34-4	1141	–	/	0.565	/	0.18	0.572	/	0.169	0.377	/	/
E13	Ethyl 3-phenylpropionate	2021-28-5	1359	Hyacinth aroma, rum, honey flavor	/	/	0.018	0.017	/	0.01	0.011	/	/	/
E14	Methyl N-hydroxybenzoimidate	67,160–14-9	1301	–	/	/	0.477	/	/	/	/	0.168	0.328	0.359
E15	Ethyl hex-4-enoate	51,368–03-7	992	fruity	0.116	0.134	0.148	0.089	0.195	0.067	0.095	0.114	/	/
Acid					2.577	2.529	11.914	10.787	5.092	13.924	15.459	4.253	28.336	31.047
C1	acetate	64–19-7	576	acidity	/	/	9.352	8.178	1.877	11.152	12.418	/	23.695	26.459
C2	isobutyric acid	79–31-2	711	Sour, cheesy, rancid butter.	0.338	0.3	0.253	0.239	0.343	/	/	0.24	0.245	/
C3	hexanoic acid	142–62-1	974	Sour, Spicy, Cheesy	0.405	0.492	0.378	0.411	0.493	0.454	0.536	0.617	1.108	1.003
C4	heptanoic acid	111–14-8	1073	Waxy, fermented flavor	0.016	0.018	0.015	0.026	0.01	0.033	0.035	0.018	0.07	0.088
C5	pungent	124–07-2	1173	Greasy, fatty, vegetable flavor	0.496	0.638	0.304	0.263	0.513	0.259	0.254	1.055	0.272	0.276
C6	sorbic acid	110–44-1	990	–	0.148	0.179	0.148	0.175	0.116	0.193	0.178	0.179	0.347	0.307
C7	dehydroacetic acid	771–03-9	1453	–	0.459	/	0.888	0.799	0.67	1.063	1.139	1.005	1.296	1.289
C8	decanoic acid	334–48-5	1372	Fatty, citrusy, rancid.	0.164	0.332	/	/	/	0.158	0.18	0.353	/	/
C9	benzoic acid C6H5COOH	65–85-0	1150	Balsamic vinegar. Urine.	0.136	0.314	0.375	0.341	0.3	0.385	0.437	0.34	0.524	0.491
C10	2-Methylbutyric acid	116–53-0	811	Fruity, sour, cheesy	0.415	0.256	/	/	/	/	/	0.236	/	/
C11	4-Methylpentanoic acid	646–07-1	910	Spicy, cheesy	/	/	0.201	0.183	/	0.202	/	/	/	/
C12	lauric acid	143–07-7	1570	Fatty, coconut oil flavor	/	/	/	0.164	0.292	/	/	/	/	/
C13	Hydrocinnamic acid	501–52-0	1349	Rose, musk, cinnamon.	/	/	/	0.008	/	0.025	0.028	/	0.041	0.033
C14	3-Methylpentanoic acid	105–43-1	910	Sour, Raw Green, Cheese	/	/	/	/	0.478	/	0.229	/	0.346	0.464
C15	myristic acid	544–63-8	1769	Oily, Waxy, Coconut Flavor	/	/	/	/	/	/	0.025	0.029	0.126	0.126
C16	palmitic acid	#######	1968	Greasy, Waxy	/	/	/	/	/	/	/	0.181	0.151	0.232
C17	4-Hexenoic acid	35,194–36-6	982	–	/	/	/	/	/	/	/	/	0.115	0.279
Ketone					0.02	0.257	0.152	0.019	0.664	0.078	0.09	0.044	0.23	0.213
K1	3-Hydroxy-2-butanone	513–86-0	717	Sweet, Butter, and Yogurt Flavors	0.02	0.257	0.021	0.019	0.664	/	0.006	0.044	0.145	0.096
K2	4-Hydroxy-2-butanone	590–90-9	798	–	/	/	0.131	/	/	0.078	0.084	/	/	/
K3	acetophenone	98–86-2	1029	Floral, sweet, cherry	/	/	/	/	/	/	/	/	0.085	0.117
Aldehyde					0.787	/	/	0.056	0.206	0.142	0.053	0.088	0.057	0.128
D1	phenylacetaldehyde	122–78-1	1081	Hyacinth aroma, sweet, cocoa notes	0.279	/	/	/	/	/	/	/	/	/
D2	benzaldehyde C6H5CHO, the simplest aromatic aldehyde	100–52-7	982	Bitter almond flavor, cherry flavor, sweet flavor	0.508	/	/	/	/	/	/	/	0.057	0.128
D3	2,4-Dimethylbenzaldehyde	15,764–16-6	1208	Almond, Cherry, and Vanilla Flavors	/	/	/	0.056	0.206	0.142	0.053	0.088	/	/
Other					0.679	0.011	/	0.006	0.538	0.015	/	0.017	0.01	0.029
T1	benzothiazole	95–16-9	1208	Coffee, nutty, meaty	0.008	0.011	/	0.006	0.008	0.015	/	0.017	0.01	0.029
T2	2-Methoxy-4-vinylphenol	7786-61-0	1293	Clove, Vanilla, Smoky	0.671	/	/	/	0.53	/	/	/	/	/
Total (Sum of relative contents)					50.165	42.893	35.051	31.954	48.769	32.71	32.791	39.659	36.122	43.922

