The dried aerial parts of TA were purchased from the Chinese pharmacy of Shuguang Hospital, affiliated with Shanghai University of Traditional Chinese Medicine (origin: Anhui, batch number: 20230513, supplied by Shanghai Yanghetang Chinese Herbal Pieces Co., Ltd.). Following the preparation method described by Wang et al. (22), 1 kg of TA was soaked in 8 L of purified water for 1 hour. The mixture was heated to boiling using an electric ceramic kettle and maintained at a boil for 30 minutes. The decoction was filtered through eight layers of gauze to obtain Filtrate A. This extraction process was repeated with 6 L of purified water to obtain Filtrate B. Filtrates A and B were combined and concentrated using rotary evaporation at 60°C. The concentrated solution was lyophilized to produce a freeze-dried powder, yielding 15.34% (w/w). The freeze-dried powder was stored in a cool, dry place, protected from light. For analysis, the freeze-dried powder was reconstituted in 10 mL of 80% methanol and analyzed using UPLC-MS/MS (chromatographic parameters, mass spectrometry conditions, and methodological details are provided in Supplementary Information S1).

Dextran sulfate sodium (DSS) with a molecular weight range of 36-50 kDa was supplied by MP Biomedicals Inc. (Solon, Ohio, USA). The positive control drug, sustained-release mesalazine granules (5-aminosalicylic acid), was purchased from Shuguang Hospital affiliated with Shanghai University of Traditional Chinese Medicine (Shanghai Ai De Pharmaceutical Co., Ltd., batch number 231018). ELISA kits for mouse total superoxide dismutase (SOD, JL12237), malondialdehyde (MDA, JL53632), interleukin-1 beta (IL-1β, JL18442), interleukin-6 (IL-6, JL20268), tumor necrosis factor-α (TNF-α, JL10484), and interleukin-10 (IL-10, JL20242) were procured from Shanghai Jianglai Biotechnology Co., Ltd. Sinigrin (RFS-H05311805017), isovitexin (RFS-Y11602204022), orientin (RFS-H04411803007), apigenin (RFS-Q00211801029), and luteolin (RFS-M02501909016) were obtained from Chengdu Refiney Biotechnology Co., Ltd. Antibodies to GAPDH, tubulin, NF-κB, p-NF-κB, IκB, IKKα/β, and p-IKKα/β were provided by Cell Signaling Technology (Massachusetts, USA), while antibodies to caspase-1, ZO-1, claudin-1, and p-IκB were supplied by Proteintech (Wuhan, China).

Forty 7-week-old C57BL/6J mice, each weighing 20 g, were randomly assigned to five groups (eight mice per group). The study comprised five groups: (1) Control group (Control): Received daily oral gavage of sterile water (vehicle) without DSS induction; (2) DSS model group (Model): Administered 3% DSS water + daily vehicle gavage; (3) Positive control group (5-ASA): 3% DSS + mesalazine (200 mg/kg/day, i.g.); (4) TA low-dose group (TA-L): 3% DSS + TA extract (2.3 g/kg/day, i.g., a clinical equivalent dose); (5) TA high-dose group (TA-H): 3% DSS + TA extract (4.6 g/kg/day, i.g.). After a 3-day pretreatment period with oral gavage, mice received 3% DSS in drinking water, with the DSS solution replaced every 48 hours, for a total DSS exposure duration of 7 days.

NCM460 cells were maintained in Dulbecco’s Modified Eagle Medium (DMEM, L110KJ, BasalMedia Co., Ltd.) supplemented with 10% fetal bovine serum (FBS, v/v), 100 U/mL penicillin, and 100 μg/mL streptomycin under standard culture conditions (37°C, 5% CO2, humidified atmosphere).

