Exploring the mechanism and key active components of Gegen Qinlian decoction combined with XELOX in the treatment of ulcerative colitis-associated colorectal cancer based on network pharmacology and multi-omics analysis

2.1 Reagents and drug preparation

1,2-dimethylhydrazine (DMH), 2,4,6-trinitrobenzenesulfonic acid (TNBS) were obtained from Sigma-Aldrich (St. Louis, MO, USA). Distilled water was got from Wahaha Group Co. Ltd. (Hangzhou, China). Ethanol, pentobarbital sodium and physiological saline were obtained from Yuwang Co. Ltd (Shandong, China). Methanol, formic acid and acetonitrile were purchased from Fisher Scientific (Newd Jersey, USA). The ELISA kit (TNF-α, VEGF, IL-6, SOD, IL-10, MDA, IL-1β) was purchased from Shanghai MLBio Biotechnology Co., Ltd. (Shanghai, China). Oxaliplatin and capecitabine was obtained from Qilu Pharmaceutical Co., Ltd (Shandong, China). Protein for Western blot: ALOX15 (ab244205, Abcam), CYP1B1 (ab185954, Abcam), PTGS2 (ab179800, Abcam); Secondary antibody (bs-40295G-HRP; Biomass), rapid blocking solution (Genefist, Oxfordshire, UK).

According to the prescription ratio (8:3:3:2), Pueraria lobata (Willd.) Ohwi (Gegen), Scutellaria baicalensis Georgi (Huangqin), Coptis chinensis French (Huanglian) and Glycyrrhiza uralensis Fisch (Gancao) were immersed in 8 times amount of water for 30 min. First, Pueraria lobata (Willd.) Ohwi (Gegen) was boiled for 20 min. Next, the remaining medicine was boiled (40 min) for twice and filtered through a 200-mesh screen. Then, the two filtrates were combined and concentrated under pressure at 50 ℃. GQD was administered to rats at 2.16 g/kg (low dose, equivalent to half of the dose of GQD commonly used in clinical patients) and 8.64 g/kg (high dose, equivalent to the twice dosage of GQD commonly used in clinical patients).

2.2 Animals experiment

Six-week-old Sprague-Dawley rats (n = 86, males, weighing 180–220 g) were purchased from the Laboratory Animal Center of Shenyang Pharmaceutical University. The animals were raised in a dedicated SPF standard room with a temperature of 22 ± 2 ℃, humidity of 40–60 %, and natural light and dark cycles from 7 am to 7 pm. All 86 rats were free to eat and drink for 7 days, and were deprived of water for 12 h before the experiment. The animal procedures were executed according to the SYPU Ethics Committee and carried out in line with the SYPU Guidelines for Animal Experimentation (SYPU-IACUC-S2022-07.28-201).

1. After 7 days of adaptive feeding, 6 rats were randomly selected to collect blank plasma in the EDTA tube (orbital blood collection). Then GQD was given intragastric administration (8.64 g /kg) once a day for 7 days. Plasma samples were collected (orbital blood collection) at 0.5 h, 1 h, 2 h and 4 h after the last administration for subsequent analysis.

2. The remaining rats were randomly divided into 10 groups (n = 8): (1) control; (2) UC (after anesthesia with 30 mg/kg pentobarbital sodium, rats were given enema by injecting TNBS (100 mg/kg)-50 % ethanol mixture into the anuses of the rats); (3) UL (low-dose GQD to treat UC); (4) UH (high-dose GQD to treat UC); (5) CRC (Intraperitoneal injection of 30 mg/kg DMH once a week for 13 weeks, continued to induce cancer based on UC); (6) RL (low-dose GQD for the treatment of CRC); (7) RH (high-dose GQD for the treatment of CRC); (8) XELOX (4 mg/kg Oxaliplatin intraperitoneal injection once a week and 150 mg/kg capecitabine intragastric injection once a day for CRC rats); (9) RXL (low-dose GQD combined with XELOX for the treatment of CRC); (10) RXH (high-dose GQD combined with XELOX for the treatment of CRC).

After 12 h of fasting, rats were given TNBS (100 mg/kg)-50 % ethanol enema to induce UC, UL and UH rats were given GQD intragastric administration for 7 days, and daily weight was registered. On the 7th day after modeling, the disease activity index (DAI) score was recorded. For specific scoring criteria, please refer to Table S1.

