Astragalus mongholicus Bunge and Wolfiporia cocos Ryvarden & Gilb were bought from Tiancheng prepared herbal medicine Co., Ltd. (Guangdong, China; batch number 221201 and 220901); Colla Corii Asini Pingstone was purchased from Shandong Dong-E E-Jiao Co., Ltd. (Shandong, China; batch number 2208003); Panax notoginseng Burkill was obtained from Baiyunshan Pharmaceutical Co., Ltd. (Guangzhou, China; batch number YPB2K0001); Strobilanthes cusia Kuntze was supplied from Qianfang Chinese Herb Slices Co., Ltd. (Sichuan, China; batch number 202204026). Mesalazine was procured from Shanghai Aidifa Pharmaceutical Co. Ltd. (Shanghai, China). Dextran sodium sulfate (DSS) was purchased from MP Biomedicals (San Diego, United States). CH223191 was obtained from MedChemExpress (New Jersey, United States). The GoldHi Plasmid Mini Kit was procured from Cowin Biotech Co. Ltd. (Jiangsu, China). Goat anti-mouse/rabbit antibodies were obtained from Bioss Biotechnology Co. Ltd. (Beijing, China). Mouse LPS, IL-22 and IL-13 ELISA kits were supplied by Enzyme Industry Co. Ltd. (Jiangsu, China). AhR, Occludin, ST2, Claudin2, ZO-1 and CYP1A1 antibodies were purchased from Proteintech (Chicago, IL, USA). CD16/32 and fluorescein isothiocyanate (FITC)-lineage antibodies were procured from Tonbo Biosciences (San Diego, United States). PE-CD45, Brilliant Violet 421-GATA3, APC-IL-13, and PerCP/Cy5.5-IL-22 antibodies were purchased from Biolegend (San Diego, United States). The PE-Cy7-RORγt antibody was purchased from Thermo Fisher Scientific (Waltham, MA, USA). All antibodies used were anti-mouse.

MJQP is composed of Astragalus mongholicus (12 g), Wolfiporia cocos (12 g), Colla Corii Asini (3 g), Panax notoginseng (6 g), and Strobilanthes cusia (2 g). MJQP was prepared as follows: Astragalus mongholicus, Wolfiporia cocos and Strobilanthes cusia were submerged in distilled water for 30 min and decocted twice for 40 min. Then, Panax notoginseng and Colla Corii Asini were added to dissolve completely, and the liquid was concentrated to produce MJQP, which had a 1 g/mL crude drug concentration.

1 mL of the sample was added to 2 mL of the extraction solution of methanol-acetonitrile (1:1, v/v) and ultrasonicated for 30 min. The supernatant was collected after centrifugation (12,000 rpm, 4 °C, 20 min). The sample was incubated at −20 °C for 1 h, then centrifuged (12,000 g, 4 °C, 10 min). The supernatant was dried in a vacuum centrifuge and then re-dissolved in 100 μL of 30% ACN for UPLC–MS/MS analysis.

UPLC-Orbitrap-MS system (UPLC, Vanquish, MS, and HFX) was used to analyses the sample extracts. The UPLC conditions were as follows: column, Waters HSS T3 (100*2.1 mm, 1.8 μm); injection volume, 2 μL; flow rate, 0.3 mL/min; column temperature, 40 °C; solvent system, water (0.1% Formic acid): acetonitrile (0.1% Formic acid); gradient program, 100: 0 V/V at 0–1 min, 5: 95 V/V at 12 min, 5: 95 V/V at 12–13 min, 100: 0 V/V at 13.1–17 min. A Q Exactive HFX Hybrid Quadrupole Orbitrap mass spectrometer system (Thermo Fisher Scientific) was used to LC-MS/MS analysis.

A total of 80 male C57BL/6 mice (approximately eight weeks age) were supplied by the Guangdong Medical Laboratory Animal Center (Guangdong, China). The experimental animals were nourished on a conventional diet and water and kept in specific pathogen-free surroundings at a temperature of 22 °C ± 2 °C, humidity of 55% ± 2% and a 12-h light/dark cycle. Animal studies were conducted in strict compliance with the Ethics Committee of Animal Experiments of The First Affiliated Hospital of Guangzhou University of Chinese Medicine. Ethical clearance number: GZTCMF1-2023004.

