Rice-Fried and Sun-Dried Ginseng: A Comparative Study of Chemical Composition and Protective Effects Against Ulcerative Colitis

TL;DR

This study compares rice-fried and sun-dried ginseng to see how their chemical differences affect their ability to protect against ulcerative colitis.

Contribution

The study reveals that rice-frying ginseng enhances its anti-inflammatory and protective effects against ulcerative colitis through chemical composition changes.

Findings

Rice-fried ginseng (RFG) showed improved anti-inflammatory effects and protective actions in ulcerative colitis models compared to sun-dried ginseng.

RFG altered gut microbiota and lipid metabolism, and reduced phosphorylation of key inflammatory signaling proteins like PI3K, Akt, and NF-κB.

RFG increased cell viability and tight junction proteins in Caco-2 cells and improved colon health in mice.

Abstract

Ginseng (Panax ginseng C. A. Mey.), a traditional Chinese medicine, exhibits spleen-fortifying, anti-inflammatory, and anti-ulcerative colitis (UC) effects. Rice-fried ginseng (RFG), prepared by stir-frying with rice together, yields a marked enrichment of rare ginsenosides, which is hypothesized to enhance its anti-inflammatory and anti-UC effects. Therefore, in this study, the chemical compositions of RFG and sun-dried ginseng (SDG) were systematically compared using LC–MS combined with MS-DIAL, and their protective effects against UC were evaluated using lipopolysaccharide (LPS)-induced Caco-2 cells and a dextran sulfate sodium (DSS)-induced UC mouse model. Rice-frying markedly altered the chemical composition of ginseng, and a total of 64 major compounds were identified, of which 31 increased and 33 decreased after processing. These compositional changes were associated with enhanced anti-inflammatory and immunomodulatory effects of RFG. Consistently, RFG enhanced Caco-2 cell viability, decreased TNF-α, IL-1β, and IL-6, and increased ZO-1, occludin, claudin-1, and E-cadherin. In DSS-induced UC mice, RFG attenuated body weight loss, reduced DAI, increased colon length, and decreased the spleen index, accompanied by improved histopathology, reduced pro-inflammatory cytokine levels, and increased expression of tight-junction proteins (TJPs) in a dose-dependent manner. In addition, RFG ameliorated DSS-induced gut microbiota dysbiosis. Metabolomics and network pharmacology analyses highlighted disturbances in linoleic acid and arachidonic acid metabolism and emphasized the involvement of the PI3K–Akt and NF-κB signaling pathways. Western blotting demonstrated decreased phosphorylation of PI3K, Akt, IKKβ, and NF-κB after RFG intervention. Overall, compared with SDG, RFG showed stronger protective effects in vitro and in vivo, accompanied by improved inflammatory readouts, altered lipid-related metabolites and gut microbiota profiles, and reduced phosphorylation of PI3K, Akt, IKKβ, and NF-κB.

Figures

Full text

1. Introduction

Ulcerative colitis (UC), a subtype of chronic inflammatory bowel disease (IBD), is characterized by lesions primarily located in the colon and rectum [1]. Clinically, UC most commonly presents with relapsing diarrhea, abdominal pain, and hematochezia. Durable remission remains difficult to achieve, and relapse rates are high. In addition to its profound impact on quality of life, UC confers an increased risk of colorectal cancer [2,3]. Current evidence attributes UC pathogenesis to the interplay among dysregulated mucosal immunity, impaired epithelial barrier function, and dysbiosis of the gut microbiota [4,5,6]. As the global incidence and prevalence of UC continue to rise, developing therapies that balance efficacy and safety has become an urgent research priority. Traditional Chinese medicine (TCM) has demonstrated distinctive advantages in the management of UC by leveraging multi-component, multi-target regulation and serving as an important pillar of complementary and alternative medicine [7,8].

Ginseng (Panax ginseng C. A. Mey.), a classical herb, reinforces vital qi and tonifies the spleen [9]. Contemporary pharmacology shows that ginseng and its constituents exert immunomodulatory, anti-inflammatory, antioxidant, and mucosal-protective actions [10,11], providing a rationale for its use in UC. Multiple studies have shown that ginsenosides and ginseng polysaccharides exert anti-UC activity [12,13]. Studies have shown that rice-frying can markedly increase the abundance of rare ginsenosides in ginseng, which may potentiate its anti-inflammatory and immunomodulatory activities. These pharmacological features align well with key mechanisms implicated in UC amelioration, suggesting promising avenues for developing therapies that combine efficacy with safety. Rice-fried ginseng (RFG) is a traditional processed form of ginseng in which the herb is co-heated with rice to moderate its medicinal nature and enhance its spleen-fortifying and anti-diarrheal effects. From a physicochemical perspective, rice can serve as a heat-buffering medium that helps distribute heat more evenly during stir-frying, potentially allowing a more controlled thermal exposure of ginseng. Under these conditions, reducing sugars and amino compounds present in rice and ginseng may enter a Maillard reaction-favorable window [14]. We note that the extent of Maillard chemistry and the formation of specific thermal by-products depend on multiple factors, such as time, temperature, moisture, and matrix composition, and thus, these considerations are presented as plausible processing contexts.

Consistent with prior reports on heat processing of ginseng, the controlled thermal treatment during rice-frying may promote structural conversions of prototype ginsenosides by deglycosylation and dehydration, thereby shifting the overall profile toward enrichment of rare ginsenosides such as Rg_3_, Rk_1_, and Rg_5_ [15]. These rare ginsenosides have been widely associated with anti-inflammatory and immunomodulatory activities and have been discussed in relation to barrier-related protection in intestinal inflammation models, providing a pharmacological rationale for evaluating RFG in UC. Overall, rather than attributing efficacy to traditional functional terms, we frame rice-frying as a processing intervention that may remodel chemical composition in a reproducible manner, thereby motivating a direct comparative evaluation of RFG versus sun-dried ginseng (SDG) in matched experimental models of UC.