After the last dose, orbital venous blood of each group was collected in coagulant tubes and centrifuged at 4000 rpm for 10 min to collect serum samples. After the colon contents were collected, the colon was cleaned with physiological saline. Then, the colonic mucosal damage index (CMDI) was scored by observing the gross morphology of colon tissues. The scoring criteria and results were shown in Table S1. Part of colon tissues were clipped for Hematoxylin-eosin (HE) staining, and the remaining were immediately frozen at −80 ℃.

Then, CRC, RL, RH, XELOX, RXL and RXH groups were given DMH (30 mg/kg) by intraperitoneal injection once a week for 13 weeks to continue to induce CRC. The rats in RL, RH, RXL and RXH groups were given GQD, while the rats in XELOX group were given the XELOX continuously for 6 weeks. Sample collection methods were the same as above.

2.3 HE staining

Colon tissues were fixed with paraformaldehyde (4 %). More than 24 h later, paraffin was adopted for embedding and section. After dewaxing, hydration and HE staining were carried out successively. Ultimately, the slices were observed under a 200-fold microscope and photographed.

2.4 Component identification experiment

1. Preparation of GQD for the analysis of vitro components: GQD was prepared according to the prescription ratio (8:3:3:2), and the filtrate was concentrated at 50 ℃ to the relative density of 1.05. A certain volume was taken and diluted with an appropriate amount of methanol, which was equivalent to 0.1 g crude drug per 1 mL. The liquid was filtered by 0.22 μm for instrumental use.

2. Preparation of plasma samples for the analysis of components in vivo: 1 mL plasma sample mixed with 3 mL methanol, then vortexed for 3 min and centrifugated at 12,000 rpm (5 min, 4 ℃). The supernatant was dry by placed under nitrogen flow and then added with 50 μL methanol, followed by vortexed for 3 min, ultrasound and centrifugated (5 min, 12,000 rpm, 4 ℃). After that, the supernatant was separated for instrumental analysis.

The chromatographic and mass spectrum conditions were shown in Table S2-3. Then, the chemical composition of GQD were analyzed and identified using PeakView 1.2.1 (SCIEX, Framingham, MA, USA) software.

2.5 Network pharmacology analysis

The target of the components in plasma was predicted based on SwissTargetPrediction website (https://www.swisstargetprediction.ch). To obtain the key targets of components in plasma, SMILES strings of these components in vivo were entered into SwissTargetPrediction (https://www.swisstargetprediction.ch), from which possible targets for these compounds were returned based on predicted probability ranking order. Meanwhile, the genes of UC and CRC were obtained on Therapeutic Target Database (TTD, https://db.idrblab.net/ttd/) and GeneCards (https://www.genecards.org), with “Ulcerative colitis” and “Colorectal cancer” as keywords. Venny (https://www.liuxiaoyuyuan.cn/) website was used to compare the differential genes of diseases and the targets of the components of GQD in plasma, and the intersection targets were selected. A protein–protein interaction (PPI) network was draw through String database (https://cn.string-db.org/). Finally, the “herb-component-target” network was constructed by Cytoscape 3.10.0 (UC, San Diego, La Jolla, CA, USA).

2.6 Elisa assay

The expression levels of VEGF, SOD, TNF-α, IL-1β, MDA, IL-6 and IL-10 in colon and serum were quantitatively determined by ELISA kit according to the product description. First, 50 μL standard products in kit with different concentrations and 50 μL samples to be tested were added in each hole, and nothing is added to the blank hole. Next, each standard and sample holes were added with 100 μL of HRP labeled detection antibody and incubated in a 37 ℃ incubator for 60 min, and then washed with washing buffer for 3 times. 50 μL substrate A and 50 μL B were added to each hole. After incubation at 37 ℃ for 15 min, 50 μL termination solution was added to each hole successively. Finally, OD value of each hole was measured at 450 nm by enzyme-labeled instrument.

2.7 Flow cytometry detected of T lymphocytes (CD4⁺ and CD8⁺)

First, 0.5 mL blood was diluted to 3 mL and placed on 3 mL lymphocyte separation solution, and centrifuge at 400 g (20 min). The lymphocyte layer was cleaned with cleaning solution centrifuged for 10 min (250 g). Next, the precipitation was mixed with RBC lysate (3 mL) and incubated on ice (10 min), followed by terminated with PBS. The underlying cells were retained after centrifugation. PBS was used for suspending the cells, and then 0.5 μg FITC-CD3⁺, 0.25 μg APC-CD4⁺ and 0.25 μg PE-CD8⁺ were added and stained at 4 ℃ for 1 h away from light. 500 μL PBS was added and centrifuged 250 g for 10 min for cleaning. The underlying cells were re-suspended with PBS and transferred to the flow tube. Flow cytometry was adopted for detecting and calculating the proportions of CD4⁺ and CD8⁺ T lymphocytes.