The digested cells were resuspended in DMEM and seeded into a 96-well plate at a density of 50,000 cells/mL, with 100 µL per well. The outermost wells were filled with an equal volume of phosphate-buffered saline (PBS) to minimize edge effects. The plate was incubated for 6 hours to allow cell attachment. After incubation, the original medium was removed, and the cells were treated with DMEM containing different concentrations of isovitexin for 24 or 48 hours. Following treatment, the medium was discarded, and 100 µL of DMEM containing 10% (v/v) CCK-8 reagent was added to each well. The plate was incubated in the dark for 30 minutes, and after which the absorbance was measured at 450 nm using a microplate reader. Cell viability was calculated based on the optical density (OD) values, with untreated cells serving as the control.

To evaluate the protective effects of isovitexin on intestinal epithelial cells, NCM460 cells were allocated into different experimental groups (n=3 replicates/group): (1) Control group: cells maintained in standard culture conditions; (2) Model group: cells treated with 10 ng/mL recombinant human TNF-α for 24 hours; (3) Co-treatment with TNF-α (10 ng/mL) and isovitexin at 25 μM, 50 μM, or 100 μM in DMEM. Post-treatment, cells were harvested for downstream analyses including immunofluorescence staining, Western blotting, and quantitative real-time PCR (qPCR; primer sequences provided in Supplementary Table S1 ).

Throughout the experiment, the mice were weighed daily at a fixed time, and their fecal characteristics were observed and recorded. Fecal occult blood tests were performed to assess fecal blood content. The Disease Activity Index (DAI) was assessed based on percentage body weight loss, stool consistency, and fecal blood score, with averaged scores determined according to Wang et al. (23). This scoring system provided a preliminary evaluation of model establishment and treatment efficacy.

After 14-day of drug administration, mice were euthanized. The distal colon (1 cm proximal to the anus) was excised, rinsed with PBS to remove residual feces, and fixed in 4% paraformaldehyde. Following paraffin embedding, the intestinal tissue was serially sectioned into 4 µm thick slices. Hematoxylin and eosin (H&E) staining was performed to evaluate histopathological changes, with scoring conducted according to Elkholy et al. (24). Alcian blue-periodic acid-Schiff (AB-PAS) staining was utilized to quantify goblet cells, and cell counts were analyzed using ImageJ software.

Paraffin-embedded intestinal tissue sections were deparaffinized with xylene and rehydrated via graded ethanol. Antigen retrieval was performed by microwave heating in citrate buffer (pH 6.0). After permeabilization with 0.25% Triton X-100, sections were blocked with 5% BSA for 1 h at room temperature. Dual labeling was achieved by incubating with ​ZO-1 antibody (Cy3 conjugate, 1:200) and ​Claudin-1 antibody (FITC conjugate, 1:200) ​overnight at 4°C. Post-washing with PBS, nuclei were counterstained with DAPI. Slides were mounted with anti-fade reagent and stored at 4°C in the dark prior to imaging using a laser-scanning confocal microscope.

Intestinal tissue sections were deparaffinized with xylene (3 × 15 min) and rehydrated through a graded ethanol series (100%, 95%, and 70%, 5 min each). Antigen retrieval was performed by microwave heating in 1× citrate-based antigen retrieval buffer (pH 6.0) at high power until boiling, followed by low-power maintenance for 15 min to restore epitope accessibility. Sections were then pre-hybridized with probe-free hybridization buffer at 38°C for 1 h in a humidified chamber to block nonspecific binding. The universal bacterial probe EUB338 (5′-GCTGCCTCCCGTAGGAGT-3′) was denatured at 95°C for 5 min, immediately chilled on ice, and hybridized to tissue sections at 38°C for 8 h in the same hybridization buffer. Post-hybridization, sections underwent stringent washes with preheated 2× SSC (46°C, 30 min) and 2× SSC/0.1% NP-40 (46°C, 30 min) to eliminate unbound probes. Nuclei were counterstained with DAPI, and slides were mounted with anti-fade reagent for fluorescence preservation. All specimens were stored at 4°C in the dark prior to imaging using a laser-scanning confocal microscope.

