S.L. was purchased from Harbin, Heilongjiang Province. DSS (MW: 36,000-50,000) and AOM was acquired from MP Biomedicals (Santa Ana, CA, USA). RIPA, Pmsf, and Removal solution of Western primary and secondary antibodies were acquired from Thermo Fisher Scientific. Formalin was purchased by Sigma-Aldrich. Hematoxylin and eosin were purchased from Hangzhou Hua ‘an Biotechnology Co., LTD. 4’, 6-diamidino-2-phenylindole was purchased from Meilen Bio. TNF-α, COX-2, NF-κB p65, and iNOS were obtained from ABclonal. The specific antibodies applied for immunohistochemistry were from Affinity Biosciences (Cincinnati, OH, USA).

S.L. processed according to the Chinese Pharmacopoeia 2020 edition. The processed and raw are crushed with a high speed grinder, then passed through an 80–100 mesh sieve, sealed and stored.

The raw and processed S.L. were extracted by heating reflux extraction method (1 hour, 3 times), the extraction solution was 70% ethanol solution, the liquid-solid ratio was 8:1, and the extract was concentrated into extractum under vacuum reduction. According to previous reports (18), the contents of TR and TP were determined by perchloric acid colorimetry. The standard curve is plotted as follows: y=0.06921x-0.0059 (R=0.9995) (where x is the mass concentration (mg/mL) value and y is the absorbance value). TP and TR are dissolved by water. Dissolved TR and TP were extracted three times with n-hexane and then three times with saturated n-butanol. The obtained n-butanol layer was concentrated to obtain enriched TR and TP. Triterpenoids in TR and TP were 62.5% and 73.9%, respectively.

C57BL/6 male mice (n=84, 6 weeks old) were obtained from “Liaoning Changsheng Biotechnology Co., LTD.”. After a 1-week period of acclimatization (12-h dark/light cycle at 25 °C), mice were randomly distributed into seven groups (n=12, each mouse used an individual cage system). Qualification certificate number: SCXK (Liao) 2023-0001. The model building scenario is shown in Figure 2A . Based on conversion and extraction rate calculation, the doses of TR and TP are as follows: the low dose is 15mg/kg and the high dose is 30mg/kg. All the mice were anesthetized and sacrificed after anesthetizing.

The DAI score is calculated according to the previous criteria (19).

Hematoxylin and eosin (H&E) staining sections were performed according to previous reports (20). Immunohistochemical staining of sections was performed according to manufacturer’s instructions. Apoptosis detection kits were used for TUNEL staining.

Experiments were conducted according to literature reports (20). Image J software is used for quantitative analysis.

Bisulfite sequencing PCR was used for DNA extraction and detection of colon cancer-related genes. Firstly, 1 μg of genomic DNA was converted to bisulfite, purified, and recovered, and relevant primers were designed. Subsequently, we purified mouse colon tissue using the DNA extraction kit, bisulfite treatment kit, etc. The purified products were amplified by PCR, and all the amplified PCR products were loaded into the agarose gel wells for electrophoresis. Generay’s pTG19-T was used as the vector, following the product instructions in the kit, and positive plasmids were sequenced. The sequencing result analysis software used in this study includes the sequencing instrument ABI3730XL, sequencing reagent BigDye V3.1, and sequencing analysis software Sequence Analysis V5.02.

