Poria powder, P. cocos polysaccharides (120 mesh, Lot No. SNT231105) was purchased from Fufeng Snout Biotechnology Co. MRS broth culture medium was purchased from Beijing Landbridge Technology Co., LTD. Sterile PBS (Cell-specific, 0.1 M, pH 7.2–7.4) was sourced from FEIMOBIO. The free radical reagent DPPH (1,1-diphenyl-2-trinitrophenylhydrazine) was sourced from Tokyo Chemical Industry Co., Ltd. (TCI). Phenanthroline, triphenylphenol, and glucose standards were purchased from Sigma. Hydrogen peroxide, hydrochloric acid, potassium ferrocyanide, and trichloroacetic acid were acquired from National Group Chemical Reagent Co., Ltd. Tris–HCl was obtained from Thermo Fisher, while ferrous sulfate was purchased from Xi’an Chemical Reagent Factory. Both sulfuric acid and phenol are of analytical reagent grade, and they were acquired from Beijing Chemical Works and Tianjin Xinyuan Chemical Co., Ltd., respectively.

L. reuteri HM-R28 (strain preservation number: CGMCC No. 23658) was isolated from the breast milk of healthy Chinese women. Its probiotic characteristics have been comprehensively evaluated by our research group. The strain was selected for this study based on its strong synergistic performance in co-fermentation with P. cocos. It is currently preserved at the National Maternal and Infant Dairy Health Engineering Technology Research Center. The strain was subcultured twice in MRS broth medium at 37°C shaking at 180 rpm. The bacterial solution was diluted with sterile PBS to a concentration of 0.500 ± 0.02 MCF/L for subsequent experiments.

P. cocos powder (25% w/w) was weighed, added to distilled water, and soaked for 12 h. This concentration corresponds to the 1:3 solid–liquid ratio for polysaccharide extraction, as determined by orthogonal optimization experiments. Subsequently, 5% cellulase was added, and the mixture was placed in a 70°C water bath for 1 h before it was set aside for later use (Cellulase was heat-inactivated prior to fermentation). Following sterilization of the MRS liquid medium at 121°C for 20 min, it was inoculated with pretreated Poria powder and diluted L. reuteri HM-R28. Thereafter, it was incubated for 24 h in a shaker incubator at 37°C and 180 rpm. This time point was selected based on preliminary time-course assays showing maximal microbial proliferation without diminishing returns. The blank control contained MRS liquid medium inoculated with the L. reuteri HM-R28 strain but without Poria. The experiment was conducted in triplicate with three parallel sets per group.

Viable bacteria in fermentation samples were quantified using the dilution gradient microplate counting method. Sample (1 mL) was incorporated to 9 mL of sterilized PBS, mixed thoroughly, and serially diluted 10-fold. Using a sterile pipette, the bacterial solution was diluted at gradients of 10⁻⁵, 10⁻⁶, and 10⁻⁷, respectively, and inoculated onto plates corresponding to each dilution. The plates were incubated invertedly at 37°C for 24–48 h, and viable bacteria counts were recorded. N1 represents the viable bacteria count in the P. cocos fermentation group, while N0 indicates the viable bacteria count in the single-strain group. The bacterial proliferation rate was calculated using the following equation (1):

The growth curve of L. reuteri HM-R28 in the fermentation broth was determined d at 600 nm using a UV spectrophotometer. The pH was determined using a digital pH meter (FE28, Mettler Toledo (Shanghai) Co., Ltd.). The total sugar content, using glucose as the standard, was analyzed using the Dubois method at 490 nm, and reducing sugars were quantified using the DNS colorimetric assay at 540 nm (11).

The overall phenolic concentration was measured using a modified Folin–Ciocalteu method (12). A mixture of 1.0 mL Folin–Ciocalteu reagent and 2 mL of the sample supernatant was incubated reacted at ambient temperature for 10 min. Subsequently 3 mL of 7.5% Na₂CO₃ solution was added, subsequently incubated in a 45°C water bath for 1.5 h. Absorbance was quantified at 765 nm. Gallic acid served as the standard. The standard curve equation was as follows: y = 0.1508x + 0.0876, with R² = 0.9924.

Total flavonoid content was quantified using a modified sodium nitrite- aluminum nitrate spectrophotometric (13). A 2 mL aliquot of the fermentation supernatant was blended with 0.5 mL of 5% NaNO₂ and maintained at ambient temperature for 6 min. Thereafter, 0.5 mL of 10% Al (NO₃)₃ was added, followed by additional incubation of 6 min. Next, 4 mL of 4% NaOH and 3 mL of 60% ethanol were added, mixed thoroughly, and allowed to settle for 15 min. Finally, the absorbance was recorded at 510 nm with a UV–Vis spectrophotometer. Rutin was used as a standard, with the standard curve equation presented as y = 0.1436 + 0.0621, R² = 0.9922.

