Enzymatic Synergy-Driven Biotransformation Generates a Postbiotic-Rich Functional Matrix That Reprograms Gut Microbiota Metabolic Pathways Under Stress Conditions

TL;DR

A fermented plant matrix rich in postbiotics helps the gut microbiome adapt to stress, improving host resilience through changes in microbial metabolism.

Contribution

This study reveals how enzymatic synergy in fermentation produces postbiotics that reprogram gut microbiota under stress.

Findings

Co-fermentation increases extractable flavonoids and generates distinct metabolite clusters.

Postbiotic-rich matrix partially normalizes stress-related neuroendocrine markers and improves behavior in mice.

Microbial functional shifts include CAZyme enrichment and reduced cytochrome P450 activity under stress.

Abstract

The physiological efficacy of plant-based matrices is often limited because bioactive compounds are sequestered within complex lignocellulosic architectures, restricting their release and downstream activity. Fermentation-driven enzymatic biotransformation can overcome these structural barriers; however, the mechanisms by which fermentation-derived, non-viable functional ingredients (postbiotics) confer benefits remain incompletely defined. Here, we examined whether a postbiotic-rich, co-fermented plant matrix enhances host resilience under metabolic stress and whether such effects are accompanied by a remodeling of gut microbial functional capacity. A functional plant matrix was produced by solid-state co-fermentation using two Lactobacillus plantarum strains selected for complementary lignocellulolytic profiles. Untargeted metabolomics and deep shotgun metagenomic sequencing were integrated with a hydrocortisone-induced murine metabolic stress model to quantify substrate remodeling, host neuroendocrine/behavioral outcomes, and microbiome functional reprogramming. Co-fermentation markedly remodeled the phytochemical landscape, increasing extractable flavonoids and generating distinct metabolite clusters. In vivo, administration of the postbiotic-rich matrix partially normalized stress-responsive neuroendocrine markers (ACTH, TRH, and testosterone) and improved behavioral outcomes in open-field and forced swim assays. These systemic changes were paralleled by a coordinated shift in microbial functional potential, including the enrichment of carbohydrate-active enzyme (CAZyme) families involved in complex polysaccharide utilization (e.g., AA9, GH129, CE14) and attenuation of phosphotransferase system modules and cytochrome P450-related functions. Enzymatic synergy-driven biotransformation yields a postbiotic-rich functional matrix that is associated with a selective remodeling of gut microbiome metabolic potential under stress and concomitant improvement in host physiological resilience. This study underscores microbial functional remodeling as a critical mechanistic interface linking fermentation-modified substrates to host physiological recovery, providing a molecular framework for the development of targeted postbiotic interventions.

Figures

Full text

1. Introduction

Plant-based functional foods have attracted increasing interest as nutritional strategies for mitigating metabolic stress and preserving systemic homeostasis [1]. A wide range of edible botanical matrices is abundant in bioactive phytochemicals, notably flavonoids and complex polysaccharides. These compounds are strongly implicated in exerting a spectrum of biological activities, including antioxidant, anti-inflammatory, and metabolic regulatory functions [2]. However, the physiological efficacy of these compounds is frequently thwarted by poor bioaccessibility, stemming from their sequestration within the rigid lignocellulosic architectures composed of cellulose, hemicellulose, and lignin. These structural barriers restrict enzymatic access during digestion, thereby limiting nutrient release and attenuating downstream biological activity [3,4,5]. In this context, increasing attention has been directed toward fermentation-derived, non-viable functional ingredients, commonly referred to as postbiotics, whose biological effects are mediated by microbial metabolites or fermentation-modified substrates rather than live microbial colonization.

Targeted bioprocessing strategies have emerged as effective approaches to overcome the intrinsic limitations of plant-based matrices [6]. Among them, enzymatic biotransformation via microbial fermentation has been widely explored for its capacity to deconstruct plant cell wall components and liberate bound phytochemicals [3]. Solid-state fermentation (SSF), in particular, provides a favorable ecological context for microbial growth and enzyme production, often enabling more efficient enzyme–substrate interactions than submerged fermentation systems [7,8,9]. Nevertheless, the chemical and structural heterogeneity of lignocellulosic substrates typically exceeds the catalytic repertoire of individual microbial strains, thereby constraining both the efficiency and specificity of biotransformation [4].

To circumvent these limitations, leveraging enzymatic synergy through co-fermentation has emerged as a robust strategy, harnessing microorganisms with complementary lignocellulolytic profiles to orchestrate multi-level substrate deconstruction [10,11]. Such synergistic systems enable the coordinated degradation of cellulose, pectin, and lignin fractions, enhancing phytochemical bioaccessibility while simultaneously generating fermentation-modified polysaccharides and oligosaccharides with altered structural and functional properties [12]. Although these approaches have shown promise in improving the compositional profiles of fermented plant products, the molecular mechanisms by which enzymatic synergy reshapes plant matrices and programs downstream biological effects remain incompletely understood.

Beyond the direct enhancement of nutrient release, accumulating evidence indicates that fermentation-derived functional matrices exert many of their biological effects through the modulation of gut microbiota function rather than through microbial colonization per se [13]. The gut microbiota plays a central role in shaping host metabolic responses, particularly under stress-associated conditions, through its extensive repertoire of metabolic enzymes and pathways [14]. Central to this host–microbe interaction are carbohydrate-active enzymes (CAZymes), which govern the microbial breakdown of dietary polysaccharides and direct carbon flux through microbial metabolic networks [15]. The balance between microbial pathways involved in carbohydrate utilization and those associated with xenobiotic or stress-related metabolism is increasingly recognized as a key determinant of metabolic resilience [16].

Recent advances in shotgun metagenomic sequencing have enabled pathway-level interrogation of gut microbial functional potential. These studies reveal that metabolic stress and metabolic disorders are frequently associated with functional dysregulation of microbial metabolic programs rather than with taxonomic shifts alone [17]. Specifically, stress-associated microbiomes are characterized by a functional shift toward xenobiotic metabolism, including cytochrome P450-associated functions and phosphotransferase systems (PTS), at the expense of complex carbohydrate utilization capacity. Importantly, these features are increasingly regarded as functional signatures of microbial metabolic adaptation to host stress or altered nutritional environments, rather than as direct causal drivers of pathology. Conversely, the restoration of metabolic homeostasis has been linked to enhanced microbial carbohydrate-degrading capacity and the rebalancing of core metabolic pathways [18].