Although the anti-inflammatory potential of ginseng and the involvement of PI3K/Akt/NF-κB signaling, gut microbiota, and lipid metabolism in UC have been widely documented [16,17], direct evidence connecting processing-driven chemical remodeling to comparative anti-UC efficacy remains limited. Here, we adopt a processing-centric design in which RFG is prepared under a defined processing technology and compared with SDG under identical extraction and dosing conditions. Using LC–MS with MS-DIAL-based feature annotation and stringent differential screening criteria, we define a chemical profile distinguishing RFG from SDG. We further evaluate whether this processing-driven chemical shift translates into reproducible differences in protective effects in both an lipopolysaccharide (LPS)-induced Caco-2 model and a dextran sulfate sodium (DSS)-induced UC model. Finally, multi-omics associations are interpreted in conjunction with phosphorylation-level validation of PI3K/Akt/IKKβ/NF-κB to support a convergent mechanistic framework. This study, therefore, provides a comparative evidence chain linking processing, chemistry, and biological outcomes rather than merely reiterating canonical inflammatory pathways.

2. Results

2.1. Analysis of Differential Components and Fragmentation Pathways

2.1.1. RFG and SDG Differential Components Analysis

The chemical constituents of RFG and SDG were analyzed using UPLC-Q-Orbitrap-MS under positive and negative ionization modes, and the total ion chromatograms (TICs) are shown in Figure 1A. Principal component analysis (PCA) revealed tight clustering of quality control (QC) injections, indicating system stability and reproducibility (Figure 1B). RFG and SDG were clearly separated on opposite sides of the Y-axis, demonstrating significant compositional differences between groups. Orthogonal partial least squares discriminant analysis (OPLS-DA) further confirmed the intergroup separation (Figure 1C). A 200-time permutation test supported model robustness, with R^2^ and Q^2^ both exceeding 0.9; permuted R^2^/Q^2^ values were lower than the original model, and the Q^2^ intercept was below zero, confirming the model’s stability and reliability (Figure 1D). Collectively, these analyses demonstrated that the chemical composition changed significantly after rice-frying. Volcano plots further highlighted differences between RFG and SDG (Figure 1E). In positive ion mode, 1787 compounds in RFG showed significantly higher peak areas than those in SDG, whereas 998 were significantly lower; in negative ion mode, 665 compounds were significantly higher, and 378 were significantly lower in RFG. Based on the criteria of fold change (FC) > 2, p < 0.05, and variable importance in projection (VIP) > 1, a total of 64 compounds were identified, of which 57 were differential compounds; among these, 31 increased and 26 decreased after rice-frying (Table S1).

2.1.2. Fragmentation Pathways of Representative Components

Using L-arginine as an example, the primary fragmentation pathways of amino acids were elucidated (Figure 2A). In positive ion mode, L-arginine exhibited at 0.74 min with formate adduct ions observed at mass-to-charge ratio (m/z) 175.1193 [M + H]^+^. The major MS/MS fragment ions were m/z 158.0932, 130.0988, and 116.0718. Specifically, the ion at m/z 158.0932 was generated by dehydration of the precursor ion, m/z 130.0988 was formed via decarboxylation, and m/z 116.0718 resulted from loss of the guanidino group.

Using ginsenoside Re and ginsenoside Rg_1_ as examples, the primary fragmentation pathways of protopanaxatriol-type ginsenosides were elucidated (Figure 2B,C). In negative ion mode, Re and Rg_1_ exhibited at 12.62 and 13.41 min, with formate adduct ions detected at m/z 991.5512 [M + HCOO]^−^ and m/z 845.4890 [M + HCOO]^−^, respectively. For ginsenoside Re, the main fragment ions were m/z 945.5413, 799.4857, 783.5028, and 637.4426. The ions at m/z 799.4857 arose from the loss of a rhamnose at C-6, corresponding to ginsenoside Rg_1_, and subsequent loss of a glucose produced m/z 637.4426, corresponding to ginsenoside F_1_. The ions at m/z 783.5028 arose from the loss of a glucose at C-20, corresponding to ginsenoside Rg_2_. For ginsenoside Rg_1_, fragment ions included m/z 845.5046, 799.4888, 637.4288, and 475.3813. The m/z 637.4288 ion reflected the loss of a glucose at C-6, corresponding to ginsenoside Rh_1_, and further loss of another glucose from the C-20 generated m/z 475.3813.

Using ginsenoside Rb_1_, Rd, and Rg_3_ as examples, the principal fragmentation pathways of protopanaxadiol-type ginsenosides were elucidated (Figure 2D–F). In negative-ion mode, the compounds exhibited at 21.52, 27.08, and 38.42 min, with formate adduct ions detected at m/z 1153.6040 [M + HCOO]^−^, m/z 991.5515 [M + HCOO]^−^, and m/z 829.4987 [M + HCOO]^−^, respectively. For ginsenoside Rb_1_, the main fragment ions were m/z 1107.6078, 945.5678, and 783.5008. The m/z 945.5678 originated from the loss of a glucose at C-20, corresponding to ginsenoside Rd; further loss of a glucose generated m/z 783.5008, corresponding to ginsenoside Rg_3_. For ginsenoside Rd, fragment ions were detected at m/z 945.5417, 783.5015, and 621.4344. The m/z 783.5015 was formed by the loss of a glucose at C-3, corresponding to ginsenoside F_2_, whereas a subsequent loss of another glucose generated m/z 621.4344, corresponding to ginsenoside CK. For ginsenoside Rg_3_, the main fragment ions were detected at m/z 783.4843, 763.4280, and 621.4523. The m/z 763.4280 was generated through dehydration, corresponding to ginsenoside Rk_1_ or Rg_5_, while the m/z 621.4523 resulted from the loss of a glucose at C-3, corresponding to ginsenoside Rh_2_.

2.2. Protective Effect of RFG and SDG Against LPS-Induced Caco-2 Cells

2.2.1. Cell Viability

Cell counting kit-8 (CCK-8) assays were used to assess the cytotoxicity of LPS and RFG in Caco-2 cells. Viability remained above 80% after exposure to 5, 10, 25, and 50 μg/mL of LPS, whereas 75 and 100 μg/mL significantly reduced viability to 74.77 ± 1.86% and 65.70 ± 2.85%, respectively. Similarly, RFG at 10, 20, 40, and 80 μg/mL also maintained viability above 80% [18], while 160 and 320 μg/mL decreased viability to 75.74 ± 2.44% and 70.11 ± 1.87%, respectively (Figure 3A). Based on these data, RFG at 20, 40, and 80 μg/mL was selected for subsequent cytokine, immunofluorescence, and WB analyses, and 50 μg/mL LPS was used to induce the inflammatory model. Notably, the 20, 40, and 80 μg/mL concentrations constitute a two-fold incremental gradient, which enables evaluation of potential dose-dependent effects while ensuring that all tested concentrations remain within a safe exposure range.

