Berberine Hydrochloride Tablets (0.1 g/Slice, NO. 2230533) and Mesalazine Enteric coated tablets (0.25 g/Slice, NO. H20103359) were respectively purchased from Northeast Pharmaceutical Group Shenyang NO.1 Pharmaceutical CO., LTD. and Heilongjiang Timehome Pharmaceutical CO., LTD. Acetonitrile (LC grage), methanol and formic acid were purchased from Merck and Co., Inc. The reference standards of clenbuterol (NO. WP23112904, ≥98.00%) and chloramphenlcol (NO. L2213370, ≥98.00%) were respectively purchased Shanghai Yuanye Bio-Technology CO., LTD and Shanghai Aladdin Biochemical Technology CO., LTD. The reference standards of L-methionine (NO. 24010251, ≥99.30%), sarcosine (NO. 24070588, ≥99.30%), L-valine (NO. 24020502, ≥99.90%), L-alanine (NO. 24010392, ≥99.80%), DL-2-aminooctanoic acid (M16GB148843, ≥98.00%), L-glutamic acid (NO. 231002016, ≥99.70%), L-threonine (NO. 24020271, ≥99.90%), L-lysine (24020360, ≥97.20%), L-histidine (NO. 23120455, ≥99.50%), L-pipecolic Acid (NO. 22060947, ≥99.90%), guanine (NO. 24030186, ≥99.90%), niacinamide (24010447, ≥99.80%), O-phosphoethanolamine (NO. Q15F8S29252, ≥98.00%), taurine (NO. 24030314, ≥98.50%), spermidine (NO.24010276, ≥99.40%) were purchased from Tanmo Quality Inspection Technology CO., LTD. A Milli-Q system (Arium bagtank 50, Sachsen, DEU) was used to produce deionized water. Carboxymethylcellulose sodium (CMC-Na) was purchased from Labgic Technology CO., LTD (Beijing, China), and Dextran Sulfate Sodium Salt (DSS) was purchased from Yeasen Biotechnology (Shanghai) CO., LTD. (Shanghai, China).

All animal experiments were approved by Institutional Animal Care and Use Committee of Shenzhen TOP Biotechnology CO., LTD (TOP-IACUC-2024-0083). Male Sprague–Dawley (SD) rats (190–210 g, 7–8 weeks), provided by Guangdong Medical Laboratory Animal Center, were randomly divided into four groups: Control group (treated with water, n = 6), Model group (treated with 4% DSS, n = 6), Berberine group (treated with 4% DSS and Berberine, n = 6) and Mesalazine group (treated with 4% DSS and Mesalazine, n = 6). DSS-induced colitis in Model group, Berberine group and Mesalazine group were established by oral intake of 4% DSS in drinking water for 7 days, while rats in Control group were treated with water. After DSS administration for 24 h, rats in Berberine group and Mesalazine group were respectively gavaged with 100 mg/kg Berberine and 100 mg/kg Mesalazine (dissolving in 0.5% CMC-Na solution) for the last 7 days, while rats in Control and Model group were gavaged with 0.5% CMC-Na solution with equal volume. During the experiment, the food and water intake were observed daily. The Disease Activity Index (DAI) of the rats was determined by body weight loss, stool consistency and the degree of stool occult blood. The scoring system was shown in Supplementary Table S1. The calculation formula for the Disease Activity Index (DAI) is as follows: D A I = B o d y   W e i g h t   L o s s + S t o o l   C o n s i s t e n c y + O c c u l t   B l o o d / 3

Data collection for the DAI was performed daily. Body weight was measured using an electronic scale, and the percentage weight loss relative to the initial weight was calculated. Stool consistency was assessed visually and scored according to the provided criteria. Occult blood was detected using the benzidine method with a fecal occult blood test kit. The DAI was evaluated on the final day of the experiment to assess overall disease severity. Higher DAI values indicated more severe disease activity, while lower values suggested improvement or remission. At the end of the experiment, all rats were anesthetized by tribromoethanol and plasma was taken through the abdominal aorta after fasted for 12 h. The colons were fixed by 4% paraformaldehyde for 48 h and then embedded in paraffin wax. Five-micron-thick sections were cut and stained with hematoxylin-eosin for histopathological evaluation. Plasma samples were separated by centrifugation at 14000 rpm for 10 min and stored at −80°C until further analysis.

