Abstract Mounting evidence suggests that dietary polyphenols exert health benefits partly through their favorable interactions with gut bacteria. However, little is known about polyphenol’s metabolic regulatory effects towards individual bacteria at the molecular level. Ellagic acid (EA), a polyphenol abundantly present in plant-based foods, was found to exhibit prebiotic properties through differential interactions with probiotic-like bacteria, including the EA-to-urolithin converting species such as Gordonibacter urolithinfaciens (G. uro). This study aimed to investigate the crosstalk between EA and EA-responsive beneficial bacteria, including both conventional and next-generation probiotics originating from the human gut, and to understand the underlying mechanism by which EA exerts prebiotic activities in vitro. The influence of EA and urolithins on probiotic bacteria was investigated at the levels of fecal microbiota and individual strains via anaerobic culturomics and metabolomics approaches. Results indicate that dietary-level EA favorably regulated gut microbial composition through the enrichment of probiotic genera (e.g., Bifidobacterium and Akkermansia) in vitro. Regarding individual bacteria, EA supplementation promoted the growth of Lacticaseibacillus rhamnosus GG and Bifidobacterium infantis. Integrated targeted and untargeted metabolomic analyses of intracellular and extracellular metabolites revealed that EA/urolithins modulated metabolic pathways associated with amino acid metabolism, energy production and oxidative stress. Furthermore, G. uro exhibited strong EA uptake ability, facilitating the urolithin bioconversion and cellular accumulation in a dose-dependent manner. Overall, this study provides in-depth understanding on how dietary polyphenols with prebiotic properties regulate the growth and metabolic functions of probiotic-like bacteria. Graphical abstract [29]Image, graphical abstract [30]Open in a new tab 1. Introduction The intricate ecosystem of human digestive tract encompasses millions of microorganisms that interact closely with dietary components and the host organisms. This intimate relationship underscores the significance of gut microbiota in modulating the health outcomes of dietary components through various mechanisms. Polyphenols, a type of plant secondary metabolites abundantly present in plant-based diets, represent an important dietary factor that intensively interact with diverse gut bacteria ([31]Kolodziejczyk et al., 2019). Although intestinal absorption and bioavailability of most polyphenols are limited, they exhibit multiple health benefits, which are partly mediated through their favorable interactions with gut bacteria ([32]Gu et al., 2019; [33]Yang et al., 2024). For example, polyphenols and phenolic metabolites possess potent antioxidant activity and thus can protect the predominant bacterial population, especially anaerobes, from oxidative stress ([34]Heinonen et al., 2003). In addition, polyphenol–microbiota interactions can lead to the production of host-beneficial bioactive metabolites, such as short-chain fatty acids (SCFAs), neurotransmitters, and bioactive ring-fission phenolic metabolites ([35]Rodríguez-Daza et al., 2021). In this context, polyphenols have been conceptualized, akin to dietary fibers, as a distinct category of prebiotics–substrates selectively utilized by microorganisms for health benefits ([36]Gibson et al., 2017; [37]Rodríguez-Daza et al., 2021). Earlier studies, including ours, have identified ellagic acid (EA) as a prebiotic-like polyphenol in view of its favorable modulatory activity towards gut microbiota composition and growth-promoting effects on beneficial bacterial groups, including conventional and novel probiotics ([38]Xie et al., 2025; [39]Yang et al., 2024). EA, commonly found in berry fruits, edible nuts and herbs, is one of the most commonly consumed polyphenols and yields multiple health benefits ([40]Kang et al., 2016). Due to its extremely low intestinal absorptivity, the health benefits of EA are believed to be closely linked to its two-way interactions with the gut microbiota, i.e., modulating the microbial compositions and being bioconverted by the microbiota. The most prominent gut-derived metabolites of EA are urolithins, the hydroxylated 6H-dibenzo[b,d]pyran-6-one derivatives, which are 25–80 fold more bioavailable than EA and have longer half-lives ([41]Cortes-Martin et al., 2020). Supplementation with EA at dietary (0.4 – 12 mg/day) or higher levels has been shown to remarkably promote the growth of many beneficial bacteria, including Akkermansia, Lacticaseibacillus, and Bifidobacterium subspecies, while inhibiting opportunistic pathogens in rodents ([42]Duan et al., 2022; [43]Yin et al., 2023). Clinical investigations demonstrate that the health benefits of consumption of EA-rich pomegranate extract are associated with modulation of gut microbiota composition and function ([44]Henning et al., 2017a; [45]Li et al., 2015). Collectively, these findings strongly suggest the prebiotic potential of EA. In addition, administration of an EA-to-urolithin converting bacterium, Gordonibacter urolithinfaciens (G. uro), led to a systemic enhancement of EA and urolithin bioavailability, with EA, urolithin C, and urolithin A identified as the major metabolites ([46]Yang et al., 2024). Our preliminary study indicated that G. uro possesses notable probiotic potential, including improving obesity phenotype and alleviating systemic inflammation in mice (data not shown). Such effects were further strengthened in combination with EA supplementation and active EA-to-urolithin conversion was observed. However, little is known about the interactions between polyphenols with prebiotic properties (e.g., EA) and individual beneficial gut bacteria, i.e., probiotic-like bacteria, at the molecular level. Although