The following groups were designed: control, model, positive drug group (mesalazine at 400 mg/(kg i.g. d), MJQP low-, medium-, and high-dose [2.5, 5, 10 g/(kg i.g. d)] groups, AhR antagonist CH223191 (10 mg/kg i.p. d) combined administration with MJQP (5 g/(kg i.g. d) (CH + M) group, and AhR antagonist CH223191 group (10 mg/kg i.p. d). All mice were randomly assigned to these groups. Chronic UC mice were induced by freely drinking water containing 2% DSS for 4 days following by water for 2 days, for 4 cycles in total, referring to the modeling procedure reported by Huang (Huang et al., 2022). All the mice were anesthetized by intraperitoneal injection of 0.3% pentobarbital sodium and then euthanized by cervical dislocation.

Throughout the experimental period, weight loss, diarrhea, and fecal blood of all mice in different groups were recorded on alternate days to monitor and evaluate the changes in the disease activity index (DAI), as previously described (Cooper et al., 1993). On the last day, the spleen and thymus of the mice were separated after euthanisation. Colonic length was measured and organ indices were computed.

The colon fraction was preserved for 24 h by submerging in 4% paraformaldehyde. Subsequently, the samples were prepared in accordance with standard procedures for hematoxylin and eosin staining (H&E) and then observed using an inverted microscope (Olympus IX73).

The colon was ground in TRIzol reagent using a cryogenic tissue grinder (SCIENTZ-48 L). The GoldHi Plasmid Mini Kit was used to extract the total RNA. Next, the RNA was mixed with 4× RT Master Mix and gDNA Remover reagent for reverse transcription to cDNA. Finally, 2× Color SYBR and primers (shown in Table 1) were added to cDNA, and the mixture was amplified by real-time qPCR (Applied Biosystems QuantStudio 5).

For extraction of colonic protein, colon tissue was treated with moderate RIPA containing 1% PMSF, homogenized in a cryogenic tissue grinder (SCIENTZ-48 L), centrifuged, and then the supernatant was collected, which contained the colonic protein. After routine electrophoresis and transfer, the membranes were sealed with 5% skim milk and incubated with the corresponding primary antibodies, such as AhR antibody (1:2000 dilution), CYP1A1 antibody (1:2000 dilution), ST2 antibody (1:2000 dilution), ZO-1 antibody (1:5000 dilution), Occludin antibody (1:5000 dilution) and Claudin2 antibody (1:2000 dilution), and secondary antibodies (1:10000 dilution). Finally, the membrane images were exposed using a chemiluminescence imager (Bio-Rad, Chemidoc MP) and analyzed using the ImageJ software.

After dewaxing the paraffin sections of colon, 30 μL alcian blue solution was added to the sections for staining; the sections were washed with distilled water three times, each time for 10 s; 30 μL nuclear fast red solution was added for staining, then washed with tap water for 5 min. After the section underwent routine dehydration process, it was covered with 20 μL of neutral balsam for sealing, and then observed under an inverted optical microscope (Olympus IX73).

The paraffin sections of colon were routinely dewaxed, immersed in 10 mM sodium citrate buffer and heated for antigen retrieval. After 0.3% Triton was added for permeabilization and 10% goat serum was added to seal, the tissues were incubated with primary antibodies. The next day, the tissues were examined using a fluorescence microscope (Olympus IX73) after being treated with the matching fluorescent secondary antibodies and coated with DAPI.