Using the Python 3.5.1 program, chemical components and targets of Thlaspi arvense (TA) were retrieved from SymMap (http://www.symmap.org/) and ETCM (http://www.tcmip.cn/ETCM/index.php/) databases with search terms “Thlaspi arvense,” “Su Baijiang,” or “Ximi”. Additional phytochemical data were extracted from CNKI and PubMed literature. Potential component targets were predicted via SwissTargetPrediction (http://www.Swisstarget-prediction.ch/), TargetNet (http://targetnet.scbdd.com/), and SEA (https://sea.bkslab.org/) using Python’s pandas and Scrapy modules. UC-related targets were acquired from DisGeNet database (https://www.disgenet.org/) and the GeneCards database (https://www.genecards.org/) using “ulcerative colitis” as the query. Duplicate targets were removed, and gene symbol standardization was performed using NCBI Batch Entrez (https://www.ncbi.nlm.nih.gov/sites/batchentrez) platform.

To identify key therapeutic targets of TAin UC, potential targets of TA’s bioactive components were intersected with UC-associated targets using the R package VennDiagram. Overlapping targets were identified and visualized through a Venn diagram to delineate shared mechanisms.

The key targets for TA against ulcerative colitis were uploaded to the STRING database (v11.5; https://string-db.org), with Homo sapiens as the organism. The full STRING network was generated using a high-confidence interaction score threshold (≥0.700) and medium false discovery rate (FDR <5%). The resultant PPI network was exported in TSV format and analyzed in Cytoscape (v3.10.0), where core targets were identified through degree centrality analysis and visualized. GO and KEGG Enrichment Analyses Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were performed for TA’s anti-UC targets using the bioinformatics platform (https://www.bioinformatics.com.cn/) (25, 26). Results were visualized through bubble plots created with R’s ggplot2 package (v3.4.4) (27).

SDF files of TA’s major bioactive components were retrieved from PubChem (https://www.ncbi.nlm.nih.gov/pccompound). Core targets were selected based on PPI network analysis (Section 1.6.4), KEGG pathway enrichment, and literature evidence. High-resolution (≤2.5 Å) protein structures were acquired from PDB (https://www.rcsb.org/) and Uniprot database (http://www.uni-prot.org/). Molecular docking was performed using AutoDock Vina 2.0 (Scripps Research Institute, USA) with ligand SDF files and protein PDB structures.

A 100 ns molecular dynamics (MD) simulation of the TNF-α-isovitexin complex was performed using GROMACS 2022. The CHARMM36 force field were applied to the protein (28), while GAFF2 parameters were used for ligand topology. The complex was solvated in a cubic box with TIP3P water molecules (29) under periodic boundary conditions. Long-range electrostatic interactions were calculated via the particle-mesh Ewald (PME) method, with neighbor searching managed by the Verlet algorithm. System equilibration included 100,000-step NVT and NPT ensembles (0.1 ps coupling time, 100 ps each). Van der Waals and Coulomb interactions used a 1.0 nm cutoff. Production simulations ran for 100 ns at 300 K and 1 bar.

Intestinal tissue was minced and homogenized in lysis buffer supplemented with 1% phosphatase/protease inhibitor cocktail for 30 min. The homogenate was centrifuged at 12,000 rpm for 10 min, and the supernatant was collected for protein concentration quantification. Levels of oxidative stress markers (SOD, MDA) and inflammatory cytokines (TNF-α, IL-1β, IL-6, IL-10) were quantified using species-specific ELISA kits per manufacturers’ protocols.