metabolomics has been widely applied as a powerful tool for investigating bacterial responses to environmental factors, including exposure to dietary polyphenols such as EA, prior studies largely focused on host metabolome rather than the metabolic shifts within bacterial cells, thereby constraining the understanding of their interactions and the associated health impact. To fill this research gap, in this work, we chose EA as a representative dietary polyphenol, together with its bacterial metabolites, i.e., urolithins, for assessing the bilateral interplay between phenolic compounds and probiotic-like bacteria originating from human gut via culturomic and metabolomic approaches (see study design in [47]Fig. 1). Fig. 1. [48]Fig 1 [49]Open in a new tab Schematic illustration of the study design. To the best of our knowledge, this is the first study on the modes of action by which EA and urolithins regulate the growth of beneficial gut bacteria and their intracellular metabolic networks, alongside the modulation of bacterial metabolic activities indicative of prebiotic effects. In addition, the method for extracting bacterial intracellular metabolites was optimized to maximize the coverage of metabolites while attenuating matrix effects in untargeted metabolomic analyses. 2. Materials and methods 2.1. Chemicals and reagents Phenolic compounds (all HPLC-grade, purity > 98 %) were purchased from various sources: EA from Macklin (Shanghai, China); urolithin A (UroA) and 2-ethylbutyric acid from Aladdin (Shanghai, China); trans-cinnamic-d[7] acid (TCA) from Sigma-Aldrich (St Louis, MO, USA); and 6,7-dihyroxycoumarin (6,7-DHC) from Yuanye Bio-Technology (Shanghai, China). Urolithin C (UroC), 8-methyl urolithin A, and 8,9-dimethyl urolithin C were kindly provided by Prof. Man-kin Wong (State Key Laboratory of Chemical Biology and Drug Discovery, The Hong Kong Polytechnic University). Chemicals for preparing bacterial culture media were obtained from Macklin and Aladdin unless otherwise specified. Solvents for liquid chromatography–mass spectrometry (LC-MS) analyses, including methanol (Duksan, Ansansi, Korea), acetonitrile (ACN), formic acid (Macklin), and ethyl acetate (Anaqua, Cleveland, OH, USA), were all of HPLC grade or above unless otherwise specified. 2.2. Bacterial strains and culture conditions Gordonibacter urolithinfaciens DSM 27213 (G. uro; DSMZ, Braunschweig, Germany) was cultured anaerobically in a modified peptone yeast glucose (PYG) broth at 37 °C ([50]Yang et al., 2024). Bifidobacterium infantis CICC 6069 (CICC, Beijing, China) was cultured anaerobically in De Man–Rogosa–Sharpe (MRS) broth (Haibo, Qingdao, China) supplemented with 0.5 % l-cysteine (Merck & Co., Inc., NJ, USA). Lacticaseibacillus rhamnosus GG (LGG; Culturelle Kids Probiotics, Cromwell, CT, USA) was cultured anaerobically in MRS broth at 37 °C. Akkermansia muciniphila BAA-835 (ATCC, VA, USA) was cultured anaerobically in brain–heart infusion broth (BD Bioscience, San Jose, CA, USA) supplemented with 0.5 % porcine mucin and 0.05 % l-cysteine. The basal medium for fecal microbiota culture was prepared according to our previous work ([51]Yang et al., 2024). Following bacterial activation, aliquots of medium (20 mL), with or without phenolic compound supplementation, were inoculated with individual bacterial strains (1 %, v/v) and incubated anaerobically in anaerobic jars at 37 °C prior to subsequent experiments. 2.3. In vitro anaerobic fermentation 2.3.1. Colon-simulated fermentation of EA by fecal microbiota For fecal sample collection, five wild-type male C57BL/6 J mice (ca. 16 weeks) were obtained from the Laboratory Animal Services Centre, The Chinese University of Hong Kong. All procedures involving animals in this study were approved by the Hong Kong Polytechnic University Animal Subjects Ethics Sub-committee (ASESC project number: 20–21/173-ABCT-R-NSFC). Fecal samples were collected from live animals by gently massaging their abdomens and placing them in clean cages for 10 min without sacrifice. Fecal pellets were diluted in saline (1:9, w/v) and centrifuged at 500 × g for 5 min to obtain a fecal suspension. EA was dissolved in dimethyl sulfoxide (DMSO) and diluted in culture medium (final DMSO concentration: 1 % v/v). Basal medium containing G. uro and either 30 μM of EA or 1 % DMSO was inoculated with either fecal suspension (10 % v/v) or autoclaved fecal suspension (10 % v/v). All samples were processed inside an anaerobic chamber (10 % CO[2], 10 % H[2], and 80 % N[2]) at 37 °C. Inoculated samples were placed in an anaerobic jar with Oxoid AnaeroGen sachets (Thermo Scientific Waltham, MA, USA) and incubated at 37 °C with continuous shaking at 200 × g for 5 days. Samples were aseptically collected from the jar at 0 h and on day 5 for subsequent analyses. 2.3.2. Fermentation of EA and urolithins with individual bacteria EA and urolithins were dissolved in DMSO and subsequently diluted in culture medium (final DMSO concentration: 1 % v/v). For fermentation, an aliquot of medium (20 mL) supplemented with EA or urolithins was inoculated with each bacterial strain (1 % v/v) and incubated anaerobically at 37 °C. The duration of fermentation was 5 days for G. uro in PYG broth supplemented with EA (10, 30, or 90 μM), UroC (30 μM), or UroA (30 μM), and 2 days for B. infantis, 24 h for LGG, and 3 days for A. muciniphila, using their respective culture medium supplemented with EA (10 μM) or UroC (30 μM). Culture medium containing 1 % DMSO served as the blank control. Samples were harvested at the indicated endpoints and centrifuged at 4500 × g for 5 min. The supernatants were collected for extracellular metabolite analyses, while the cell pellets were reserved for intracellular metabolite analyses. 