Lactobacillus casei Fermentation Enhances the Sensory Quality of Shoumei Tea Infusion: Based on Flavor and Metabolic Analysis

ABSTRACT

Shoumei white tea is widely consumed, yet its limited product diversity constrains its economic value. This study used Shoumei tea infusion as a fermentation substrate to examine how Lactobacillus casei modulates flavor and underlying biochemical pathways. During 72 h of fermentation, the pH decreased from 5.0 to 4.1, accompanied by a 95.5% reduction in soluble sugars (from 10,056 to 453 mg/L). The accumulation of organic acids, particularly lactic acid (rising from 873.90 to 1268.17 µg/mL) and malic acid (rising from 150.83 to 725.18 µg/mL), drove the increase in sourness. Concurrently, substantial reductions in L‐theanine (by 47.66%), catechins, flavonoids, and polyphenolic pigments resulted in markedly decreased bitterness and astringency. Headspace solid‐phase microextraction–gas chromatography–mass spectrometry (HS‐SPME–GC–MS) combined with partial least squares discriminant analysis (PLS‐DA) identified 51 volatile flavor compounds (VFCs), among which phenylethyl alcohol, linalool, 1‐heptanol, 1‐hexanol, ethyl hexanoate, and ethyl octanoate (variable importance in projection [VIP] > 1) were established as major contributors to aroma enhancement. Integrated metabolomics and enzyme activity analyses indicated that elevated β‐glucosidase and polyphenol oxidase activities, together with shifts in amino acid biosynthesis, 2‐oxocarboxylic acid metabolism, and galactose metabolism, drove the formation of characteristic fermentation aromas. These findings provide mechanistic insight and technical support for developing value‐added fermented Shoumei tea beverages.

1. Introduction

White tea (Camellia sinensis) is one of the six major types of tea in China, produced mainly through withering and drying processes. The finished tea is distinguished by its dense white trichomes and its silvery‐white to gray‐green appearance. The characteristic aroma of white tea is primarily attributed to volatile flavor compounds (VFCs) such as hexanal, (E)‐2‐hexenal, benzaldehyde, phenylacetaldehyde, (E)‐geraniol, phenylethyl alcohol, linalool, and their oxides (Q. Chen et al. ²⁰¹⁹). In addition, white tea is rich in bioactive constituents including catechins, hydrolyzable tannins, phenolic acids, theanine, caffeine, and kaempferol (Tan et al. 2017). As a significant category of white tea, Shoumei is appreciated for its mellow taste and health benefits. However, its wider consumer acceptance is often constrained by inherent flavor limitations. Newly processed Shoumei is frequently described as overly bitter and astringent, mainly due to its high concentrations of catechins and other polyphenolic compounds derived from mature leaves and stems, which vary across grades and production regions (Z. Chen et al. ²⁰²⁴). Moreover, the aroma of fresh Shoumei tea tends to be weak and dominated by grassy or “green” notes rather than a complex, pleasant fragrance. The formation of desirable aroma and the reduction of undesirable taste depend heavily on precise control of processing parameters, particularly withering, which governs the transformation of key precursor compounds (Wu et al. 2024). Ultimately, the overall flavor quality of Shoumei results from a series of intricate biochemical reactions during processing (Tian et al. 2024). Therefore, mitigating bitterness and astringency while enhancing aroma complexity remains a crucial challenge and research focus in tea chemistry.

Fermentation has been extensively utilized in dairy products to improve nutritional and functional properties and to establish characteristic flavor profiles through integrated chemical analyses (Khazzar et al. 2025; Muñoz et al. ²⁰²⁵; Zacometti et al. ²⁰²⁴). Similarly, fermentation has proven to be an effective strategy for enhancing the flavor, nutritional value, and functional properties of tea beverages, and more broadly for improving the technological and functional value of diverse botanical and beverage substrates (Battistini et al. 2023; Horn et al. ²⁰²⁴; Praia et al. ²⁰²²). Kombucha, an ancient Chinese fermented tea, exemplifies this process, as it involves a symbiotic culture of yeasts and bacteria that carry out alcoholic, lactic, and acetic fermentation, producing organic acids such as acetic, gluconic, tartaric, and malic acids (Bortolomedi et al. 2022; de Miranda et al. ²⁰²²). However, spontaneous fermentation may pose safety risks due to microbial contamination (Jayabalan et al. 2014). Controlled fermentation using single lactic acid bacteria (LAB) strains can ensure safety and consistent quality, while allowing for targeted strain performance across complex, matrix‐dependent beverage systems (Borges Martins da Silva et al. ²⁰²⁴; M. E. Herkenhoff et al. ²⁰²³; M. E. Herkenhoff ²⁰²⁴). Our previous study found that Lactobacillus plantarum fermentation of black tea improved flavor characteristics and sensory acceptability, although it reduced some nutrients and bioactive compounds (R. Li et al. ²⁰²²). During fermentation, microbial enzymes convert carbohydrates, nitrogenous compounds, and other substrates into micromolecular bioactives such as organic acids, amino acids, and volatiles (Y. Wang et al. ²⁰²³), while producing beneficial postbiotics like vitamins, γ‐aminobutyric acid, and extracellular polysaccharides (Nataraj et al. 2020). Globally, LAB are recognized as safe (GRAS) and play a central role in various fermented foods (Peng et al. 2020). Recent studies using multiomic approaches in fermented foods have revealed that microbial community succession and metabolic cross‐talk contribute substantially to flavor development and quality formation during fermentation (Tata et al. 2024). Among them, Lactobacillus casei, commonly found in fermented beverages (Bancalari et al. 2020), has emerged as a promising strain for tea fermentation, given its demonstrated adaptability to challenging, nondairy beverage environments (Borges Martins da Silva et al. ²⁰²⁴; M. E. Herkenhoff et al. ²⁰²³; M. E. Herkenhoff ²⁰²⁴). Despite the well‐documented advantages of LAB in beverage processing, there is a notable scarcity of research utilizing microbial fermentation specifically to modulate the flavor profile of Shoumei white tea.

Despite the well‐documented advantages of LAB in beverage processing, there is a notable scarcity of research utilizing microbial fermentation specifically to modulate the flavor profile of Shoumei white tea. While previous studies have demonstrated that LAB fermentation can improve the sensory and functional properties of tea by enhancing aroma, reducing bitterness and astringency, and promoting the formation of bioactive compounds, thereby improving the overall quality of tea beverages (R. Li et al. ²⁰²²; Liao et al. ²⁰²⁴), we hypothesize that controlled fermentation with L. casei will trigger specific enzymatic bioconversions, effectively degrading bitter‐tasting phenolic compounds while generating highly desirable aromatic volatiles. Therefore, the primary objective of this investigation is to elucidate these material transformations and structurally map the flavor evolution mechanisms during L. casei fermentation.