2.2.2. Anti-Inflammatory Activity

The anti-inflammatory effects of RFG on Caco-2 cells were assessed by quantifying the levels of TNF-α, IL-1β, and IL-6. As shown in Figure 3B, LPS stimulation significantly increased three pro-inflammatory cytokines (p < 0.01). Intervention with RFG-L significantly reduced TNF-α and IL-1β (p < 0.05, p < 0.01), while RFG-M, RFG-H, and SDG-H groups significantly decreased TNF-α, IL-1β, and IL-6 (p < 0.01). Overall, RFG suppressed pro-inflammatory cytokine production in a dose-dependent manner, and RFG-H exerted a stronger effect than SDG-H (p < 0.05), while showing no significant difference versus the positive control (p > 0.05).

2.2.3. Effects of RFG on the Expression of Tight-Junction Proteins (TJPs) In Vitro

Tight-junction proteins (TJPs) are crucial for maintaining intestinal epithelial barrier integrity, and their dysregulation leads to epithelial dysfunction and increased permeability [19]. Since TJPs are primarily localized on the cell membrane, their spatial distribution was visualized by immunofluorescence microscopy. In the control group, four TJPs exhibited the strongest fluorescence signals, forming a continuous reticular distribution between adjacent Caco-2 cells, localized predominantly at the cell junctions (Figure 3C). Consistently, WB analysis showed higher expression of ZO-1 and occludin proteins (Figure 3D). The model group showed weakened or absent fluorescence signals of TJPs, with significantly decreased ZO-1 and occludin expression (p < 0.01), indicating that LPS disrupted the structure of TJPs and compromised epithelial barrier integrity. In contrast, intervention with RFG markedly increased the fluorescence intensity of TJPs. Notably, the expression of ZO-1 and occludin was significantly increased in RFG-M, RFG-H, and SDG-H groups (p < 0.01), as confirmed by western blot (WB). Moreover, RFG displayed a dose-dependent enhancement of TJP expression. Collectively, RFG and SDG mitigated LPS-induced epithelial barrier injury in Caco-2 cells by upregulating TJPs and improving their junctional distribution, suggesting preservation of epithelial junctional structure.

2.3. Protective Effect of RFG and SDG Against DSS-Induced UC Mice

2.3.1. Disease Activity Index (DAI) Score, Body Weight, Colon Length, and Spleen Index

During DSS induction, body weight initially increased slightly with a mild rise in DAI, followed by marked weight loss and a sharp Disease activity index (DAI) increase in the model group during the middle phase (Figure 4B). In the late phase, weight loss slowed in all treated groups, except the RFG-L group. By day 11, the model group exhibited marked weight loss and increased DAI, accompanied by shortened colon length and an increased spleen index (p < 0.01). RFG-L significantly restored colon length (p < 0.05), whereas RFG-M, RFG-H, and SDG-H significantly attenuated weight loss and suppressed the increase in DAI while also restoring colon length and reducing the spleen index (p < 0.05, p < 0.01). Notably, RFG-H exhibited a stronger effect than SDG-H in improving the spleen index (p < 0.05). RFG-H and positive groups demonstrated comparable therapeutic outcomes in body weight, DAI, colon length, and the spleen index (p > 0.05).

2.3.2. Cytokine Content Measurements

Pro-inflammatory cytokines and immunoglobulins are integral to UC pathogenesis [21,22]. As shown in Figure 5A,B, DSS induction significantly enhanced TNF-α, IL-1β, IL-6, IgM, and IgA levels (p < 0.01). Intervention with the RFG-L group significantly decreased TNF-α, IL-1β, IgM, and IgA levels (p < 0.05, p < 0.01), while the RFG-M, RFG-H, and SDG-H groups significantly decreased TNF-α, IL-1β, IL-6, IgM, and IgA levels (p < 0.05, p < 0.01). Notably, RFG-H produced a stronger regulatory effect than SDG-H (p < 0.05) and was comparable to the positive group (p > 0.05).

2.3.3. Histopathological Examination

H&E staining allows direct visualization of intestinal mucosal structural destruction, inflammatory infiltration, ulceration, and glandular abnormalities, thereby enabling histological assessment of UC severity [23]. In the control group, the colonic architecture remained intact, with orderly glandular arrangement, abundant goblet cells, and no apparent inflammatory cell infiltration, representing typical healthy histological features (Figure 5C). In the model group, crypt architecture was largely lost, and goblet cells were markedly depleted, accompanied by substantial inflammatory infiltration. RFG improved histopathology in a dose-dependent manner, with RFG-H and SDG-H largely restoring crypt structure and reducing inflammatory infiltration. Compared with SDG-H, RFG-H showed less inflammatory infiltration and better preservation of crypt and goblet cell features, suggesting stronger mucosal repair.

2.3.4. Effects of RFG on the Expression of TJPs In Vivo

Immunohistochemistry localizes TJPs within tissue and provides a structural indicator of epithelial junctional disruption associated with barrier injury in UC [24]. In the control group, staining was continuous and widespread with strong yellow–brown immunoreactivity, consistent with well-preserved epithelium (Figure 5D). WB likewise showed high expression of ZO-1 and occludin (Figure 5E). In the model group, staining was discontinuous and irregular, accompanied by significantly reduced ZO-1 and occludin levels (p < 0.01), indicating disruption of epithelial TJPs. Following intervention, positive expression of TJPs was partially restored, along with improved cellular morphology. In particular, RFG-M, RFG-H, and SDG-H groups showed significantly increased ZO-1 and occludin expression, as confirmed by WB (p < 0.05, p < 0.01). RFG displayed an overall dose-dependent enhancement of TJP expression, and RFG-H and SDG-H showed similarly improved junctional staining patterns (p > 0.05).

2.4. Metabolic Profiling and Differential Metabolites

DSS induced UC in mice and elicited systemic metabolic perturbations. Accordingly, we profiled the serum metabolome by LC–MS to characterize these alterations. Chromatographic peaks from the control, model, RFG-H, and SDG-H groups displayed clear separation, with distinct metabolic profiles observed across the chromatograms (Figure 6A). PCA and OPLS-DA demonstrated clear separation among groups across pairwise comparisons, and permutation testing supported model robustness (Figure 6A–D). Potential differential metabolites were further screened by volcano plot analysis (Figure 6E). Based on the combined criteria of VIP > 1, FC > 2, and p < 0.05, a total of 70 serum biomarkers of the therapeutic effects of RFG in UC were identified, and these were further annotated using the Human Metabolome Database (HMDB) (Table S2). The differential metabolites primarily involved amino acids, bile acids, phospholipids, lysophospholipids, triglycerides, ceramides, and polyunsaturated fatty acids (PUFAs) and their oxidized derivatives. The model group significantly increased levels of arachidonic acid (AA) and linoleic acid (LA); there were widespread increases in LysoPC, LysoPE, and LysoPI and general decreases in PS, PI, PG, PC, and PE. Bile acids and amino acid levels were also increased. RFG or SDG partially reversed these shifts, with RFG showing a more pronounced restorative effect.