2.4. Analysis of bacterial groups by real-time quantitative PCR Cell pellets from fermented and non-fermented fecal samples were collected for extracting total genomic DNA using a Stool DNA Isolation Kit (Norgen, Thorold, Canada). Real-time quantitative PCR (qPCR) was then conducted for semi-quantification of the main bacterial groups in the fecal microbiota as described in our previous work ([52]Yang et al., 2024). The data were analyzed using the [ΔΔ]CT method to determine the changes in bacterial levels between the fermented and non-fermented samples. 2.5. Analysis of bacterial intracellular metabolites 2.5.1. Extraction of intracellular metabolites The bacterial cell pellets were sequentially washed twice with cold phosphate-buffered saline (PBS) and ice-cold double-distilled water (DDW) before being suspended in 600 μL of pre-chilled methanol containing trans-cinnamic-d[7] acid (TCA) and 6,7-dihyroxycoumarin (6,7-DHC) as internal standards (ISs). Following brief vortex mixing, each suspension was immediately flash-frozen in liquid nitrogen for 30 s, then thawed at room temperature and vigorously vortexed for 30 s to lyse the cells. An aliquot of pre-chilled chloroform (150 μL) was added to cell suspension, and the freeze–thaw process was repeated as described above. DDW (450 μL) was then added to the mixture, followed by vigorous vortex mixing for 30 s. The mixture was centrifuged at 14,000 × g and 4 °C for 1 min, after which the upper organic layer was collected, centrifuged at 17,000 × g and 4 °C for 10 min, and used for LC-MS/MS analysis. 2.5.2. Untargeted profiling of intracellular metabolites The extracted intracellular metabolites were analyzed using an Agilent 1290 Infinity II ultra-HPLC (UHPLC) system coupled with an Agilent 6540 Quadrupole-TOF Mass Spectrometer with an electrospray ionization (ESI) source (Agilent Technologies, Santa Clara, CA, USA). Chromatographic separation was achieved using a Waters Acquity HSS T3 column (2.1 mm  × 100 mm, 1.7 µm; Milford, MA) with a Waters VanGuard Acquity HSS T3 Pre-column (2.1 mm × 5 mm, 1.8 µm). The binary mobile phase system consisted of (A) water with 0.1 % formic acid and (B) ACN with 0.1 % formic acid. The gradient elution (noted as B %), at a flow rate of 0.3 mL/min, was programed as follows: 1 % from 0 min to 1 min, 99 % at 10 min, 99 % from 10 min to 13 min, return to the initial condition in 0.5 min, and equilibrate for 2.5 min prior to the next injection. The column was thermostatted at 40 °C, and the autosampler was maintained at room temperature. The injection volumes for both positive and negative modes were 10 μL. Metabolites were identified by comparing the data with the Human Metabolome Database (HMDB), METLIN, and ChemSpider databases using Progenesis QI software, version 3.0 (Nonlinear Dynamics, Newcastle, UK). 2.5.3. Targeted analysis of intracellular amino acids The bacterial intracellular amino acids were derivatized using a Kairos Amino Acid Kit (Waters, Milford, MA, USA), followed by quantification via UHPLC-triple quadrupole (QqQ)-MS/MS. Amino acids were analyzed on an Agilent 1290 Infinity II UHPLC system coupled with an Agilent 6460 Triple Quadrupole Mass Spectrometer with an ESI source (Agilent Technologies Inc.). Chromatographic separation was achieved using a Waters CORTECS UPLC C18 column (2.1 mm × 150 mm, 1.6 μm) with a Waters CORTECS C18 VanGuard Pre-column (2.1 mm × 150 mm, 1.6 μm). The binary mobile phase system consisted of (A) water with 0.1 % formic acid and (B) ACN with 0.1 % formic acid. The gradient elution (noted as B %), at a flow rate of 0.3 mL/min, was as follows: 2 % from 0 min to 1.5 min, 9 % at 4 min, 11 % at 5 min, 13 % at 18 min, 95 % from 18.5 min to 20.5 min, return to the initial condition in 0.1 min, and equilibrate for 4.4 min prior to the next injection. The column was thermostatted at 55 °C, and the autosampler was maintained at room temperature. The injection volume was 2 µL. 2.6. Analysis of SCFAs in bacterial broth Short-chain fatty acids (SCFAs) were extracted from fermented bacterial broths in the presence or absence of fecal microbiota and analyzed by gas chromatography-flame ionization detection (GC-FID). Briefly, an aliquot of fermented sample (500 μL) was spiked with 2-ethylbutyric acid as an IS to a final concentration of 10 μg/mL, followed by acidification with 6 M HCl to pH 2.0. The acidified sample was extracted by adding 500 μL of ethyl acetate, vigorously shaken for 1 min, and sonicated in ice water for 5 min. The mixture was then centrifuged at 8000 × g and 4 °C for 5 min, and the top layer was collected into an autosampler vial. The sample was extracted with ethyl acetate twice, and all the top layers of each sample were pooled for subsequent GC-FID analyses. Gas chromatography analysis was performed on an Agilent 7980B GC instrument equipped with a flame ionization detector, split injector, and a DB-FFAP 123–3232 column (30 m × 0.32 mm × 0.25 μm; Agilent Technologies Inc.). The injector and detector temperatures were both 250 °C, with the column temperature programmed to increase from 90 °C for 5 min to 200 °C for 3 min at a rate of 10 °C/min. Nitrogen served as the carrier gas, at a flow rate of 2 mL/min, and samples (1 μL) were injected in split mode (1:1 split). Peak areas were recorded, and all subsequent data analyses were completed using ChemStation Software A.10.02 (Agilent Technologies Inc.). 