100 μL of plasma sample was spiked with 20 μL of clenbuterol as internal standard (CL, 5 μg/mL) in positive mode and 20 μL of chloramphenicol (CH, 5 μg/mL) as internal standard in negative mode, followed by vortex mixing for 30 s. Then, 400 μL of precipitate solvents (Acetonitrile: Methanol = 1: 1) was added and vortexed for another 3 min. After centrifugation at 14000 rpm for 30 min, 5 μL aliquot of the supernatant was injected into LC-MS/MS system for further analysis. A quality control (QC) sample was produced by mixing and blending equal volumes (10 µL) of each sample. To ensure system equilibrium at the beginning of the analysis, a QC sample was injected 6 times. Subsequently, to monitor system stability throughout the sample testing, a QC sample was injected every six samples. Additionally, a blank sample was run every eight samples to ensure the absence of carryover and contamination. All samples were tested in randomized order to minimize systematic biases and further validate the robustness of the analytical method. All standards were dissolved in 50% methanol water and finally diluted to 50 ng/mL.

The HPLC-MS/MS systems consisted of HPLC separation system (Shimadzu, Kyoto, JPN) and an API 6500+ Qtrap mass spectrometer equipped with an ESI interface (AB Sciex, Framingham, MA, United States). Chromatographic separation was conducted using an ACQUITY UPLC HSS T3 column (2.1 mm × 100 mm, 1.8 μm, Waters Corp., Milford, MA, United States) with column temperature maintained at 40°C. The 0.1% formic acid in water as mobile phase A and 0.1% formic acid in acetonitrile as mobile phase B. The gradient elution program with a flow rate of 0.3 mL/min was used and as follows: 0–2 min, 0% B; 2–14 min, 0%–95% B; 14–16 min, 95% B; 16–20 min, 0% B.

Low-resolution MS and MS/MS analysis were performed on an API 6500+ Qtrap mass spectrometer equipped with an ESI interface (AB Sciex, Framingham, MA, United States). Equipment control was performed using Analyst software ver.1.7.2 (AB Sciex). The data analysis was performed by using SCIEX OS software ver.2.1.6.59781 (AB Sciex). The conditions of the MS detector were set as follows: ion spray voltage, 5.5 kV (Positive mode), −4.5 kV (Negative mode); capillary temperature, 550°C; ion source GS1, 70 psi; ion source GS2, 80 psi; curtain gas, 35 psi. The mass spectrum was respectively recorded in the m/z range of 100–1000 in positive mode and negative mode. The dwell time of each ion pair was 2 m. Nitrogen was used in all cases. The MS/MS experiments were conducted following the methodology reported (Chen et al., 2020), with a critical threshold of approximately 3000 counts set for the intensity of precursor ions to ensure high-quality data acquisition and minimize background noise. The mass spectrometric conditions for various metabolites under the Dynamic MRM mode are provided in Supplementary Table S2.

All data were acquired using Analyst software version 1.7.2 (AB Sciex) and subsequently imported into SCIEX OS software version 2.1.6 (AB Sciex) for peak area integration. Prior to statistical analysis, the raw data were filtered to remove noise and artifacts. Specifically, ions with a signal-to-noise ratio below three were excluded to minimize background interference. The peak areas of the identified compounds were then imported into MetaboAnalyst 6.0 for Pareto scaling normalization. Multivariate statistical analyses (PCA and OPLS-DA) were used to the different groups and determine the altered compounds. OPLS-DA models were validated by confusion matrix, cross validation and CV-ANOVA, while OPLS-DA models were validated by cross validation and CV-ANOVA. The altered compounds between each two groups were screened by variable importance in the projection (VIP) ≥1 from the cross-validated OPLS-DA models. The altered metabolites were analyzed by Metabolomics Pathway Analysis (http://www.metaboanalyst.ca/) and were related to potential pathways. KEGG database (http://www.kegg.jp/) was used to further determine the location and function of these altered metabolites in various metabolic pathways.