2.8 Intestinal microflora detection and data processing methods

DNA was extracted from every 0.1 g of frozen colon contents and genomic DNA integrity (including purity and concentration) was examined by 1 % agar-gel electrophoresis. The PCR amplification system was 20 μL, and the V3-V4 region was amplified using the universal primers 338F (5′-ACTCCTACGGGAGGCAGCAG-3′) and 806R (5′-GGACTACHVGGGTWTCTAAT-3′). Next, the products were identified, purified and quantified. Sequencing was then performed on the Illumina MiSeq PE250 system. Trimmomatic (v0.33) was used to filter the sequenced Raw Reads. And cutadapt 1.9.1 software was used to identify and remove primer sequences, then Clean Reads without primer sequences were obtained. After a series of preprocessing, the original data were obtained with high quality sequences, and then OUT clustering and species annotation were carried out. According to the annotation results, distribution histogram of the top 10 community in abundance at different levels were drawn by QIIME2 (https://qiime2.org/). In addition, linear discriminant analysis (LDA, https://huttenhower.sph.harvard.edu/lefse/) was used to estimate the influence of microflora abundance on the difference effect. Finally, the significance analysis of inter-group differences (Metastats, https://metastats.cbcb.umd.edu/) was used to measure the differences in microflora abundance composition in disease group, control group and drug intervention groups, and to find biomarkers with statistical differences.

2.9 Metabolomics detection and data processing methods

1 mL of physiological saline and 100 mg of colon tissue were combined and homogenized for 3 min. Then, 500 μL of homogenate supernatant was measured and added with 10 μL internal standard solution (10 µg/mL of L-2-Chlorophenylalanine and heptadecanoic acid) and 1.5 mL methanol. The mixture was vortexed for 3 min before being centrifuged (12,000 rpm for 10 min at 4 °C). The residue dried by nitrogen was then redissolved with 100 μL methanol, swirled for 3 min, ultrasonically for 3 min, then centrifuged (5 min, 12,000 rpm). The supernatant was collected for instrument analysis and detailed chromatographic and mass spectrum conditions were presented in the Table S4-5.

The original data was imported into XCMS (https://xcmsonline.scripps.edu/) platform for peak identification, peak matching and other processing. Principal component analysis (PCA) was performed on the processed data using SIMCA-P 14.0 (Umetrics, Malmo, Sweden), followed by orthogonal partial least squares discriminant analysis (OPLS-DA) and 200 permutation tests. Subsequently, a combination of multiple indicators was used to determine the differential metabolites. Subsequently, based on the precise mass/charge ratio and MS/MS information provided by HPLC-Q-TOF-MS/MS, the differential metabolites were identified using HMDB (https://hmdb.ca/), MetDNA (https://metdna.zhulab.cn/) and Mzcloud (https://www.mzcloud.org) databases. Then, MetaboAnalyst (https://www.metabo-analyst.ca/) platform was used for the analysis of KEGG pathway enrichment of differential metabolites. In addition, Spearman analysis between differential microflora and metabolites were conducted and visualized by heatmap. Finally, Metscape was used to obtain compound-reaction-enzyme-gene interaction networks to realize integrated analysis of network pharmacology and metabolomics.