Cell suspensions of mesenteric lymph nodes were prepared by mechanical grinding and filtered through 100, 70 μm cell filtration membranes for collection a flow tube. The CD16/32 antibody blocks nonspecific binding of Fc receptors before cell labeling. The surface markers of ILCs include lneage⁻ (which includes CD11b, CD3, Ly-6G, and B220) and CD45⁺, which were stained using FITC-lineage antibody and PE-CD45 antibody. Cells were then fixed and permeabilized, and nuclear staining was performed with antibodies, including Brilliant Violet 421-GATA3, APC-IL-13, PerCP/Cy5.5-IL-22, and PE-Cy7-RORγt. Finally, the proportions of IL-13⁺ ILC2 and IL-22⁺ ILC3 were analyzed by flow cytometry (FACS LSRFortessa).

The secretion levels of TNF-α, IL-1β, LPS, IL-13 and IL-22 were determined using ELISA kits.

16S rDNA sequencing and targeted tryptophan metabolite detection in fecal samples were conducted by the Shanghai BIOTREE Biomedical Technology Co., LTD (Shanghai, China). Briefly, the 16S rRNA gene V4 was targeted for DNA extraction from fecal samples, and an amplified library was generated for bilateral sequencing using Illumina MiSeq technology. Trimmomatic was used to filter the sequenced raw reads. Next, to obtain high-quality reads, primer sequences were located and eliminated. The overlapping sequences were subsequently spliced using FLASH v1.2.7 software to acquire clean reads. To obtain the final valid data for further analysis (splitting features, diversity analysis, species analysis, difference analysis, and correlation analysis), chimera sequences were examined and eliminated using UCHIME software.

For metabolite extraction, fecal samples were weighed and approximately 50 mg was placed into a tube. The samples were homogenized (35 Hz, 4 min) and sonicated for 5 min after 500 μL of the extract solution was added. After repeating the homogenate and sonication cycle twice, the temperature was lowered to −40 °C. Following 15-min centrifugation (12,000 rpm, 4 °C), 320 μL of the supernatant was transferred to another tube. Subsequently, the supernatant was dried off under a nitrogen stream and then mixed with 80 μL of 0.1% formic acid for reconstitution. After a 15-min centrifugation (12,000 rpm, 4 °C), UHPLC–MS/MS analysis was conducted on the supernatants.

All data were analyzed with GraphPad Prism 8.3 software and shown as mean ± standard deviation (SD). Two separate groups were compared using Student’s t-test, and group comparisons were conducted using one-way analysis of variance (ANOVA). The threshold for a statistically significant difference was p < 0.05.

To characterize the material basis of MJQP, UPLC-MS/MS analysis was used. Figures 1A,B showed the positive and negative extracting ion chromatogram of MJQP. The compounds were characterized by matching with retention time, molecular mass, and secondary fragmentation spectra in databases. The results showed that prenol lipids, fatty acyls, organooxygen compounds, flavonoids, carboxylic acids and derivatives were main components with percentage as 25.7%, 11.19%, 8.92%, 7.17%, and 6.12%, respectively. Among them, the main chemical components were flavonoids, such as quercetin, ononin, calycosin-7-O-β-D-glucoside, formononetin, liquiritigenin, kaempferol. The full lists of flavonoids in MJQP were shown in Supplementary Table S1.

To assess the effect of MJQP in chronic colitis mice, we treated them with three different doses of MJQP and mesalazine. Compared with mice in the control group, DSS resulted in classic UC symptoms in model mice, such as bloody stool, significant weight-decrease, increased DAI score, decreased thymus index, increased spleen index, constricted colon (a marker of inflammation), and the elevated levels of pro-inflammatory cytokines TNF-α and IL-1β in the serum. By sharp contrast, MJQP and mesalazine administration improved the weight loss, constricted colon and thymus index, reduced DAI score and spleen index, and decreased the levels of TNF-α and IL-1β in the serum (Figures 2A–E,H–L). In addition, UC-related colonic histological alterations were found in model mice, including erosion and ulceration, epithelial cell depletion, and the muscularis with deep mucosal lymphocytosis (Figures 2F,G), which were relieved in medium and high-dose of MJQP as well as mesalazine-treated mice, along with a lower histological score as compared to the model mice (Figures 2F,G).