The accurate TP weighing of 2.1 mg obtained in 2.3 was contained in a 10 ml volumetric bottle. The TP solution was filtered through 0.22 μm membrane and stored in refrigerator at 4°C. The chromatography was performed on Waters ACQUITY UPLC BEH C¹⁸ column (2.1 mm×100 mm, 1.7 μm, American Waters Company). The liquid phase conditions were column temperature 40°C, sample chamber temperature 4°C, sample 3 μl, flow rate 0.3 mL·min^-1. The mobile phases were distilled water with 0.1% formic acid (A) and acetonitrile (B). The program was applied to gradient elution: 0–2 min, 95%-95% (A); 2–25 min, 95%-50% (A); 25–42 min, 50%-5% (A). Electrospray ion sources (ESI) are used. High-purity N₂ is used as atomizing gas and cone-hole gas. The mass spectrum parameters were as follows: atomized gas pressure, 650 kPa; scanning range, 50~1500 Da; temperature, 550°C; positive ion spary voltage, 5500 V; negative ion spary voltage, -4500 V; curtain gas, 35 psi; ion source gas 1, 50 psi; and ion source gas 2, 50 psi; declustering potential, ± 80eV; collision energy, ± 35 eV.

Drug targets from SwissTargetPrediction (https://www.swisstargetprediction.ch/) in the database access (probability>0). colon cancer targets were collected from GeneCards (https://www.genecards.org/) and OMIM databases (https://www.omim.org/). Subsequently, potential active ingredient targets and colon cancer targets were imported into the bioinformatics to make Venn diagram (https://bioinformatics.psb.ugent.be/webtools/Venn/) to collect common targets. The common targets were then imported into STRING database (https://cn.string-db.org) to obtain protein-protein interaction (PPI) and visualize by Cytoscape 3.10.2. DAVID (https://david.ncifcrf.gov/home.jsp) was employed to perform Gene ontology (GO) enrichment and Kyoto Encyclopedia of genes and Genomes (KEGG) pathway analysis. Finally, “component-target-pathway” network of network pharmacology was constructed using Cytoscape 3.10.2.

The crystal structure of the core targets were obtained from the protein database (pdb) (https://www.rcsb.org/). The small molecule structure is converted into pdb format by Chem3D software. The ligands will be dehydrated and hydrotreated. The ligands and receptors were converted to pdbqt format using Autodock 1.5.7. Molecular docking Performed by Autodock Vina to obtain the Active ingredients and the key targets. Finally, PyMOL 4.6.0 and Discovery Apply Studio 4.5 to visualize the results.

Compound 27 and TNF-α were simulated by molecular dynamics according to the previously reported method (21). Finally, we analyzed the root mean square deviation (RMSD), root mean square fluctuation (RMSF), radius of gyration (Rg), the number of hydrogen bonds between protein and ligand, relative free energy distribution and 0,25,50,75,100 ns in the molecular dynamics simulation trajectory of the TNF-α - compound 27 complex. In addition, we used the MM/GBSA method to calculate the average binding free energy between protein and ligand.

Data were presented as mean ± standard deviation (SD). Student’s t-test and one-way ANOVA were used for comparisons between and among different groups, respectively. P ≤ 0.05 was considered statistically significant. Histograms were plotted using Graph Pad Prism (version 6.0; GraphPad Software Inc., CA, USA).

To investigate the potential therapeutic effects of TR and TP on colon cancer, a model was established by administering 2% DSS over three cycles to induce colon cancer. The symptoms associated with chronic colitis, including weight loss, diarrhea, and hematochezia, were recorded for 100 days to calculate the DAI. ST treatment significantly reduced the DAI compared to the untreated model group ( Figure 2B ), indicating a mitigation of colitis severity. Furthermore, ST administration attenuated body weight loss and even promoted weight recovery in treated mice, suggesting a protective role against colitis-associated colon cancer progression ( Figure 2C ). Additionally, as illustrated in Figures 2D, E , the control group maintained stable food and water intake throughout the study. In contrast, compared with the treatment group, the food consumption and water intake of the model group were significantly reduced. These findings further support the conclusion that ST ameliorates DSS-induced colon cancer by improving metabolic and clinical parameters.