Despite these advances, a fundamental question remains unresolved: how does the enzymatic synergy inherent to the fermentation process itself program the functional output of the gut microbiota? Specifically, the molecular pathways through which fermentation-driven biotransformation of plant matrices reshapes microbial CAZyme repertoires and metabolic pathway landscapes—and how this functional remodeling translates into host metabolic resilience under stress—remain poorly defined.

In this study, we engineered a functional plant matrix leveraging an enzymatic synergy-driven solid-state co-fermentation strategy, employing Lactobacillus plantarum strains selected for their complementary lignocellulolytic profiles. We hypothesized that this targeted biotransformation would not only liberate sequestered phytochemicals to enhance their bioaccessibility but also yield a postbiotic-rich substrate capable of realigning gut microbial functional landscapes under metabolic stress. To validate this hypothesis and elucidate the underlying molecular mechanisms, we integrated multi-scale chemical profiling with a hydrocortisone-induced murine model and deep shotgun metagenomic sequencing. Special emphasis was placed on quantifying microbial metabolic flux through the interrogation of carbohydrate-active enzyme (CAZyme) repertoires and core metabolic pathways. Ultimately, this work seeks to delineate a mechanism-informed framework linking fermentation-mediated substrate remodeling to microbiome functional reprogramming and host physiological resilience, thereby advancing our understanding of postbiotic-driven host–microbiota interactions in stress-associated contexts.

2. Results

2.1. Screening for Complementary Enzymatic Profiles and Construction of the Synergistic Co-Fermentation System

A screening of 44 strains using Azurine cross-linked (AZCL) assays showed that most tested strains exhibited detectable pectinase and cellulase activities. Among these, strain 177 displayed high levels of both pectinase and cellulase production. Substantial variation in enzymatic activity was observed across different strains of the same species when assessed on multiple AZCL substrates, as reflected by large standard deviations, indicating pronounced intraspecies variability in enzyme activity profiles. For example, Lactobacillus plantarum strains 177 and 130 exhibited markedly different levels of pectinase and cellulase production, while Lactobacillus plantarum 191 produced pectinase but did not exhibit detectable cellulase activity (Figure 1A,B). Growth assays on lignin-containing agar plates further revealed that only strains 191, 180, and N1 were able to utilize lignin as a carbon source (Table 1). Subsequent AZCL-based screening for ligninolytic enzymes demonstrated that all three strains produced lignin peroxidase, whereas manganese peroxidase activity was not detected. Among these strains, Lactobacillus plantarum 191 exhibited the highest lignin peroxidase activity (Table 2).

2.2. Synergistic Co-Fermentation Enhances Microbial Growth and Modulates Phytochemical Profiles

To evaluate the performance of the synergistic fermentation strategy, the plant-based matrix was fermented with strain 177 or strain 191 alone, or with their combination (177 + 191). Compared with both monocultures, the co-fermentation condition produced a significantly higher final viable cell count (CFU/mL) (Figure 2A), indicating improved overall growth yield in the mixed culture. Strain-resolved plating counts in the co-culture—based on reproducible colony morphology discrimination on MRS agar—showed that both strains remained highly viable at the endpoint, with no significant difference in their relative abundance (p > 0.05) (Figure 2B). These results support the notion that strains 177 and 191 can co-exist without evident competitive exclusion under the tested fermentation conditions, supporting their suitability for co-fermentation.

Regarding substrate transformation, total polyphenol content remained stable across all groups. In contrast, polysaccharide content decreased markedly in monocultures, whereas it remained stable in the co-fermentation system (Figure 2C,E). The total flavonoid content in the co-fermentation group was significantly higher than that in the unfermented control group (p < 0.05), a change not observed in monocultures (Figure 2D). The stability of total polysaccharide content observed during co-fermentation may reflect a dynamic balance between enzymatic degradation and polysaccharide-associated processes within the microbial consortium.

To gain preliminary chemical insights into this flavonoid enhancement, we performed an exploratory untargeted metabolomic profiling (UHPLC–Q–Orbitrap–MS) on representative samples from the non-fermented (NF) and co-fermented (FNF) groups (Figure S1A). Given the screening nature of this analysis, no statistical inference was performed and changes are reported as relative ion responses. The overall feature landscape suggested fermentation-associated remodeling, with multiple putatively annotated metabolites showing higher relative signals in FNF compared with NF (Figure S1B), including aromatic/phenolic-related compounds and glycosylated derivatives (e.g., benzylacetic acid, multifidol glucoside, and 6-feruloylglucose-related features) (detailed annotation parameters including retention times, accurate mass, and database identifiers are provided in Table S2). In addition, several features were detected in FNF but were not detected or fell below the preset intensity threshold in NF, including leucic acid and benzyl glucopyranoside (Figure S1C). Focusing on flavonoid-related annotations, the dataset indicated a compositional shift with both increased and decreased flavonoid candidates (Figure S1D), consistent with fermentation-driven transformation of phytochemicals rather than a uniform change in a single compound. Collectively, these exploratory MS features are compatible with the observed increase in total flavonoid content (Figure 2D) and support the possibility that co-fermentation may enhance the extractability and/or biotransformation of specific phytochemicals.

Based on this chemical evidence, we hypothesized that this enzyme-driven, compositional remodeling of the substrate would translate into beneficial systemic effects via the gut axis. To test this hypothesis, we employed a hydrocortisone-induced metabolic and intestinal dysfunction (model) of stress-induced metabolic dysfunction. This model reliably induces phenotypes—such as weight loss, reduced mobility, and gut dysbiosis—that are relevant targets for postbiotic interventions, allowing us to evaluate whether the FNF-fermented product could ameliorate these whole-organism disturbances.

2.3. FNF Administration Is Associated with Improved Physical Performance and Physiological Parameters in a Hydrocortisone-Induced Stress Model

General physiological and behavioral conditions were systematically monitored throughout the experimental period. By day 14, mice in the control group displayed normal activity levels, coordinated movement, intact fur condition, regular excretion, and prompt responsiveness. In contrast, hydrocortisone-exposed mice exhibited a range of stress-associated physiological alterations, including reduced locomotor activity, lethargy, delayed responsiveness, altered posture, and decreased intestinal motility. In a subset of male animals, reproductive-related abnormalities were observed (Figure 3B).