To test this hypothesis, the Shoumei tea infusion was utilized as the substrate for producing a tea infusion through L. casei fermentation. The flavor characteristics of the white tea and the changes in the contents of VFCs, catechins, soluble sugars, and amino acids during the fermentation process were examined. Furthermore, the underlying mechanism of flavor enhancement was preliminarily explored based on enzyme activity and metabolomics. These findings may provide a theoretical basis for quality control (QC) of white tea fermented with L. casei.

2. Materials and Methods

2.1. Materials

Shoumei‐grade Dabai tea was purchased from Fujian Baizhang Rock Tea Co., Ltd.; L. casei was a strain maintained by the Institute of Food Science and Technology, Fuzhou University (We previously screened and found that L. casei is the most suitable strain for fermenting white tea based on its fermentation and sensory properties [Figures S1 and S2]). Analytical‐grade anhydrous glucose, ninhydrin, gallic acid, sodium hydroxide, sodium carbonate, concentrated sulfuric acid, methanol, rutin, phenol, potassium acetate, and aluminum chloride hexahydrate were purchased from Sinopharm Chemical Reagent Co., Ltd.; analytical‐grade disodium hydrogen phosphate and sodium dihydrogen phosphate were purchased from Xilong Science Co., Ltd.; analytical‐grade stannous chloride was purchased from Shanghai Chemical Reagent Purchasing and Supply Station; MRS medium and MRS broth were purchased from Guangdong Huankai Microbiology Technology Co., Ltd.; authentic standards (purity ≥98%) were purchased from two vendors. Ethyl hexanoate, 1‐hexanol, 1‐heptanol, 1‐nonanol, citronellol, phenylethyl alcohol, and nerolidol were obtained from Shanghai Yuanye Bio‐Technology Co., Ltd. (Shanghai, China). Ethyl octanoate, linalool, and nerol were obtained from Shanghai Macklin Biochemical Co., Ltd. (Shanghai, China). All other reagents were of analytical grade.

2.2. Preparation of the Fermented White Tea Infusion

Dried white tea (Shoumei class) was finely ground using a high‐speed grinder to obtain tea powder. The powder was mixed with distilled water at a ratio of 1:100 (w/v) and extracted at 100°C for 90 min with continuous stirring. The extract was filtered through a 300‐mesh sieve, and 200 mL of filtrate was dispensed into 250‐mL screw‐cap bottles. Subsequently, 1% (w/v) glucose was added, and the tea infusion was sterilized at 121°C for 20 min. Lactobacillus casei was used as the starter culture. The strain was first activated in MRS broth (1:100, v/v) and incubated statically at 37°C for 12 h. Then, 1 mL of the culture was transferred into 100 mL of fresh MRS broth and incubated at 37°C until the optical density at 600 nm (OD₆₀₀) reached 0.6. Cells were harvested by centrifugation at 8000 × g for 5 min, washed three times with sterile saline, and resuspended in saline to OD₆₀₀ = 0.6 (∼10⁸ CFU/mL). The bacterial suspension was inoculated into the sterilized tea infusion at a 2% (v/v) inoculation rate, followed by static incubation at 37°C for 72 h. Samples were collected at 0, 12, 24, 36, 48, and 72 h for subsequent analysis.

2.3. Measurement of Reducing Sugars

Reducing sugars were determined using the PMP‐derivatization HPLC method. Briefly, 1 mL of fermented tea supernatant was dried under nitrogen at 70°C, reacted with 0.5 mL of 0.5 mol/L PMP (in methanol) and 0.5 mL of 0.3 mol/L NaOH at 70°C for 30 min, and then neutralized with 0.5 mL of 0.3 mol/L HCl. After triple extraction with chloroform, the aqueous phase was filtered (0.22 µm) for HPLC analysis. Separation was performed on an Acclaim 120 C18 column (4.6 × 250 mm, 5 µm; Agilent, USA) using 82% phosphate buffer (0.1 mol/L, pH 7.0) and 18% acetonitrile at 1.0 mL/min and 25°C. Detection was at 245 nm with a 10 µL injection.

2.4. Determination of Organic Acids

Organic acids were analyzed by HPLC following a reported method with minor modifications. Briefly, 2 mL of sample was centrifuged at 12,000 × g for 5 min, and the supernatant was filtered through a 0.22‐µm membrane. Analysis was performed on an UltiMate 3000 HPLC system (Thermo Fisher Scientific, USA) using a Hypersil Gold aQ C18 column (4.6 × 100 mm, 5 µm). The mobile phase was 0.01 M KH₂PO₄ (pH 2.7) at a flow rate of 0.8 mL/min, a column temperature of 27°C, a detection wavelength of 210 nm, and an injection volume of 20 µL.

2.5. Determination of Catechins

The fermented tea samples were centrifuged at 12,000 × g for 5 min, and the supernatant was filtered through a 0.22‐µm membrane before HPLC analysis. Chromatographic separation was performed on a reversed‐phase C18 column (4.6 × 200 mm, 5 µm; Agilent) at 35°C. Mobile phase A was 0.2% acetic acid in water (with 0.2% EDTA), and mobile phase B was acetonitrile (Merck). The liquid‐phase gradient elution program for the determination of catechin is shown in Table S1. The flow rate was 1 mL/min, the injection volume was 10 µL, and detection was at 278 nm.

2.6. The Examination of VFCs and Quantitative Analysis of Key Aroma Components

Fermented tea infusions collected at different fermentation stages were analyzed for volatile organic compounds (VOCs) following a modified method based on previous research (R. Li et al. ²⁰²²; M. E. Herkenhoff et al. ²⁰²⁴). Briefly, 5 mL of sample was transferred into a 20‐mL headspace vial containing 1 g NaCl and 10 µL of 2‐octanol (10 mg/L in ethanol) as the internal standard. The vial was incubated and stirred at 450 rpm and 45°C for 20 min, followed by solid‐phase microextraction (SPME) for 50 min prior to GC–MS analysis (7890B/5977A, Agilent Technologies, USA). Chromatographic separation was achieved on an HP‐INNOWAX capillary column (30 m × 0.25 mm, 0.25 µm; Agilent Technologies). The GC oven temperature was programmed as follows: initial 40°C (held for 1 min), ramped to 80°C at 4°C/min, then to 160°C at 2°C/min (held for 10 min), and finally to 220°C at 10°C/min. High‐purity helium (99.999%) was used as carrier gas at a constant flow rate of 1.0 mL/min. The temperatures of the ion source, transfer line, and quadrupole were maintained at 230°C, 250°C, and 150°C, respectively. Electron ionization (70 eV) spectra were recorded in full scan mode (m/z 30–500) with a solvent delay of 2.5 min.

Volatile compounds were rigorously identified by integrating mass spectral matching, retention index (RI) calculation, and reference standards: mass spectra were matched against NIST 14/WILEY databases (match factor ≥85%), and experimental retention indices (RI_calc) were determined relative to a series of n‐alkanes (C11–C23) analyzed under identical chromatographic conditions using the Van den Dool and Kratz equation. These indices were further cross‐verified against reference literature values (RI_lit​) reported for polar (WAX) capillary columns.

