Network Pharmacology-based Strategy for Studying the Mechanism of Gegen Qinlian Decoction for the Treatment of Acute Colitis

Background: Gegen Qinlian Decoction(GQD) has been used to treat acute colitis (AC) for several years in China and it has shown good ecacy. However, the active components and target proteins of its anti-AC effects remains to be deciphered. Methods: In this study, serum pharmacochemistry and network pharmacology strategy were integrated to identify the constituents in blood and the mechanism of GQD for the treatment of AC. Ultra-performance liquid chromatography and LTQ-Orbitrap mass spectrometry(UPLC-LTQ-Orbitrap-MS) was used to identify the absorbed components of GQD in rat serum; molecular docking and compound-target network analysis were used to predict candidate targets and critical components in GQD responsible for ecacy; In addition, the Kyoto Encyclopedia of Genes and Genomes(KEGG) pathway analysis and Gene Ontology(GO) enrichment analysis were used to predict the related pathways and biological process respectively; Finally, the model rats with acute colitis were induced by DSS(Dextran Sulfact Sodium) in order to verify the effects and potential mechanism of baicalein, which is an important component of GQD. Results: Based on our comprehensive systematic approach, 23 components were successfully identied in rat serum after oral administration of GQD. The predicted results of molecular docking indicated that these 23 active components closely interacted with 41 protein targets associated with inammation, immunity and enteric mucos. Among the 23 compounds identied, baicalin, baicalein, wogonoside , liquiritin and daidzin may be the most important components of GQD. Futhermore, according to GO enrichment analysis, the 41 candidate targets identied were mainly involved in two biological process, immune system process and inammatory response. The KEGG pathway analysis revealed that 41 candidate targets were associated with 62 biological pathways, including HIF-1 signaling pathway and PI3K/Akt signaling pathway. Animal experiments found that baicalein could inhibit the activation of PI3K/Akt/HIF-1 signaling pathway and signicantly reduce pro-inammatory cytokines, such as IL-1β, IL-6, IL-8 and TNF-α to alleviate intestinal mucosal KEGG pathway enrichment analysis predicted that GQD could suppress AC by regulating the HIF-1 signaling pathway, PI3K-Akt signaling pathway and Toll-like receptor signaling pathway. And the animal experiments conrmed that baicalein could ameliorate intestinal mucosal damage by regulate PI3K/Akt/HIF-1 signaling pathway. These results demonstrate that multi-component synergistic system are the advantage of TCM when play a therapeutic role in the treatment of AC, which deserves further exploration.


Molecular docking
The interaction between compounds and protein targets were preliminarily determined using discovery Studio 2016 for molecular docking. All docking studies were performed using a protein-ligand complex with a crystal structure present, similar to the method described previously in Chen et al [31] .
Based on the target prediction results, candidate targets were selected based on the following three criteria: i) low binding energy between compounds and proteins by virtual screening (better match degree), ii) the higher docking scores for target proteins (more likely to be potential targets), iii) proteins that closely associated with colitis.

Compound-target network construction
A compound-target network to investigate the anti-AC mechanism of GQD was constructed using Cytoscape 3.6.1 http://www.cytoscape.org/ ). Cytoscape is widely used in network pharmacology for visualizing molecular and component interaction networks [32]. In the graphical network, compounds and proteins were represented as nodes and intermolecular interactions between molecules (compound-target) were denoted with links [33]. The analysis of network parameters were performed by Network Analyzer (a Cytoscape plug-in).

PPI network construction
The candidate proteins were submitted to the Search Tool, STRING, http://string.embl.de/ database, to explore the functional interactions between the proteins. The interactions with combined score > 0.4 were selected as signi cant. The mapped proteins were imported into Cytoscape software to construct a protein-protein interaction (PPI) network. The Network Analyzer of Cytoscape was used to further analyze network parameters, such as the average degree of freedom and degree of freedom etc. Then, set the size and color of the nodes according to the degree of freedom, the bigger the node size and the brighter the color with the greater degree value.

GO enrichment analysis and KEGG pathway analysis
The GO enrichment analysis and KEGG pathway analysis were performed using DAVID online tool. p< 0.01 was considered statistically signi cant. The GraphPad Prism software and Omicshare cloud platform were used to achieve visualization.

Experimental validation DSS-induced AC models and drug treatments
Animal information and breeding environment are the same as above. The AC rat model was established by administrating DSS in their drinking water. After environmental adaption, forty rats were randomly divided into the following ve groups: Normal group, DSS group, baicalein (15 mg/kg) +DSS group, baicalein (20 mg/kg) +DSS group and SASP (400 mg/kg) +DSS group.
Rats in the normal group received distilled water, while rats in the other groups received 7% (w/v) DSS in their drinking water. Seven days later, the AC model was fully established. In the model therapy groups, SASP (40 mg/ml), the reference drug, and baicalein (1mg/ml) were suspended in CMC-Na. Drugs were administered to rats in their respective groups once a day for 7 days. The remaining groups were administered an equal volume of CMC-Na.
Disease activity index(DAI) assessment and histological analysis of colonic Stool consistency, gross bleeding and body weight of rats were recorded, and were used to assess DAI. The scoring criteria were based on the method previously described [34]. Rats were euthanized at the end of the study and the entire colon was harvested. The colon was opened longitudinally and quickly washed in saline buffer and the morphology of the colon was examined using a microscope (Motic, China). Colonic sections were xed in 4% polyformaldehyde and embedded in para n for histopathological analysis by Hematoxylin & eosin (H&E) staining. Photomicrographs were captured using a digital research photographic microscope (Thermo Corporation, America).
Enzyme-Linked immunosorbent Assay (ELISA) analysis Blood was collected from abdominal aorta of rats after intraperitoneal anesthesia with 2% pentobarbital sodium, then the whole blood was placed at 4 ℃ for 2 hours and centrifuged at 3 500 × g for 10 minutes to obtain serum. The enzyme linked immunosorbent assay was used to determine the levels of TNF-α, IL-1β, IL-6 and IL-8 in rat serum.