Regarding basic physiological parameters, both the control and hydrocortisone-treated (HYD) groups exhibited body weight gain during the early phase of the experimental period; however, the rate of weight gain in the HYD group was significantly lower than that in the control group. By day 12, the average body weight of the control group became significantly higher than that of the HYD group, a difference that persisted until the end of the experimental period (Figure 3C). No significant differences in food intake were observed between groups throughout the study (Figure 3F).

In subsequent intervention experiments, hydrocortisone-exposed mice received different formulations. JCJ-HYD denotes the historical reference formulation, NF-HYD represents the non-fermented plant-based matrix, FNF-HYD refers to the fermented formulation (a fermentation-derived, non-viable, postbiotic-like matrix), and FNF/2-HYD indicates administration of the fermented formulation at half dosage.

2.4. Behavioral Analysis of Mice

The open field test (OFT) was used to assess locomotor activity and exploratory behavior under stress-challenged conditions. As shown in Figure 3D,E,G, compared with the control group, hydrocortisone-exposed mice exhibited significant reductions in total locomotor distance and swimming endurance (p < 0.01). Administration of JCJ-HYD, NF-HYD, FNF-HYD, and FNF/2-HYD was associated with significant improvements in both total movement distance and endurance performance compared with the hydrocortisone-treated group.

2.5. Organ Coefficients

Organ coefficients were assessed as physiological indices reflecting systemic responses to hydrocortisone-induced stress. As shown in Figure 4A–D, hydrocortisone exposure was associated with significantly reduced coefficients of the thymus, spleen, kidneys, and testes compared with the control group. Relative to the hydrocortisone-treated group, both the NF-HYD and FNF/2-HYD groups exhibited significant increases in thymus coefficient (p < 0.05). In addition, the FNF-HYD group showed a significant increase in testis coefficient compared with the hydrocortisone-treated group (p < 0.05).

2.6. Histopathological Observation of Testicular Tissue

Histopathological examination was performed to assess testicular structural alterations associated with hydrocortisone exposure. As shown in Figure 4F, testicular tissue from hydrocortisone-exposed mice displayed marked histological changes, including thinning of the seminiferous epithelium, reduced numbers of germ cell layers, disorganized arrangement of spermatogenic cells, decreased sperm content within the lumen, and reduced interstitial cell density. In contrast, mice receiving JCJ-HYD, NF-HYD, FNF-HYD, or FNF/2-HYD exhibited varying degrees of attenuation of these histological alterations. Compared with the hydrocortisone-treated group, the untreated groups showed thicker seminiferous epithelia, increased germ cell layers, higher luminal sperm content, and partially normalized interstitial cell density.

2.7. Biochemical Analysis of ACTH, TRH, and Testosterone (T)

Following hydrocortisone exposure, serum levels of adrenocorticotropic hormone (ACTH), thyrotropin-releasing hormone (TRH), and testosterone (T) were significantly lower in hydrocortisone-treated mice compared with the control group (p < 0.01). ACTH levels were also significantly reduced (p < 0.05) (Figure 4J–L).

Administration of the fermented formulation (FNF-HYD) was associated with significantly higher serum levels of ACTH and TRH compared with the hydrocortisone-treated group (p < 0.05), with values approaching those of the control group. Testosterone levels were also significantly increased relative to the hydrocortisone-treated group (p < 0.01). Similar trends were observed in the FNF/2-HYD group, although the magnitude of change was generally lower.

2.8. Effects on Intestinal Motility-Related Parameters in a Hydrocortisone-Induced Stress Model

As shown in Figure 4E,G–I, hydrocortisone-exposed mice exhibited significantly reduced fecal moisture content, decreased numbers of defecated pellets, and lower small intestinal transit rates (p < 0.05), along with a prolonged time to first black stool excretion (p < 0.05), compared with control mice. These changes collectively indicate impaired intestinal motility and altered fecal properties under hydrocortisone-induced stress conditions.

Following the administration of the different formulations, varying degrees of change were observed in intestinal function-related parameters. Fecal moisture content showed an increasing trend in the JCJ-HYD, NF-HYD, FNF-HYD, and FNF/2-HYD groups; however, these changes did not reach statistical significance (Figure 4H). In contrast, the number of defecated pellets and the small intestinal transit rate were significantly higher in all treatment groups compared with the hydrocortisone-treated group (p < 0.05) (Figure 4E,G).

With respect to defecation latency, all treatment groups exhibited significantly shorter times to first black stool excretion compared with the hydrocortisone-treated group (p < 0.05). Notably, the FNF-HYD and FNF/2-HYD groups showed greater reductions in defecation latency than the JCJ-HYD and NF-HYD groups (Figure 4I).

2.9. FNF Administration Is Associated with Altered Gut Microbiota Structure and Taxonomic Composition

To assess the impact of the fermented formulation (FNF-HYD) on gut microbiota composition under hydrocortisone-induced stress, metagenomic sequencing data were analyzed (Figure S2). Alpha diversity analysis revealed distinct patterns among groups (Figure 5A and Figure S3). Compared with the control group, the hydrocortisone-treated group exhibited significantly higher Chao1 and ACE indices (p < 0.05), whereas the Observed species count remained unchanged. Following FNF-HYD administration, the Chao1 index decreased to values comparable to the control group, while the Observed species index was significantly higher than that of the hydrocortisone-treated group (p < 0.05). Shannon and Simpson indices did not differ significantly among groups (Figure S3).

Beta diversity analysis based on Bray–Curtis distances demonstrated clear separation among groups (PCoA; Figure 5B), which was supported by PERMANOVA (R^2^ = 0.346, p = 0.001). Samples from the FNF-HYD group formed a distinct cluster positioned between the control and hydrocortisone-treated groups. At the phylum level, compositional shifts were observed (Figure 5C). The hydrocortisone-treated group showed a lower relative abundance of Bacillota (formerly Firmicutes) and a higher abundance of Bacteroidota, resulting in a reduced Bacillota/Bacteroidota (F/B) ratio. Following FNF-HYD administration, Bacteroidota abundance decreased, and the F/B ratio increased relative to the hydrocortisone-treated group.

