Materials and reagents

The herbal concentrate-granules of four different plants—Scutellalria baicalensis Georgi (21,072,421), Paeonia lactiflora Pall (21,072,371), Glycyrrhiza uralensis Fisch (21,073,631), and Ziziphus jujuba Mill (21,074,391)—which were obtained from Jiangyin Tian Jiang Pharmaceutical Co., Ltd. CPT-11 (MB1126-2) was purchased from Dalian Meilun. Other chemicals and reagents used in the study, including D-Saccharic acid 1,4-lactone hydrate (SCL, GC43576-100, GLPBIO), 4-Nitrophenyl β-D-glucuronide (PNPG, MB0588-1, Meilunbio), hexadecyl trimethyl ammonium bromide (CTAB, MB3224-1, Meilunbio). Competent cells E. coli DH5α (CD201-02) and BL21(DE3) (Ec1002S) were purchased from TransGen Biotech. Loperamide Hydrochloride (HY-B0418A) was obtained from Xian Janssen Pharmaceutical Ltd. The AKP activity assay kit (A059-2–2) was purchased from Boxbio Science & Technology. The standard compounds baicalein (A0018), baicalin (A0016), and paeoniflorin (A10133) were purchased from Chengdu Mansite and had a purity of 98.0% or higher. Elisa kits for the detection of protein levels of the inflammatory factors IL-1β (MM-0905M2), IL-6 (MM-1011M2), TNF-α (MM-0132M2) were purchased from Jiangsu Meimian Industrial Co., Ltd. Antibodies for Western blotting and immunohistochemistry, including β-actin (T40104M, Abmatr), Claudin-1 (PA2349, Abmatr), Occludin (GB111401, Servicebio), and zonula occludens 1 (ZO-1, #A11417, ABclonal Technology).

Network pharmacological analysis of drug and diarrhea targets

To identify potential targets for the active ingredients, we used the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP) (https://tcmsp-e.com/tcmsp.php) (Ru et al. 2014), the Swiss Target Prediction databases (http://www.swisstargetprediction.ch/) for drug target prediction. GeneCards (https://www.genecards.org/), Therapeutic Target Database (TTD) (https://ngdc.cncb.ac.cn/databasecommons/) (Zhou et al. 2022), and Online Mendelian Inheritance in Man (OMIM) databases (https://www.omim.org/) were used to collect disease targets. Targets were filtered by defining "CPT-11 induced diarrhea" as the keyword. We then integrated the target data and eliminated duplicated genes to obtain a therapeutic target database. Venny2.1 software was used to capture the common target of the compound and CPT-11 induced diarrhea, and these were imported the data into Cytoscape 3.9.1 software to construct an interaction network map of the potential target-compound-pathway.

HPLC

2 g HQD granules were dissolved with 10 mL methanol. One part was centrifuged and the other part was not centrifuged. The two samples were put on the solid phase extraction column, 0.1 mL of each sample solution was taken, and the methanol of chromatographic grade was added to 1 mL. The sample solution was obtained by millipore filtration membrane. The control solution was prepared by dissolving 1 mg of baicalein, baicalin and paeoniflorin with 2 mL of chromatographic grade methanol. InertSustain C18 (4.6 × 150 nm, 5 μm), the mobile phase was 0.1% formic acid-acetonitrile (acetonitrile 0–30 min,10%—70%), the flow rate was 0.4 mL/min, the column temperature was 35 °C, and the injection volume was 5 μL.