2.7. Analysis of phenolic compounds in fermentation broth and bacterial cells Extracellular EA and urolithins were extracted from the fermented broth following the method based on [53]Zhao et al. (2018) with modifications. In brief, an aliquot of fermented broth (800 µL) was spiked with two ISs (TCA and 6,7-DHC) to a final concentration of 100 ng/mL. The sample was extracted three times with ethyl acetate containing 1.5 % formic acid. The mixture was centrifuged at 8000 × g and 4 °C for 10 min, and the upper organic phase was collected in a glass tube. The pooled top layer from all extractions was dried under nitrogen, reconstituted in 80 % methanol containing 0.1 % formic acid, and centrifuged at 17,000 × g and 4 °C for 10 min before LC-MS/MS analysis. Phenolic compounds from fermentation broth and intracellular metabolites (prepared as stated in [54]Section 2.5.1) were analyzed on an Agilent 1290 Infinity II UHPLC system coupled with an Agilent 6460 Triple Quadrupole Mass Spectrometer with an ESI source (Agilent Technologies Inc.). Chromatographic separation was achieved using a Waters Acquity UPLC BEH C18 column (2.1 mm × 50 mm, 1.7 µm) with a Waters VanGuard Acquity C18 guard column (2.1 mm × 5 mm, 1.7 µm). The UHPLC instrumentation and MS parameters were set according our previous study ([55]Yang et al., 2024). The MS/MS parameters of the target metabolites and ISs are detailed in Supplementary Table 1. The precursor-to-product ion transitions and retention times were compared with those of the authentic standards for compound identification. Quantitation was achieved with calibration curves established using the analyte-to-IS peak area ratio of the quantifier ions. 2.8. Statistical analysis All experiments were carried out at least in triplicate. MassHunter Quantitative Analysis B.06.00 (Agilent Technologies Inc.) and GraphPad Prism 9 software (GraphPad Software Inc., La Jolla, CA, USA) were used for raw data processing and statistical analyses. The data were subjected to a two-tailed Student’s t-test, with a p value < 0.05 considered statistically significant between groups. Principal component analysis (PCA) visualization was generated using MetaboAnalyst 6.0 ([56]https://www.metaboanalyst.ca/). Heatmap visualization was performed using the R package ComplexHeatmap v2.12.1 ([57]https://jokergoo.github.io/ComplexHeatmap-reference/book/). 3. Results 3.1. Influence of EA on fecal microbiota composition To investigate the effects of EA supplementation on the fecal microbiota, fold changes in the major fecal bacterial groups from mouse fecal samples were measured in comparison to baseline samples ([58]Fig. 2A). Following a fermentation period of 3 days, the abundances of G. uro increased to 62.0-fold and 78.9-fold relative to baseline level in the EA(-) and EA(+) groups, respectively. Notably, the increase in G. uro in the EA(+) group was considerably higher than that in the EA(-) group. Additionally, a moderate but significant expansion of the Bifidobacterium population was observed in fermented samples relative to the baseline. The abundance of Bifidobacterium in the EA(+) group was significantly higher than that in the EA(-) group, exhibiting 4.81-fold and 2.60-fold relative to the baseline, respectively (p < 0.05). Furthermore, Akkermansia population in EA(+) group was significantly elevated than that in EA(-) group. In contrast, the relative abundances of other genera were decreased to varying degrees after fermentation, while no significant change was noted in the relative abundance of Lacticaseibacillus. However, we also observed substantial increases in Escherichia populations in both EA(+) and EA(-) groups following the 3-day fermentation. Overall, the supplementation of EA led to the enrichment of the G. uro, Bifidobacterium and Akkermansia populations after in vitro fermentation, while limiting the excess proliferation of other bacterial groups of interest. Fig. 2. [59]Fig 2 [60]Open in a new tab Changes in the main gut bacterial groups and G. uro abundance after 3 days of in vitro fermentation with (yellow bars) or without (blue bars) EA supplementation (A). *, p < 0.05, compared between EA(-) and EA(+) groups by student’s t-test. The scale of the right Y-axis shows the abundances of the Escherichia genus. Data are expressed as mean ± SEM (n = 4). The bacterial cell density (B - E) and viable bacterial cell number (F - I) of different bacteria after co-incubation with EA or urolithins. *, p < 0.05, **, p < 0.01 and ***, p < 0.001, compared with control by student’s t-test. Data are expressed as mean ± SEM (n = 3 – 5). 3.2. Influence of EA and urolithins on individual bacterial growth The effects of varying dietary levels of EA and urolithins on the growth of individual bacteria strains during in vitro fermentation were assessed after incubating individual bacteria in their corresponding culture media supplemented with or without EA or urolithin ([61]Fig. 2B–I and Supplementary Fig. 1). Cell density was recorded regularly to monitor bacteria growth ([62]Fig. 2B–E). Our preliminary study suggested that supplementation with EA and urolithins at concentrations exceeding dietary levels (>30 μM) inhibited bacterial growth. Consequently, we applied moderate dietary levels of EA and urolithins (10–30 μM) in this study. The addition of 10 μM EA had little effect on G. uro cell density, while higher concentrations of EA resulted in a significant enhancement of bacterial density (p < 0.001) relative to the control (Supplementary Fig. 1A). For other bacterial species, significant increases (p < 0.01) in LGG and B. infantis were observed in the 10 μM EA group ([63]Fig. 2A). Additionally, 30 μM UroC promoted the growth of LGG and A. muciniphila (p < 0.001) ([64]Fig. 2B-E). In addition to cell density measurement, viable cell count assay was conducted ([65]Fig. 2F–I). In contrast to the changes in bacterial density, most bacterial strains examined did not change significantly after incubation with or without EA and UroC, with the exception of LGG, which exhibited a significantly enhanced growth when supplemented with 10 μM EA. These findings indicate that dietary levels of EA and urolithins may facilitate the proliferation of some probiotic-like bacterial growth in vitro. 