At the termination of the experiment, mice were euthanized, and colon tissues were collected for macroscopic and morphometric analysis. The control group exhibited normal colonic morphology, characterized by a smooth, intact surface and typical coloration. In contrast, the model group displayed pronounced pathological alterations, including reddish-brown discoloration, disorganized architecture, severe congestion, and significant colon shortening ( Figure 3A ). Quantitative assessment revealed that the mean colon length in the control group was 9.10 ± 0.15 cm, whereas the model group demonstrated a marked reduction (5.50 ± 0.56 cm, P<0.01). Treatment with TR and TP significantly attenuated this shortening (P<0.05), indicating a protective effect against DSS-induced structural damage ( Figure 3C ). Tumor incidence and morphology were further evaluated ( Figure 3B ). The number of tumors in the model group was 9.00 ± 0.89, the number of tumors in the positive group was 1.16 ± 0.51, the number of tumors in the TRL group was 5.00 ± 0.63, the number of tumors in the TRH group was 3.00 ± 1.09, and the number of tumors in the TPL group was 2.00 ± 0.63. The number of tumors in the TPH group was 1.33 ± 0.40, significantly different from that in the model group (P<0.05), and the statistical results are shown in Figure 3D . Notably, tumors in treated mice were predominantly solitary and smaller in diameter (<2 mm), with a minority measuring 2–4 mm and rare instances exceeding 4 mm ( Figure 3E ). These findings suggest that both TR and TP exert dose-dependent chemopreventive effects against colon cancer, albeit with varying efficacy.

In the model group, the glandular structures exhibited back-to-back and cribriform patterns, with severe inflammatory infiltration, crypt loss, and near-total depletion of the epithelial mucus layer. Goblet cells were absent, nuclei were enlarged and hyperchromatic, the nucleus-to-cytoplasm ratio was increased, nuclear polarity was lost, and multinucleation occurred. Mucosal folds were disrupted, and the colon contour became irregular. TR and TP can protect the integrity of the colonic mucosal structure. Due to the differences in the dosage and composition of the drug, the degree of damage to the colonic mucosa varies slightly, and the results are shown in Figure 4A . Histological scoring revealed significant differences between the model group and the control group (P<0.005), as well as between the model group and the drug administration groups (P<0.05). The statistical results are presented in Figure 4D .

In the model group, the expression of Ki67 in colon tissues was significantly elevated, with a positive cell index of 58.50 ± 0.60% ( Figure 4B ). In contrast, the number of Ki67-positive cells decreased in a dose-dependent manner in the TRL, TRH, TPL, TPH, and positive control groups, with positive cell indices of 37.70 ± 1.20%, 28.30 ± 0.40%, 21.20 ± 0.95%, 13.50 ± 0.35%, and 12.02 ± 0.25%, respectively. Semi-quantitative analysis of Ki67 protein expression ( Figure 4E ) revealed that, compared to the model group, Ki67 expression was significantly reduced in the TRL, TRH, TPL, and TPH groups (P<0.01). Notably, the efficacy of TPH in reducing Ki67 expression was comparable to that of the positive control group. Similarly, the expression of PCNA was significantly higher in the model group compared to the control group, with a positive cell index of 49.50 ± 0.40% ( Figure 4C ). The expression of PCNA was also reduced in a dose-dependent manner in the TRL, TRH, TPL, TPH, and positive control groups, with positive cell indices of 29.00 ± 1.45%, 21.50 ± 1.30%, 16.40 ± 0.75%, 12.20 ± 0.88%, and 10.01 ± 0.25%, respectively ( Figure 4F ). The reduction in PCNA expression was statistically significant (P<0.01). TUNEL staining was used to assess apoptosis in colon tissues. As shown in Figures 5A, B , the apoptotic fluorescence intensity was significantly higher in the TRL, TRH, TPL, TPH, and positive control groups compared to the model group (3.50 ± 0.40%). The fluorescence intensities in these groups were 61.60 ± 0.80%, 71.50 ± 1.30%, 80.40 ± 0.26%, 85.50 ± 0.35%, and 89.90 ± 0.40%, respectively (P<0.01). In the control and model groups, few green fluorescent cells were observed, indicating minimal apoptosis. In contrast, a significant increase in apoptotic cells was observed in the colon tissues of mice treated with TR and TP. These findings demonstrated that both TR and TP could inhibit the proliferation of colon tumor cells in a dose-dependent manner, with TP exhibiting a more pronounced effect than TR.