Mice diarrhea modeling and administration

Male ICR mice (4—6 weeks) weighing between 20 and 22 g were purchased from the Medical Animal Center of Guangdong (Guangzhou, China) with approval of the Guangzhou University of Chinese Medicine Animal's Care and use committee (Ethics number: 20220720 and 20,221,211). All mice were housed in a specific pathogen-free facility (License number: SYXK (Guangzhou) 2019–0144), at standard temperature (22 ± 2 °C), and humidity (55% ± 2%) and maintained under a 12 h light/dark cycle. The above animal experiments were conducted under the guidelines for laboratory animals drawn up by the National Research Council. Mice were provided with normal drinking water throughout the experiment. The mice were randomly divided into eight groups, with 8 mice in each group: control, CPT-11, Loperamide, HQD, Baicalein, Baicalin, Paeoniflorin, and BBP. From day 1 to day 3, HQD (10 g/kg), Baicalin (100 mg/kg), Baicalin (100 mg/kg), Paeoniflorin (100 mg/kg), BBP (33.33 + 33.33 + 33.33 mg/kg), Loperamide (0.55 mg/kg) were administrated by gavage. On the fourth day, except for the control group, the diarrhea model was established by intraperitoneal injection of CPT-11 (100 mg/kg) for four days. The control group was given 0.9% saline (i.g.) of the same volume every day. The mice were monitored for weight, stool concentration, and stool occult blood daily. On the 10th day, all the mice were euthanized. For analysis of gut microbiota and β-GUS activity, feces from each mouse were collected on day 10 and stored at − 80 °C. The effective doses of HQD, baicalein, baicalin, and paeoniflorin were based on previous studies on the treatment of experimental colitis in mice (Li et al. 2022a; Liu et al. 2015; Yu et al. 2021).

Disease activity index

Disease activity index (DAI) was evaluated by counting data on weight loss, fecal viscosity, and degree of bloody stools in mice (Luo et al. 2021). The degree of bloody stools in mice was assayed with a fecal occult blood test kit. Specific standards are given in Table 1.

Table 1 Disease activity index (DAI) scoring criteriaHematoxylin–eosin (HE) staining

As in previous studies (Yao et al. 2021), we washed the distal colon tissue with 0.9% sodium chloride and fixed it with 4% paraformaldehyde before the HE stain. First, we sectioned the colon tissue embedded in paraffin into a certain thickness (commonly 4–6 μm) and then stained it according to the method of hematoxylin–eosin staining. Histological scores were calculated based on the previous methods (Chen et al. 2011). All photographs were taken at random fields under 100 × and 400 × light microscopy (Nikon Eclipse, Japan).

Real-time PCR analysis

We took 20 ± 2 g of colon tissue, and isolated total RNA from colon samples using a tissue RNA purification kit. Next, we reverse-transcribed RNA using a reverse transcription kit to reverse the transcription of the RNA extracted from the above steps. cDNA was fivefold diluted and gene expression levels were quantified using 2 × SYBR Green Color qPCR Mix. Using β-actin as an internal reference, we analyzed the transcriptional differences of ZO-1, Occludin, Claudin-1, TNF-α, IL-6, and IL-1β (Table 2).

Table 2 Real-time PCR primer sequencesCytokine measurement in colonic tissue

The colons of mice were collected and fully homogenized and then centrifuged at 2000 rpm at 4 °C for 10 min. The supernatants were collected and the levels of interleukin IL-1β, IL-6, and tumor necrosis factors TNF-α in the tissue supernatants were detected by ELISA kit (Meimian, Jiangsu, China).

Alcian blue-periodic acid schiff (AB-PAS) staining

Slides were first dewaxed and rehydrated and then incubated in a 1% iodate solution. After the above steps, the slides were stored in Matt Heath reagent under dark and cold conditions and dried at 25 °C. Stained goblet cells were analyzed under 100 × and 400 × light microscopy (NIKON Eclipse, Japan).

MPO activity assay

As mentioned previously, MPO activity in colon tissue can assess the extent of neutrophil accumulation (Elian et al. 2015). Colon tissues were homogenized with a mixed buffer (pH = 4.7) containing 0.015 M NA2-EDTA, 0.02 M NaH2PO4·1H2O, and 0.1 M NaCl. The homogenate was centrifuged (12,000 rpm, 10 min), and the precipitate was lysed (12,000 rpm, 10 min) for 10 min and then centrifuged (12,000 rpm, 10 min). The suspension was resuspended at 0.05 M sodium bisulfate (pH = 5.4) containing 0.5% cetyltrimethylammonium bromide, frozen in liquid nitrogen for three times and then centrifuged (12,000 rpm, 10 min). The optical density of the supernatant at 450 nm was determined.