2.5. Gut Microbiota

2.5.1. Composition of the Gut Microbiota

Fecal microbiota were characterized using high-throughput sequencing of the 16S rRNA gene. Rarefaction curves indicated adequate sequencing depth across all groups (Figure 7A). The Shannon index was markedly reduced in the model group, suggesting inflammation-induced loss of microbial diversity (Figure 7B). PCA showed complete separation between the model and control groups, whereas RFG-H clustered closer to the control group, indicating that RFG more effectively restored overall microbial community structure (Figure 7C).

At the phylum level, the gut microbiota of control mice was dominated by Bacteroidota (63.50%), Firmicutes (23.03%), and Verrucomicrobiota (10.20%). DSS induction caused a decline in Bacteroidota (47.44%) and Verrucomicrobiota (4.05%) and a striking expansion of Proteobacteria from 0.69% to 18.84%, accompanied by increases in Deferribacterota and Desulfobacterota, which is consistent with the expansion of pathobionts secondary to mucosal injury. RFG-H effectively restored Bacteroidota (60.37%), Proteobacteria (2.07%), and Verrucomicrobiota (8.30%) to nearly normal levels. SDG-H also increased Bacteroidota (54.43%) and Verrucomicrobiota (7.02%) and decreased Proteobacteria (9.63%), though its restorative effect was weaker than RFG-H (Figure 7D). At the genus level, similar trends were observed. In the model group, beneficial genera such as Akkermansia (from 5.00% to 2.05%) and Lachnospiraceae_NK4A136_group (from 4.55% to 2.36%) were significantly decreased, while Bacteroides (from 14.18% to 33.78%) increased sharply. RFG-H markedly increased Akkermansia (4.30%) and Lachnospiraceae_NK4A136_group (3.81%) and decreased Bacteroides (22.63%), whereas SDG-H achieved only partial recovery (Figure 7E).

LEfSe analysis further corroborated these findings from a phylogenetic perspective: the model group was mainly enriched in inflammation-related lineages such as Enterobacteriaceae, Escherichia–Shigella, Klebsiella, and Helicobacteraceae. The RFG-H group was significantly enriched in Firmicutes branches conducive to short-chain fatty acid production and mucosal repair, including Lachnospiraceae, Lachnoclostridium, and Clostridia_UCG_014 (Figure 7F).

Collectively, these multi-level data consistently indicate that DSS-induced UC mice developed a typical inflammation-associated intestinal dysbiosis. Both SDG and RFG alleviated this imbalance, with RFG exerting the more pronounced regulatory effect.

2.5.2. Correlation Analysis

Spearman correlation analysis was performed to further elucidate the relationships between gut microbiota and UC evaluation indicators, including body weight, DAI, colon length, the spleen index, TNF-α, IL-1β, IL-6, IgM, IgA, ZO-1, and occludin (Figure 7G). The analysis revealed that inflammation-related microbiota, such as p_Proteobacteria, p_Desulfobacterota, g_Bacteroides, g_Klebsiella, p_Deferribacterota, and p_Campylobacterota, were positively correlated with DAI, the spleen index, and pro-inflammatory cytokines. Conversely, they were negatively correlated with body weight, colon length, TJPs, and immunoglobulin levels. These results suggest that the expansion of these taxa is closely associated with increased disease activity and impaired mucosal barrier integrity, highlighting their key role in UC. In contrast, p_Verrucomicrobiota, g_Akkermansia, p_Bacteroidota, g_Lachnospiraceae_NK4A136_group, g_Parabacteroides, g_Alistipes, g_Alloprevotella, and g_Blautia were positively correlated with colon length, body weight, ZO-1, occludin, IgM, and IgA levels and negatively correlated with DAI, the spleen index, and pro-inflammatory cytokines. These microbes, which mainly include mucin-degrading bacteria, short-chain fatty acid (SCFA)-producing bacteria, and symbiotic anaerobes, may promote epithelial barrier repair, inhibit inflammatory responses, and reduce pro-inflammatory lipid metabolism. Overall, p_Verrucomicrobiota, g_Akkermansia, p_Bacteroidota, and g_Lachnospiraceae_NK4A136_group showed significant correlations with most UC-related indicators, while g_Bacteroides exhibited strong correlations with nearly all UC indicators. These five microbes are, therefore, proposed as key microbial targets mediating the anti-UC effects of RFG.

Furthermore, Spearman correlation analysis was conducted to explore the relationships between these five key taxa and differential metabolites, revealing a pronounced functional coupling between microbial shifts and lipid metabolic remodeling (Figure 7H). Specifically, p_Verrucomicrobiota, g_Akkermansia, p_Bacteroidota, and g_Lachnospiraceae_NK4A136_group showed overall positive correlations with multiple structural lipid species, including PG, PE, PI, PS, and several TGs, suggesting their potential involvement in membrane lipid homeostasis and barrier repair-related metabolic regulation. By contrast, g_Bacteroides exhibited stronger associations with inflammation-related lipid mediators, showing positive correlations with linoleic acid, arachidonic acid, and the oxidative derivative 9-hydroxylinoleic acid and significant associations with LysoPI(18:1). These results suggest that g_Bacteroides may contribute to the activation of unsaturated fatty acid-derived oxidative lipid pathways and the propagation of inflammatory responses. Based on correlation magnitude, statistical significance, and their central roles within the pro-inflammatory lipid network, linoleic acid, arachidonic acid, LysoPI(18:1), and 9-hydroxylinoleic acid were ultimately defined as four key metabolites, which may act as critical metabolic mediators through which RFG exerts anti-UC effects via the “microbiota–lipid metabolism–inflammatory signaling” axis.