Furthermore, 10 key target aroma compounds (ethyl hexanoate, 1‐hexanol, ethyl octanoate, 1‐heptanol, linalool, 1‐nonanol, citronellol, nerol, phenylethyl alcohol, and nerolidol) were selected for absolute quantification using calibration curves constructed with authentic commercial standards. These specific volatiles were chosen because they represent the dominant odor‐active components responsible for the characteristic sensory profile and aroma evolution in the fermented tea. Semiquantitative analysis of nontarget flavor compounds was performed by normalizing their respective peak areas against the internal standard.

2.7. Electronic Nose (E‐Nose) Analysis

The olfactory profiles of fermented tea infusions were assessed using a PEN3 E‐nose (AIRSENSE, Germany) equipped with 10 metal oxide sensors (sensor types and characteristics listed in Tables S2 and S3) (L. Chen et al. ²⁰²³). The system was preheated for 24 h, and operational parameters were set as follows: sample interval 1 s, flush time 60 s, zero point 5 s, presampling 5 s, measurement 60 s, chamber flow 400 mL/min, and initial injection flow 400 mL/min.

Samples were diluted 1:1 with ultrapure water, and 5 mL was transferred to a 10‐mL headspace vial sealed with PTFE‐silicone septa. The vials were equilibrated at 60°C for 30 min. Gas from the headspace was adsorbed via ventilation and injection needles and injected into the E‐nose. Sensors were cleaned with dry air at 400 mL/min for 1 min between measurements. Data were collected for 60 s at 0.5 s intervals, digitized, and analyzed using PCA for pattern recognition. Each sample was measured in triplicate.

2.8. Electronic Tongue (E‐Tongue) Analysis

The taste profiles of fermented tea infusions were evaluated using an E‐tongue system (SA402B, Intelligent Sensor Technology, Japan) with five taste sensors: sourness, saltiness, bitterness, astringency, and umami. Sensors and reference electrodes were pre‐immersed in their respective solutions for 24 h. Fermented tea samples from different time points were centrifuged at 8000 × g for 8 min, and the supernatant was diluted 1:2 with ultrapure water. Then, 35 mL of the solution was placed into 50 mL testing cups and analyzed following system calibration and cleaning (L. Chen et al. ²⁰²³). Four measurements were conducted in parallel, and the average of the last three was used for analysis. Sensor data were processed using PCA.

2.9. Determination of Enzyme Activity

2.9.1. β‐Glucosidase Activity

β‐Glucosidase activity was measured as described previously. Samples were centrifuged at 5000 × g for 10 min, and 25 µL of supernatant was mixed with 100 µL of 5 mM pNPG (preheated) and incubated at 50°C for 30 min. The reaction was stopped with 125 µL of 1 M Na₂CO₃, and absorbance was measured at 405 nm (Varioskan LUX, Thermo Fisher). Activity was expressed as the amount of enzyme producing 1 µM pNP per minute (see formula 1):

$$\beta -\mathrm{Glucosidase}\mathrm{Activity}\left(\mathrm{U}/\mathrm{mL}\right)=\frac{\left(Y-b\right)\times V_1}{K\times V_2\times t\times 1000},$$ (1)

where Y represents the absorbance value at 405 nm; K denotes the slope of the pNP standard curve equation (as shown in Supporting Information); b indicates the y‐intercept of the pNP standard curve equation; V ₁ represents the total volume of the reaction system (mL); t refers to the reaction time; and V ₂ indicates the sample volume.

2.9.2. Polyphenol Oxidase (PPO) Activity

PPO activity was measured as described by Szczepańska et al. (2020). Samples were centrifuged at 4000 rpm for 10 min, and 10 µL of supernatant was mixed with 70 mM catechol in 50 mM phosphate buffer (pH 6.5). Absorbance at 420 nm was recorded at 0 and 2 min. One unit of enzyme activity (U/mL) was defined as a 0.01 increase in absorbance per minute per milliliter.

2.9.3. Peroxidase (POD) Activity

POD activity was measured following previous protocols with minor modifications (Loh and Lim 2018). Fermented tea supernatant (100 µL) was mixed with 1 mL of 50 mM phosphate buffer (pH 6.5) containing 12 mM guaiacol and 3 mM H₂O₂. Absorbance at 418 nm was recorded at 0 and 2 min. Activity (U/mL) was defined as a 0.01 increase in absorbance per minute per milliliter.

2.10. Nontargeted Metabonomic Analysis

Nontargeted metabolomic analysis of fermented tea infusions was performed following Y. Liu et al. (2023) with minor modifications. Samples collected at 0, 36, and 72 h were centrifuged at 12,000 rpm for 10 min at 4°C. Supernatant (100 µL) was mixed with 400 µL methanol:acetonitrile (1:1, v/v) containing isotopic internal standards, vortexed for 30 s, sonicated for 10 min in an ice‐water bath, and stored at −40°C for 1 h to precipitate proteins. After centrifugation (12,000 × g, 15 min, 4°C), the supernatant was collected for UHPLC–Orbitrap MS analysis. QC samples were prepared by pooling equal volumes from all individual samples and injected every six analytical runs to monitor system stability. Chromatography was performed on a Kinetex C18 column (2.1 × 50 mm, 2.6 µm) using water containing 0.01% acetic acid (A) and isopropanol:acetonitrile (1:1, v/v) (B), with an injection volume of 2 µL and a tray temperature of 4°C. MS data were acquired on an Orbitrap Exploris 120 equipped with an ESI source (capillary temperature 320°C, sheath gas 50 Arb, auxiliary gas 15 Arb, full MS resolution 60,000, MS/MS resolution 15,000, collision energies 20/30/40 eV, spray voltages +3.8 kV/−3.4 kV). Raw data were processed using XCMS for peak alignment and extraction, and metabolites were putatively annotated at Metabolomics Standards Initiative (MSI) Level 2 by matching accurate precursor mass (m/z error <5 ppm) and MS/MS spectra against mzCloud, HMDB, KEGG, and BiotreeDB databases. To minimize false positives, features with relative standard deviation (RSD) >30% in QC samples or signal‐to‐noise ratio <10 were filtered out, and chemically implausible annotations (e.g., synthetic pharmaceuticals or environmental contaminants resulting from automated database matches) were manually curated and excluded from biological pathway analyses. Pareto‐scaled data were analyzed using SIMCA (v16.0.2) for principal component analysis (PCA) and partial least squares discriminant analysis (PLS‐DA). The PLS‐DA model was evaluated using sevenfold cross‐validation and rigorously validated against overfitting via a 200‐iteration permutation test (Q ₂ intercept <0). Finally, differential metabolites were selected based on a strict multicriteria threshold: variable importance in projection (VIP) >1.0, Student's t‐test (p < 0.05), and fold change (FC) >1.2 or <0.83.