Immunohistochemical (IHC) staining

The colonic tissues were evenly sectioned (usually 4–6 μm) in each group. It was repaired in an antigen retrieval buffer containing citric acid (pH 6.0) after the removal of paraffin and rehydration. 0.3% H2O2 was used to suppress the endogenous peroxidase activity for 20 min at 25 °C. It was then blocked with 3% BSA at 25 °C for 30 min. Finally, it was incubated with ZO-1 at 4 °C overnight. The histological section was probed with newly prepared 3,3′-N-diaminobenzidine tetrahydrochloride solution and counterstained with hematoxylin after incubation with the peroxidase-biotinylated secondary antibody. All photographs were taken at random fields under 100 × and 400 × light microscopy (NIKON Eclipse, Japan).

Immunofluorescence staining

The distal tissues of the colon were fixed with paraformaldehyde, and then the paraffin sections were dewaxed to water. After antigen retrieval, the sections were dried, drawn a circle around the tissue with a histochemical pen, and then blocked with 3% BSA for 30 min. Primary antibody was added to tissue sections and incubated at 4 °C overnight. The first antibody was washed and the second antibody was added. The nuclei were counterstained with Dapi to quench the tissue autofluorescence. After sealing, the sections were observed under a fluorescence microscope (Pannoramic MIDI, 3D histech, HUN).

Intestinal permeability analysis

We used FITC-dextran (MW 4000) as a tracer to study the effect of CPT-11, HQD, and active components on intestinal permeability in vivo. Mice were fed with FITC-dextran (440 mg/kg) in the morning after fasting for 12 h before the experiment. Four hours later, we immediately euthanized the mice and quickly collected their blood. Each sample was diluted in PBS and divided into three equal parts for analysis. FITC-dextran concentration was then determined by spectrophotofluorometry (Fluoroskan Microplate Fluorometer, Thermo Fisher Scientific) at 490 nm excitation wave and 520 nm emission wave. The sample was prepared by continuous dilution method and determined its fluorescence intensity to establish the FITC-dextran standard curve.

Western blot

Mouse tissue proteins were extracted using RIPA lysates. Proteins were separated on 10% polyacrylamide gels (SDS-PAGE) and then transferred to polyvinylidene difluoride (PVDF) membranes, blocked with 5% BSA for 2 h at room temperature, and incubated with specific primary antibodies overnight at 4 ℃. The blots were detected with an HRP-conjugated secondary antibody. The proteins were visualized by ECL detection reagents.

Mendelian randomization analysis

eQTLL(expression quantitative trait loci) of TOP1MT was retrieved from eqtlgen database (https://www.eqtlgen.org) as exposure data. GWAS data "ebi-a-GCST90016936" for Enterobacteriaceae, "ebi-a-GCST90011328" for Clostridium XIVa and "ebi-a-GCST90016996" for Eubacteria were used as outcome data. The 2SMR analysis was performed using the "TwoSampleMR" package, version 0.5.11. First, the European population was selected with r2 = 0.3 and kb = 100, and linkage disequilibrium was handled to ensure that the instrumentality variables met independence. The inverse-variance weighted (IVW) method was used as the main evaluation method, and the weighted median (Wmedian) was used as the main evaluation method. WM), weighted method and MR-Egger to improve the reliability of causal inference. P < 0.05 was used as the criterion for statistical significance of 2SMR studies. Mr-presso was used to test for horizontal genetic pleiotropy of exposure factors and outcome variables, Leave-one-out method was used for sensitivity analysis to test for the presence of SNPS with excessive influence on MR Estimates, and Cochran Q test was used to evaluate heterogeneity.