The metaX computational pipeline (v8.2) executed supervised learning protocols through PLS-DA algorithms coupled with unsupervised PCA clustering for metabolic pattern differentiation. The t-test was applied to establish statistical significance. The fold-change (FC) was calculated based on the quantitative profiles of all experimental replicates for each metabolite in the compared cohorts. Differentially abundant metabolites were identified based on a variable importance in the projection (VIP) score > 1.0, an FC threshold of > 1.2 or FC < 0.833, and significance was defined as p < 0.05 (13). The functional and metabolic pathways of these metabolites were examined using the KEGG database. For the untargeted metabolomics analysis, the false discovery rate (FDR) correction was applied to six independent experiments, retaining only differential metabolites with a q-value < 0.05. Other experimental analyses were conducted with three independent replicates. Basic data were organized using Microsoft Excel 2010, and data analysis and visualization were performed Graph Pad Prism software. Experimental outcomes are reported as mean ± standard deviation (SD). A two-tailed independent t-test was conducted for pairwise comparisons, whereas analysis of variance (ANOVA) was applied to analyze differences across multiple groups.

To investigate the relationship between P. cocos polysaccharides and L. reuteri HM–R28, we plotted glucose content on the x-axis and optical density (OD) values on the y-axis. The resulting linear regression relationship, Y = 6.2383X + 0.0789 (R² = 0.999), was used for the determination of polysaccharide content in subsequent experiments.

The proliferation of L. reuteri HM–R28 derived from breast milk, varied with the addition of different concentrations of P. cocos powder polysaccharides dissolved in MRS broth medium. At concentrations of 1, 2.5, and 5% P. cocos polysaccharide (Figure 1), the proliferation rates of strain L. reuteri HM-R28 were 21.37, 35.96 and 54.71%, respectively, indicating a positive correlation between the P. cocos polysaccharide concentration and the proliferation rate of L. reuteri HM-R28. The water-soluble polysaccharides likely served as a key carbon and energy source for L. reuteri HM-R28, as indicated by the significantly higher viable bacterial count observed in the experimental group compared to the control. This finding is consistent with previous studies demonstrating that exogenous polysaccharides can promote the proliferation of lactic acid bacteria (LAB) by providing fermentable substrates and modulating metabolic activity (19). Similarly, supplementing Roy’s mucopramium culture medium with coix seed polysaccharides resulted in a viable bacterial count of 12.24 log10 CFU/mL, demonstrating a significant statistical advantage over the control group (9.16 log10 CFU/ml, p < 0.05) (20). A similar phenomenon has been previously reported in the bergamot juice fermentation by L. plantarum, in which sugar serves as a carbon source and an energy source for bacterial growth (21). These findings suggest that P. cocos polysaccharides may enhance biomass accumulation in L. reuteri HM-R28 by activating glycolytic pathways or inducing extracellular enzyme production.

Figures 2A,B show the changes within the total sugar and reduced sugar matrix in fermentation broth over different fermentation periods. The total sugar content demonstrated an initial reduction, later increasing, and then another decline. However, as seen in Figure 2A, the total sugar content showed a general upward trend increasing. We hypothesized that in the initial stage of fermentation, L. reuteri HM-R28 would metabolize fermentable sugars present in P. cocos, leading to a reduction in the total sugar content (11). At 18–24, the growth of L. reuteri HM-R28 may have been restricted by the accumulation of high concentrations or other inhibitory factors, causing, a subsequent decline in total sugar content. We further hypothesized that the overall increase within the total sugar during fermentation was due to the production of lactic acid by L. reuteri HM-R28. This acidification could have facilitated the breakdown of P. cocos components, releasing additional sugars under specific conditions (22). The continuous increase in reduced sugar content, as seen in Figure 2B, may be attributed to the enzymatic hydrolysis of polysaccharides into smaller sugar molecules during fermentation, thereby increasing the concentration of reducing sugars in the fermentation broth.