According to the ELISA results, model mice had higher levels of LPS in response to DSS stimulation, which decreased after treatment with MJQP and mesalazine (Figure 3A). Moreover, mice in model group showed significantly reduced Muc2 mRNA expression and mucin secretion (Figures 3B,C), as well as the epithelial TJ rupture presented as decreased levels of ZO-1, Occludin mRNA and protein, and increased level of Claudin2 (Figures 3D–K). The intestinal epithelial integrity was effectively enhanced by MJQP and mesalazine compared with the model group, which promoted Muc2 mRNA expression and mucin secretion (Figures 3B,C), down-regulated the mRNA and protein expression of Claudin2, and up-regulated the expression of ZO-1 and Occludin (Figures 3D–K). Transmission electron microscopy (Figure 3L) consistently showed TJ repair in the intestinal epithelium of MJQP-treated mice, indicating that MJQP is capable of restoring the intestinal barrier integrity. Together, the aforementioned findings showed that MJQP repaired intestinal barrier damage to alleviate chronic colitis in mice.

Innate lymphocytes are crucial participants in mediating pathogen immunity and maintaining intestinal immune homeostasis (Saez et al., 2021). ILC2/ILC3 imbalance leads to aberrant release of the cytokines IL-13 and IL-22, which causes intestinal disruption. Therefore, we examined how ILC2, ILC3 and their effectors change in the colons of chronic colitis mice. The proportion of IL-22⁺ ILC3 decreased dramatically, whereas that of IL-13⁺ ILC2 increased noticeably according to the flow cytometry results (Figures 4A–C). Correspondingly, the levels of IL-22 significantly reduced while IL-13 contents raised in colon (Figures 4D–G). Following medium and high-dose of MJQP intervention, there was a discernible increased ratio of IL-22⁺ ILC3, as well as a corresponding decreased ratio of IL-13⁺ ILC2 in chronic colitis mice (Figures 4A–C). Moreover, MJQP dose-dependently decreased IL-13 contents, and increased IL-22 contents (Figures 4D–G).

A previous study reported that AhR controls the function of ILC2 and ILC3 to maintain their balance (Li et al., 2018). Therefore, we investigated the effects of MJQP on AhR signaling in mice with chronic colitis. Compared to the control group, the expression of AhR and CYP1A1 (AhR’s downstream target gene) in the colon of the model group were significantly decreased, whereas ST2 expression was increased. MJQP could reverse these changes in a dose-dependent fashion (Figures 5D–G). Correspondingly, MJQP-treated groups displayed decreased ST2 mRNA expression and elevated AhR and CYP1A1 mRNA expression in comparison to the model group (Figures 5A–C). These findings indicated that MJQP regulates the AhR signaling-mediated ILC2/ILC3 balance, which be related to its beneficial effects on the intestinal barrier.

We then added the AhR antagonist CH223191 as a pretreatment before MJQP administration (CH + M) to investigate whether the effect of MJQP on intestinal barrier repair is AhR-dependent. The result showed that mice in the CH + M group showed more severe UC symptoms compared to the MJQP-treat group. In particular, AhR antagonists resulted in increased DAI score, constricted colons, considerable spleen swelling, and the elevated levels of pro-inflammatory cytokines TNF-α and IL-1β in the serum (Figures 6A–E,H–L). Meanwhile, H&E staining result showed that CH223191 weakened MJQP’s effect on eliminating ulcerative erosion, reducing mucosal lymphocyte infiltration and improving mucosal structure (Figures 6F,G). In addition to the observations above, CH223191 also increased the contents of LPS in serum of mice (Figure 7A). Besides, disrupted intestinal barrier structures were observed in CH + M group mice, which were correlated with considerably decreased mucin expression (Figures 7B,C), lower levels of ZO-1, Occludin expression, and higher levels of Claudin2 expression in comparison to the MJQP administration group (Figures 7D–K). Consistent with the protein results, transmission electron microscopy revealed the TJ of intestinal epithelium to be breaking down (Figure 7L). These results implied that AhR is necessary for MJQP to repair intestinal barrier damage in chronic colitis mice.