16S rRNA gene sequencing and analysis of gut microbes

Feces of mice in each group were collected on day 9 and transferred to sterile tubes, flash frozen in liquid nitrogen, and stored at − 80 °C. After genomic DNA extraction from the samples, the conserved region of rDNA was amplified with specific primers with barcode. The PCR amplified products were cut and recovered, and quantified using a QuantiFluorTM fluorometer. The purified amplification products were mixed in equal amounts, connected to a sequencing adapter, and the sequencing library was constructed and sequenced on an Illumina PE250. After raw reads were obtained by sequencing, the low-quality reads were first filtered and then assembled, the double-end reads were spliced into tags, and then the tags were filtered. The obtained data was called Clean tag. Next, clustering was performed based on Clean tag to remove the chimera tag detected during the clustering process, and the final data obtained was Effective tag. After OTU was obtained, OTU abundance statistics were performed based on Effective tag. Species annotation, α diversity analysis, β diversity analysis, and community function prediction were performed according to the analysis process.

AKP analysis

Bacteria were collected via centrifugation, and then 5 million bacteria were added into 1 mL of the extract solution for ultrasonic crushing. After centrifugation (8000 rpm, 10 min) at 4 °C, the supernatant was collected and detected. The supernatant was taken out and placed on the ice for testing after centrifugation at 4 °C (8000 rpm, 10 min). Preheat the photometer for at least 30 min, adjust the wavelength to 680 nm, and set the distilled water to zero. An AKP activity assay kit was used to detect AKP content.

In vitro inhibition of E. coli

E. coli was cultured to OD = 0.8 and diluted ten times. The diluted liquid of 2 μl was applied to LB solid medium, and then the same volume of the LB medium was added to the liquid as the negative control. Similarly, 25 mg/ml Amp was used as the positive control. We then used 0.25 ~ 1.5 g/ml HQD, 10 mg/ml baicalin, 10 mg/ml baicalin, and 10 mg/ml paeoniflorin for the bacteriostasis test.

Microscopic characteristics

We first expanded the culture of E. coli in the Luria–Bertani medium. We Gram-stained with a light microscope to differentiate cell types. In short, we immobilized E. coli in a super-clean table and then stained it with crystal violet oxalate for one minute, then washed it with distilled water and dried it with absorbent paper. We covered the coating with an iodine solution for about one minute, then washed and dried it as before. We then added alcohol for decolorization, 20 s after washing and drying. After dying with Safranin o solution (0.5%) for 1 min, we rinsed cells with distilled water. Finally, we examined specimens with a dry mirror.

SEM analysis

E. coli cells with OD = 0.6 were mixed with 1000 mg/mL HQD, 10 mg/mL baicalein, 10 mg/mL baicalin, and 10 mg/mL paeoniflorin and then centrifuged for precipitation (4000 rpm, 5 min). Control cells were incubated with an equal amount of medium. All the samples were resuspended washed with saline solution buffer 2 ~ 3 times and then fixed with 1 ml of 2.5% em fixative solution at 4 °C overnight. The fixed samples were resuspended and washed 2–3 times in saline solution again beforehand dehydrated in gradient ethanol. followed by drying in the Critical Point Dryer (Quorum). Finally, they were observed under the Scanning Electron Microscope (Hitachi).

Imaging of intestinal β-GUS activity

We used FDGLcU to detect the difference in intestinal β-GUS activity in mice. The mice were gavaged with 7.3 μmol/kg FDGLcU after the last CPT-11 administration(Chen et al. 2017). The mice were euthanized immediately before imaging. The product of β-GUS-catalyzed hydrolysis of FDGlcU generated fluorescence at 520 nm and was imaged using an IVIS spectral imaging system (Bruker). We measured intestinal fluorescence to quantify the activity of β-GUS in the gut.

β-GUS-producing bacteria detection in feces

To detect changes in β-GUS-producing bacteria in the intestines of mice after administration, we collected feces on the tenth day. Treated feces were inoculated on a 4-Methylumbelliferyl beta-D-glucuronide agar plate and cultured for 12 h. Briefly, we collected 60–80 mg of feces, homogenized it, and centrifuged it (3000 rpm, 10 min). To collect the sediment, we re-centrifuged the supernatant (12,000 rpm, 15 min). Finally, all samples were resuspended in saline solution and coated on a 4-MUG culture plate. The abundance of β-GUS-producing bacteria was detected via UV.