Intestinal tissue lysates were centrifuged, and protein concentrations were quantified. Samples were mixed with 5× Laemmli buffer, denatured by boiling (95°C, 5 min), and resolved via SDS-PAGE (10% gel). Proteins were electrophoretically transferred onto 0.44 μm PVDF membranes. Membranes were incubated overnight at 4°C with primary antibodies against: p-NF-κB p65 (1:1000), NF-κB p65 (1:1000), p-IKKα/β (1:1000), IKKα/β (1:1000), p-IκBα (1:1000), IκBα (1:1000), GAPDH (1:1000), cleaved caspase-3 (1:1000), and β-tubulin (1:1000) in TBST containing 5% BSA. After TBST washes, membranes were incubated with HRP-conjugated goat anti-rabbit IgG secondary antibody (1:5000) for 1 hr at 25°C. Protein bands were visualized using ECL substrate and quantified by ImageJ software.

Genomic DNA was extracted from fecal samples, and the hypervariableV3-V4 regions of bacterial 16S rRNA gene were amplified using specific primers. Sequencing libraries were prepared by ligating Illumina adapters to purified amplicons, followed by paired-end sequencing on an Illumina MiSeq platform. Sequencing libraries were prepared by ligating Illumina adapters to purified amplicons, followed by paired-end sequencing on an Illumina MiSeq platform. Sequencing libraries were prepared by ligating Illumina adapters to purified amplicons, followed by paired-end sequencing on an Illumina MiSeq platform (http://www.omicsmart.com).

Data analysis was conducted using SPSS 26.0 software, and graphs were generated with GraphPad 9.0. Quantitative data are presented as mean ± SD. Student’s t-test was used for two-group comparisons, and one-way ANOVA was applied for multi-group analyses. Data violating normality or homogeneity of variance assumptions were analyzed with the Mann-Whitney U test. P < 0.05 was considered statistically significant.

UPLC-MS/MS quantification of five representative phytochemicals in Thlaspi arvense (TA) revealed distinct compositional patterns ( Figures 1A, B ). Quality-controlled analysis demonstrated the following constituent concentrations per gram of crude TA extract: isovitexin (691.73 μg/g), orientin (636.78 μg/g), sinigrin (439.65 μg/g), cynaroside (47.79 μg/g), and apigenin (12.97 μg/g). These values were calculated based on extraction yield normalization, with isovitexin emerging as the predominant constituent.

To investigate the therapeutic potential of TA in DSS-induced ulcerative colitis, with comparative analysis against the standard therapy 5-ASA. 7-day administration of DSS in drinking water was used to induce ulcerative colitis (UC) in mice, with the pharmacological intervention regimen detailed in Figure 1C . The results showed that low and high dose TA demonstrated therapeutic effects comparable to 5-ASA. Specifically, TA ameliorated DSS-induced body weight loss from day 9 onward ( Figure 1D ) and reduced disease activity index (DAI) scores ( Figure 1E ). Given that splenomegaly represents a hallmark of systemic inflammation in UC, we quantified the spleen coefficient, calculated as (spleen weight/body weight)×100%, to assess immune status. Following 14 days of treatment, TA significantly attenuated splenomegaly and reduced spleen coefficients ( Figures 1F, G ), paralleling 5-ASA’s efficacy. Furthermore, TA mitigated UC-associated colonic shortening ( Figures 1H, I ), confirming its anti-inflammatory activity comparable to standard therapy.

Building on the observed anti-inflammatory effects, we further investigated TA’s capacity to preserve intestinal barrier integrity in DSS-induced colitis. The intestinal barrier, a critical defense mechanism against bacterial translocation and systemic inflammation, demonstrated severe compromise in DSS-treated mice. After 7 days of 3% DSS intervention, mice in the model group exhibited severe intestinal epithelium destruction, crypt architecture loss, and submucosal inflammatory cell infiltration ( Figure 2A ). TA administration significantly attenuated these pathological alterations, with reduced epithelial damage and lower pathological index scores ( Supplementary Material Figure A ). Furthermore, TA treatment markedly increased intestinal goblet cell numbers compared to the model group ( Figure 2B ; Supplementary Material Figure B ). To evaluate barrier structural integrity, we assessed colonic tight junction proteins, key components of epithelial paracellular sealing. Compared to the control group, DSS-induced colitis significantly reduced ZO-1 and claudin-1 expression. And both TA and 5-ASA treatments effectively restored these critical barrier components to near-normal levels ( Figure 2C ). In recent years, the bacterial translocation due to intestinal mucosal damage has garnered increasing attention. To investigate bacterial translocation, a functional consequence of barrier compromise, we performed fluorescence in situ hybridization using the universal 16S rRNA-targeted EUB338 probe (30, 31). The results showed that TA treatment substantially reduced bacterial infiltration into intestinal tissues ( Figure 2D ), confirming its mucosal protective efficacy.