2.6. Network Pharmacology Analysis

Network pharmacology was performed to prioritize candidate targets and pathways for validation. SwissTargetPrediction yielded 419 putative targets for differential compounds, and integrated disease databases identified 1420 UC-related targets; intersection analysis produced 130 overlapping targets (Figure 8A,B). protein–protein interaction (PPI) analysis highlighted high-degree nodes including HSP90AA1, STAT3, SRC, ESR1, HIF1A, PIK3CA, JUN, MAPK1, JAK2, and NFKB1 (Figure 8C,D). Furthermore, integrated pathway enrichment analysis of the metabolites and core targets highlighted significant enrichments in the PI3K-Akt and NF-κB signaling pathways, as well as in LA and AA metabolism (Figure 8E,F). These analyses are hypothesis-generating and were used to prioritize candidate pathways for experimental validation rather than to establish causality.

2.7. Integrated Analysis

This Sankey diagram depicts the key multi-layered association network through which RFG intervenes in UC, spanning “phenotypes–microbiota–metabolites–target genes–pathways” (Figure 8G). On the left, UC-related indicators, including body weight, colon length, the spleen index, DAI, TNF-α, IL-6, IL-1β, IgA, IgM, ZO-1, and occludin, were closely connected to core microbial features. Among them, p_Verrucomicrobiota, g_Bacteroides, g_Akkermansia, and g_Lachnospiraceae_NK4A136_group served as major hub nodes, suggesting that shifts in these taxa may constitute a critical microbial basis for improved disease phenotypes and attenuated inflammation. Notably, these microbes showed stronger linkages with lipid-associated metabolites, which were mainly enriched in AA and LA, together with oxidative lipid species, such as LysoPI(18:1) and 9-hydroxylinoleic acid. These findings indicate that RFG may participate in UC modulation by reshaping fatty acid metabolism and the network of pro-inflammatory lipid mediators. On the right, metabolite-associated signals converged on key molecular targets, including NOS2, PTGS2, NFKB1, STAT3, JUN, MAPK1, MAPK14, PIK3CA, PIK3R1, and MTOR, and ultimately mapped to the NF-κB and PI3K-Akt pathways, which are central to inflammatory and immune regulation. Collectively, this integrated multi-layer network summarizes cross-layer associations among phenotypes, microbial features, lipid-related metabolites, and inflammation-related targets. It is presented as a hypothesis model to facilitate interpretation, while mechanistic conclusions are anchored by the phosphorylation-level validation of PI3K/Akt/IKKβ/NF-κB.

2.8. Validation of Key Pathways

To validate the prioritized signaling pathways, WB was performed to quantify total and phosphorylated PI3K, Akt, IKKβ, and NF-κB in colon tissues and in LPS-stimulated Caco-2 cells (Figure 9A,B). DSS induction in mice and LPS stimulation in Caco-2 cells significantly increased the phosphorylation of PI3K, Akt, IKKβ, and NF-κB (p < 0.01), indicating activation of the PI3K/Akt/IKKβ/NF-κB pathway. RFG intervention dose-dependently attenuated phosphorylation in both models. In vivo, the RFG-L group reduced p-PI3K and p-Akt (p < 0.05, p < 0.01). The RFG-M, RFG-H, and SDG-H groups markedly decreased phosphorylation of all four proteins (p < 0.01). In vitro, the RFG-L, RFG-M, RFG-H, and SDG-H groups similarly decreased phosphorylation of PI3K, Akt, IKKβ, and NF-κB (p < 0.05, p <0.01). Importantly, the inhibition of RFG-H was comparable to that in the positive group both in vivo and in vitro (p > 0.05).

These results indicate that PI3K/Akt/IKKβ/NF-κB represents a convergent inflammatory signaling node that is significantly attenuated under RFG intervention, with RFG-H showing stronger suppression than SDG-H in both vivo and vitro (p < 0.05). This is consistent with the observed improvements in inflammatory readouts and the supportive multi-omics patterns.

3. Discussion

Ginseng is traditionally used for tonifying the spleen, a function that has been reported in various pharmacological studies [25]. In the context of rice-frying, the enhancement of spleen-tonifying effects is often suggested and may be consistent with its increased ability to modulate immune responses and gut-related inflammation [26]. However, it is important to note that the classic functional markers related to “spleen-tonifying” have not been directly tested in this study. The molecular evidence from our experiments supports improvements in intestinal inflammation and barrier function, but the linkage to “spleen- tonifying” requires further validation and is currently considered a speculative interpretation. In addition, the rice-frying conditions used in the present study were established through preliminary, in-laboratory process exploration, with the primary aim of achieving consistent organoleptic properties and a reproducible chemical profile of RFG [27]. However, no industrial-scale upscaling was conducted; therefore, these parameters should not be interpreted as definitive industrial standards. Before RFG can be advanced as a standardized candidate therapeutic product, scale-up validation under good manufacturing practice (GMP)-like conditions will be required, together with rigorous inter-batch quality assessment to ensure manufacturing robustness and product consistency.

Rice-frying can dynamically alter the chemical constituents of ginseng, thereby leading to differences in pharmacological effects between SDG and RFG [28]. During rice-frying, the short-term high temperature first induces the cleavage of macromolecules, such as proteins and polysaccharides, in both rice and ginseng, releasing small molecules including free amino acids and monosaccharides or oligosaccharides. With the onset of the Maillard reaction, highly reactive amino acids such as arginine and tryptophan are preferentially consumed, while simultaneously triggering the Strecker degradation of phenylalanine to generate aromatic precursors [29]. This leads to a reduction in the content of arginine, tryptophan, and phenylalanine, along with the enhancement of the aroma. Citrulline may increase because it is released from proteins and can be partially formed from arginine under high-temperature conditions. In the early stage of the Maillard reaction, polysaccharides degrade more rapidly, resulting in an increase in reducing sugars [30]. In parallel, ginsenosides undergo marked transformations during rice-frying. Malonyl ginsenosides undergo decarboxylation and demalonylation to yield the corresponding neutral ginsenosides [31]. Specifically, ginsenoside Rg_1_ can lose glucose moieties at the C-20 and C-6 positions, forming Rh_1_ and F_1_, respectively; ginsenoside Re can lose the C-20 glucose unit to generate Rg_2_. Likewise, ginsenosides Rb_1_, Rb_2_, Rb_3_, and Rc can undergo the removal of a disaccharide group at C-20 to form Rg_3_, or alternatively lose a C-3 glucose and an additional C-20 glucose or arabinose, leading to the formation of F_2_ [32]. Moreover, ginsenoside Rg_3_ may undergo dehydration at the C-20 position to form Rg5 and Rk1, or lose a C-20 glucose unit to yield Rh_2_ and compound K. Similarly, ginsenoside Rg_2_ can form Rg6 through the loss of one molecule of water [33]. This is consistent with our observation that ginsenosides Rh_1_, Rh_2_, Rh_3_, Rh_4_, F_1_, F_2_, Rs_1_, Rs_2_, Rg_4_, Rg_5_, Rg_6_, Rg_8_, Rg_9_, compound K, Rk_1_, Rk_2_, 20(S)-Rg_3_, 20(R)-Rg_3_, 20(S)-Rs_3_, and 20(R)-Rs_3_ were significantly increased in the rice-fried ginseng group. This indicates that the rice-frying process induces a transformation window that favors the enrichment of rare ginsenosides. Structure–activity relationship studies further indicate that reducing glycosylation at C-3, C-6, and C-20 can enhance the biological activity of ginsenosides, while introducing a C-20 double bond and losing a C-25 double bond can further increase their bioactivity [34]. While these findings are consistent with the TCM theory that rice-frying enhances the spleen-tonifying effects of ginseng, it is important to note that further experimental validation is needed to fully confirm the mechanisms behind these transformations.