To identify taxa contributing to these differences, LEfSe analysis (LDA score > 3.0; Figure 5E) combined with multi-criteria screening (Figure 5F) was performed. Several genera, including Muribaculum, were significantly reduced in the hydrocortisone-treated group but exhibited higher relative abundance following FNF-HYD administration (Figure S5). In addition, genera such as Dysosmobacter and Enterococcus were enriched in the FNF-HYD group (p < 0.05). To resolve whether the observed genus-level changes reflected potential pathogenic enrichment, we further profiled Enterococcus at the species level using shotgun metagenomic taxonomic assignment (Figure S4). The results indicated a compositional shift within the genus rather than a uniform increase across species. Specifically, the model group was dominated by Enterococcus_E-annotated features, whereas the FNF-HYD group showed a redistributed species profile with reduced dominance of Enterococcus_E and relatively higher contributions from other Enterococcus species (e.g., E. gallinarum). These data suggest that FNF-HYD does not simply “enrich” Enterococcus at the genus level, but is associated with restructuring within the genus under stress conditions. Conversely, taxa enriched in the hydrocortisone-treated group, including UBA7173, CAG-485, and Enterococcus_E, showed significantly lower relative abundance following FNF-HYD administration (p < 0.01; Figure S5B).

Collectively, these results demonstrate that FNF-HYD administration was associated with pronounced alterations in gut microbiota composition and structure under hydrocortisone-induced stress conditions.

2.10. FNF Administration Remodels Gut Microbiome Functional Potential and CAZyme Repertoires

To systematically assess the impact of FNF on gut microbial functional potential, metagenomic functional annotation was performed. A large fraction of annotated genes was shared across groups, suggesting a conserved core functional repertoire (Figure S6A). At the pathway level, the FDR-aware visualization based on KEGG Level 2 categories indicated a directional remodeling associated with FNF administration (Figure 6A). Specifically, several disease- and stress-related categories that showed higher enrichment scores in the hydrocortisone-treated group relative to controls (Mod vs. Ctrl; e.g., xenobiotics biodegradation and metabolism, drug resistance, and infectious disease-related categories) tended to shift toward the opposite direction after FNF intervention (FNF vs. Mod), consistent with a restoration-like trend in functional potential. In parallel, multiple metabolism-related categories exhibited opposite directional patterns across the two contrasts, suggesting partial recovery of metabolic processing capacity.

Analysis of KEGG Level 3 modules revealed coordinated differences in multiple metabolic pathways between groups (Figure 6B). Pathways such as the phosphotransferase system (PTS) and cytochrome P450-related metabolism displayed higher relative abundance in the hydrocortisone-treated group and lower relative abundance following FNF administration. In contrast, pathways including estrogen signaling exhibited lower relative abundance in the hydrocortisone-treated group and higher relative abundance in the FNF group. In addition, starch and sucrose metabolism exhibited distinct abundance patterns between groups. Supplementary analyses indicated similar trends in galactose metabolism, the pentose phosphate pathway, and retinol metabolism (Figure S7A).

CAZy profiling provided higher-resolution insight into differences in polysaccharide-degrading enzyme repertoires among groups. At the enzyme class level, the auxiliary activity (AA) category exhibited marked differences in relative abundance (Figure 6C). At the family level (Figure 6D), carbohydrate esterase CE14 and glycoside hydrolase GH129 showed significantly higher relative abundance in the FNF group compared with the hydrocortisone-treated group (p < 0.05). The lytic polysaccharide monooxygenase family AA9 also displayed an increasing trend following FNF administration. Conversely, enzyme families including AA10, GH25, and GH73 exhibited lower relative abundance in the FNF group (p < 0.01).

Collectively, metagenomic functional profiling demonstrates that FNF administration was associated with pronounced differences in microbial metabolic pathway distribution and carbohydrate-active enzyme composition under hydrocortisone-induced stress conditions.

2.11. Integrated Association Analysis Identifies Distinct Correlation Modules Among Microbial, Functional, and Host Parameters

To examine associations among gut microbiota composition, microbial functional features, and host physiological parameters, key differentially abundant bacterial genera, metagenomic features (CAZy enzyme families and KEGG pathways), and representative host indices were integrated into a correlation-based network analysis (Figure 7). Spearman correlation analysis revealed two major co-variation modules with distinct correlation patterns (Figure 7A).

The first module, located in the lower portion of the heatmap, comprised genera that were more abundant in the FNF-HYD group, including Muribaculum, Dysosmobacter, and Enterococcus. These taxa showed positive correlations with multiple carbohydrate-active enzyme families involved in polysaccharide degradation, such as AA9, CE14, and GH129, as well as with selected KEGG pathways including estrogen signaling. Variables within this module also exhibited positive correlations with host physiological parameters, including serum testosterone levels, swimming endurance, and intestinal transit rate (p < 0.05).

In contrast, the second module, positioned in the upper portion of the heatmap, included taxa enriched in hydrocortisone-treated mice, such as UBA7173, Bacillus_AD, and Enterococcus_E. These taxa were positively correlated with enzyme families including AA10, GH25, and GH73, as well as with metabolic pathways such as the phosphotransferase system and cytochrome P450-related metabolism. Variables within this module showed positive correlations with constipation-related indices and negative correlations with serum testosterone levels.

To visualize the overall structure of these associations, a weighted Sankey network was constructed based on significant pairwise correlations (|r| > 0.45, p < 0.05; Figure 7B). The network depicts distinct correlation patterns linking bacterial taxa, functional features, and host parameters, with edge width reflecting the strength of statistical significance and color indicating correlation direction. Several carbohydrate-active enzyme families showed positive correlations with fiber-degrading genera such as Enterococcus, suggesting potential taxon–function associations.

3. Discussion

This study presents a rationally designed, enzymatic synergy-driven solid-state co-fermentation (SSF) framework that enhances the functional output of complex plant-based matrices [19]. By orchestrating targeted, enzyme-mediated deconstruction of lignocellulosic structures, the co-fermentation system combining Lactobacillus plantarum strains 177 and 191 achieved a higher degree of substrate biotransformation than single-strain fermentations [20]. Importantly, rather than emphasizing process optimization alone, this framework illustrates how enzymatic complementarity can be leveraged to reshape substrate bioaccessibility, thereby establishing a mechanistically grounded basis for fermentation-driven functional enhancement [21]. In this context, the fermentation process can be viewed as generating a non-viable, postbiotic-like functional matrix, in which biological effects are mediated by fermentation-modified substrates rather than by live microbial colonization.