2.10 Western blot analysis of crucial target proteins

Using the Pierce™BCA Protein Quantification Kit (Thermo, San Jose, CA, USA), the total proteins in the colon tissues were measured. Colon samples were diluted and added to 5 × protein loading buffer (Epizyme, Shanghai, China) to achieve a final protein concentration of 4 g/L (the volume ratio of samples to loading buffer was 4:1). For further examination, all samples were then denatured for 10 min (100 °C). After adding 2.5 μL of marker (Fisher, San Jose, CA, USA), 40 μg samples were added, sodium dodecyl sulfate–polyacrylamide gel electrophoresis from 60 V to 110 V were carried out. Then, the electric transfer was performed at 240 mA. Membranes were then blocked for 30 min using blocking solution (Genefist, Oxfordshire, UK), and then eluted for 3 times with TBST (Solarbio, Beijing, China). Next, the bands were incubated in ALOX15 (1:1000), CYP1B1 (1:5000), PTGS2 (1:1000) and β-actin (colon loading control, 1:1000) primary antibodies overnight at 4 ℃. Then, the protein bands were re-elution of TBST for 3 times, followed by incubated with 1:5000 secondary antibodies (bs-40295G-HRP; Biomass) away from light for 1 h. After coated with ECL developer solution evenly (US Everbright, Suzhou, China), the target bands were visualized by Tanon 5200 Multi fully automated chemiluminescence image analysis system (Tanon, Shanghai, China). Finally, ImageJ (Rawak Software Inc., Stuttgart, Germany) software was utilized for the purpose of quantification.

2.11 Molecular docking

The protein files of ALOX15 (PDB: 2p0m), CYP1B1 (PDB: 3 pm0) and PTGS2 (5kir) were acquired from the RCSB PDB protein Crystal Structure database (https://www.pdbus.org/). PyMOL 2.4 (PyMOL software, Schrödinger, NewYork, NY, USA) was applied to remove water and ligands. AutoDock 4.2.6 (AutoDock software, Scripps Research, La Jolla, CA, USA) was used for hydrogenation and charging. Mol2 structures of key pharmacodynamic ingredients were then obtained from TCMSP (https://old.tcmsp-e.com/tcmsp.php), and the two receptors were semi-flexibly docked with the active ingredients by AutoDock 4.2.6 (AutoDock software, Scripps Research, La Jolla, CA, USA). In the end, PyMOL 2.4 (PyMOL software, Schrodinger, NY, USA) was used to visualize the molecular docking results.

2.12 Statistical analysis

All calculated experimental values were presented as mean ± SD. Statistical analysis was conducted with Student’s t-test using SPSS 26.0 (SPSS software, SourceForge, San Diego, CA, USA).

3.1 Composition identification results of GQD in vitro and in vivo

Through analysis of the fragmentation pathways obtained from literature sources and public data, as well as database matching, comparison of mass accuracy, isotope patterns, and retention times of reference standards, 93 and 34 compounds in GQD were preliminarily identified in the positive and negative ion mode, respectively. And a total of 35 components in vivo were detected in GQD. The total ion chromatograms (TICs) were shown in Fig. S1. The specific information of each chemical component was shown in Table 1.

3.2 Network pharmacology predictive analysis

43 targets were obtained in total from the intersection of 35 in vivo component targets (the top 5 targets with the highest Probability value) and disease targets. Then, PPI network diagram of 43 targets was constructed through String database (Fig. 1A). With the degree value > 4 as the standard, 31 key targets were screened out and the interaction network diagram was drawn (Fig. 1B). The list of targets was shown in Table S6. Finally, using Cytoscape 3.9.1 (UC, San Diego, La Jolla, CA, USA) software, a network diagram of 4 herb, 35 components and 31 targets was constructed (Fig. 1C). The top 4 components with the highest degree of each herb were identified as the key active components that might play important roles in drug efficacy. As a result, Ononin, genistein, puerarin, daidzein were the key components in Pueraria lobata (Willd.) Ohwi (Gegen); Skullcapflavone II, chrysin 7-O-beta-gentiobioside, baicalin, wogonoside were the key components in Scutellaria baicalensis Georgi (Huangqin); DeMethyleneberberine, p-coumaric acid, coptisine, berberine were the key components in Coptis chinensis French (Huanglian); Isoliquiritigenin, glyasperin A, isoononin and glycyrrhizin were the key components in Glycyrrhiza uralensis Fisch (Gancao).

Close

Fig. 1

Prediction results of network pharmacology study. PPI diagram network (A); Network diagram of 31 key targets (B); “Herb-component-target” multivariate network (C).

Export to PPT

3.3 GQD intervention mitigated histopathologic injury in colon tissue caused by UC and CRC and reduce gastrointestinal adverse reactions of XELOX

The colon structure of control group was complete, and goblet cells were arranged neatly (Fig. 2A). However, colonic mucosa ulceration was serious in UC group, and pathological phenomena such as absence of colonic mucosal epithelial cells, destruction of glandular structure and disappearance of crypt structure were found. Severe colon erosion was observed in CRC group, and the pathological features were more obvious than UC group. After GQD treatment, the pathological phenomena of colonic tissue were recovered to a certain extent. Moreover, the colonic lesions of rats treated with GQD combined with XELOX were more improved than those treated with XELOX alone, indicating that GQD could effectively reduce colonic mucosal injury, assist XELOX to enhance the effect of improving intestinal lesions and reduce gastrointestinal adverse reactions of chemotherapy drugs.