The spleen is a major secondary lymphoid organ involved in systemic immune regulation, and its status can reflect inflammatory activity during UC progression [35]. In this study, DSS induction increased the spleen index, whereas both RFG and SDG reduced it, with a more pronounced effect observed for RFG. Importantly, the spleen index is used here as a biomedical indicator of DSS-induced systemic inflammation, rather than as a direct functional surrogate for the TCM concept of “spleen tonification”. Consistent with the systemic and mucosal inflammatory responses, DSS activated the PI3K–Akt/IKKβ/NF-κB pathways, as evidenced by elevated phosphorylation of PI3K, Akt, IKKβ, and NF-κB [36]. In parallel, pro-inflammatory cytokines (TNF-α, IL-6, and IL-1β) and immunoglobulins (IgM and IgA) were increased [21,22]. Following treatment, both RFG and SDG significantly reduced the phosphorylation of these signaling proteins, and RFG produced stronger inhibition than SDG at the same dose. Together, these data support attenuation of PI3K–Akt/IKKβ/NF-κB phosphorylation as a key validated signaling event linked to the superior anti-inflammatory and immunomodulatory effects of RFG in the UC model. In addition, the reductions in IgA and IgM after treatment are consistent with the alleviation of immune-inflammatory activation in UC [22,37].

Beyond signaling validation, our metabolomics and microbiota analyses offer a supportive context for how systemic lipid remodeling and dysbiosis may accompany disease attenuation. LA and AA metabolic pathways are among the most influential metabolic routes involved in the pathogenesis and progression of UC [38]. Upon external stimulation, phospholipase A_2_(PLA_2_) is activated and releases LA and AA as free fatty acids, thereby initiating downstream inflammatory cascades [39]. AA is further metabolized by cyclooxygenases (COXs), lipoxygenases (LOXs), and cytochrome P450 (CYP) enzymes to generate pro-inflammatory mediators, including prostaglandin E_2_ (PGE_2_), leukotrienes, and hydroxyeicosatetraenoic acids (HETEs) [39,40]. LA can also be converted via COX and LOX into hydroperoxyoctadecadienoic acid (HPODE) and hydroxyoctadecadienoic acid (HODE), which contribute to inflammatory progression [41]. In this study, DSS was associated with disturbed ω-6 lipid signatures and altered microbial composition, whereas RFG and SDG partially normalized these patterns, with RFG showing a greater tendency than SDG to regulate LA- and AA-related metabolites and to remodel the gut microbiota. Correlation analysis further suggested that taxa linked to a higher inflammatory burden, including LPS-immunogenic lineages, such as g_Bacteroides, were positively associated with ω-6 lipid mediators [42]. By contrast, taxa often discussed in relation to mucosal homeostasis [43], including g_Akkermansia, p_Verrucomicrobiota, and g_Lachnospiraceae_NK4A136_group, tended to associate with improved outcomes. Overall, these patterns support a microbiota-related lipid metabolism that parallels disease severity and recovery. Because these omics layers are inherently association-based, we present them as supportive evidence that strengthens biological plausibility rather than as stand-alone mechanistic proof. Notably, improvements in TJPs, including ZO-1, occludin, claudin-1, and E-cadherin, observed after intervention are consistent with reduced junctional disruption in UC [44]. However, because functional permeability assays, such as TEER or FITC–dextran were not performed, barrier-related conclusions are restricted to molecular and histological indicators.

In summary, our data indicate that rice-frying-driven chemical remodeling is associated with stronger anti-inflammatory activity in RFG than in SDG. This superior effect is accompanied by coordinated changes in immune indicators, TJP markers, lipid-associated metabolites, and gut microbiota structure. Finally, although the current study focused on inflammation and immunity as primary endpoints, potential differences in other dimensions, such as oxidative stress, were not examined and warrant future investigation.

There are still several limitations to this study. Firstly, the transformation of polysaccharides during rice-frying was inferred mainly from outcome-level observations, and targeted compositional analyses were lacking, making it difficult to fully elucidate the Maillard reaction. Furthermore, restoration of the intestinal barrier was evaluated primarily by the expression of TJPs, while functional indicators, such as TEER and FITC–dextran permeability, were not included. Therefore, our barrier-related interpretation is restricted to TJP expression and histological indicators and should not be considered a direct demonstration of restored epithelial permeability function. In addition, although this study confirmed that RFG exerts UC-alleviating effects both in vivo and in vitro, the pharmacokinetic characteristics of the major active ginsenosides and their distribution in target organs, such as the colon, remain unclear. Future work could, therefore, focus on detailed characterization of polysaccharide changes after rice-frying, incorporation of direct barrier function assays, determination of the in vivo distribution of key constituents, and assessment of oxidative stress so as to provide a more systematic and comprehensive basis for the use of RFG in alleviating UC.