A central molecular outcome of this engineered synergy was the pronounced enhancement of flavonoid bioaccessibility, accompanied by distinct polysaccharide dynamics [22]. The integration of cellulolytic and pectinolytic activities associated with strain 177 with the lignin-modifying capacity attributed to strain 191 is consistent with a coordinated, multi-faceted deconstruction of plant cell wall architectures, operating as an enzymatic consortium rather than as the sum of individual strains [23]. Notably, total polysaccharide content remained relatively stable despite clear evidence of enzymatic activity. This observation may reflect a dynamic molecular turnover involving concurrent degradation of native plant polysaccharides and microbial synthesis of fermentation-derived glycans, such as exopolysaccharides [24]. While this biosynthetic offset remains hypothetical, such in situ remodeling suggests the generation of structurally modified polysaccharides with altered bioaccessibility, highlighting a distinctive advantage of multi-strain SSF over simple enzymatic hydrolysis [25]. These structurally remodeled carbohydrates likely constitute a key component of the postbiotic effector pool generated during fermentation.

Importantly, our exploratory untargeted metabolomics (n = 1 per group) provides a qualitative chemical context for this substrate remodeling (Figure S1). Specifically, multiple putatively annotated phenolic conjugates and flavonoid-related features showed differences in relative ion responses between FNF and NF (Figure S1B,D). For example, a feature putatively annotated as a secoiridoid glycoside (oleuropein/oleuropein-like) and several phenolic–carbohydrate conjugates (e.g., 6-feruloylglucose-related features) displayed higher relative signals in FNF, which may reflect enhanced extractability and/or fermentation-associated biotransformation of bound phytochemicals within the complex matrix. Concurrently, the detection of fermentation-associated small molecules such as leucic acid and benzylacetic acid (Figure S1C) is suggestive of microbial amino acid and aromatic metabolism under the co-fermentation setting. While these screening-level data do not support statistical inference, they are compatible with a model in which co-fermentation facilitates the release and transformation of phytochemicals into more readily extractable forms, complementing the observed increase in total flavonoid signal (Figure 2D). Future work incorporating targeted LC–MS/MS quantification (using authentic standards, MSI level 1), EPS quantification, and polysaccharide molecular-weight profiling (e.g., HPSEC–MALLS) will be required to validate these mechanistic hypotheses.

Beyond direct substrate transformation, intake of the fermented product was associated with a restoration-like directional shift in gut microbial functional potential under stress conditions [26]. Shotgun metagenomics suggested functional realignment characterized by attenuation of stress-adaptive signatures and relative recovery of nutrient-processing capacities; pathways related to microbial stress adaptation, including the phosphotransferase system and cytochrome P450-associated metabolism, showed higher representation in the model condition and a tendency to decrease following FNF administration, whereas carbohydrate metabolism-related functions displayed an opposite directional pattern [9]. Importantly, the enrichment of xenobiotic-related pathways and phosphotransferase systems in stressed microbiomes has been interpreted as a functional signature of metabolic adaptation to chemically or nutritionally challenging environments rather than a direct pathogenic mechanism [27]. In this context, the observed attenuation of these stress-associated signatures, together with the increased representation of carbohydrate-active enzyme (CAZyme) families involved in complex polysaccharide utilization (e.g., GH129 and CE14), is consistent with a shift toward a nutrient-processing mode [28]. Collectively, these findings support the possibility that fermentation-modified substrates and their associated postbiotic components may contribute to reshaping microbiome functional potential.

Multi-omics integration further identified associations between specific CAZy families (e.g., GH129 and CE14) and bacterial taxa, including Enterococcus [29]. Given that the ecological role of Enterococcus is highly context-dependent, we further examined this genus at the species level and observed a restructuring rather than a uniform genus-wide expansion (Figure S4). This within-genus shift suggests that the Enterococcus signal in the intervention group should not be interpreted as a generic enrichment of potentially opportunistic taxa, but instead as a context-dependent redistribution under altered substrate availability [30]. Importantly, the associations between Enterococcus and CAZy families are best interpreted at a correlative, community-level functional layer, rather than as a direct attribution of CAZyme functions to a single genus [31,32]. In parallel, although abundance changes in canonical saccharolytic commensals were variable, their known polysaccharide-degrading capacities and their alignment with the remodeled CAZyme repertoire support a model of collective functional redundancy [33]. Together, these observations underscore that host benefits are more closely linked to the pathway-level activation of a carbohydrate-degrading microbial guild than to the expansion of any single taxon [34], reinforcing a functional (rather than purely taxonomic) view of postbiotic action.

At the ecosystem level, these coordinated microbial and functional shifts depict a transition from a stress-adapted configuration toward a more cooperative, nutrient-metabolizing state—a feature increasingly recognized as a hallmark of microbiota resilience [35]. The observed correlations between this reprogrammed microbial metabolic capacity and improvements in host physiological parameters, including physical endurance and hormonal balance, are interpreted as associative and do not imply direct causality [36]. To comprehensively capture the systemic impact of this microbiota remodeling, we focused our correlation analysis on a specific set of physiological indices representing the ‘gut–brain–motility’ axis. We selected neuroendocrine markers (ACTH, TRH, testosterone) and behavioral metrics (swimming/open-field tests) to reflect the central stress response, while gastrointestinal motility parameters (time to black stool, intestinal transit rate) were chosen as direct indicators of gut functional integrity. The observed strong correlations between beneficial microbial modules (e.g., Dysosmobacter, GH129) and improved intestinal motility—manifested as shortened time to black stool and accelerated transit rates—are particularly meaningful. They suggest that the fermentation-driven microbiome restoration extends beyond metabolic signaling to physically ameliorate stress-induced gastrointestinal stagnation, a common comorbidity of neuroendocrine dysregulation. Collectively, these findings support a microbiota-centered framework in which host benefits emerge from the integrated molecular output of a functionally remodeled microbial community [37].

Collectively, our findings support a microbiota-centered mechanism through which enzymatic synergy-driven fermentation enhances host metabolic resilience under stress. Based on the integrated multi-omics and physiological evidence, we propose a molecular model in which solid-state co-fermentation acts as a priming step that converts complex plant matrices into postbiotic-like functional substrates. These fermentation-modified polysaccharides and associated phytochemicals serve as selective molecular cues that reshape gut microbial CAZyme repertoires and metabolic pathway utilization. The resulting functional reprogramming favors carbohydrate-processing networks while attenuating stress-associated signatures such as cytochrome P450 and phosphotransferase systems, thereby enhancing microbial ecosystem metabolic resilience. In this framework, host physiological benefits—including restored neuroendocrine balance and improved physical endurance—arise indirectly from the stabilized and rebalanced metabolic output of a functionally remodeled gut microbiota. This model highlights postbiotic-mediated functional remodeling as a key mechanistic interface linking targeted fermentation processes to host metabolic adaptation, offering a rationale for developing precisely designed, microbiota-targeted nutritional interventions against stress-induced dysfunction.