Close

Fig. 2

H&E staining results of colon tissues (200 × magnification) (A); Bar chart of ELISA kit results for inflammatory factors, tumor factors and antioxidant indicators (B). Values shown are means ± SD. *p < 0.05, **p < 0.01 UC vs Control group; *p < 0.05, **p < 0.01 CRC vs Control group; #p < 0.05, ##p < 0.01 vs UC group; #p < 0.05, ##p < 0.01 vs CRC group.

Export to PPT

3.4 GQD intervention enhanced anti-inflammatory, antitumor and antioxidant activities

The activities of anti-inflammatory, anti-tumor as well as anti-oxidative of GQD were evaluated by ELISA. As shown in Fig. 2B, the pro-inflammatory factors TNF-α, IL-6 and IL-1β in colon and serum increased substantially in the development of UC to CRC. After GQD treatment, significant callback effect was observed in both UC and CRC period. Meanwhile, the inflammatory factors in RXL and RXH groups were more significantly decreased than those in XELOX group. In addition, GQD also greatly increased the content of IL-10, the anti-inflammatory factor. And there was a more significant correction trend in the RXL and RXH groups than that in the XELOX group, which declared that GQD had a very strong anti-inflammatory effect, and combined treatment with XELOX might enhance the anti-inflammatory effect. Moreover, the detection results of tumor factor VEGF, antioxidant factor SOD and MDA also proved that GQD had excellent anti-tumor and antioxidant effects, and might assist chemotherapy drugs to enhance its anti-tumor and antioxidant activities. Inflammation and oxidative stress are the key factors in the toxicity and side effects of XELOX. The results above indicated that GQD had great potential in alleviating the toxicity and side effects of XELOX.

3.5 GQD intervention regulated immune response to resist the adverse reactions of XELOX

It is well known that the balance of the CD4⁺/CD8⁺ is essential for immunomodulatory effects. In this study, CD4⁺ and CD8⁺ positive cells in T lymphocytes were detected by flow cytometry (Fig. 3). The proportion of CD4⁺/CD8⁺ cell subsets in UC and CRC groups were prominently reduced compared with the control group (p＜0.05), indicating that inflammation and tumor could damage immune function. However, after GQD intervention, the ratio of CD4⁺/CD8⁺T lymphocytes showed a dose-dependent correction and was similar to that of the control group, enunciating that GQD treatment could normalize the immune system’ balance and facilitate to play an anti-tumor role. In addition, after treatment of GQD combined with XELOX, a further reversed trend in the T lymphocytes proportion (CD4⁺/CD8⁺) was found refer to the XELOX treatment group. The data above manifested that the intervention of GQD combined with XELOX could have a positive effect on the immune system's ability to fend off unfavorable reactions.

Close

Fig. 3

Flow cytometric analysis of CD4⁺ and CD8⁺ cells in each group of rats. *p < 0.05 UC vs control group; *p < 0.05 CRC vs Control group; #p < 0.05 vs UC group; #p < 0.05 vs CRC group.

Export to PPT

3.6 Intestinal microflora analysis

To study the impacts and molecular underpinnings of the microbial consortia on colorectal carcinogenesis, 16S rDNA sequencing technology was applied to detect the gut microbial community of rats in each group. As shown in Fig. 4A, the community distribution histogram of each group at various levels was drawn. Moreover, LDA distribution histogram and cladogram of dominant microorganisms in different groups were displayed both in Figs. 4 and 5. It could be seen from the diagram that during the evolution from UC to CRC, the intestinal microflora was obviously disturbed, manifested with Firmicutes' abundance dramatically dropped, while Proteobacteria's abundance dramatically grew at the phylum level. The microflora with high abundance were Escherichia_Shigella and Lactobacillus, and the two bacteria were common beneficial bacteria and harmful bacteria in vivo, respectively (genus level). The results showed that from UC to CRC, the abundance of Escherichia_Shigella gradually increased, and that of Lactobacillus gradually decreased. After GQD intervention, the abundance of various microflora including the above microflora was regulated towards normal levels, indicating that GQD could treat UC and CRC by remodeling intestinal microflora homeostasis and inhibit the progression of UC to CRC. What’s more, the decline trend of Escherichia_Shigella in RXL and RXH groups was more than that in XELOX group, the lactobacillus’ abundance in the RXL and RXH groups was much greater than that in XELOX group, suggesting that GQD could help restore the balance of intestinal microflora on the basic of XELOX treatment.