The survival of lactic acid bacteria is a crucial prerequisite for their probiotic functionality. Figure 2C shows that the growth rate of L. reuteri HM-R28 in the fermentation broth was the highest within the first 0–6 h of fermentation, likely due to the initial were abundance of nutrients. By the 12 h mark, the growth rate of L. reuteri HM-R28 began plateau. We speculate that L. reuteri HM-R28 underwent rapid proliferation throughout the fermentation process, and the continuous nature of its growth curve reflects a stable and healthy fermentation state (11). The pH value of the biotransformation broth exhibited overall downward trend as fermentation progressed (Figure 2D). This decline may be attributed to the metabolic characteristics of L. reuteri HM-R28. In the early stages of fermentation, L. reuteri HM-R28 produces lactic acid as it ferments total sugar (23). Additionally, throughout the fermentation process, lysine in the broth was degraded, and alkaline compounds such as histidine, were metabolized. This led to a continuous decrease in pH, further indicating active microbial fermentation and organic production.

Foods rich in total phenols and flavonoids positively influence multiple physiological functions in humans, thereby promoting overall health (24). The total phenol and flavonoid content in fruits and vegetables exist in both bound and free states. During fermentation, lactic acid bacteria (LAB) hydrolyze complex molecules into free or simpler forms, enhancing the bioavailability of phenol and flavonoids. Different strains of lactic acid bacteria secrete distinct enzymes during fermentation, each exerting unique effects on phenols and flavonoids (25). The total flavonoid levels decreased during the first 0-6 h of fermentation, rapidly, increased at 6–12 h, and showed a slight increase 12–24 h, ultimately demonstrating an overall upward trend throughout the entire fermentation process (Figure 2E). The initial decline in flavonoid levels observed during the first 0–6 h is likely attributable to microbial utilization of flavonoids as a carbon source or the enzymatic hydrolysis of glycosylated flavonoids into aglycones, which may not be fully detected by total flavonoid assays (26). In contrast, the subsequent increase between 6 and 24 h likely reflects microbial biotransformation processes, including the deglycosylation of flavonoid precursors by bacterial β-glucosidases or the action of polyphenol oxidases or both. These enzymatic activities may produce bioactive derivatives such as smaller phenolic compounds or quinones, which could have contributed to the observed rise in total flavonoid content (26, 27). Additionally, co-fermentation with Monascus Anka and Saccharomyces cerevisiae enhances flavonoid content in citrus peel, thereby improving its characteristic compounds ingredients and overall quality. Flavonoid compounds may be utilized by microorganisms as nutrients or transformed into other bioactive compounds through enzyme, leading to a reduction in their content (13, 28). We speculated that during the initial 0–6 h (Figure 2F), L. reuteri HM-R28 in the fermentation broth utilized certain phenolic precursor compounds present in P. cocos itself to convert them into phenolic compounds through enzymatic reactions (29). After 6 h, the phenol-type compounds formed in the fermentation solution may undergo further conversion into quinone compounds or participate in polymerization reactions, producing more complex polymers. These complex compounds may not be detected as total phenols, leading to an apparent decline in the measured phenolic content (25). This hypothesis was confirmed when we noted an increase in levels of lipids, polyketides, and organic heterocyclic compounds in the fermentation broth after fermentation. These findings support our hypothesis, although the effects of L. reuteri HM-R28 and P. cocos fermentation on total flavonoids and total phenolic compounds require further investigation.

Lactic acid bacteria fermentation generalizability in enhancing the antioxidant activity of medicinal and edible homology (30, 31). To evaluate the antioxidant activity of P. cocos fermented by L. reuteri HM-R28, we measured its DPPH, hydroxyl, and superoxide anion radical scavenging activities (Figures 3A–C). The removal rates of these radicals followed a similar trend across the different groups, with the fermented P. cocos exhibiting higher scavenging activity than non-fermented P. cocos and the L. reuteri HM-R28 solution. The scavenging activities of DPPH, hydroxyl, and superoxide radicals of fermented P. cocos solution were higher than those of the P. cocos solution alone by 3.77, 9.34, and 2.77 times, respectively; these values were also higher than those observed for the L. reuteri HM-R28 strain by 1.20, 1.09 and 1.32 times, respectively. Our findings align with those of a previous study, which demonstrated that the radical-neutralizing capacity of P. cocos improves significantly after fermentation (21). Polysaccharides are considered key contributors to the antioxidant properties of P. cocos. During fermentation, Lactobacillus releases β-glucosidase, an enzyme that degrades polysaccharides and may release their bioactive components (32). Generally, L. reuteri HM-R28 fermentation significantly improved both the antioxidant capacity and nutritional value of P. cocos, although its underlying antioxidant mechanisms require further investigation.