Several limitations should be acknowledged. Firstly, while the integrated multi-omics approach revealed robust correlations among microbial taxa, functional features, and host physiological parameters, causal directionality cannot be established from the current data. Future studies employing germ-free models or fecal microbiota transplantation will be required to directly test the sufficiency of fermentation-remodeled microbiota in conferring metabolic benefits. Secondly, regarding metabolic outputs, although untargeted profiling detected several fermentation-associated organic acids (e.g., leucic acid, a branched-chain amino-acid-derived hydroxy acid), we did not perform a targeted quantification of volatile short-chain fatty acids (SCFAs) because appropriately preserved aliquots for SCFA extraction were not available. Accordingly, the “metabolic reprogramming” described here is defined primarily at the level of microbial functional potential—namely, a coordinated shift toward complex carbohydrate deconstruction and utilization (supported by CAZyme enrichment and fermentation-linked taxonomic changes) together with upstream substrate biotransformation—rather than by absolute concentrations of SCFA end-products. Future studies integrating targeted SCFA metabolomics (e.g., GC–MS with standardized preservation) will be essential to identify the specific metabolites that mechanistically connect microbiome functional remodeling to host physiological outcomes.

4. Materials and Methods

4.1. Materials

Food Ingredients, Reagents, and Test Kits: Food-compatible botanical materials, including Cistanche deserticola (batch no. 250901, origin: Inner Mongolia Autonomous Region, China), Semen Raphani (batch no. 251101, origin: Chengdu, China), Semen Persicae (batch no. 251101, origin: Jinan, China), Astragalus membranaceus (batch no. 251101, origin: Lanzhou, China), Codonopsis pilosula (batch no. 251101, origin: Lanzhou, China), and Pericarpium Citri Reticulatae (batch no. 250701, origin: Chengdu, China), were purchased from Sichuan Qianyuan Chinese Herbal Decoction Pieces Co., Ltd. (Guanghan, China). TRH, ACTH, and T were obtained from Shanghai Jianglai Biotechnology Co., Ltd. (Shanghai, China). NaNO_2_, Al(NO_3_)3, NaOH, rutin, carboxymethyl cellulose, pectin, and soluble starch were purchased from Shanghai Maclean Biochemical Technology Co., Ltd. (Shanghai, China). Hydrocortisone injection was purchased from Tianjin Lisheng Pharmaceutical Co., Ltd. (Tianjin, China). Jichuanjian Granules (JCJ) were obtained from Jiangsu Kanion Pharmaceutical Co., Ltd. (Lianyungang, China).

4.2. Rationale for Substrate Design and Reformulation

The plant-based matrix used in this study was rationally designed by abstracting functional principles from a classical multi-herb formulation traditionally associated with gastrointestinal regulation, while reformulating it exclusively with food-compatible botanical ingredients. Rather than reproducing the original prescription, the reformulation focused on constructing a fermentation-adapted substrate enriched in polysaccharide- and glycoside-containing plant materials suitable for microbial biotransformation. Ingredient selection was guided by three primary criteria: (i) compliance with food safety and regulatory frameworks, (ii) enrichment of fermentable polysaccharides and phenolic glycoside precursors, and (iii) compatibility with probiotic solid-state fermentation. Non-food medicinal components present in the historical formulation were excluded and replaced with botanicals of comparable compositional profiles, such as Astragalus membranaceus and Codonopsis pilosula, which are known to be rich in fermentable polysaccharides and support microbial enzymatic activity [38,39]. Auxiliary components were incorporated to enhance matrix heterogeneity and provide diverse carbohydrate structures, thereby facilitating enzyme–substrate interactions during co-fermentation. Notably, the specific ratios of these ingredients were determined to preserve the functional architecture of the reference formulation while optimizing the carbon-to-nitrogen balance for microbial growth. The final formulation represents a function-oriented, fermentation-compatible plant-based substrate rather than a traditional medicinal prescription. A compositional correspondence between the historical reference and the reformulated plant-based matrix is summarized in Supplementary Table S1 [38,39,40,41,42,43].

4.3. Strains, Media, and Growth Conditions

In this study, a preliminary functional screening was performed on hundreds of isolates from the in-house library, from which 44 candidate strains exhibiting stable activity in herbal matrices were selected for further enzymatic evaluation (Table 3). To prepare the inoculum and verify strain identity, each isolate was spot-inoculated on MRS agar and incubated at 37 °C for 48 h. All strains underwent three consecutive subcultures prior to fermentation experiments and were cultivated in sterile MRS broth with a 5% inoculation rate at 37 °C for 24 h [44].

Based on integrated enzymatic screening results, Lactobacillus plantarum strains 177 and 191 were selected for subsequent co-fermentation studies. This selection was guided by their complementary enzymatic profiles, with strain 177 exhibiting strong cellulase and pectinase activities and strain 191 displaying ligninolytic capacity. In addition, strain selection considered growth performance and potential metabolic complementarity under solid-state fermentation conditions. Preliminary qualitative co-cultivation tests indicated that strains 177 and 191 exhibited compatible growth kinetics without observable antagonistic effects, supporting their suitability for synergistic co-fermentation.

4.4. Enzyme Assays and Screening

Monoclonal strains were inoculated into 14 mL test tubes containing 2.5 mL of MRS liquid medium and cultured at 37 °C with 180 rpm agitation for 24 h to obtain fresh fermentation broth. The fresh fermentation broth was then inoculated onto lignin isolation and screening medium plates [45,46], which were labeled and subsequently placed in a constant temperature incubator set at 37 °C for 3–7 days. Observations for colony growth on each plate were conducted every 12 h, with records made upon the appearance of colonies.

A medium containing 0.5% yeast extract, 1% peptone, and 1% agar was prepared, to which different substrates of carboxymethyl cellulose, pectin, and soluble starch were added. The mixture was poured into 9 cm diameter Petri dishes [47,48]. After solidification, four 9 mm diameter wells were cut into the agar, and 20 μL of fresh fermentation broth was added to each well. The plates were incubated at 37 °C for 24 h. Enzyme activity was measured and expressed in terms of the diameter of the wells.