Close

Fig. 4

Distribution histogram of the top 10 community in abundance at the level of phylum, class, order, family and genus (A); Distribution histogram based on LDA of dominant microorganisms in the control, UC, UL, UH groups (B); Cladogram of dominant microorganisms in the control, UC, UL, UH groups (C).

Export to PPT

Close

Fig. 5

Distribution histogram based on LDA of dominant microorganisms in different groups (A); Cladogram of dominant microorganisms in different groups (B).

Export to PPT

3.7 Non-targeted metabolomics analysis and correlation analysis with intestinal microflora

To further explore the mechanism of GQD regulation of disease at the small-molecule level, non-targeted metabolomics study was performed on colon samples by HPLC-Q-TOF-MS/MS, with.

proven precision, stability, and reproducibility (Table S7). The Control, UC and CRC groups showed significant separation without overfitting (Fig. S2). Subsequently, metabolites satisfying multiple conditions were screened out (P＜0.05, FC＞2.0 or FC＜0.5, VIP > 1.0) as potential biomarkers. In the end, 45 biomarkers were screened in the Control and UC groups, and 38 biomarkers were examined in the Control and CRC groups, among which 21 biomarkers were examined in both the UC and CRC groups. The detailed information of each biomarker was shown in Table 2. Then, the abundance of these metabolites in each group was visualized using heat maps (Figs. 6 and 7) to intuitively investigate the variation of metabolites after GQD intervention. Moreover, among the 21 metabolites associated with UC and CRC, linoleic acid, 7-ketodeoxycholic acid and other metabolites showed obvious trend of correction after treatment (p < 0.05), and the changes of these metabolites were shown in Fig. 8.

Close

Fig. 6

Heat maps of differential metabolites between the control and UC groups (A); Heat maps of differential metabolites between the control and CRC groups (B); Heat maps of differential metabolites in both the UC and CRC groups (C); circle graphs represented the classification and proportion of differential metabolites.

Export to PPT

Close

Fig. 7

Heat maps of differential metabolites between the control and UC groups after treatment (A); Heat maps of differential metabolites between the control and CRC groups after treatment (B).

Export to PPT

Close

Fig. 8

Column chart of different metabolite concentrations in the control, UC, UL and UH groups (A); Column chart of different metabolite concentrations in the control, UC, CRC, RL, RH, XELOX, RXL and RXH groups (B).

Export to PPT

Furthermore, the MetaboAnalyst (https://www.metabo-analyst.ca/) platform was used to enrich the KEGG pathway (Fig. 9). The results showed that 15 metabolic pathways were significantly enriched during UC development. 14 metabolic pathways were significantly enriched during CRC development. 7 pathways might be affinitive to the transition from UC to CRC.

Close

Fig. 9

KEGG enrichment pathway maps of differential metabolites between the control and UC groups (A); KEGG enrichment pathway maps of differential metabolites between the control and CRC groups (B); KEGG enrichment pathway maps of differential metabolites related to UC and CRC (C).

Export to PPT

Whereafter, Spearman correlation analysis was adopted to evaluate the interaction between altered metabolites and microflora, in order to find out the key microflora and metabolites (Fig. 10). According to the Metastats analysis, among the differential microflora in control vs UC and control vs CRC, there were 22 differential microflora both in the UC and CRC groups at phylum and genus level, including Enterococcus, Allobaculum, Escherichia_Shigella, Dubosiella, Ligilactobacillus, Lactobacillus etc. And the correlation analysis was performed between them and 21 differential metabolites. The abundance of Proteobacteria, Enterococcus, Allobaculum, Escherichia_Shigella, Dubosiella and uncultured_Clostridiales_bacterium showed a gradually increasing change pattern in the development process from UC to CRC. On the contrary, the abundance of Marvinbryantia, Monoglobu, Lachnospiraceae_UCG_006, [Bacteroides]_pectinophilus_group, Gordonibacter, Ligilactobacillus and Lactobacillus showed a gradual decrease. The above bacteria were considered to have connections to the evolution of UC into CRC. Correlation analysis showed that Proteobacteria, Escherichia_Shigella and several bacteria had a significant positive correlation with unsaturated fatty acids (oleic acid, alpha-linoleic acid, linoleic acid, etc.). Ligilactobacillus, Lactobacillus and other bacteria showed significant negative correlation with unsaturated fatty acids.