*Relative contents (μg/mL equivalent, normalized to 2-octanol) of alcohols, esters, acids, and other volatiles are compared to highlight the impact of inoculation methods on the wine's aromatic composition. * (Note: “/” indicates not detected).

The synthesis of higher alcohols is governed by the availability of free amino acids. In our study, the increased concentrations of leucine and phenylalanine in LAB-supplemented groups provided enhanced substrates for S. cerevisiae metabolism. Specifically, these amino acids entered into the Ehrlich pathway transformed into key aroma-active higher alcohols. For example, leucine was converted to 3-methyl-1-butanol (imparting fruity, cognac-like notes), and phenylalanine to phenethyl alcohol (contributing floral, rose, and honey aromas) (Hazelwood et al., 2008; Jaimand et al., 2023; Nabhan & Piñera, 2023).

L. casei YH80JZ-3 and L. fermentum JDYJZ-3 seemed to play an indirect and primary role within this process. Through their potent proteolytic activity, LAB hydrolyzes proteins, thereby expanding the release of free amino acids for S. cerevisiae uptake and metabolism. The microbial cross-feeding critically shapes the final concentration of these higher alcohols in the wine (Luo et al., 2024). It was also to note that these compounds could enrich sensory complexity at concentrations below approximately 300 μg/mL, while excessive levels would lead to harshness and off flavors, highlighting the necessity for balanced synthesis (Fejzullahu, 2024).

The formation of LAB-characteristic esters is a direct outcome of co-metabolism. Lactic acid, the primary metabolic end product of homo-fermentative LAB like L. casei, serves as the crucial acyl donor for ethyl lactate synthesis in the presence of yeast-derived ethanol (Liu et al., 2004). This ester, along with γ-nonalactone (imparting a coconut-like aroma), were specific markers of LAB-supplemented fermentation in the present study, with recent research confirming that LAB co-fermentation actively enhances the production of lactones (Sun et al., 2025; Yu et al., 2021). Similarly, other organic acids produced by LAB become precursors for various ethyl esters. The carbon chain length of these esters determines their aroma profile and persistence, with longer chains generally yielding more intense and enduring fragrances (Hu et al., 2018; Sumby et al., 2010).

The comprehensive analysis also revealed a coherent biochemical narrative linking the compositional data to the final sensory outcome. The S—C protocol notably elevated the levels of precursor amino acids such as leucine, valine, and phenylalanine, which entered into metabolic pathways during co-fermentation as substrates for the Ehrlich pathway to yield the corresponding higher alcohols such as 3-methyl-1-butanol (fruity, malty), 2-methyl-1-propanol (winey), and phenethyl alcohol (floral, rose-like), respectively (Hazelwood et al., 2008). Concurrently, the organic acids (e.g., lactic acid) produced by L. casei provided acyl donors for esterification, leading to the formation of characteristic esters like ethyl lactate. This co-fermentation in the S—C group resulted in a more complex and balanced volatile profile with an optimal ratio of beneficial alcohols and esters to suppress undesirable off-flavors. From this point, the harmonized aroma profile, derived from the directed transformation of abundant precursors, directly correlated with and explained the highest sensory evaluation scores of the S—C group, particularly in terms of aroma complexity and taste balance.