3.3. Metabolism of EA by G. uro in the presence or absence of fecal microbiota To evaluate the influence of the fecal microbiota on the metabolism of EA beyond the activities of G. uro, EA was incorporated to basal media, which was subsequently inoculated with either an active or heat-inactivated fecal microbial suspension and allowed to ferment. The concentrations of EA and urolithins in the fermented samples are shown in [66]Table 1. Following fermentation, the concentrations of EA in the samples inoculated with live fecal microbiota (FM(+) group: 4.4 ± 1.3 μM) were markedly lower than those in samples inoculated with heat-inactivated microbiota (FM(-)13.3 ± 3.6 μM). Reflecting the trend observed in EA metabolism, the production of urolithins was significantly higher in the FM(+) group than in the FM(-) group. Specifically, the concentration of UroC in the FM(+) group was two-fold greater than that observed in the FM(-) group (p < 0.05). UroA was exclusively detected in the FM(+) group at a concentration of 0.19 ± 0.01 μM. Furthermore, other intermediate urolithins, such as UroM6 and UroD, were detected at lower concentrations in the FM(+) group (0.002 ± 0.000 μM) compared to the FM(-) group (0.13 ± 0.06 μM). These results demonstrate the synergism between G. uro and other fecal bacteria on EA metabolism. Table 1. Concentrations of bacteria metabolites in fecal microbiota-fermented broth supplemented with EA and G. uro. Treatments __________________________________________________________________ Metabolites __________________________________________________________________ EA FM Phenolics (μmol/L) __________________________________________________________________ SCFAs (μg/mL) __________________________________________________________________ EA UroM6 & UroD UroC UroA Acetic acid Propionic acid Isobutyric acid Butyric acid Isovaleric acid Valeric acid + + 4.4 ± 1.3 0.002 ± 0.000 19.1 ± 2.1 0.19 ± 0.01 821 ± 21 103.1 ± 8.0 2.77 ± 0.11 270 ± 8 25.9 ± 0.5 201 ± 17 + – 13.3 ± 3.6 0.13 ± 0.06 9.81 ± 1.75[67]^* n.d.[68]^⁎⁎⁎ 161 ± 13[69]^⁎⁎⁎ n.d. n.d. n.d. n.d. 69.2 ± 1.8[70]^⁎⁎⁎ – + n.d. 747 ± 51 59.2 ± 9.7[71]^* 0.68 ± 0.39[72]^⁎⁎ 266 ± 7 20.9 ± 1.1[73]^⁎⁎ 161 ± 41 [74]Open in a new tab Data are expressed as mean ± SEM (n = 4). ^⁎ , p < 0.05. ^⁎⁎ , p < 0.01. ^⁎⁎⁎ , p < 0.001, compared with the group supplemented with EA and FM by student’s t-test. Abbreviations: FM, fecal microbiota; SCFA, short-chain fatty acid; EA, ellagic acid; UroM6, urolithin M6; UroD, urolithin D; UroC, urolithin C; UroA, urolithin A. n.d., not detected. Other urolithin metabolites including urolithin M7, 8,9-dimethyl urolithin C and 8-methyl urolithin A were also surveyed but not detected. 3.4. Influence of EA on the production of SCFAs by fecal microbiota In addition to the EA-to-urolithin conversion, we evaluated the SCFA levels in fecal samples fermented with or without EA to further elucidate the effect of EA on the metabolic activities of gut microbiota ([75]Table 1). Supplementation with EA significantly enhanced propionic acid production in the fermented samples compared with the those lacking EA (p < 0.05). In addition, the production of isobutyric acid and isovaleric acid was also significantly increased in samples incubated with EA (fecal microbiota + EA + G. uro) compared with those without EA supplementation (p < 0.01). Slight increases in the levels of acetic acid, butyric acid, and valeric acid were also observed when incubated with EA. Collectively, these results suggest that EA supplementation remarkably enhances SCFA production, with a significant effect observed for propionic acid, isobutyric acid and isovaleric acid. 3.5. Metabolism of EA or urolithins by individual bacteria To explore the metabolism of EA and urolithins by individual bacterial of interest, the concentrations of EA and urolithins in bacteria cells and fermented broth were quantified using UHPLC-QqQ-MS/MS ([76]Table 2). Analyses of intracellular metabolites revealed a dose-dependent decrease in EA concentrations upon incubation with three different concentrations of EA, which correlated with the pronounced increases in UroC levels. Notably, the intermediate urolithins M6 and D were detected only in broth supplemented with 90 μM EA. For the fermented broth, lower levels of EA were detected in the broth following fermentation with other bacteria compared to G. uro. Regarding UroC, the lowest concentration was observed in the broth inoculated with A. muciniphila (7.47 ± 0.34 μM) when supplemented with 30 μM UroC, followed by G. uro (11.3 ± 1.7 μM). Higher concentrations of UroC were observed in the broths inoculated with B. infantis and LGG (21.2 ± 0.3 μM and 17.3 ± 2.2 μM, respectively). Interestingly, varying concentrations of 8,9-dimethyl UroC (DMeUroC, up to 0.81 ± 0.24 μM) were detected in UroC-supplemented broth fermented with different bacterial strains, a finding reported for the first time. Following UroA exposure, 8-methyl UroA (MeUroA) was also detected at a concentration of 0.43 ± 0.01 μM in the fermented broth inoculated with G. uro. In summary, the bacterial strains evaluated exhibited diverse capabilities in metabolizing EA and urolithins. Table 2. Concentrations of EA and urolithins accumulated in bacterial cells (intracellular) or fermented broths (extracellular). Bacteria Group Analytes __________________________________________________________________ EA UroM6 & UroD UroC UroA DMeUroC MeUroA Intracellular metabolites (nmol/g) G. uro 10μM EA 3.43 ± 0.12 n.d. 0.18 ± 0.04 n.d. n.d. n.d. 30μM EA 34.1 ± 2.9 n.d. 0.57 ± 0.05 n.d. n.d. n.d. 90μM EA 15.7 ± 2.4 0.16 ± 0.01 2.63 ± 0.05 n.d. n.d. n.d. 30μM UC n.d. n.d. 60.9 ± 6.7 n.d. n.d. n.d. 30μM UA n.d. n.d. n.d. 315 ± 110 n.d. 0.27 ± 0.02 A. muc 10μM EA n.d. n.d. n.d. n.d. n.d. n.d. 30μM UC n.d. n.d. 64.2 ± 4.60 n.d. n.d. n.d. LGG 10μM EA n.d. n.d. n.d. n.d. n.d. n.d. 30μM UC n.d. n.d. 0.14 ± 0.01 n.d. n.d. n.d. B. inf 10μM EA n.d. n.d. n.d. n.d. n.d. n.d. 30μM UC n.d. n.d. 6.82 ± 0.62 n.d. n.d. n.d. Metabolites in broth (μmol/L) G. uro 10μM EA 10.3 ± 0.7 n.d. 0.002 ± 0.000 n.d. n.d. n.d. 30μM EA 9.24 ± 0.32 n.d. 0.01 ± 0.00 n.d. n.d. n.d. 90μM EA 30.1 ± 0.4 0.34 ± 0.01 1.00 ± 0.07 n.d. n.d. n.d. 30μM UC n.d. n.d. 11.3 ± 1.7 n.d. 0.002 ± 0.000 n.d. 30μM UA n.d. n.d. n.d. 23.5 ± 0.2 n.d. 0.43 ± 0.01 A. muc 10μM EA 0.14 ± 0.02 n.d. n.d. n.d. n.d. n.d. 30μM UC n.d. n.d. 7.47 ± 0.34 n.d. 0.23 ± 0.00 n.d. LGG 10μM EA 0.37 ± 0.13 n.d. n.d. n.d. n.d. n.d. 30μM UC n.d. n.d. 17.3 ± 2.2 n.d. 0.81 ± 0.24 n.d. B. inf 10μM EA 0.51 ± 0.26 n.d. n.d. n.d. n.d. n.d. 30μM UC n.d. n.d. 21.2 ± 0.3 n.d. n.d. n.d. [77]Open in a new tab Data are expressed as mean ± SEM (n = 3 - 5). Abbreviations: n.d., not detected. Please refer to [78]Table 1 for the analyte abbreviations. 