2.11. Statistical Analysis

All experiments were performed in triplicate, and results are presented as mean ± standard deviation. Statistical analyses were conducted using SAS 9.4 (SAS Institute, Cary, NC, USA) via one‐way ANOVA followed by Tukey's HSD test (α = 0.05). Instrumental data were analyzed using Laboratory Analytical Viewer (LAV, G.A.S., Dortmund, Germany) with three plug‐ins and the GC × IMS Library Search. PCA analysis of VOCs was performed using the Dynamic PCA plug‐in in Vcal software (G.A.S.), while PCA of E‐nose and E‐tongue data was conducted in MATLAB 2013a using the princomp function.

3. Results and Discussion

3.1. Lactobacillus casei Fermentation Gradually Changed the Basic Physicochemical Parameters of White Tea

Figure 1 shows the dynamic changes in the major nonvolatile chemical components of Shoumei white tea during 72 h of fermentation with L. casei. Throughout the fermentation process, the soluble sugar content continued to decline significantly, from an initial concentration of approximately 10,056 mg/L to approximately 453 mg/L at 72 h (Figure 1A), indicating that L. casei efficiently utilized the soluble sugars. A similar significant decrease was also observed in the total free amino acid content (Figure 1B). As both carbon and nitrogen sources serve as essential substrates for microbial growth and metabolism, these decreases align with previous findings showing that L. casei fermentation elevates total acid content while reducing pH in tea systems (Figure S1) (C. Chen, Guo, et al. ²⁰²⁵). Furthermore, the concentration of total tea polyphenols remained relatively stable within the first 24 h of fermentation before gradually declining (Figure 1C). In contrast, total flavonoids and catechins declined significantly throughout fermentation (Figure 1D,F). Catechins are well recognized as major bioactive compounds in tea, exhibiting antioxidant and anti‐inflammatory properties (Chandran et al. 2025; Yin et al. ²⁰²⁵). However, it is undeniable that they also bring obvious bitterness to the tea soup, reducing the sensory acceptability of the tea soup. The observed reduction in total flavonoids and catechins during L. casei fermentation likely contributed to a milder, less bitter taste in the resulting tea infusion. In addition, the characteristic amino acid l‐theanine in the tea leaves also decreased significantly after the fermentation (by about 47.66%) (Figure 1E). As l‐theanine is a key contributor to the umami and sweet taste of tea, its reduction may partially weaken these sensory attributes; however, this change should be considered together with the concurrent reductions in bitterness and astringency, representing a sensory trade‐off (Zhang et al. 2020). Regarding tea pigments, theaflavin content declined sharply within the first 24 h, whereas thearubigins and theabrownins decreased more gradually throughout the 72‐h process (Figure 1G). Normally, catechins are oxidized to theaflavins, thearubigins, and theabrownins during tea processing, enhancing pigment accumulation (M. Liu et al. ²⁰²⁵; Zhao et al. ²⁰²²). While Cheng et al. (2025) reported that LAB fermentation increased catechin and phenolic acid contents in black tea (Cheng et al. 2025), our findings indicate that L. casei fermentation of white tea resulted in the degradation of catechins and oxidation‐derived pigments (theaflavins and thearubigins). This contrast implies that the metabolic impact of LAB on tea polyphenols is highly dependent on tea matrix composition and strain‐specific enzymatic activities. In addition, Liu et al. documented that gut microbiota could metabolize complex tea pigments such as theaflavin‐3,3′‐digallate (TFDG), producing unique metabolites like theanaphthoquinone and phenylcarboxylic acids, thereby demonstrating microbial capability in transforming theaflavins and related polyphenolic pigments (Z. Liu et al. ²⁰²¹).

FIGURE 1.

Changes in basic physical and chemical indicators of white tea during fermentation. (A) Soluble sugar. (B) Total free amino acids. (C) Tea polyphenols. (D) Total flavonoids. (E) l‐Theanine. (F) Caffeine. (G) Relative content of tea pigments. Different lowercase letters indicate significant differences (p < 0.05).

To elucidate the carbohydrate metabolism of L. casei, seven reducing sugars were profiled during fermentation (Table 1; Figure S3). Glucose was initially the dominant carbon source (12,034.808 ± 197.659 µg/mL) but rapidly declined and was nearly depleted after 72 h. Unlike kombucha, where sucrose, fructose, and glucose dominate and are interconverted through sucrose degradation (S. Wang, Li, et al. ²⁰²⁴), our system showed a distinct pattern. The pH dropped from 5.0 to 4.2 within 48 h and stabilized around 4.0 thereafter, indicating active acid production followed by metabolic balance (Figure S1). Meanwhile, mannose increased markedly from 4.70 to 9.74 µg/mL at 48 h, suggesting release from polysaccharides or microbial synthesis. Ribose and galactose remained stable in the early stage but declined later, while xylose and rhamnose continuously decreased after 36 and 24 h, respectively. Arabinose showed no clear trend. These results indicate dynamic sugar utilization and metabolic adaptation of L. casei to acidifying conditions. Similarly, Y. L. Lau and Tang (2025) found that the addition of the parabiotic strain K‐1 to the kombucha system did not significantly alter the pH, acidity, and Brix of any kombucha variant, suggesting that carbohydrate utilization by the microorganisms is highly strain specific. Organic acids, key contributors to the flavors of fermented beverages, were quantified (Table 2; Figure S4). Lactic acid, the primary metabolite of L. casei, increased significantly, with its concentration rising from 873.90 µg/mL at 0 h to 1268.17 µg/mL at 72 h. This confirms the robust metabolic activity of L. casei. Concomitantly, the concentrations of malic acid and formic acid also showed a significant upward trend throughout the fermentation process, increasing from 150.83 and 51.64 µg/mL to 725.18 and 81.67 µg/mL, respectively. Similarly, the concentrations of tartaric acid and succinic acid increased significantly toward the end of fermentation. In contrast, citric acid was consumed during fermentation, with its concentration decreasing significantly by approximately 46.45% between 0 and 48 h before recovering slightly. Interestingly, a slight recovery to 36.181 mg/L was observed at 72 h. Since L. casei generally lacks a complete tricarboxylic acid (TCA) cycle for de novo citrate biosynthesis, this late‐stage increase is likely nonmetabolic. We hypothesize that this recovery may be attributed to partial bacterial cell autolysis releasing intracellular citrate, or the continuous drop in pH promoting the dissociation of citrate–metal complexes in the tea matrix, thereby increasing detectable free citric acid. Acetic acid levels fluctuated but did not show a significant net increase within 72 h. These changes in the organic acid profile indicate a typical heterofermentative metabolic pathway, which is crucial for the development of the characteristic acidic and complex flavor of fermented white tea beverages (Cheng et al. 2025).

TABLE 1.

The dynamic variation in the levels of reducing sugars during the different stages of tea infusion fermentation.