Organic acids play a dual sensory role such as direct impact and ester precursor. Acids such as acetic, hexanoic, and octanoic acid are produced via microbial metabolism, including fatty acid degradation and amino acid conversion (Tufariello et al., 2012). These acids synergize with other volatiles to shape the unique aroma profile (Samakradhamrongthai, 2024). At low levels, acetic acid provides a sharp, tangy note, hexanoic acid contributes cheesy and spicy nuances, and longer-chain acids like octanoic acid exhibit milder, oil-like aromas (Da Conceicao Neta et al., 2007; El Hadi et al., 2013). However, excessive synthesis in the F—S and CF-S samples would impart sharp, vinegary, or rancid off-notes, which could explain the inferior sensory scores of these groups (Luo et al., 2024). Beyond their direct flavor impact, their abundance created a reservoir for ester synthesis, highlighting the necessity of balanced microbial activity to promote positive aroma formation while preventing negative sensory impacts. These findings align with observations by Gao et al. that co-culture of Eurotium cristatum and kefir significantly reshaped microbial succession of dark tea, thereby enhancing typical aroma compounds such as linalool and methyl salicylate while suppressing off-flavor metabolites like 3-methylbutanoic acid (Yao Gao et al., 2025). Additionally, during proposed S—C inoculation strategy in the present study, S. cerevisiae initially establishes a foundation to produce ethanol and higher alcohols (Styger et al., 2011), followed by L. casei providing lactic acid and moderateing the organic acid profile (Luo et al., 2024), which further underscored that orchestrated microbial succession was a common lever for directing flavor complexity and sensory balance across diverse fermented beverages. From this point, control of microbial communities during production of the fermented beverages such as rice wine and dark tea emerges as a key unifying strategy for targeting the favorite flavor.

3.7. Comprehensive quality evaluation of P. sibiricum-rice wine with different inoculation and fermentation methods

Building upon the analysis of the quality attributes, a comprehensive quality evaluation was performed using PCA. Eighteen key quality indices were selected for this multivariate analysis including total phenolic content (X1), alcohol content (X2), pH (X3), DPPH radical scavenging ability (X4), ABTS radical scavenging ability (X5), hydroxyl radical scavenging ability (X6), total phenolic acid content (X7), umami amino acid content (X8), W5S response value (X9), W1S response value (X10), W1W response value (X11), W2S response value (X12), 3-methyl-1-butanol content (X13), phenylethanol content (X14), total ester content (X15), octanoic acid content (X16), 3-hydroxy-2-butanone content (X17), and total aromatic compound content (X18). The cumulative contribution rate of the first five principal components was 96.49%, effectively synthesizing the key information for a comprehensive assessment. Y1, Y2, Y3, Y4, and Y5 were selected to evaluate the quality of P. sibiricum-rice wine instead of the screened indexes. The scores of the first five principal components (Y1 to Y5) were calculated using their respective analytical expressions (Eqs. 4.1–4.5). A composite score (Y) for each co-fermentation protocol was then calculated by weighing these component scores by their variance contribution (Eq. 4.6). As shown in Table 3, the protocols were ranked by composite score as follows: S-C > S > SC > S-F > S-CF > C-S > SCF > SF > F-S > CF-S, which aligned closely with the results of the sensory evaluation (Table 3 and confirmed the reliability of the PCA-based evaluation model. The S—C protocol achieved the highest comprehensive quality score integrating the favorable performance across all measured indices and consistent with its optimal physicochemical balance, its significant enhancement of phenolic acids and umami amino acids, and the harmonious aroma profile. On this basis, it could be confirmed that that S—C group yields P. sibiricum-rice wine with the best overall quality.

Table 3.

Quality evaluation of P. sibiricum-rice wine.

Group	Y1	Y2	Y3	Y4	Y5	Model Evaluation		Sensory Evaluation
						Y	Sort	Score	Sort
S	114.83	47.07	3.23	−23.72	−8.86	70.66	2	83.14 ± 3.90b	2
S-C	111.15	47.22	15.87	−24.72	5.36	70.70	1	87.42 ± 3.08a	1
S-F	112.68	15.80	16.74	−27.16	−4.82	65.25	4	75.14 ± 3.62cde	5
S-CF	97.83	13.70	7.42	−19.38	−13.06	55.69	5	77.08 ± 2.71cd	4
SC	111.58	40.06	18.14	−29.66	−6.51	69.00	3	79.80 ± 1.82bc	3
SF	81.60	19.89	2.72	−13.00	−15.23	47.61	8	65.72 ± 3.72f	8
SCF	85.25	19.46	3.33	−12.18	−17.40	49.57	7	72.82 ± 2.49de	6
C-S	78.91	41.82	9.21	−9.82	−8.70	51.39	6	71.02 ± 2.82e	7
F-S	19.24	72.16	12.50	−5.93	−0.01	24.93	9	56.80 ± 4.09g	10
CF-S	−22.27	97.28	7.18	7.16	4.60	7.00	10	58.22 ± 3.62g	9

*Scores from principal component analysis (Y1-Y5), the composite score (Y), and actual sensory evaluation are ranked, demonstrating that the sequential inoculation Y—P (S. cerevisiae → L. casei) yielded the highest overall quality. *.