4.5. In Vitro Fermentation Assessment and Chemical Profiling

4.5.1. Probiotic Viability Assay

Herbal components were weighed according to the prescribed formula ratio and ground into a fine powder. The powder was mixed with distilled water at a ratio of 1:11 (w/v) to prepare the fermentation substrate, followed by sterilization at 121 °C for 20 min. After cooling to room temperature, suspensions of strains 177 and 191 were adjusted to 10^8^ CFU/mL, respectively. Each sterilized herbal substrate was separately inoculated with 6% (v/v) of Lactobacillus plantarum strains 177, 191, or a 1:1 mixture of strains 177 and 191, and then incubated at 37 °C for 24 h [49]. After incubation, cultures were serially diluted in sterile diluent and spread-plated onto MRS agar. Plates were incubated at 37 °C for 72 h, and viable counts were determined as CFU/mL. Total viable counts were calculated from all colonies on each plate. For co-culture samples, strain-resolved CFU were obtained by morphology-based colony discrimination, leveraging reproducible differences in colony appearance between strains 177 and 191 (e.g., colony size, transparency, and/or edge morphology) under identical plating and incubation conditions [50].

4.5.2. Determination of Total Flavonoid Content

Two milliliters of the sample was pipetted into a 10 mL volume flask, mixed with 0.3 mL of 5% NaNO_2_ solution, and left to stand for 6 min. Then, 0.3 mL of 10% Al(NO_3_)3 solution was added and mixed well, followed by standing for another 6 min. Finally, 4 mL of 1 mol/L NaOH solution was added and left to stand at room temperature for 15 min. Its absorbance was determined at 510 nm. The content of total flavonoid was expressed as ml rutin equivalent/mg dry sample (ml rutin/mg DW). A standard curve was employed for calculations by using a standard rutin solution [51,52].

4.5.3. Determination of Total Phenolic Content

A 5 mL measure of the sample was accurately weighed and mixed with 9.5 mL of distilled water. Then, 0.5 mL of Folinol reagent was added and mixed, followed by the addition of 1.5 mL of 20% Na_2_CO_3_ solution within 0.5–8 min. The volume of the solution was fixed, placed in the dark at 30 °C for 5 h, and the absorbance was measured at 760 nm. The content of total phenolic content is expressed as the equivalent mg of gallic acid per 1 mg sample (ml GAE/mg DM). The standard curve is obtained by using a standard gallic acid solution ranging from 10 to 100 μg/mL [52].

4.5.4. Determination of Soluble Polysaccharides

A 1.0 mL measure of the sample solution was pipetted into a test tube. The total volume was adjusted to 50 mL with distilled water. Three milliliters were pipetted into another test tube, followed by the addition of 1.0 mL of 5% phenol solution. Subsequently, 5.0 mL of concentrated sulfuric acid was quickly added, vigorously shaken, and incubated in a water bath at 37 °C for 20 min. The absorbance was measured at 490 nm. The content of soluble polysaccharide in the sample was determined by a standard curve based on a glucose standard solution [53].

4.5.5. Untargeted Metabolomic Profiling

To explore the fermentation-induced chemical remodeling, representative samples were subjected to untargeted metabolomic analysis at Sanshu Biotech (Shanghai, China). Briefly, metabolites were extracted using a methanol/acetonitrile/water system and analyzed on a Vanquish UHPLC system coupled with a Q Exactive HFX mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA). Chromatographic separation was performed on an ACQUITY UPLC HSS T3 column (Waters, Milford, MA, USA) using a water/acetonitrile gradient containing 0.1% formic acid. Raw data were processed using Progenesis QI software (version 2.3, Waters Corporation, Milford, MA, USA) for peak alignment and metabolite annotation against public and in-house databases.

4.6. Preparation of the Final Fermented Formulation (FNF) for Animal Experiments

To maximize biomass density and metabolite accumulation for in vivo administration, a solid-state fermentation strategy was adopted.

4.6.1. Fermentation Process

A food-compatible, multi-component plant-based matrix was formulated using the following edible botanical materials: Cistanche deserticola (30 g), Semen Raphani (30 g), Semen Persicae (15 g), Radix Astragali (20 g), Radix Codonopsis (20 g), and Pericarpium Citri Reticulatae (10 g). The mixture was ground into a fine powder and moistened with distilled water at a substrate-to-moisture ratio of 1:3 (w/v) to prepare a solid-state fermentation substrate. This ratio was selected to balance microbial metabolic activity and oxygen diffusion, consistent with reported solid-state fermentation practices [54].

The substrate was sterilized at 121 °C for 20 min and cooled to room temperature prior to inoculation. A mixed suspension of Lactobacillus plantarum strains 177 and 191 was added at a total inoculation rate of 6% (v/v) with a 1:1 strain ratio. The mixture was thoroughly homogenized to ensure even microbial distribution and incubated statically at 37 °C for 24 h. The fermentation duration was selected to allow sufficient enzymatic biotransformation while avoiding substrate depletion or excessive acidification. No mechanical agitation was applied during fermentation, consistent with typical static solid-state fermentation systems [55]. The initial pH was not externally adjusted and was allowed to evolve naturally as a function of microbial metabolism during fermentation.

4.6.2. Post-Fermentation

Following fermentation, the entire fermented mash was suspended in 1000 mL of distilled water. To terminate microbial activity and obtain a non-viable, fermentation-derived product enriched in microbial metabolites and transformed plant components, the suspension was boiled for 45 min. This step also served to extract bioactive compounds. The resulting decoction was filtered through multi-layered gauze to remove insoluble residues. The filtrate was concentrated under reduced pressure to a final volume of 200 mL (corresponding to a crude medicinal liquid concentration of 0.625 g/mL), yielding the NFN sample, which was stored at −20 °C until use. NFN/2 represents a half-dose of NFN. The unfermented control formulation (NF) was prepared using the same extraction and concentration procedure but without the microbial inoculation and incubation steps.

4.7. Animal Experiments

Healthy male KM mice (18–20 g, 6–8 weeks old, animal license No. SCXK-(e) 2021-0027, SPF grade) were commercially purchased from the Hubei Bentai Biotechnology Co., Ltd. (Wuhan, China). All mice were housed in clean polypropylene cages under standardized housing conditions (12/12 h light/dark cycle, temperature 22 ± 2 °C, relative humidity 50 ± 5%) and were provided food and water ad libitum, and five male mice were housed per cage [56]. Before the experimental treatment, the animals were first acclimatized to the facilities for 3 days [57]. All efforts were made to minimize the suffering of the animals used in this study [58].