To delineate the anti-inflammatory potential of TA beyond structural barrier repair, we quantified oxidative stress and cytokine profiles. As illustrated in Figures 3A, B , DSS exposure induced significant oxidative imbalance, with MDA levels significantly elevated and SOD activity markedly reduced (P < 0.005) compared to controls. Notably, TA administration dose-dependently ameliorated these alterations. Concurrently, the inflammatory cascade manifested through upregulated pro-inflammatory mediators, including tumor necrosis TNF-α, IL-6, and IL-1β (P < 0.005). TA treatment significantly attenuated these cytokine elevations. ( Figures 3C–E ). Furthermore, the anti-inflammatory cytokine IL-10, which was significantly decreased in the model group (P < 0.005), showed restoration following TA intervention ( Figure 3F ).

Given the established role of gut microbiota in intestinal homeostasis, we investigated whether TA-mediated improvements correlate with microbial modulation. As demonstrated in Figure 4A , DSS-induced colitis significantly reduced microbial richness (Sob index), whereas TA treatment effectively restored this parameter. Furthermore, TA administration enhanced microbial evenness (Shannon index) and species diversity compared to the model group, with taxonomic distribution shifts visualized in Figure 4B . To assess community structural changes, multivariate analyses (PCA/PCoA) revealed distinct clustering patterns: the model group microbiota deviated markedly from controls, while TA intervention partially normalized this dysbiosis ( Figures 4C, D ). Importantly, ANOSIM validation confirmed intergroup dissimilarity (R = 0.816, P = 0.001; Figure 4E ).

At the taxonomic level, TA suppressed pathobiont expansion, specifically reducing Proteobacteria phylum and Enterobacteriaceae genus abundances ( Figures 5A, B ). Through LEfSe analysis (LDA >3, P <0.05), Muribaculaceae and Alistipes were identified as control-associated symbionts, in contrast to model-enriched pathogens (Escherichia-Shigella, Acinetobacter). Notably, TA dosage influenced commensal selection: Lactobacillus dominated low-dose regimens, whereas Lachnospiraceae typified high-dose treatment ( Figure 5C ). Complementing taxonomic alterations, PICRUSt2 prediction associated TA with downregulated pathogenic pathways, particularly infectious disease and transmembrane signaling mechanisms ( Figure 5D ).

To elucidate host-microbe interactions, Spearman correlation mapped significant bacterium-cytokine relationships ( Figure 6A ). Consistent with pro-inflammatory effects, TA dose-dependently diminished Escherichia-Shigella (TNF-α-correlated, P <0.001), Erysipelatoclostridium, and Romboutsia loads ( Figure 6B, D, E ). Conversely, anti-inflammatory taxa (Lachnospiraceae_NK4A136_group, Alistipes) exhibited inverse TNF-α associations alongside model group depletion ( Figures 6C, F ). Despite maintaining stable abundance across groups, Turicibacter showed correlations with IL-1β, IL-6, and SOD ( Figure 6G ).