We further examined whether AhR was required for MJQP-modulated ILC2/ILC3 balance. As shown in Figures 8A–D, AhR antagonist CH223191 combined administration with MJQP (CH + M) could significantly elevate IL-13⁺ ILC2 proportion and IL-13 contents, while reduced IL-22⁺ ILC3 proportion and IL-22 levels when compared to MJQP-treated group. Moreover, CH223191 significantly increased the protein and mRNA expression of ST2, while obviously decreased the expression of CYP1A1 in the colon of the CH + M group compared with the MJQP-treated group (Figures 8E–I). The results collectively suggested that the role of MJQP in modulating ILC2/ILC3 balance to repair intestinal barrier in chronic colitis mice is dependent on AhR signaling pathway.

The primary endogenous ligand sources of AhR are indole derivatives with AhR activity, which originate from tryptophan, metabolized by the intestinal microbiota. To explore whether MJQP enhances the AhR signaling pathway by restoring tryptophan metabolites from the gut microbiota, we examined variations in the intestinal microbiota composition and tryptophan metabolites. As illustrated in the Venn diagram, the control, model, and MJQP groups had 1,634, 2,341, and 1812 individual operational taxonomic units (OTUs) respectively, and shared 489 identical OTUs (Figure 9A). The rarefaction curve directly reflects the sufficiency and richness of a sample. As shown in Figure 9B, the curve tended to be flat, indicating that the sample size was reasonable and species-rich. Besides, principal coordinate analysis (PCoA) showed that the intestinal microbiota structure varied considerably between the control and colitis groups, while there was little difference between the two colitis groups (Figure 9C). We further examined the bacterial composition. As depicted in Figures 9D–H, compared to the control group at the phylum level, the reduced abundance of Firmicutes and Bacteroidetes, and the elevated abundance of Actinobacteria and Proteobacteria were observed in the model group. However, MJQP markedly augmented the abundance of Firmicutes and diminished that of Actinobacteria and Proteobacteria compared to the model group. When compared with the model group, the abundance of Lachnospiraceae_NK4A136_group, Lactobacillus, Lachnospiraceae_unclassified, Clostridiales_unclassified, Clostridia_UCG-014_unclassified, Clostridium were considerably rised in MJQP group at the genus level (Figures 9I–O), and that of Escherichia-Shigella, Coriobacteriaceae_unclassified, Bifidobacterium, Erysipelatoclostridium and Bacteroides were observably reduced (Supplementary Figure S1). Noteworthily, Lachnospiraceae, Lactobacillus, Clostridia and Clostridiales are essential for the metabolism of tryptophan.

We next observed alterations in tryptophan metabolites (particularly indole derivatives) associated with the intestinal microbiota in the mouse feces. In the model group, the level of Indole-3-acetic acid (IAA), Indole-3-propionic acid (IPA), Indole-3-carboxaldehyde (ICA), Indole-3-acetamide (IAM), and kynurenic acid (KYNA) were decreased, and that of L-tryptophan (Trp) was upregulated compared to the control group (Figures 10A–F), implying that DSS-induced chronic colitis mice suffered tryptophan metabolism disturbances related to the gut microbiota. After MJQP intervention, the concentrations of IAA, IPA, and KYNA were increased, and that of Trp was reduced in comparison with the model group, indicating that MJQP may enhance AhR signaling by restoring tryptophan metabolites (as endogenous AhR ligands) associated with the gut microbiota. There were no significant changes in the levels of the other tryptophan metabolites (Supplementary Figure S2). Additionally, the correlation analysis of IAA, IPA, KYNA, IL-13⁺ ILC2, IL-22⁺ ILC3 were executed, respectively. The results indicated that IL-22⁺ ILC3 were positively correlated with IAA, IPA, and KYNA (Figures 11A–C). On the other hand, negative correlations were discovered between IL-13⁺ ILC2 and the aforementioned metabolites (Figures 11D–F). Collectively, MJQP restored intestinal microbiota-driven tryptophan metabolites, which could act as potential endogenous ligands of AhR to enhance the AhR pathway and regulate ILC2/ILC3 balance to achieve therapeutic effects in chronic colitis.