The dynamic changes in metabolites during P. cocos and L. reuteri HM-R28 co-fermentation, as well as L. reuteri HM-R28 single-strain fermentation were analyzed using non-targeted metabolomics technology. In the positive ion mode, 890 metabolites were identified (Figure 4A), including 33.07% of lipid and lipid analogs, 22.57% of organic acids and their compounds, 15.56% of organic heterocycles, and 8.75% of benzene compounds. In the negative ion mode, 490 metabolites were identified (Figure 4B), including 32.23% lipid and lipid analogs, 25.32% organic acids and their compounds, 12.28% organic heterocycles, 8.44% organic compounds with oxygen, and 8.44% nucleosides, nucleotides, and related compounds.

Figure 5A illustrates a significant classification effect between samples from the L. reuteri HM-R28 fermentation (group B) and L. reuteri HM-R28 fermented-P. cocos group (group PB) groups. Samples within each group clustered closely, indicating the co-fermentation of P. cocos and lactic acid bacteria. Unlike single fermentation metabolites of lactic acid bacteria, the total contribution rates in positive and negative ion settings reached 57.82 and 64.91%, respectively. Additionally, the Pearson correlation coefficient for QC samples was determined using the relative quantification values of metabolites (11). In the present study, the QC samples were highly concentrated in both modes (R² > 0.98), confirming the robustness of the detection process and superior data quality. Therefore, the metabolic group was suitable for biological analysis.

Orthogonal partial least squares discriminant analysis (OPLS-DA) eliminates unrelated variables, enhancing to improve model efficiency and facilitating the identification of reliable differential metabolites (11). Herein, the positive and negative ion modes reached 55.54 and 62.42%, respectively, with pronounced inter-group differences (Figure 5B) and an R² of 1.00 and Q² > 0.98, confirming the stability and reliability of the model (33). Model accuracy was evaluated using the parameters R²X, R²Y, and Q². R²X indicates the ability of the model to explain changes in the X variable, R²Y represents the fit of the model to the Y variable, and Q² reflects the model’s ability to predict outcomes. The R²X, R²Y, and Q² values range between 0 and 1, with values close to 1 indicating at better model fit. A model is considered stable and interpretable when R² exceeds Q², and Q² remains above 0, confirming the absence of overfitting (Figure 5C). Therefore, the OPLS-DA model was deemed suitable for identifying differentiated metabolites pre- and post-fermentation.

Multivariate statistical analysis was conducted using the obtained VIP of the OPLS-DA model, applying a threshold of VIP > 1.0, FC > 1.2 or FC < 0.833 and p < 0.05. To better visualize the changes in the differential metabolites, volcano plots were generated (Figure 5D). In the positive ion mode, 348 metabolites were annotated, with 156 up-regulated and 192 down-regulated. In the negative ion mode, 201 metabolites were annotated, with 98 up-regulated and 103 down-regulated. These findings indicate significant differences in the metabolic composition of the fermentation broth before and after fermentation. Based on this, a clustered heat map analysis of the differential metabolites was performed (Figure 5E), which was dominated by lipid analog, organic acid derivatives, and organic heterocyclic compounds in the positive ion mode. Specifically 89 lipid analog, 46 organic acid derivatives, and 30 organic heterocycles were identified. Additionally, 10 organic oxygen compounds, 11 organic nitrogen compounds, 15 benzene ring-type compounds, 13 nucleotide-type differential metabolites, and 5 phenylpropanoids and polyketides were detected. Lipids are generally challenging for microorganisms to metabolize during fermentation, and their content usually increases over time (34). L. buchneri uses xylose as a carbon source to produce short-chain fatty acids under anaerobic conditions, facilitating its growth and lipid accumulation (35). Oleanolic acid, the primary triterpenoid in P. cocos, showed a significant increase in the fermentation broth (p = 8.2278E-06, VIP = 1.511939271) after co-fermentation with L. reuteri HM-R28 compared with that in the reference group. The findings indicate that L. reuteri HM-R28 fermentation from breast milk could improve the medicinal and health benefits of P. cocos. However, the specific mechanism underlying the increased oleanolic acid content during fermentation require further investigation. In the negative ion mode, 63 types of lipids, 43 types of organic acids and their derivatives, 16 types of organic heterocyclic compounds, 12 types of nucleoside and nucleotide analogs, 12 types of benzene ring compounds, 11 types of organic oxidation compounds, and 5 types of phenylpropanoids and polyketones were identified.