Alterations in amino acid, energy, and lipid metabolism are linked to ulcerative colitis (UC) (Wang et al., 2019). We developed an LC-MS/MS method using Dynamic MRM to monitor these changes. Figure 1A shows the distribution of compound classes: amino acids and peptides (32%), organic phosphoric acid derivatives (12%), carboxylic acids and derivatives (7%), fatty acyls (7%), benzene derivatives (6%), purine nucleotides (5%), pyrimidine nucleosides (5%), phenylpropanoic acids (5%), and steroids (4%). Other categories each contribute less than 5%. Quality control assessment, based on the relative standard deviation (RSD) of peak areas in six QC samples, showed that 81.31% of the metabolites had an RSD of less than 20% for peak area in the QC samples (n = 6) (Figure 1B), confirming the reliability and precision of the analytical methodology employed in this study.

Throughout the 7-day treatment period, rats in the Control group exhibited consistent weight gain, whereas those in the Model group experienced a notable weight decline. Both the Mesalazine and Berberine groups initially faced weight loss during the first 2 days of treatment; however, this trend decelerated over the subsequent 2 days and was succeeded by a steady increase during the final 3 days. These observations suggest that both berberine and mesalazine have the capacity to mitigate or potentially reverse the weight loss associated with ulcerative colitis (Figure 2A). The Disease Activity Index (DAI) was employed to evaluate the impact of UC on the rats. As depicted in Figure 2B, the DAI in the Model group showed a progressive increase relative to the Control group. Conversely, the DAI scores for both the Mesalazine and Berberine groups began to diminish by the second day of treatment, indicating that both interventions effectively alleviated the clinical symptoms associated with UC.

As illustrated in Figure 2C, the colon lengths of the Model group were markedly shorter compared to those of the Control, Berberine, and Mesalazine groups (Control group: 16.97 ± 2.48 cm; Model group: 13.22 ± 2.02 cm; Berberine group: 13.38 ± 0.90 cm; Mesalazine group: 13.67 ± 1.97 cm). Beyond the reduction in length, the colons in the Model group presented with evident signs of redness, swelling, and ulceration, contrasting sharply with those in the other groups (Figure 2D).

Histological analysis was conducted to evaluate the extent of damage to the colon tissues. As shown in Figure 3A, hematoxylin and eosin (HE) staining of the Control group revealed no evidence of damage to the colonic mucosal epithelium. In stark contrast, the Model group exhibited extensive inflammatory cell infiltration within the submucosa, along with signs of crypt deformities (Figure 3B). However, the Mesalazine (Figure 3C) and Berberine (Figure 3D) groups demonstrated a marked reduction in congestion and edema compared to the Model group. Moreover, while the Model group displayed prominent inflammatory cell infiltrates and crypt abnormalities, the colons in both the Control and Berberine groups exhibited relatively intact crypt structures and epithelial integrity. These findings indicate that berberine can effectively attenuate the inflammation and colonic damage associated with ulcerative colitis.

As depicted in Figure 4, principal component analysis (PCA) score plots reveal a tendency for the plasma metabolite profiles of the four groups to segregate. The tight clustering of QC samples in both positive (Figure 4A) and negative (Figure 4B) electrospray ionization (ESI) modes indicates excellent instrument stability and reproducibility, ensuring that the observed differences in metabolite profiles are attributed to biological variations rather than technical artifacts. Orthogonal projections to latent structures-discriminant analysis (OPLS-DA) models effectively distinguished between the Control and Model groups (Figures 4C, D), the Model and Mesalazine groups (Figures 4E, F), and the Model and Berberine groups (Figures 4G, H). Validation parameters of the OPLS-DA models, including Q² and R ² values, are provided in Supplementary Figure S1S, indicating a low risk of overfitting and high model reliability.