β-GUS enzyme activity assay

To explore the effects of HQD, baicalein, baicalin, and paeoniflorin on intestinal β-GUS, we performed PNPG hydrolysis experiments on mouse fecal samples (Weng et al. 2017). First, 60 ± 5 mg of feces were taken from each group. And then, fecal particles were homogenized in PBS buffer and then centrifuged (12,000 rpm, 20 min). The supernatant was collected and the protein content was determined by the Bradford method. The hydrolysis activity of PNPG by fecal β-GUS was detected at 405 nm. After mixing 1.0 mg/mL β-GUS from E. coli (EcGUS), 250 μm PNPG and 100 μm compound, the reaction was carried out. To stop the reaction, we added 0.2 M Na2CO3 immediately after 30 min of incubation at 37 °C, and absorbance was measured at 405 nm. All analyses are performed in triplicate.

Mutant construction, protein expression, purification and enzyme activity assays

The EcGUS gene (Gene ID: WP_270671058.1) was successfully cloned into the pET28a ( +) expression vector (Novagen, Germany) with an 8xHis tag. We transferred the recombinant plasmid into E. coli BL21 (DE3). EcGUS overexpression was induced by the addition of 0.2 mM isopropyl β-D-1-thiopyranoside. Then, the target protein was purified by a nickel-nitrilotriacetic acid column. The protein was dialyzed into the PBS buffer. Nine mutants of N412A, N412T, E413, E412L, Y468A, Y468F, Y468K, E504A, and G569A were constructed by site-directed mutagenesis using EcGUS as a template. The enzyme activity assay method can be referred to 2.21.

Molecular docking and dynamic simulation

We obtained the crystal structure of EcGUS (PDB ID: 3LPF, derived from E. coli K12) by searching the protein database, and the molecular docking of EcGUS with PNPG, baicalein, and baicalin was performed using AutoDockTools. In brief, the receptor (PNPG, baicalein, and baicalin) is dehydrated and the unwanted hetero atoms are removed by using Pymol software, and the state of charge and protonation of the receptor molecule EcGUS is confirmed, molecular docking was then performed using AutoDockTools. The Protein–Ligand Interaction Profiler online pages (https://plip-tool.iotec.tu-dresden.de/plip-web/plip/index) were used to graph the docking of complexes. The structures of apo EcGUS, complex EcGUS with baicalein, and complex EcGUS with baicalin were used for the dynamic simulation. First, the small molecule's force-field parameters were compiled by Acpype programs (Sousa da Silva and Vranken 2012), GAFF force field was applicated. Protein parameters were prepared by AMBER99SB protein field parameters. Gromacs v2020.4 software was used in Molecular dynamics simulations. A module with a preparation parameter is loaded onto the protein receptor and ligand molecule addition. A dynamic simulation was carried out in the TIP3P explicit water model after setting the number of periodic boundary conditions. After layer-by-layer optimization, each system submitted a 50 ns dynamic simulation. The trajectory coordinates are saved every 10 s during the simulation. The root means square deviation (RMSD) and root mean square fluctuation (RMSF) was calculated by the Gromacs v2020.4 software (Spoel et al. 2005). The structure of the last frame (50 ns) was obtained in pdb format. IGM (Independent Gradient Model) software was used to provide insight into concrete interaction between Y468 and baicalein/baicalin particularly, Multiwfn v3.8 (dev) software was used to perform a calculation (Lu and Chen 2012).

Data analysis

We used GraphPad Prism 9.0 (San Diego, CA) software for statistical analysis. Student's t-test was used to compare the differences between the two experimental groups. When multiple sets of independent sample data need to be compared, analysis of variance is used for comparison. All data in this study were presented as mean ± SD. p < 0.05 was considered to be significantly different between the two groups.