Metabolites associated with antioxidant activity, including sedanolide, punicic acid, epigallocatechin and homocysteine, were identified in the fermentation broth of the P. cocos group fermented by L. reuteri HM-R28 (Figures 6A–I). These metabolites are effective antioxidants that enhance cellular and tissue protection. In this study, the substantial increase in the contents of sedanolide and punicic acid (Figures 6A,B) may have significantly increased hydroxyl radical scavenging rate observed in the P. cocos group fermented with strain L. reuteri HM-R28 compared with that of the supernatant of strain L. reuteri HM-R28. These findings indicate that the synergistic fermentation of P. cocos and Lactobacillus from breast milk can enhance the antioxidant effect of strain L. reuteri HM-R28. Epigallocatechin gallate, a flavonoid-like compound, protects the biofilm from free radical-induced oxidized damage, although the stability is poor (36). Lactic acid bacteria can significantly enhance the biotransformation and metabolism of catechin and catechin derivatives (37). This may also contribute to the observed decline of the epigallocatechin gallate content in the female L. reuteri HM-R28 in this study (Figure 6D). Furthermore, homocysteine undergoes auto-oxidization generating reactive oxygen species that cause cellular causing damage (38), particularly affecting human cardiovascular and nervous system health negatively. However, homocysteine levels were significantly reduced during co-fermentation (Figure 6I), thereby reducing the risk of disease. In this study, the content of most metabolites with antioxidant effects in strain L. reuteri HM-R28 fermented Poria group showed a significant increase. Additionally, antioxidant assays showed a significant increase in antioxidant capacity following fermentation with L. reuteri HM-R28. Moreover, foods rich in antioxidants generally promote human health and aid in disease prevention (39). Therefore, the synergistic fermentation of L. reuteri HM-R28 and P. cocos can develop functional, healthy food with an antioxidant effect.

Poria polysaccharides alone can stimulate the growth of Lactobacillus and Bifidobacterium (40). In this study, the contents of raffinose and other oligosaccharides in the fermentation medium of strain L. reuteri HM-R28 increased, while that of oligosaccharides in the fermentation medium of strain L. reuteri HM-R28 was very low. This suggests that strain L. reuteri HM-R28 promotes its growth and proliferation by converting pachyman into various oligosaccharides (Figures 7A–C). During the co-fermentation of Saccharomyces cerevisiae with S. cerevisiae and Hanseniaspora uvarum Yun268 respectively, the aroma of cider and wine is enhanced by adjusting the polyphenol and ethyl ester content thereby reducing sourness and bitterness (31, 41). The production of these aromatic compounds not only enhances the flavor and taste of P. cocos but also improves its health benefits, which can help diversify P. cocos products to meet the needs of various consumers.

To further elucidate the relevant metabolic channels, the KEGG pathway database was used to analyze the difference in metabolites. The results showed that the strain L. reuteri HM-R28 fermented Poria group and strain L. reuteri HM-R28 group of different metabolites participated in 41 metabolic channels (Figure 8A). The effect of L. reuteri HM-R28 derived from breast milk on the metabolic pathways of P. cocos before and after fermentation was analyzed based on the p-value and influence value of each metabolic pathway (Figure 8B). Key metabolic pathways identified included purine, arachidonic acid metabolism, lysine degradation, β-alanine, and sphingolipid metabolism, in which the number of metabolites associated with purine metabolism, arachidonic acid metabolism, pyrimidine metabolism, and histidine metabolism increased significantly, which may be a key pathway for the synergistic effect of L. reuteri HM-R28 and P. cocos in vivo.

The metabolic alterations in the fermentation broth of P. cocos are regulated by multiple substances and reactions. In this study, the purine metabolic pathway was significantly different (p = 0.0165), with 17 different metabolites identified in this pathway (Figure 8C). Key precursors of uric acid synthesis, including guanine, xanthine, hypoxanthine, and guanylic acid, were significantly reduced (Figures 9A–D), which may explain the observed decrease in uric acid content. While this study presents in vitro evidence demonstrating a reduction in purine metabolites following fermentation of P. cocos with L. reuteri HM-R28, the potential suitability of this product for patients with hyperuricemia is further supported by in vivo findings. For instance, a recent study (42) reported that a Gardenia jasminoides–P. cocos extract (GPE) significantly lowered serum uric acid levels in hyperuricemic rats by inhibiting xanthine oxidase activity and mitigating oxidative stress and inflammation. GPE treatment also improved renal function indicators (such as BUN and creatinine) and decreased pro-inflammatory cytokines (including TNF-α and IL-6), offering preclinical validation of the anti-hyperuricemic properties of Poria-based formulations. Although our work is limited to in vitro metabolic analysis, these in vivo results provide supporting evidence for the potential dietary application of P. cocos fermented with L. reuteri HM-R28 as a low-purine functional food for individuals with hyperuricemia.