3.8. Comparison with P. sibiricum-rice wine produced with traditional method

Compared with those of P. sibiricum-rice wine produced with traditional method, the co-fermentation strategy developed in this study significantly enhanced the overall quality. According to Wang, conventional P. sibiricum-rice wine contains approximately 1.90 mg/mL of polysaccharides and received a sensory score around 82.5 points (Wang et al., 2020). The optimized S—C sample preserved a higher level of polysaccharides and markedly enriched the profile of esters (e.g., γ-nonalactone, ethyl lactate) and phenolic acids resulting in a more harmonious flavor balance. Moreover, the S—C group achieved the highest sensory evaluation score of 87.42 ± 3.08 characterized by well-balanced sweetness and acidity, a pronounced aromatic complexity, and a pleasant aftertaste. The P. sibiricum-rice wine of S—C group also exhibited significantly elevated levels of functional components such as P. sibiricum polysaccharides and phenolic acids, combining the sensory foundation of traditional rice wine with enhanced nutritional functionality. These findings demonstrate that the liquid-state fermentation system combined with co-culture of S. cerevisiae and L. casei YH80JZ-3 improved the utilization of P. sibiricum bioactive compounds and substantially enhanced the organoleptic and nutritional quality, offering a viable pathway for developing the functional P. sibiricum-rice wine products.

4. Summary

A co-fermentation strategy to develop the functional P. sibiricum-rice wine was well investigated by optimizing the sequential and/or simultaneous inoculation of S. cerevisiae and L. casei YH80JZ-3 and/or L. fermentum JDYJZ-3. Compared to the control fermented solely with S. cerevisiae, the sequential inoculation of S. cerevisiae followed by L. casei (S—C protocol) achieved a 1.69-fold increase in total phenolic acids, a 16.5% rise in total free amino acids, and improved the volatile aroma profile by increasing esters and acids (e.g., γ-nonalactone and ethyl lactate) with the reduced higher alcohol contents. Multivariate and sensory analyses confirmed that S—C group was the optimal strategy since S. cerevisiae provided the alcoholic and ester precursors, while subsequent L. casei contributed to balance the acidity and liberate antioxidant phenolic compounds. The outputs of the present study demonstrated that sequential co-fermentation with S. cerevisiae and L. casei was a promising approach to developing a high-quality, functional P. sibiricum-rice wine with enhanced organoleptic and nutritional properties. Future studies would employ advanced multivariate data analysis such as multiple factor analysis or Partial Least Squares regression to integrate the full spectra of physicochemical, volatile, sensory, and potentially metagenomic data to systematically understand the microbial processes shaping the final product profile and targetedly design the functional fermented beverages.

CRediT authorship contribution statement

Yu-Tong Han: Writing – original draft, Methodology, Formal analysis, Data curation, Conceptualization. Yan-Shu Li: Writing – review & editing, Methodology, Data curation, Conceptualization. Sanabil Yaqoob: Writing – original draft, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Aysha Imtiaz: Writing – review & editing, Software, Investigation, Formal analysis, Data curation. Chunlai Zeng: Writing – review & editing, Visualization, Validation, Software, Investigation. Yong-Kun Ma: Writing – original draft, Supervision, Project administration, Funding acquisition, Conceptualization. Feng-Jie Cui: Writing – review & editing, Supervision, Resources, Methodology, Conceptualization. Qing Shen: Writing – original draft, Supervision, Project administration, Funding acquisition, Conceptualization.

Ethical approval

In the sensory evaluation component of this study, only voluntary adult participants were recruited to assess the sensory attributes (appearance, aroma, taste, and typicality) of P. sibiricum–rice wine. Informed, written consent was obtained from each participant in the study. Each of them could withdraw their consent without providing any justification. Each participant also consented to the processing of their personal data in accordance with the relevant guidelines and regulations. Herein the ethical permission, to conduct a human sensory study, is not a requirement of Jiangsu University.

Funding

No funding has been received for this research.

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

Supplementary material

Data availability

Data will be made available on request.