3.6. Influence of EA and urolithins on the bacterial intracellular metabolome To assess the influence of EA and UroC exposures on bacterial growth and metabolic activities, intracellular metabolites from individual bacteria strains were evaluated using integrated untargeted and targeted metabolomics. PCA score plots indicated that for G. uro and A. muciniphila, the groups treated with phenolic compounds (EA/UroC) were clustered closely together and exhibited clearly separation from the control group, suggesting distinct metabolome characteristics between phenolic-treated and control groups. In contrast, for LGG and B. infantis, the EA-treated and control groups displayed more similar metabolite profiles, which were divergent from the UroC-treated group (Supplementary Fig. 2). Global metabolite profiling showed that, following fermentation with EA, the production of the most significantly differential metabolites (p < 0.05, control vs. EA-treated group) was reduced in G. uro and A. muciniphila ([79]Fig. 3A & C), while an increasing trend in their production was observed in LGG and B. infantis ([80]Fig. 3B & D). In contrast, UroC exposure remarkably reduced the production of the significantly different metabolites (p < 0.05, control vs. UroC-treated group) across all four strains. Of note, those differentially abundant metabolites mainly belonged to three categories: amino acids, phenolics (xenobiotics), and nucleotides and their derivatives (endogenous metabolites). Regarding nucleotides and their derivatives (e.g., flavin mononucleotide [FMN] and flavin adenine dinucleotide [FAD]), significantly increased production was only observed in LGG after exposure to EA, with no significant differences between-group in B. infantis. Additionally, a clear reduced production of nucleotides and their derivatives were clearly observed after supplementation with EA/UroC. Fig. 3. [81]Fig 3 [82]Open in a new tab Heatmaps showing the abundances of differential intracellular metabolites (log-transformed) in four bacteria strains (A-D) after co-incubation with EA or UroC. Data were integrated following targeted and untargeted metabolomic analyses (n = 3 – 5). As we observed marked changes in amino acid metabolism and tentative accumulation of EA and urolithins, targeted metabolomic analyses were conducted to quantify amino acids and phenolic compounds in bacterial cells (Supplementary Table 2). Following EA exposure, a total of four amino acids in LGG and five amino acids in A. muciniphila were significant altered (p < 0.05). Notably, lysine levels varied in both bacterial strains, while no significant changes were observed for the other amino acids. Regarding UroC supplementation, seven amino acids were significantly altered in G. uro and A. muciniphila, particularly histidine, lysine, methionine and proline. In contrast, no significant changes in amino acid concentrations were observed in LGG in response to UroC supplementation. Regarding the phenolic metabolites, intracellular and extracellular levels of EA and urolithins following fermentation were quantified by UPLC-QqQ-MS/MS ([83]Table 2). In general, EA was only detected in G. uro cells, whereas UroC was found at quantifiable levels in all four bacteria. Notably, no dose-dependent accumulation of EA was observed in G. uro cells following exposure to 10–90 μM EA. Conversely, UroC production by G. uro increased dose-dependently in response to elevated EA concentrations in the broth, leading to higher intracellular levels of UroC. In addition, several urolithin intermediates, specifically UroM6 and UroD, were detected in G. uro cultured in the presence of high concentrations of EA. The highest accumulating levels of UroC were observed in A. muciniphila (64.2 ± 4.6 nmol/g), closely followed by G. uro (60.9 ± 6.8 nmol/g). In contrast, lower intracellular UroC levels were detected in B. infantis and LGG. Furthermore, a substantial amount of UroA (315 ± 110 nmol/g) and trace amounts of methylated UroA (0.27 ± 0.02 nmol/g) were detected in G. uro cells ([84]Table 2). These observations indicate that supplementation with EA and UroC exerts distinct impacts on the intracellular metabolome of bacteria, with remarkable variation in phenolic accumulation across different bacterial strains. 3.7. Metabolic pathways involved in bacterial responses to phenolic exposure To identify significant alterations in the metabolic pathways of bacterial cells in response to phenolic exposure, pathway enrichment analysis was conducted using MetaboAnalyst 6.0 ([85]Fig. 3). Among the four strains examined, the metabolic network in A. muciniphila was the most extensively regulated. Specifically, a total of 12 pathways were significantly altered (p < 0.05), with more notable changes observed in those related to amino acid metabolism and one-carbon pool by folate. Conversely, exposure to EA/UroC resulted in significant modulation (p < 0.05) of three pathways in B. infantis, with riboflavin metabolism showing the most pronounced alteration. Additionally, eight pathways were significantly altered in both G. uro and LGG. Notably, several pathways were common among these strains, including riboflavin metabolism, as well as alanine, aspartate, and glutamate metabolism (Supplementary Table 3). These findings indicate that EA/UroC exposure differentially altered bacterial metabolic networks, with several metabolic pathways shared among the tested bacterial strains. 