With the aim of expounding the molecular basis of the therapeutic effects observed in colitis models, a network pharmacology approach was systematically applied. Target profiling identified 865 bioactive targets associated with TA and 1,461 disease targets related to UC, with 220 overlapping candidates constituting potential therapeutic nodes ( Figure 7A ; Supplementary Material Tables S2 - 3 ). Subsequent PPI network analysis through Cytoscape topology optimization identified TNF, ALB, AKT1, IL-6, TP53, and EGFR as hub targets mediating the anti-colitis effects of TA ( Figures 7B, C ). Functional enrichment analysis revealed three mechanistic dimensions: biological processes involving oxidative stress response and redox homeostasis ( Figure 7D ), molecular functions encompassing kinase regulation and nuclear receptor activation ( Figure 7E ), and cellular localization to membrane microdomains including caveolae ( Figure 7F ). Finally, KEGG pathway mapping further confirmed TNF signaling pathway modulation as the principal mechanism ( Figure 7G ; Supplementary Material Figure C ), consistent with the cytokine regulatory capacity of TA demonstrated experimentally.

Complementing the network pharmacology findings, molecular docking and dynamics simulations were employed to characterize interactions between TA components and therapeutic targets. Based on PPI and KEGG analyses identifying AKT1 and TNF-α as key mediators, five phytochemicals enriched in TA were computationally docked with nine candidate targets. As shown in Figure 8A , the compounds exhibited strong binding affinities to TNF-α, with binding energies below -7.0 kcal/mol. Isovitexin, the predominant constituent identified through quantitative chemical analysis, demonstrated optimal binding geometry with TNF-α ( Figure 8B ). Molecular dynamics simulations further validated complex stability through conformational equilibrium attainment in the isovitexin-TNF-α system, evidenced by root mean square deviation (RMSD) values stabilizing below 2 Å ( Figure 8C ). Radius of gyration (Rg) analysis revealed minimal structural fluctuations (20.462 Å - 20.925 Å), confirming maintained protein compactness ( Figure 8D ). Solvent-accessible surface area (SASA) variations indicated microenvironmental adaptations during ligand binding ( Figure 8E ). Crucially, dynamic hydrogen bond formation (0–9 bonds) between isovitexin and TNF-α ( Figure 8F ) substantiated robust molecular interactions underlying the therapeutic efficacy of TA.

To clarify the molecular interaction evidence, the anti-inflammatory mechanisms of TA were further investigated through TNF-α signaling pathway analysis in colonic tissues. The results showed that high-dose TA administration significantly suppressed phosphorylation ratios of IKKα/β (p-IKKα/β/IKKα/β), IκB (p-IκB/IκB), and NF-κB (p-NF-κB/NF-κB) compared to model controls (P < 0.01; Figure 9A ). Concomitant with NF-κB signaling inhibition, both TA and 5-ASA significantly reduced cleaved-Caspase-3 expression in colonic tissues compared to the model group ( Figure 9A ). This suppression of caspase-dependent apoptosis paralleled the anti-inflammatory efficacy observed with clinical standard therapy, confirming the functional linkage between NF-κB pathway inhibition and apoptosis regulation in colitis management. Complementing these in vivo observations, the principal bioactive constituent isovitexin identified in TA was systematically evaluated for therapeutic efficacy in vitro.

Initial cytotoxicity assessment using CCK-8 assay revealed that 400 μM isovitexin reduced NCM460 cell viability within 24 hours, with 200 μM exhibiting cytotoxic effects after 48-hour exposure ( Figure 9B ). In parallel, immunofluorescence imaging demonstrated that TNF-α treatment (24 hours) significantly diminished Claudin-1 and ZO-1 expression in intestinal epithelial cells. Notably, isovitexin co-treatment concentration-dependently restored tight junction protein localization ( Figure 9C ). Mechanistic interrogation further showed that 100 μM isovitexin effectively suppressed TNF-α-induced phosphorylation of NF-κB (p-NF-κB/NF-κB ratio) and caspase-3 activation (cleaved-Caspase-3), with statistical significance (P < 0.01; Figure 9D ). Transcriptional profiling via qPCR confirmed dose-responsive regulation of TNF-α-dysregulated genes, including pro-inflammatory mediators (IL-1β, IL-6), anti-inflammatory cytokine IL-10, and antioxidant enzyme SOD ( Figure 9E ).