The underlying inhibitory mechanisms of TR and TP on colon cancer cells were further elucidated through Western blot analysis ( Figure 6 ). Compared with the control group, the model group exhibited a significant upregulation in the protein levels of TNF-α, NF-κB p65, COX-2, and iNOS (P<0.005). Upon treatment with TRL, TRH, TPL, TPH, and the positive control, a dose-dependent decrease in the expression of these proteins was observed. Specifically, significant reductions were noted when comparing the treatment groups with the model group (P<0.05). These findings indicate that the triterpenoid components of both TR and TP effectively inhibited the expression of TNF-α, NF-κB p65, COX-2, and iNOS in a dose-dependent manner. Furthermore, TP demonstrated a more significant inhibitory effect compared to TR, which might imply that certain specific triterpenoids compound components in TP have stronger biological activity, or that the combination of triterpenoids compound components in TP has a synergistic effect in inhibiting inflammatory signaling pathways.

As shown in Figure 7A , a total of 7 CPG sites within the TNF-α gene were methylated. The methylation rates were 40.0% in the control group, 17.1% in the model group, 31.4% in the positive group, and 30.0% in the TPH administration group. According to Figure 7C , the methylation rates in the positive group and the TPH group were significantly higher than that of the model group (P<0.05) and comparable to the control group. At this methylation level, the protein cannot bind to DNA, thereby silencing transcription and preventing tumor gene expression, which contributes to the therapeutic efficacy of the drug.

As shown in Figure 7B , a total of 29 CPG sites within the p65 gene were methylated. The methylation rates were 40.9% in the control group, 14.7% in the model group, 31.5% in the positive group, and 27.6% in the TPH administration group. After drug administration, the methylation status of the genes was altered. As shown in Figure 7D , the methylation rates in the positive group and the TPH group were significantly higher than that of the model group (P<0.05). This hypermethylation of the CPG islands inhibits the expression of tumor-related genes, thereby reducing tumor development.

To identify the main triterpenoid components in TP, qualitative analysis was performed using UPLC-Q-TOF-MS. The base peak chromatograms (BPC) in both positive and negative ion modes are shown in Figure 8 . Through comparison and analysis, a total of 26 triterpenoids were identified ( Table 1 ), predominantly pentacyclic triterpenoids. These include 23 ursane-type triterpenoids, 2 oleanane-type triterpenoids, and 1 lupane-type triterpenoid. Most of these compounds have been previously reported in S. L. Notably, compounds 14 and 15 share the same retention time as compounds 25 and 26, respectively, indicating that they are identical compounds. The chemical structures of these compounds are illustrated in Figure 9 .

To elucidate the mechanism of action and identify the potential active compounds of TP, we conducted network pharmacological analysis following the characterization of triterpenoids. According to the SwissTargetPrediction database, 24 of the 26 identified triterpenoids in TP were predicted to have potential therapeutic effects. After removing duplicate targets, a total of 279 target genes were obtained. Additionally, using “colon cancer” as the keyword, we screened 1475 targets from the OMIM and GeneCards databases. By intersecting the drug targets and disease targets, we identified 85 genes as potential cross-targets for TP against colon cancer ( Figure 10A ). Figure 10B visually illustrates the PPI network among these targets. The size of each node represents the importance of its core target. The results showed that TP53, EGFR, CASP3, IL-6, TNF, and STAT3 were centrally located in the PPI network, indicating their involvement in the anti-colon cancer process of TP. GO enrichment analysis revealed that TP is associated with molecular functions such as protein binding and ATP binding, as well as biological processes like chromatin remodeling and positive regulation of transcription by RNA polymerase II ( Figure 10C ). KEGG pathway analysis further identified the key pathways through which TP compounds exert their anti-colon cancer effects, with the top 20 pathways including the PI3K-AKT signaling pathway and TNF signaling pathway ( Figure 10D ). Based on these findings, we constructed a “component-target-pathway” network ( Figure 10E ). This network illustrates that the core components of TP (compounds 1, 8, 13, 15, 16, 18, 19, 22, 27, and 28) act against colon cancer by targeting TP53, EGFR, CASP3, IL-6, TNF, STAT3, and key signaling pathways such as PI3K-AKT and TNF signaling. Given the results of network pharmacology and protein-level analysis, TNF-α was selected as the focus for further study.