No.	Compound name	Average content (µg/mL)
		0 h	12 h	24 h	36 h	48 h	72 h
1	Mannose	4.702 ± 0.187c	5.653 ± 0.703c	5.866 ± 0.238c	7.624 ± 0.725b	9.741 ± 0.318a	9.321 ± 0.772a
2	Ribose	4.956 ± 0.234a	4.989 ± 0.481a	4.971 ± 0.340a	4.308 ± 0.205ab	3.512 ± 0.527bc	2.510 ± 0.329c
3	Rhamnose	7.753 ± 0.179	7.909 ± 0.232	7.616 ± 0.332	6.253 ± 1.278	6.041 ± 0.346	4.599 ± 0.192
4	Glucose	12,034.808 ± 197.659a	12,058.633 ± 136.876a	11,864.732 ± 97.230a	10,527.473 ± 786.318a	4752.254 ± 2239.753b	138.506 ± 59.320c
5	Galactose	6.302 ± 0.190a	6.397 ± 0.209a	6.552 ± 0.190a	6.255 ± 0.304a	6.429 ± 0.284a	4.844 ± 0.552b
6	Xylose	4.269 ± 0.149a	3.723 ± 0.319ab	3.349 ± 0.330bc	3.912 ± 0.328ab	2.927 ± 0.205c	2.219 ± 0.253d
7	Arabinose	3.408 ± 0.186a	3.466 ± 0.114a	3.053 ± 0.297a	2.778 ± 0.438a	3.459 ± 0.119a	3.111 ± 0.276a

TABLE 2.

The dynamic changes in the levels of various organic acids during the fermentation of tea infusions.

No.	Compound name	Average content (µg/mL)
		0 h	12 h	24 h	36 h	48 h	72 h
1	Tartaric acid	24.035 ± 3.455bc	20.456 ± 4.164c	23.433 ± 3.801bc	30.835 ± 8.629bc	35.601 ± 4.105ab	45.643 ± 5.039a
2	Formic acid	51.641 ± 1.566bc	47.673 ± 9.015c	48.893 ± 4.638c	69.191 ± 12.460ab	77.845 ± 5.308a	81.667 ± 1.989a
3	Malic acid	150.838 ± 26.168b	111.052 ± 14.856b	288.544 ± 35.969b	626.070 ± 140.785a	647.227 ± 53.103a	725.182 ± 140.432a
4	Lactic acid	873.899 ± 156.256a	922.688 ± 107.305a	913.237 ± 171.154a	1073.470 ± 253.133a	1043.552 ± 143.718a	1268.170 ± 168.665a
5	Acetic acid	146.655 ± 30.080a	121.058 ± 35.199a	106.770 ± 26.140a	103.271 ± 18.912a	101.046 ± 12.666a	108.451 ± 30.371a
6	Citric acid	47.224 ± 1.253	49.132 ± 6.078	36.640 ± 4.322	31.731 ± 3.448	25.290 ± 4.752	36.181 ± 1.543
7	Succinic Acid	0.851 ± 0.057	0.874 ± 0.115	0.804 ± 0.138	0.856 ± 0.176	1.185 ± 0.143	0.971 ± 0.104

3.2. Lactobacillus casei Gradually Metabolized Catechins

Catechins are the major polyphenols in white tea, contributing to its characteristic bitterness, astringency, and bioactivity. In this study, the dynamic changes in four primary catechins, catechin (C), epicatechin (EC), epigallocatechin (EGC), and epigallocatechin gallate (EGCG), were monitored during 72 h of L. casei fermentation (Figure 2). The nongallated catechins C and EGC showed a sharp decline within the first 12 h (p < 0.05), followed by partial recovery at 24–48 h and another decrease by 72 h. EC content decreased slightly after 12 h but remained largely stable thereafter. EGCG, the predominant ester‐type catechin, declined significantly after 12 h and continued to decrease gradually at 72 h. The degradation of EGCG during fermentation is consistent with previous reports, where LAB utilize or transform catechins (Gaur and Gänzle 2023). Lactobacillus casei and related species are known to produce tannase, which hydrolyzes the ester bond in gallated catechins, converting EGCG into EGC and gallic acid (Jana et al. 2014). This enzymatic conversion likely accounts for the transient increase in EGC observed around 24–36 h. Similar decreases in ester‐type catechins during LAB fermentation have been associated with reduced bitterness in tea (M. Liu et al. ²⁰²⁵). Additionally, acid accumulation during fermentation may promote catechin degradation and epimerization, further modifying the polyphenolic profile (N. Li et al. ²⁰¹²). Although catechins can inhibit bacterial growth, lactobacilli exhibit remarkable tolerance and metabolic capacity for these phenolics (L. Y. J. Lau et al. ²⁰²⁵). This biotransformation not only mitigates bitterness and astringency but also potentially enhances the bioavailability and activity of catechins (Sidhu et al. 2024). Overall, L. casei fermentation significantly reshapes the catechin composition of white tea.

FIGURE 2.

Dynamic changes in the content of catechins during different fermentation stages. (A) Catechin. (B) Epicatechin. (C) Epigallocatechin. (D) Epigallocatechin gallate. Different lowercase letters indicate significant differences (p < 0.05).

3.3. Lactobacillus casei Fermentation Gradually Changed the Flavor of White Tea

To comprehensively evaluate the sensory evolution of white tea infusion during fermentation, aroma and taste were analyzed using an E‐nose and an E‐tongue.

The E‐nose radar plot (Figure 3A) revealed that L. casei fermentation markedly altered the aroma profile of white tea. Response intensities of sensors corresponding to VFCs such as short‐chain alkanes, alcohols, ethers, aldehydes, and ketones (W5S), as well as sulfur‐containing organics (W1S, W1W, W2S, W2W), increased notably after 24 h, suggesting enhanced generation of VFCs. This observation is consistent with the metabolic behavior of LAB, in which amino acids and carbohydrates are converted into diverse alcohols, aldehydes, and ketones that impart fruity and mellow notes. Similarly, R. Li et al. (2022) reported increased levels of volatile alcohols and carbonyl compounds in Lactobacillus plantarum‐fermented tea, contributing to improved aroma quality (R. Li et al. ²⁰²²). Conversely, the decline in W1C and W3S responses suggested degradation or transformation of long‐chain hydrocarbons initially present in white tea. The linear discriminant analysis (LDA) plot (Figure 3B) further visualizes temporal separation of aroma profiles. Samples from 0–24 h clustered closely, indicating a stable early‐phase profile, whereas those from 36–72 h formed distinct groups, demonstrating substantial aromatic differentiation after 24 h. This clear temporal shift highlights the E‐nose's ability to discriminate fermentation‐driven aroma remodeling.

FIGURE 3.

Determination of tea flavor and aroma at different fermentation stages using electronic tongue and electronic nose. (A) Radar chart of sensor response intensity in aroma. (B) Principal component analysis of sensor response intensity in aroma. (C) Radar chart of IA value differences in taste. (D) Principal component analysis of IA value differences in taste.