4. Discussion EA and urolithins have been widely reported for their health benefits, including those on gut microbiota composition and function. However, their metabolic modulatory effects on individual bacteria remain unclear. In this study, leveraging culturomics and metabolomics techniques, we found that prebiotic phenolic compounds, i.e., EA and urolithins, influenced gut bacterial growth by regulating the metabolic pathways primarily associated with energy production, amino acid metabolism, and redox balance. As reported in previous studies including ours, EA and urolithins may confer health benefits via favorably modulating commensal gut bacteria, including Akkermansia, Bifidobacterium, Clostridium, and Lacticaseibacillus ([86]He et al., 2024; [87]Xian et al., 2023; [88]Yang et al., 2024). Consequently, in the current study, we selected four representative human-originated beneficial bacterial strains from the Akkermansia, Bifidobacterium, Lacticaseibacillus genera, as well as G. uro, to provide mechanistic insights into how EA/urolithins regulate probiotic-like bacterial growth and exert prebiotic activities. To this end, we purposely introduced G. uro, a well-known EA-to-UroC converting gut bacterium, to a gut environment lacking Gordonibacter spp. ([89]Yang et al., 2024) to investigate how the microbial community responds to phenolic exposure. Consistent with our earlier findings in mice ([90]Yang et al., 2024), the addition of EA at dietary levels had no significant influence on the proliferation of G. uro in the indigenous microbiota ([91]Fig. 2A). Following 3 days of fermentation, supplementation of EA promoted Akkermansia growth compared to vehicle control, which is consistent with previous studies demonstrating the Akkermansia-promoting effects of EA and ellagitannins ([92]Duan et al., 2022; [93]Selma et al., 2017; [94]Yang et al., 2024). Regarding other commensal bacterial groups, aligning with established findings, supplementation with EA or EA-rich preparations did not significantly influence the abundance of Lacticaseibacillus in fecal microbiotas ([95]Bialonska et al., 2009; [96]Sanchez-Patan et al., 2012). However, discrepant results between short-term in vitro fermentation studies and in vivo or long-term in vitro studies have been reported ([97]Duan et al., 2022; [98]Sanchez-Patan et al., 2012). Regarding the substantial increases in Escherichia in fermented samples with or without EA compared to baseline levels ([99]Fig. 2A), it indeed aligns with previous in vitro fermentation studies involving EA or ET supplementation ([100]Chan et al., 2023; [101]Reddy et al., 2007; [102]Romo-Vaquero et al., 2015). The accelerated growth of Escherichia spp. may account for the suppression of growth of other bacteria following EA exposure. Consistent with the changes in bacterial composition upon exposure to EA or urolithins, the pattern of EA metabolism was also substantially altered in the presence of fecal microbiota ([103]Table 1). We previously demonstrated that the co-administration of EA and G. uro promoted the metabolism of EA into urolithins in vivo ([104]Yang et al., 2024). To further investigate the interaction between EA and gut bacteria, we conducted separate fermentation experiments to determine SCFA production, the most well-known bioactive bacterial metabolites resulting from prebiotic fermentation. Although SCFAs are well-known products of the bacterial fermentation of non-digestible soluble fibers, several studies also reported the increased production of SCFAs following polyphenol fermentation ([105]Cao et al., 2023; [106]Pulido-Mateos et al., 2022; [107]Wang et al., 2020). In our study, the addition of EA enhanced the production of propionic, isobutyric, and isovaleric acids to different degrees ([108]Table 1), which was also reported by a previous study ([109]Duan et al., 2022). A moderate increase in Lacticaseibacillus abundance may contribute to the production of propionic acid as Lacticaseibacillus spp. were previously found to produce propionate and butyric acids ([110]Cheng et al., 2022). Subsequently, at an individual bacterial level, we explored the effect of EA or urolithin supplementation on bacterial growth and the intracellular metabolic activities. In agreement with previous studies, we observed that EA supplementation enhanced the growth of Lacticaseibacillus and Bifidobacterium, while UroC improved the growth of Lacticaseibacillus and Akkermansia ([111]Bialonska et al., 2009; [112]Xian et al., 2023; [113]Zhang et al., 2022). Furthermore, our results demonstrate that the growth of A. muciniphila was not significantly affected by the presence of dietary levels of EA ([114]Fig. 2H), similar as previously reported ([115]Henning et al., 2017b). The effects of EA and urolithins on bacterial metabolic activities were further assessed through targeted profiling of intracellular and extracellular metabolites. Although no dose-dependent accumulation of EA was observed, intracellular UroC concentrations appeared to correlate with EA level in the broth ([116]Table 2). As such, it is highly possible that EA is first taken up by G. uro cells before catabolized into urolithins, with UroC being the metabolic end product prior to extracellular export. Moreover, the higher lipophilicity of UroC may contribute to the higher intracellular accumulation of UroC compared with EA ([117]Alfei et al., 2020). Additionally, following exposure of EA to all four strains, remarkable intracellular accumulations of EA and urolithins were observed exclusively in G. uro ([118]Table 2). Consistent with our