4. Materials and Methods

4.1. Materials

Fresh 5-year-old ginseng was obtained from the Changbai Mountain cultivation base in Jilin Province, China. A voucher specimen (No. 20241211) has been deposited at the Herbarium of Changchun University of Chinese Medicine. HPLC-grade methanol and acetonitrile were obtained from Fisher Scientific (Waltham, MA, USA). Mesalazine granules were obtained from Shanghai Aide Pharmaceutical Co., Ltd. (Shanghai, China). DSS (molecular weight 40,000 Da) was obtained from Macklin Biochemical Co., Ltd. (Shanghai, China). LPS, 5-aminosalicylic acid (5-ASA), and a bicinchoninic acid (BCA) protein assay kit were obtained from Shanghai Yuanye Biotechnology Co., Ltd. (Shanghai, China). The enzyme-linked immunosorbent assay (ELISA) kits were obtained from Shanghai Youxuan Biotechnology Co., Ltd. (Shanghai, China). CCK-8, high-efficiency RIPA lysis buffer, phenylmethylsulfonyl fluoride (PMSF), and protease inhibitor cocktail were obtained from Beijing Solarbio Science & Technology Co., Ltd. (Beijing, China). Enhanced chemiluminescence (ECL) reagent was obtained from Shandong Sparkjade Biotechnology Co., Ltd. (Shandong, China). Antibodies against PI3K, p-PI3K, Akt, p-Akt, IKKβ, p-IKKβ, NF-κB, and p-NF-κB were obtained from Beijing Biosynthesis Biotechnology Co., Ltd. (Beijing, China). Tris-buffered saline with Tween^®^ 20 (TBST), bovine serum albumin (BSA), antibodies against ZO-1, occludin, claudin-1, and E-cadherin were obtained from Wuhan Servicebio Technology Co., Ltd. (Wuhan, China).

4.2. Characterization of Differential Components by UPLC–MS Analysis

SDG was obtained by natural air-drying of fresh ginseng followed by slicing. RFG was prepared by a traditional rice-frying procedure: 50 g of rice was heated at 200 °C for 3 min, then 20 g of SDG slices were added and fried for 3.5 min, followed by cooling to room temperature [27]. For chemical extraction, 5 batches each of RFG and SDG were macerated with 10 volumes of 70% ethanol at room temperature, repeated three times for 24 h each. Combined filtrates were concentrated under reduced pressure until ethanol disappeared and then freeze-dried. Next, 100 mg of lyophilized powder was dissolved in 1 mL of 80% methanol, vortexed, and filtered through a 0.22 μm microporous membrane for LC–MS analysis.

Mass spectrometric (MS) and tandem mass spectrometric (MS/MS) data were acquired in full-scan mode. Compound annotation and identification were performed using database matching and data preprocessing with MS-DIAL 4.70 software to generate a dataset comprising sample groups, retention time (RT), m/z, and normalized peak area for subsequent statistical analyses. Based on this dataset, PCA and OPLS-DA models were constructed using SIMCA 14.1 software. The VIP values were computed to assess the contribution of each variable. Differential compounds between RFG and SDG were screened according to the combined criteria of VIP > 1, FC > 2, and p < 0.05. Detailed LC–MS analytical conditions and MS-DIAL parameter settings are described in the Supplementary Materials.

4.3. Effects of RFG and SDG on LPS-Induced Caco-2 Cells

4.3.1. Cell Cultures

Caco-2 cells, a human colorectal adenocarcinoma cell line, were obtained from Procell Life Science & Technology Co., Ltd. (Wuhan, China). Cells were cultured in Minimal Essential Medium (MEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS) and 1% penicillin–streptomycin under standard culture conditions (37 °C, 5% CO_2_).

4.3.2. Assay of Caco-2 Cell Viability

For the CCK-8 assay, cells were plated in 96-well plates at 1 × 10^4^ cells/mL (200 µL per well) and allowed to adhere for 24 h. After removing the supernatant, cells were exposed to various concentrations of LPS (5, 10, 25, 50, 75, 100 μg/mL) or RFG (10, 20, 40, 80, 160, 320 μg/mL) for 24 h. Subsequently, 10 μL of CCK-8 reagent was added to each well, and the plates were further incubated at 37 °C for 1 h. The optical density at 450 nm was then recorded using a microplate reader.

4.3.3. Cytokine Content and Immunofluorescence Assessment

Caco-2 cells were seeded onto glass coverslips in 6-well plates at 1 × 10^5^ cells/mL (2 mL per well). After they had reached complete adherence and confluence, the cells were divided into seven groups: control group, model group, positive group, RFG-L group, RFG-M group, RFG-H group, and SDG-H group. Supernatants were collected by centrifugation after 24 h, and TNF-α, IL-1β, and IL-6 were measured using ELISA kits. Cells were rinsed with ice-cold phosphate-buffered saline (PBS), fixed by 4% paraformaldehyde for 20 min, permeabilized with 0.1% Triton X-100 for 10 min, and blocked with 3% BSA for 30 min. Primary antibodies were applied overnight at 4 °C in a humidified chamber for dual staining: ZO-1 (1:500, rabbit) + occludin (1:500, rabbit) and E-cadherin (1:500, mouse) + claudin-1 (1:500, mouse). After PBS washes, cells were incubated in the dark for 50 min with secondary antibodies, goat anti-rabbit IgG (Alexa Fluor 488, 1:400) and goat anti-mouse IgG (Cy3, 1:300), and then counterstained with DAPI for 10 min. Finally, an anti-fluorescence quenching mounting medium was applied, and fluorescence images were captured using a fluorescence microscope (NIKON ECLIPSE C1, Tokyo, Japan).

4.4. Effects of RFG and SDG on DSS-Induced UC Mice

4.4.1. Animals and Experimental Design

Male BALB/c mice (20 ± 2 g) were obtained from Liaoning Changsheng Biotechnology Co., Ltd. (certificate number: SCXK [Liao] 2020-0001; batch number: No. 210726250100288424; Shenyang, China). Mice were maintained under specific pathogen-free (SPF) conditions at Changchun University of Chinese Medicine (22 ± 2 °C, 55 ± 5% humidity, 12 h light/dark cycle) with food and water ad libitum. After one week of acclimatization, mice were randomized into groups.

Mice were randomly divided into seven groups (n = 6): control group, model group (3.5% DSS), positive group (mesalazine, 200 mg/kg), RFG-L group (RFG, 200 mg/kg), RFG-M group (RFG, 400 mg/kg), RFG-H group (RFG, 800 mg/kg), and SDG-H group (SDG, 800 mg/kg). To facilitate dose conversion, we provided an estimated approximate human equivalent dose (HED) based on inter-species dose equivalence relationships. The detailed calculation methodology is described in the Supplementary Materials Table S3. Except for the control group, all mice received 3.5% (w/v) DSS in drinking water for 7 days to induce colitis. From day 1 to day 10, treated groups received mesalazine, RFG, or SDG suspended in 0.5% carboxymethylcellulose sodium (CMC-Na) by oral gavage once daily. Control and model groups received vehicle CMC-Na. On day 11, blood samples were collected from the orbital plexus under anesthesia and centrifuged at 4000 rpm for 20 min to isolate serum to be used for biochemical and metabolomics analyses (Figure 4A). Subsequently, mice were euthanized, and spleens, colonic tissues, and cecal contents were collected for the determination of the spleen index, WB analysis, and gut microbiota analysis, respectively.