E‐tongue analysis revealed parallel evolution in taste attributes (Figure 3C; Table S4). Sourness increased continuously over 72 h (p < 0.05), consistent with lactic acid accumulation, a hallmark of L. casei fermentation and corroborated by organic acid measurements (Table 2). In contrast, umami, bitterness, and astringency declined significantly (p < 0.05), reflecting improved palatability. The reduction in bitterness and astringency is attributed to microbial degradation of polyphenols, particularly ester‐type catechins like EGCG, via tannase‐mediated hydrolysis into nongallated catechins and gallic acid (Figure 2) (Ye et al. 2022). Liao et al. (2024) similarly emphasized the role of LAB enzymes in softening tea's harsh taste (Liao et al. 2024). Moreover, the decline in umami likely stems from the consumption of free amino acids such as l‐theanine for bacterial metabolism (Figure 1E). Furthermore, PCA of E‐tongue data (Figure 3D) provided an integrated view of taste evolution. PC1 and PC2 explained 99.6% of total variance, with PC1 (96.85%) capturing most fermentation‐related differences. The 0–24 h samples clustered tightly, while 36–72 h samples shifted progressively along PC1, indicating distinct and stabilized taste profiles in later fermentation stages.

Overall, E‐nose and E‐tongue analyses collectively demonstrate that L. casei fermentation profoundly remodels the sensory characteristics of white tea, attenuating bitterness and astringency, enhancing sourness, and enriching aroma complexity, thus yielding a more balanced and appealing flavor profile.

3.4. Evolution of VFC Profiles During Fermentation

The aroma of tea beverages is primarily shaped by the composition and concentration of VFCs. To clarify the influence of L. casei fermentation on the aroma profile of Shoumei white tea, VOCs were analyzed at multiple time points using HS‐SPME–GC–MS (Figure 4; Table S5).

FIGURE 4.

Changes in VFCs in white tea infusion fermented by Lactobacillus casei during the fermentation process. (A) Stacked chart of the number of volatile compound categories. (B) HCA of volatile components. (C) PCA of volatile components. (D) Heatmap of VFCs.

A total of 72 VOCs were identified, covering eight classes: alcohols, esters, aldehydes, ketones, phenols, hydrocarbons, acids, and heterocyclic compounds. Fermentation markedly remodeled the VOC landscape (Figure 4A). Over the 72‐h process, the relative abundances of alcohols, esters, and acids increased steadily, while the relative abundances of aldehydes, which were associated with the raw or “green” aroma of tea, decreased (X. H. Chen, Liang, et al. ²⁰²⁵). Alcohols, the primary contributors to mellow and floral notes, expanded from 17 to 20 species, while esters associated with fruity aromas also increased. These trends indicate that L. casei effectively biotransformed native volatiles into novel fermentation‐derived compounds. Multivariate analysis further illustrated this dynamic. Both hierarchical cluster analysis (HCA; Figure 4B) and PCA (Figure 4C) revealed distinct temporal separation. Samples from 0–24 h clustered tightly, while those from 36–72 h formed distinct groups. PCA revealed that the first two components explained 99.67% of total variance (PC1: 98.64%), and sample trajectories along PC1 demonstrated a pronounced metabolic shift in VOC production after 24 h. The heatmap (Figure 4D) highlights specific compound transitions. Unfermented tea (0 h) contained abundant hexanal, linalool, geraniol (floral), and trans‐2‐hexenal (leafy), which were key contributors to the inherent aroma of white tea (Qi et al. 2025). As fermentation progressed, these compounds decreased, while new volatiles emerged, particularly after 48 h.

Notably, higher alcohols such as phenylethyl alcohol (rose‐like), 1‐octanol, and benzyl alcohol increased markedly, formed via the Ehrlich pathway, where L. casei might catabolize amino acids like phenylalanine and leucine (Fan et al. 2024). Concurrently, esters including ethyl hexanoate, ethyl octanoate, and ethyl 2‐hexenoate accumulated through enzymatic esterification between alcohols and fatty acids, enhancing fruity and sweet notes (Qi et al. 2025). At the same time, short‐chain and medium‐chain acids such as hexanoic acid, caprylic acid, and nonanoic acid produced through lipid or amino acid metabolism during fermentation may make the aroma more intense (Hu et al. 2024). Other key volatiles, including β‐damascenone (floral), 2‐acetylpyrrole (roasted), and γ‐nonalactone (coconut‐like), also increased, further diversifying the aroma spectrum (Qi et al. 2025).

Overall, L. casei fermentation transformed the white tea's aroma from a fresh, green‐floral profile to a complex, multilayered one dominated by fruity esters, mellow higher alcohols, and characteristic organic acids. This bioconversion of intrinsic precursors into novel aroma compounds underlies the distinctive sensory identity of fermented Shoumei tea. Similar transitions from green aldehydes to alcohols, esters, and organic acids have also been reported in fermented beverage systems using integrated flavoromics approaches, highlighting the value of combining chromatographic profiling with sensory interpretation to elucidate aroma evolution (M. E. Herkenhoff et al. ²⁰²³; M. Herkenhoff et al. ²⁰²⁶).

3.5. Evolution of Key Volatile Substances During Fermentation

Based on PLS‐DA and quantitative GC–MS analyses, L. casei fermentation induced pronounced compositional and sensory shifts in the VFCs of the tea infusion. A total of 51 discriminant compounds (VIP >1) were identified (Figure 5), among which cymol, 1‐hexanol, indole, trans‐2‐hexenoic acid, 3,5‐dimethylbenzaldehyde, phenylethyl alcohol, linalool, n‐nonanol, hexanoic acid, and ethyl octanoate exhibited the highest VIP scores, confirming their pivotal roles in shaping the evolving aroma profile. Quantitative analysis of key aroma‐active compounds and their odor activity values (OAVs) (Table 3) further supported this trend. The concentrations of higher alcohols and esters rose sharply after 24 h, consistent with the PLS‐DA trajectory separating early and late fermentation phases. Phenylethyl alcohol, the most abundant VOC throughout fermentation, increased from 1072.22 µg/L at 0 h to 9057.87 µg/L at 72 h, with its OAV rising from 10.83 to 91.49, indicating a dominant contribution to the sweet, floral aroma typical of fermented teas (Y. Wang, Huang, et al. ²⁰²⁴). Similarly, linalool, responsible for floral and citrus notes, exhibited both high concentration (185.77 to 242.73 µg/L) and high odor potency (OAV = 40.46 at 72 h), confirming its persistence and sensory relevance (Qi et al. 2025).

FIGURE 5.

Key volatile components in white tea infusion fermented by Lactobacillus casei during the fermentation process. (A) PLS‐DA of volatile components in the tea infusion during Lactobacillus casei fermentation. (B) The top 10 volatile components with the highest VIP scores. (C) The bottom 10 volatile components with the lowest VIP scores.