To identify the main active compounds in TP, molecular docking analysis was performed based on the results of the PPI and “component-target-pathway” network. Figure 11A presents the heat map and binding energy values of the compounds. The results indicated that 10 compounds exhibited binding energies lower than -5 kcal/mol, with most having binding energies below -7 kcal/mol. Notably, compound 27 had the highest average binding energy, particularly with TNF-α, reaching -10.2 kcal/mol. The interaction forces were further characterized using PyMOL and Discovery Studio, and the molecular model with the lowest binding energy was visualized. As shown in Figure 11B , the binding energy between TNF-α and compound 27 was -10.2 kcal/mol, indicating a strong docking affinity. The interactions between TNF-α and compound 27 were primarily mediated by hydrogen bonds and hydrophobic interactions. Specifically, LYS112 (F chain), GLU110 (F chain), LYS98 (D chain), and GLU116 (D chain) on TNF-α formed hydrogen bonds with compound 27. Additionally, PRO106 (F chain) on TNF-α formed unconventional hydrogen bonds with compound 27, while ARG103 on the TNF-α (F) chain formed a hydrophobic bond with compound 27. In summary, the molecular docking results reveal that compound 27 is a key active component in TP’s prevention of colon cancer, with TNF-α likely playing a crucial role in this process.

To investigate the stability and dynamic interactions of protein-ligand complexes, molecular dynamics simulations were performed to validate the docking results. The stability analysis is shown in Figure 11C . The RMSD and RMSF curves of the complex fluctuated within a range of 1 nm, indicating stable conformational dynamics. The Rg curve fluctuated within a range of 2.15 nm and remained stable throughout the simulation. To further explore the hydrogen bond properties of the complex binding sites, the number of main hydrogen bonds stabilizing the ligand-protein interactions was calculated. The number of hydrogen bonds between TNF-α and compound 27 stabilized at 3–5 within 0–50 ns and 3–6 within 50–100 ns, with the hydrogen bond curve fluctuating steadily. The SASA curve of the TNF-α - compound 27 complex remained stable throughout the simulation, with no significant fluctuations and a variation range of 200 nm². Collectively, these results indicate that the TNF-α - compound 27 complex exhibits high stability. Figure 11D shows the free energy distribution of the TNF-α - compound 27 complex, revealing a single minimum energy cluster with a concentrated energy distribution. This suggests that the complex formed between TNF-α and compound 27 is highly stable. The binding stability of the complex was further examined by comparing the conformations at five time points (0, 25, 50, 75, and 100 ns). As shown in Figure 11E , TNF-α and compound 27 remained bound to the same position at all five time points without significant changes, indicating excellent binding stability. After the complex system reached equilibrium, the average binding free energy of TNF-α - compound 27 was calculated using the MM/GBSA method. Figure 11F shows that the average binding free energy of TNF-α - compound 27 is -54.4 kcal/mol, indicating a strong binding affinity. As depicted in Figure 11G , compound 27 formed favorable interactions with specific amino acid residues in the TNF protein, including GLN102 in the E chain (-2.49 kcal/mol), GLU116 in the F chain (-2.30 kcal/mol), and ARG103 in the F chain (-2.37 kcal/mol). These residues play a crucial role in stabilizing the TNF-α - compound 27 complex. Overall, the results demonstrate that TNF-α and compound 27 exhibit a stable binding interaction.