To elucidate the progression of UC and identify endogenous metabolites that exhibit significant positive changes following treatment. The preprocessing of the raw data was performed as described in Section 2.5. The VIP values, p-values, and fold change (FC) values were calculated using the statistics analysis functions in MetaboAnalyst 6.0. Three p-values was calculated: p1 for the comparison between the Control group and the Model group (C/M), p2 for the comparison between the Model group and the Mesalazine group (M/E), and p3 for the comparison between the Model group and the Berberine group (M/B). ‘Mesalazine-Positive Impact Compounds’ were defined as those with both p1 and p2 were less than 0.05, with consistent trends in C/M and M/E comparisons. This indicated that the concentrations of these metabolites changed significantly in UC rats, suggesting that mesalazine tablets effectively modulate their levels to exert a protective effect. Similarly, ‘Berberine-Positive Impact Compounds’ were identified by both p1 and p3 were less than 0.05, with consistent trends between C/M and M/B comparisons. This suggested that berberine tablets can also directly regulate the levels of these metabolites to provide protection (Figure 5A). To ensure that no beneficial compounds from the treatment groups were overlooked, we screened for compounds under the criteria of VIP >1, p < 0.05, and fold change (FC) > 1.5 or FC < 2/3. The details of these compounds, which can be adjusted to exert protective roles, are summarized in Table 1.

After excluding non-endogenous compounds, a total of 33 altered metabolites were identified in the plasma, primarily categorized as amino acids, pyrimidines, organic phosphoric acids, fatty acyls, and organonitrogen compounds. Heat maps were employed to illustrate the trends in metabolite changes (Figures 5B, C). Disruptions in amino acid metabolism, purine metabolism, vitamin metabolism, and lipid metabolism in both urine and feces are closely associated with the onset of UC (Liao et al., 2019). Notably, while amino acid compounds in plasma exhibited the greatest variability, changes in the concentrations of pyrimidines, organic phosphoric acids, fatty acyl groups, and organonitrogen compounds have been less frequently reported in prior studies. Despite challenges in obtaining standards, a standard reference solution was used to compare the chromatographic peaks of the 16 compounds (Supplementary Figures S2S, S3S).

Endoscopy combined with biopsy remains the gold standard for diagnosing and managing UC. However, imaging techniques have also been developed for monitoring the condition. These methods can be costly and invasive, often causing patient discomfort (Liu et al., 2022). Sensitive and specific biomarkers are therefore essential for the diagnosis and treatment of UC. Notably, metabolites such as 3-hydroxyproline, homocysteic Acid, L-threonine, L-lysine, carbamoyl phosphate, O-phosphoethanolamine, taurine, leucinic acid, and phosphorylcholine exhibited significant differences between the Treatment group and the Model group (p < 0.001), with a trend indicating a return toward Control group levels (Figure 6). These findings suggest that these compounds may serve as potential plasma biomarkers for UC.

Based on the plasma metabolomics results (Table 1) and KEGG pathway analysis (Table 2), both berberine and mesalazine exert therapeutic effects against UC via several metabolic pathways, including arginine biosynthesis, glutathione metabolism, alanine, aspartate, and glutamate metabolism, nicotinate and nicotinamide metabolism, glycine, serine, and threonine metabolism, taurine and hypotaurine metabolism, pyrimidine metabolism, and glyoxylate and dicarboxylate metabolism. However, since specific metabolites such as L-carnitine, spermidine, and shikimic acid were only restored in the berberine treatment group, this suggested that pathways such as the citric acid cycle (TCA cycle) may be uniquely involved in berberine’s anti-UC mechanism.