Close

Fig. 10

Correlation heat map of key differential microflora and differential metabolites. *p < 0.05, **p < 0.01, ***p < 0.001.

Export to PPT

3.8 Integrated analysis of network pharmacology and metabolomics

Using the Metscape plug-in to form compound-reaction-enzyme-gene networks, 31 crucial targets from network pharmacology results and 21 distinct metabolites from metabolomics results were imported to the Cytoscape 3.10.0 (UC, San Diego, La Jolla, CA, USA) software, to visually related genes and metabolites of interaction (Fig. 11). Six metabolites (linoleic acid, cis-8,11,14-eicosatrienoic acid, sphingolipid, α-linoleic acid, oleic acid, citric acid) as well as three targets (ALOX15, CYP1B1, and PTGS2) were identified in the network. Among them, linoleic acid and cis-8,11,14-eicosatrienoic acid were key metabolites that had metabolic relationship with the targets of components in vivo. ALOX15 and CYP1B1 were core genes involved in the metabolism of linoleic acid, while ALOX15 and PTGS2 were important genes involved in the metabolism of cis-8,11,14-eicosatrienoic acid. These distinct targets and metabolites were chosen as the main metabolites and targets.

Close

Fig. 11

The compound-reaction-enzyme-gene networks of the key metabolites and targets (A); the network of interactions between key metabolites (B).

Export to PPT

3.9 Western blot analysis of ALOX15, CYP1B1 and PTGS2 proteins

As shown in Fig. 12A, ALOX15, CYP1B1, and PTGS2 were all activated (p < 0.05) in UC and CRC groups, and each protein in CRC group was expressed more than the UC group. In addition, the expression of ALOX15, CYP1B1, and PTGS2 proteins were effectively inhibited by different concentrations of GQD (p < 0.05), especially the higher doses of GQD. Particularly, the administration of GQD combined with XELOX showed a stronger advantage in reversing the expression level of the above proteins than that of XELOX administration alone. These results indicated that the pathogenesis of UC and CRC might be accompanied by the activation of ALOX15, CYP1B1, and PTGS2 proteins, GQD exerted anti-inflammatory and anti-tumor effects by down-regulating these proteins, which was also a key mechanism for GQD to enhance the efficacy of XELOX.

Close

Fig. 12

Typical protein bands and relative expression of ALOX15, CYP1B1 and PTGS2 proteins in western blot analysis; Values shown are means ± SD. *p < 0.05, **p < 0.01 UC vs Control group; *p < 0.05, **p < 0.01 CRC vs Control group; #p < 0.05, ##p < 0.01 vs UC group; #p < 0.05, ##p < 0.01 vs CRC group (A); Molecular docking interaction diagrams of key active components with macromolecular ligands (B).

Export to PPT

3.10 Molecular docking result

In order to explore the pharmacodynamic substance basis of GQD in the therapy of UC and CRC, molecular docking was conducted between ALOX15, CYP1B1, PTGS2 and active components, respective. Firstly, the macromolecular ligands of ALOX15 (2p0m), CYP1B1 (3 pm0) and PTGS2 (5kir) were downloaded from the PDB database (https://www.pdbus.org/), respectively. Then, the potential active components were semi-flexible docked with the receptor proteins respectively. RS7, BHF and rofecoxib (RCX) were the positive drug of ALOX15, CYP1B1 and PTGS2, which were semi-flexible docked to 2p0m, 3 pm0 and 5kir, respectively. The results showed that GLU357 was the main active sites of 2p0m, GLN332, ASN228 were the main active sites of 3 pm0, and PHE518, ILE517, GLN192, SER530 were the main active sites of 5kir. Moreover, the binding energies to proteins of ononin, genistein, puerarin, daidzein, skullcapflavone II, baicalin, wogonoside, p-coumaric acid, coptisine, isoliquiritigenin, glyasperin A were less than −5 kcal/mol and the binding sites were similar to positive drugs, so that the above components were identified as key active components. Details of binding energy were shown in Table S8, and the results of protein–ligand interaction profile were presented in Fig. 12B and Fig. S3-5.