The hydrocortisone-induced stress model (HYD model) was established following previously reported protocols that reproducibly induce metabolic and gastrointestinal dysfunction [59,60]. Mice were randomly allocated to six experimental groups (n = 10 per group). The experimental design—including disease induction protocol and therapeutic intervention schedule—for each group is summarized in Figure 3A. The JCJ group was designated as the positive control. During the intervention period, general physical condition was monitored daily, and body weight and food intake were recorded.

At the end of the experimental period, behavioral assessments were conducted, followed by blood collection for analysis of selected physiological parameters, including adrenocorticotropic hormone (ACTH), thyrotropin-releasing hormone (TRH), and testosterone. Mice were subsequently euthanized for organ coefficient determination and histopathological examination.

4.8. Behavioral Observation

OFT was performed as described previously [61,62]. Briefly, mice were gently placed in an open field, a white plastic box (46 × 46 × 40 cm^3^). The area of the center was 3/5 of the length and width. The other area was defined as the periphery along the walls. Before the formal behavioral tests, all mice were allowed to acclimate to the test room for at least 2 h prior to starting the test [57]. The movements of the subject mice were digitally recorded and analyzed using SMART 3.0 video tracking software.

The anti-fatigue experiment in mice was conducted the next day. A glass tank measuring 50 cm × 30 cm × 25 cm was filled with water to a depth of 20 cm, maintained at a temperature of (20 ± 0.5) °C. Each mouse was loaded with a 5.5 g weight attached to its tail and placed in the glass tank to swim. Timing began immediately, and when the mouse’s head submerged underwater for 10 s without resurfacing, it was considered physically exhausted. The timing was stopped immediately, and this duration was recorded as the mouse’s swimming time [63].

4.9. Constipation Index Test

Weigh an empty sterile EP tube, collect fecal samples from the same time period, weigh the wet feces, dry them in a constant temperature oven, and weigh again as dry weight. The calculation method is as follows: fecal relative water content (%) = (wet weight − dry weight)/wet weight) × 100% [64]. Time to first black stool and stool pellet count measurement: Administer 250 µL of activated carbon solution to mice by gavage, record the time of first black stool excretion and the number of stool pellets in 6 h. Small intestinal transit rate measurement: On the last day of the experiment, administer 250 µL of 10% activated carbon suspension to mice by oral gavage. After 30 min, euthanize the mice, remove the small intestine, measure the total intestinal length, and the distance traveled by activated carbon in the intestine, and calculate the intestinal transit rate = (distance traveled by activated carbon in the intestine/small intestinal length) × 100% [65,66].

4.10. Metagenomic Sequencing and Data Processing

All samples were sequenced on the DNBSEQ-T7 platform (MGI Tech Co., Ltd., Shenzhen, China) by Shanghai Biozeron Biotechnology Co., Ltd. (Shanghai, China). Raw reads were quality-filtered and adapter-trimmed using Trimmomatic v0.39. Host-derived sequences were removed by aligning reads to the Mus musculus reference genome (GRCm39) using Bowtie2 v2.4.5. Clean reads were assembled de novo using MEGAHIT v1.2.9. Open reading frames (ORFs) were predicted with Prodigal v2.6.3, and redundant sequences were clustered using CD-HIT v4.8.1 at 95% identity and 90% coverage to construct a non-redundant gene catalog. Gene abundances were quantified using Salmon v1.10.0 and normalized to transcripts per million (TPM) [67]. Functional and taxonomic annotations were performed using DIAMOND v0.9.24 against the NCBI NR, KEGG, and CAZy databases with an e-value cutoff of 1 × 10^−5^.

Microbial community analysis was performed using R v4.3.1. Alpha-diversity indices were calculated using the vegan package. Principal Coordinate Analysis (PCoA) based on Bray–Curtis distances was utilized to visualize community structure, with group differences assessed by PERMANOVA (adonis) and ANOSIM (999 permutations). Differentially abundant taxa were identified using LEfSe (Linear Discriminant Analysis Effect Size) with an LDA score threshold of >3.0. Metagenomic functional profiling was summarized to KEGG Level 2/3 categories. Group-wise differences were evaluated with multiple-testing correction using the Benjamini–Hochberg procedure. For visualization (Figure 6A), selected KEGG Level 2 categories were displayed as a bubble plot emphasizing directional changes (log_2_ fold change) and relative abundance. For multi-omics integration, Spearman correlations between microbial features and host phenotypes were computed using the psych package. To elucidate robust associations between the gut microbiome and host physiology independent of individual variance, a group-mean mapping approach was applied for the correlation analysis. This strategy effectively captures intervention-driven functional modules across the cohort [68].

4.11. Statistical Analysis

All data are expressed as mean ± standard error of the mean (SEM). Statistical analyses and graphical representation were performed using GraphPad Prism v9.0 (GraphPad Software, San Diego, CA, USA). Differences between two groups were analyzed using an unpaired two-tailed Student’s t-test. For comparisons involving three or more groups, a one-way analysis of variance (ANOVA) was performed, followed by Tukey’s post hoc test for multiple comparisons. A p-value < 0.05 was considered statistically significant.

5. Conclusions

In conclusion, this study demonstrates that enzymatic synergy-driven biotransformation constitutes an effective strategy for enhancing the functional output of complex plant matrices. By overcoming lignocellulosic constraints, this approach markedly increased phytochemical bioaccessibility and was associated with improved physiological resilience under metabolic stress conditions. Integrated multi-omics analyses further indicate that these effects are not attributable to probiotic colonization, but rather to functional remodeling of the gut microbiota. This remodeling is characterized by coordinated pathway-level shifts, including enrichment of carbohydrate-active enzyme (CAZyme) families involved in complex polysaccharide utilization and attenuation of stress-associated metabolic signatures such as cytochrome P450-related functions and phosphotransferase systems.

Collectively, these findings support a postbiotic-centered, mechanism-informed framework in which fermentation-derived, non-viable functional matrices modulate host-associated physiological responses through reprogramming of gut microbial metabolic pathways. By emphasizing microbial functional output over taxonomic composition, this work advances molecular understanding of postbiotic–microbiota interactions and provides a rational foundation for the design of fermentation-derived functional matrices targeting stress-associated metabolic perturbations.