4.4.2. Body Weights, DAI, Colon Length, and Spleen Index

Body weight was documented every day during the experiment. Weight loss, diarrhea, and fecal bleeding were monitored and scored; scores were averaged to derive the DAI [45]. Colon length and spleen weight were measured, and the spleen index was calculated.

4.4.3. Cytokine and Immunoglobulin Content Measurements

The concentrations of inflammatory cytokines (TNF-α, IL-1β, IL-6) and immunoglobulins (IgM, IgA) in serum were measured using commercial ELISA kits.

4.4.4. Histological Analysis

Colon tissues were fixed in 4% paraformaldehyde, dehydrated through a graded ethanol series, embedded in paraffin, sectioned, deparaffinized, and rehydrated. Sections were stained with hematoxylin and eosin (H&E), followed by differentiation, clearing, and mounting. Histopathological changes were examined and imaged using a light microscope.

4.4.5. Immunohistochemical Assessment

Colon tissue sections were blocked with 3% BSA for 30 min, subsequently rinsed with ice-cold PBS, and incubated overnight at 4 °C in a humidified chamber with primary antibodies: ZO-1 (1:800, rabbit), occludin (1:500, rabbit), claudin-1 (1:500, rabbit), and E-cadherin (1:2000, mouse). After additional PBS rinses, sections were incubated with the appropriate secondary antibodies at room temperature for 50 min. After a final PBS rinse, DAB chromogenic solution was applied to visualize immunoreactivity. The expression and localization of ZO-1, occludin, claudin-1, and E-cadherin were assessed by light microscopy and imaged.

4.5. Metabolomics

A total of 100 μL of serum was deproteinized with 300 μL of methanol, vortexed, and then centrifuged at 15,000 rpm for 15 min at 4 °C. The supernatant was dried under a stream of nitrogen, reconstituted in 100 μL of chromatographic-grade methanol, vortexed, and filtered through a 0.22 μm microporous membrane. Subsequently, 5 μL of the filtrate was injected into the LC–MS system for metabolomic profiling; detailed analytical procedures are described in the Supplementary Materials. Raw data were preprocessed in MS-DIAL to generate a comprehensive dataset, followed by multivariate statistical analysis in SIMCA 14.1, where PCA and OPLS-DA models were established in positive and negative ion modes. Metabolites with VIP > 1, FC > 2, and p < 0.05 were identified as potential metabolic biomarkers of DSS-induced UC. Metabolite annotation was performed based on accurate m/z and MS/MS fragmentation data with reference to the HMDB.

4.6. Gut Microbiota

Genomic DNA was extracted from mouse cecal feces by a magnetic bead-based method. Quality was assessed by electrophoresis and NanoDrop (NanoDrop Technologies, Inc., Wilmington, DE, USA). The V3–V4 region of the 16S rRNA gene was amplified with primers 338F/806R. PCR conditions: 95 °C for 5 min; 25 cycles of 95 °C for 30 s, 50 °C for 30 s, 72 °C for 40 s; and a final 72 °C for 7 min. The PCR products were purified and recovered; detailed sample processing methods are described in the Supplementary Materials. Sequencing was performed on the Illumina NovaSeq 6000 platform, with sequencing services provided by Beijing Biomarker Technologies Co., Ltd. (Beijing, China). Diversity analyses were performed to evaluate microbial community structure differences between samples, and statistical analyses of differential species were conducted.

4.7. Network Pharmacology Analysis

Potential targets of differential compounds were predicted using the SwissTargetPrediction database. UC-related targets were retrieved from GeneCards, OMIM, TTD, PharmGKB, and DrugBank, with “ulcerative colitis” as the keyword and duplicates removed. Overlapping targets between compound targets and UC-related targets were identified using Venny 2.1.0 to generate a Venn diagram, thereby determining the potential targets of RFG in UC intervention. The intersecting targets were imported into the STRING database to construct a PPI network with a confidence score > 0.9. The TSV-formatted network was imported into Cytoscape 3.9.1 for visualization, topological analysis, and core target identification. The core targets and HMDB identifiers were simultaneously imported into MetaboAnalyst 6.0 for pathway enrichment analysis, in order to predict the signaling pathways and metabolic pathways involved in the protective effects of RFG against UC.

4.8. Western Blotting

Cells and colon tissue homogenates were centrifuged at 4 °C to collect the supernatants, and protein concentrations were determined by BCA assay. Equal amounts of protein were separated by SDS–PAGE and transferred to PVDF membranes. The membranes were blocked with 5% BSA and incubated overnight at 4 °C with primary antibodies against ZO-1, occludin, PI3K, p-PI3K, AKT, p-AKT, IKKβ, p-IKKβ, NF-κB, p-NF-κB, and β-actin, followed by incubation with HRP-conjugated secondary antibodies. Detailed sample processing methods are described in the Supplementary Materials. Signals were detected using ECL, and band intensities were quantified in triplicate using Image J 1.48 software. Statistical analysis was performed on replicate-level densitometric values rather than visual inspection of representative blots. Notably, the figure shows representative WB bands.

4.9. Statistical Analysis

Statistical analyses were carried out using GraphPad Prism 9.5.1. Data are reported as mean ± SD. Differences between the two groups were assessed with Student’s t-test, whereas comparisons among more than two groups were evaluated by one-way ANOVA. Statistical significance was defined as p < 0.05.

5. Conclusions

Compared with SDG, RFG exerted more robust protective effects against UC. This superiority is supported by a processing-centric evidence chain showing that a defined rice-frying procedure induces a distinct and reproducible chemical signature, which translates into consistent anti-UC efficacy across both in vitro and in vivo models. These benefits were accompanied by coordinated improvements in inflammatory readouts and TJP expression, normalization of lipid-associated metabolites and gut microbiota profiles, and suppression of PI3K/Akt/IKKβ/NF-κB phosphorylation.