TABLE 3.

Dynamic changes in the absolute content and OAV of some characteristic aroma compounds in tea infusion samples during different fermentation stages.

No.	VOCs	RIcalc	RIlit	Average content (µg/L)						Threshold value (µg/L)	OAV
				0 h	12 h	24 h	36 h	48 h	72 h		0 h	12 h	24 h	36 h	48 h	72 h
1	Ethyl hexanoate	1283	1245	n.d.	n.d.	n.d.	n.d.	32.51	44.81	14	n.d.	n.d.	n.d.	n.d.	2.32	3.20
2	1‐Hexanol	1396	1359	69.81	156.95	900.96	1018.37	731.14	592.25	96.77	0.72	1.62	9.31	10.52	7.56	6.12
3	Ethyl octanoate	1460	1448	n.d.	n.d.	n.d.	n.d.	21.35	28.74	12.9	n.d.	n.d.	n.d.	n.d.	1.65	2.23
4	1‐Heptanol	1481	1455	15.17	16.69	27.01	22.38	20.64	20.24	3	5.06	5.56	9.00	7.46	6.88	6.75
5	Linalool	1554	1550	185.77	175.14	161.80	174.33	190.57	242.73	6	30.96	29.19	26.97	29.06	31.76	40.46
6	1‐Nonanol	1638	1655	n.d.	0.60	1.28	1.14	1.19	1.79	50	n.d.	0.01	0.03	0.02	0.02	0.04
7	Citronellol	1722	1725	n.d.	n.d.	n.d.	0.92	3.14	6.96	40	n.d.	n.d.	n.d.	0.02	0.08	0.17
8	Nerol	1752	1755	1.62	1.47	1.74	1.95	3.22	6.04	40	0.04	0.04	0.04	0.05	0.08	0.15
9	Phenylethyl alcohol	1853	1855	1072.22	1010.40	1372.25	2079.82	4753.79	9057.87	99	10.83	10.21	13.86	21.01	48.02	91.49
10	Nerolidol	1925	1932	n.d.	n.d.	n.d.	n.d.	3.89	9.67	10	n.d.	n.d.	n.d.	n.d.	0.39	0.97

Note: “n.d.” means that this compound was not detected in the sample. RIcalc: Linear retention index calculated on the HP‐INNOWAX column using n‐alkanes (C11–C23) under identical GC–MS conditions; RIlit: Reference retention index reported in the literature or NIST database for polar (WAX) capillary columns.

Newly formed esters such as ethyl hexanoate and ethyl octanoate appeared after 36 h, with OAVs exceeding 1 at 72 h (3.20 and 2.23, respectively), signifying their perceptible contribution to the emerging fruity and creamy notes in the later stages of fermentation. Medium‐chain fatty alcohols (1‐hexanol, 1‐heptanol, and 1‐nonanol) also accumulated markedly, reflecting enhanced fatty acid metabolism. Notably, 1‐hexanol peaked at 1018.37 µg/L (OAV = 10.52 at 36 h), imparting mild fruity undertones (S. Liang et al. ²⁰²³). Meanwhile, trace amounts of monoterpene alcohols such as nerolidol and citronellol appear or slightly increase in the later stage, which may contribute positively to the overall aroma.

Collectively, the integration of PLS‐DA discrimination and OAV‐based quantification demonstrates that L. casei fermentation selectively enhances the formation of sensory‐active volatiles, especially higher alcohols, esters, and monoterpenols, while reducing or transforming native green aldehydes and ketones. This metabolic reprogramming shifts the aroma profile of Shoumei white tea from fresh and grassy to a more complex, fruity–floral and mellow character, aligning with the sensory improvements observed in the E‐nose and E‐tongue analyses. This integrated interpretation based on multivariate statistics and OAVs is consistent with recent flavoromics studies, which demonstrated that combining chromatographic profiling, chemometric analysis, and sensory evaluation provides a more comprehensive understanding of aroma‐active compounds in fermented beverages (M. E. Herkenhoff et al. ²⁰²³; M. Herkenhoff et al. ²⁰²⁶).

3.6. Changes in Flavor‐Related Enzyme Activities During Fermentation

The transformation of nonvolatile precursors into characteristic VFCs in fermented tea is largely governed by the activity of specific enzymes. To understand the biochemical mechanisms underlying the observed changes in the flavor profile, the activities of β‐glucosidase, PPO, and POD were monitored throughout the fermentation process (Figure 6). The activity of β‐glucosidase exhibited a continuous and significant increase, rising by approximately 25.5% after fermentation (Figure 6A; p < 0.05). This enzyme plays a pivotal role in aroma formation by hydrolyzing nonvolatile glycosidic precursors in tea leaves and releasing free volatile aglycones such as benzyl alcohol, (Z)‐3‐hexen‐1‐ol, and 1‐hexanol (S. Liang et al. ²⁰²³). The sustained increase in β‐glucosidase activity may result from the secretion or synthesis of the enzyme by L. casei during its proliferation in the tea infusion. This enzymatic enhancement is consistent with the enrichment of alcohols and other VOCs identified by GC–MS analysis (Figure 4D). Microbial glycosidases are well known to enhance aroma complexity in fermented beverages by liberating bound aroma compounds, with species such as Bacillus, Pseudomonas, Lactobacillus, and Microbacterium recognized as potent glycoside hydrolase producers (Lee et al. 2025; Magwaza et al. ²⁰²⁴). Similar findings have been reported in blackcurrant wine, where Penicillium species secrete β‐glucosidase capable of hydrolyzing β‐primeverosides, thereby enhancing aromatic complexity (Kim and Park 2017).

FIGURE 6.

Changes in flavor‐related enzyme activities during fermentation. (A) Dynamic changes in β‐glucosidase activity of tea infusion during fermentation. (B) Dynamic changes in polyphenol oxidase activity of tea infusion during fermentation. (C) Dynamic changes in peroxidase activity of tea infusion during fermentation. Different lowercase letters indicate significant differences (p < 0.05).

In contrast, PPO exhibited a distinct biphasic pattern, initially increasing and reaching its maximum activity at 36 h before declining sharply by 72 h (Figure 6B). As an endogenous enzyme in tea leaves, PPO catalyzes the oxidation of polyphenols to form theaflavins and thearubigins, thereby contributing to changes in both color and taste (Tang et al. 2023). The transient activation of PPO may be linked to the slight changes in the physicochemical environment during early fermentation. However, the subsequent decline is likely due to acid accumulation from organic acid production, which lowers the pH beyond the optimal range for PPO activity. This inhibition is beneficial, as it prevents excessive oxidation of catechins, thereby preserving antioxidant potential and mitigating the formation of highly astringent polymeric compounds.