observations, a recent bioRxiv preprint from Harvard University researchers identified a cluster of cellular enzymes responsible for catalyzing the stepwise conversion of EA to UroC ([119]Bae et al., 2025). Intriguingly, we also detected methylated UroC in A. muciniphila and LGG cells, suggesting the potential capacity of these bacteria to methylate UroC. Similarly, G. uro appears to methylate both UroC and UroA when exposed to high concentrations of these urolithins ([120]Table 2). To elucidate the modulatory effects of EA and urolithins on the metabolic network in probiotic-like bacteria, we examined bacterial intracellular metabolome by integrating untargeted and targeted metabolomic approaches. For cellular metabolite profiling, sample pretreatment is essential to ensure the efficient extraction of a broad range of metabolites while limiting the interference of compounds with MS data acquisition. Lipids, especially the phospholipids enriched in cell membranes, are a major cause of matrix effects in metabolomics. In this study, lipid removal from gram-positive bacteria was particularly problematic due to the unique waxy lipoteichoic acids extracted into methanol from the outer cell membranes ([121]Wicken et al., 1973). This problem was partly solved by adding chloroform to the aqueous phase to uncover more distinct MS features. Using the optimized method, global metabolite profiling was successfully conducted, resulting in the annotation of over 100 intracellular metabolites ([122]Fig. 3). Pathway enrichment analysis of the four bacterial strains revealed several shared and distinct metabolic pathways altered by supplementation of EA or UroC. Notably, common pathways included riboflavin metabolism as well as alanine, aspartate, and glutamate metabolism ([123]Fig. 4), suggesting that these pathways may play a generic role in bacterial responses to EA/UroC. Riboflavin metabolism is known to be crucial for energy production and cellular function, highlighting the influence of phenolics on bacterial energy dynamics, which may ultimately affect host metabolism ([124]Rowland et al., 2018). The regulation of alanine, aspartate, and glutamate metabolism underpins cellular survival and adaptive stress responses, thus underscoring the interplay between EA/UroC supplementation and microbial resilience in varying growth conditions ([125]Brown et al., 2013; [126]Cani et al., 2019; [127]Percy and Gründling, 2014). Fig. 4. [128]Fig 4 [129]Open in a new tab Representative intracellular metabolic pathways of individual bacteria (A-D) regulated after exposure to EA and UroC (n = 3 – 5). In this study, despite new evidence regarding the mechanisms underlying the prebiotic effects of EA and urolithins and their regulatory effects towards probiotic-like gut bacteria were successfully obtained, we encountered two major difficulties. First, no comprehensive metabolome database is designed especially for bacterial endogenous metabolites, thereby making compound annotation extremely challenging. Secondly, during pathway enrichment analysis, the identification of altered metabolism pathways, which involves matching differential metabolites to known pathways in the database, is hindered by the limited number of annotated differential metabolites. Besides, the intricate and dynamic metabolic changes in bacteria also pose substantial challenges in accurately capturing the trend or magnitude of change in specific metabolic pathways. To address these challenges, research in this field may be advanced through the use of multi-omics approaches i.e., metatranscriptomics, metagenomics, metabolomics and proteomics, to provide more comprehensive perspectives on microbial metabolic networks. Additionally, the application of artificial intelligence (AI) methodologies could enhance the identification and annotation of metabolites within complex biological samples. These strategies shall pave the way for more in-depth understanding of the metabolic influences of prebiotic substances on gut bacteria. 5. Conclusion In this study, the influence of EA and urolithins on the growth and metabolic networks of probiotic-like gut bacteria were examined using culturomic and metabolomic approaches. Through these investigations, we demonstrate that the growth-promoting effects on beneficial gut bacteria following phenolic exposure were primarily associated with the regulation of amino acid metabolism, energy homeostasis, redox balance. These underscore EA/urolithin’s prebiotic properties, alongside the production of bioactive metabolites. Furthermore, G. uro exhibited strong capacity to internalize EA for subsequent catabolic processing. Besides, for the first time, we showed that G. uro, A. muciniphila, and LGG were able to methylate urolithins. In summary, our study provides novel insights into the mechanisms through which EA and urolithin interact with probiotic-like gut bacteria and exert prebiotic effects. These findings warrant further investigation into the xenobiotic–gut bacteria interactions to better understand the prebiotic-like properties of dietary phenolic compounds. Funding This work was supported by the Guangdong-Hong Kong/Macau S&T Cooperation Program (2022A0505030027), the Hong Kong Special Administrative Region Government and InnoHK, and the RiFood Interdisciplinary Project Fund (P0038706). CRediT authorship contribution statement Yang Yang: Conceptualization, Methodology, Investigation, Data curation, Writing – original draft, Writing – review & editing. Ke Wang: Conceptualization, Methodology, Investigation, Data curation, Visualization, Writing – original draft, Writing – review & editing. Jia-Chi Chiou: Writing – review & editing, Funding acquisition. Danyue Zhao: Conceptualization, Methodology, Writing – review & editing, Funding acquisition. 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. Acknowledgments