The activity of POD, another key oxidative enzyme, demonstrated a pronounced and consistent decrease throughout the fermentation (Figure 6C). The rapid inactivation of POD is a characteristic outcome of lactic acid fermentation. This is attributable to two primary factors: the creation of a microaerophilic or anaerobic environment by the metabolic activity of L. casei, and the strong inhibitory effect of the low‐pH environment on enzyme structure and function. The suppression of POD activity is crucial for the sensory quality of the final product, as it minimizes the oxidative degradation of sensitive flavor compounds and bioactive molecules (Huang et al. 2025), thus helping to maintain the sensory quality and nutritional value of the white tea infusion. Similar results were also observed in other fermented plant materials (Y. Liang et al. ²⁰²⁴).

In summary, the enzymatic dynamics reveal a carefully orchestrated biochemical process. Fermentation promotes the aroma‐releasing activity of β‐glucosidase while simultaneously suppressing the oxidative activities of PPO and POD. This synergistic action allows for the development of a complex and desirable aroma profile while protecting the beverage from excessive oxidation, ultimately defining the unique sensory characteristics of the fermented white tea.

3.7. Metabolomic Analysis of Fermentation‐Induced Chemical Transformation

3.7.1. Global Metabolic Remodeling and Pathway Enrichment

To comprehensively elucidate the biochemical transformations occurring in white tea infusion during fermentation, a nontargeted metabolomic approach was employed. A total of 2039 metabolites were identified and classified into 12 superclasses, with lipids and lipid‐like molecules, benzenoids, organic acids, and organoheterocyclic compounds being the most predominant (Figure 7A). Volcano plot analysis revealed pronounced, time‐dependent metabolic remodeling (Figure 7B–D). The number of differentially abundant metabolites (DAMs) increased from 294 at 36 h (106 upregulated, 188 downregulated) to 948 at 72 h (233 upregulated, 715 downregulated), indicating a progressive intensification of the metabolic activity of L. casei as fermentation proceeded.

FIGURE 7.

Metabolome analysis of tea infusion at different fermentation stages. (A) A donut plot of metabolite categories in tea infusion samples at all fermentation time points. (B–D) The volcano plots of metabolites between three groups of tea soup samples at different fermentation stages (0 vs. 36 h, 0 vs. 72 h, and 36 vs. 72 h). (E, F) The KEGG enrichment bubble chart and classification chart of all tea infusion samples. (G) The metabolic pathway mechanism diagram of flavor‐related substances in tea infusion fermented by Lactobacillus casei.

KEGG pathway enrichment analysis (Figure 7E,F) demonstrated that these DAMs were primarily enriched in central metabolic pathways, notably amino acid metabolism (e.g., phenylalanine, tyrosine, and tryptophan biosynthesis; valine, leucine, and isoleucine biosynthesis) and carbohydrate metabolism (e.g., galactose metabolism, pentose phosphate pathway [PPP]). These findings suggest that L. casei primarily altered the abundance of key carbon and nitrogen metabolites, driving a broad reorganization of the tea's metabolic profile.

3.7.2. Mechanistic Insights Into Key Metabolic Fluxes

Fermentation by L. casei reshaped multiple metabolic pathways that jointly underpin the observed sensory transformation of the tea infusion (Figure 7G). Glucose, the main fermentable sugar, was rapidly depleted through the glycolytic (EMP) pathway, resulting in lactate accumulation and a concomitant pH decrease from 5.0 to 4.0, which contributed to the characteristic sourness of the fermented tea. Meanwhile, activation of the PPP, as evidenced by elevated ribose‐5‐phosphate, provided both nucleotide precursors and reducing power (NADPH) for biosynthetic processes such as fatty acid synthesis—a typical metabolic feature of heterofermentative lactobacilli (Y. Wang et al. ²⁰²¹). In parallel, extensive amino acid catabolism generated key aroma precursors (Zhang et al. 2025). Phenylalanine was degraded via the Ehrlich pathway into phenylethyl alcohol and phenyllactate, imparting a pleasant rose‐like floral aroma, while valine and leucine degradation yielded α‐keto acid intermediates (2‐oxoisovalerate, 2‐oxoisocaproate), which serve as precursors for branched‐chain aldehydes and fusel alcohols that contribute to malty and fruity notes (Maoz et al. 2022). Moreover, acetyl‐CoA, derived from pyruvate, acted as a central metabolic hub for de novo fatty acid biosynthesis, producing medium‐chain fatty acids (octanoic, dodecanoic, tetradecanoic acids) that serve as substrates for ester formation. Microbial esterases catalyzed the condensation of these fatty acids with alcohols such as ethanol, generating ethyl octanoate and related esters responsible for fruity and floral aromas (Pradal et al. 2025).

Overall, L. casei modulated the flavor profile of white tea through coordinated reprogramming of carbon and nitrogen metabolism. During fermentation, glucose catabolism through glycolysis and the PPP produced lactate and lowered pH, while generating reducing power for biosynthesis. Concurrently, amino acid degradation via the Ehrlich and branched‐chain amino acid pathways yielded higher alcohols and aldehydes that imparted floral, fruity, and malty notes. Pyruvate‐derived acetyl‐CoA fueled fatty acid synthesis and ester formation catalyzed by microbial esterases. These pathways transformed sugars and amino acids into organic acids, alcohols, and esters, enhancing acidity and aroma complexity in fermented tea.

4. Conclusion

This study demonstrated that L. casei fermentation substantially enhances the sensory quality of Shoumei white tea through targeted metabolic shifts. Reductions in sugars, amino acids, catechins, and flavonoids led to increased sourness and reduced bitterness, resulting in a more balanced taste profile. The notable enrichment of aroma‐active volatiles, such as phenylethyl alcohol, linalool, 1‐hexanol, and ethyl octanoate, contributed to pronounced floral and fruity notes. These changes were strongly correlated with elevated β‐glucosidase and PPO activities and associated with the activation of amino acid biosynthesis, 2‐oxocarboxylic acid metabolism, and galactose metabolism. Overall, L. casei fermentation offers a promising bioprocessing approach for developing value‐added fermented white tea beverages.

Despite the significant flavor enhancements observed, this study has several limitations that should be acknowledged. First, the current work utilized a laboratory‐scale fermentation model under a single set of processing conditions (37°C for 72 h), testing only one bacterial strain (L. casei) and one specific tea type (Shoumei). Therefore, the findings may not directly translate to large‐scale industrial product development or long‐term storage stability. Furthermore, while our conclusions rely heavily on microbial enzymatic activity, direct microbial growth kinetics and viability curves during fermentation were not presented in this study. Future research should address these gaps by exploring multistrain co‐fermentations, optimizing industrial processing parameters, evaluating shelf‐life stability, and precisely correlating microbial growth kinetics with real‐time metabolic fluxes.

Author Contributions

Yuhan Wu: writing – review and editing, writing – original draft. Jinzhi Han: writing – review and editing. Zhibin Liu: writing – review and editing. Wangxin Liu: writing – review and editing. Li Ni: writing – review and editing.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supplementary Materials: jfds71448‐sup‐0001‐SuppMat.doc