Western blot analysis
The proteins samples were prepared from the colon tissue by RIPA lysis buffer, and its concentrations were detected by BCA protein concentration assay kit. The proteins were exposed to electrophoresis on 8% to 12% SDS-PAGE gel and transferred to a PVDF membrane. After blocking with 5 % BSA at room temperature for 1 h, the membrane were incubated with different primary antibodies at 4 ℃ overnight. Then the PVDF membrane were washed with TBST three times and incubated with HRP conjugated secondary antibodies at room temperature for 1 h. Finally, after washing with TBST three times, the membrane were added with ECL solution and were placed in a gel imager for photographic analysis.

Data analyses
Statistical analysis was performed using GraphPad Prism 6.0 statistical software. Experimental data were expressed as mean±standard deviation (mean±SD). Statistical signi cance was determined using One-way ANOVA test, and p< 0.05 was considered statistically signi cant.

UPLC-LTQ-Orbitrap-MS analysis of rat serums after oral administration of GQD
In a previous study, we performed chemical analysis of GQD and 67 constituents were identi ed. In this study, UPLC-LTQ-Orbitrap-MS was used to detect and identify differential peaks in serum samples from rats administered saline versus GQD in the positive mode. Based on the retention time, MS/MS results and precise molecular mass, 23 components were identi ed. Among the 23 compounds, seven compounds were fromCoptidis rhizome,six compunds were from Puerariae Lobatae radix, ve compounds each were from Scutellariae radixand Glycyrrhizae Radix et Rhizoma. 13 of them were avonoids, 7 of them were alkaloids, 2 of them were triterpenoids and triterpenoid saponins, and 1 was coumarin. Detailed information of the 23 blood constituents in GQD administered rat serum are shown in Table1. Structural data was obtained from the NCBI PubChem database (https://pubchem.ncbi.nlm.nih.gov/). Total ion chromatographs of UPLC-LTQ-Orbitrap-MS using positive mode are shown in Fig.2.
Network pharmacology analysis of GQD Putative protein targets collection 44 protein targets (Table2) were obtained, and their x-ray crystallographic structures were downloaded from the (RCSB) Protein Data Bank (https://www.rcsb.org/) for high-throughput molecular docking studies. These proteins are associated with in ammation, immunity and enteric mucosa. The pathogenesis of AC is not yet clear, but studies showed that it may be related to immune dysfunction, infectious factors, environmental factors, and other factors [35].  [36]. Molecular docking can predict the binding sites of candidate drugs to disease proteins [37]. Using the molecular docking method, 41 targets were mapped out from 44 targets.

Compound-target network construction and analysis
A "compound-target" network was constructed to determine the molecular mechanism of GQD using Cyscape3.6.1. The network consisted of 64 nodes and 667 edges (Fig.3). Nodes denote drug molecules and target proteins, while an edge represents the interaction between a speci c molecule and a protein [38]. The results of network analysis are shown in Table3. The "degree" indicates the number of protein targets that a compound can bind to [39]. In the network of GQD, the top 5 components with the highest degree were baicalin, baicalein, wogonoside , liquiritin and daidzin, baicalin can interacted with 37 potential proteins, baicalein, wogonoside, liquiritin and daidzin can interacted with 36 potential proteins.

Components-targets interactions analysis
After docking the compounds in Discovery studio 2016, the pymol software is used to provide details about the interplay between proteins and active compound. Here, daidzin and ICAM-1, wogonoside and JAK2 were selected to study the action modes between compounds and proteins,which had low binding energy and high docking scores.
The interplay between daidzin and ICAM-1 is shown in Fig.4-A. The carbonyl can form a hydrogen bond with Lys142,and glycosyl group can form a hydrogen bond with Asp85, Ala84, and Gln166, respectively.
The interaction between wogonoside and JAK2 is shown in Fig.4-B. The carbonyl group on the parent nucleus can form a hydrogen bond with Pro1114, and the glycosyl group can form a hydrogen bond with Phe1116, Arg1117, Gly968, respectively. Lin's research con rmed the interaction between wogonoside and protein JAK2. He discovered that wogonoside could limit the phosphorylation process of JAK2 and reduce the production of in ammatory factor IL-8 [40].

PPI network of GQD anti-AC construction and analysis
The construction and the most signi cant module analysis of PPI network were performed by STRING and Cytoscape software (Fig.5.). The interactions between the proteins with combined score > 0.4 were selected in this study. The PPI network consist of 36 nodes (genes) , 186 edges(interactions) and the average degree of freedom is 9.07. The greater the degree value , the more the biological function of the node in the PPI network. thus, the top 15 proteins with degree greater than 11 were selected as the core targets of GQD against AC (Table 4.). The results were basically consistent with the molecular docking analysis.  Figure 6 according to P value. We found that the therapeutic effect of GQD on AC may involve in Biological Processes such as in ammatory response and innate immune response, Molecular Function such as protein kinase activity and ATP binding, Cellular Component such as cytosol and blood microparticle.  The body weights of DSS-induced rats decreased during the research, while treatment with baicalein (20 mg/kg) or SASP could signi cantly improve DSS-induced weight loss (Fig.8A). In addition to changes in body weight, diarrhea and blood stool of rats were measured daily. On the third day of modeling, presence of soft stool and perianal infection were observed. On the fourth day, blood was present in the stool. The DAI scores of DSS-induced rats signi cantly increased. Administration of baicalein (20 mg/kg) and SASP had a signi cant therapeutic effect and reduced DAI scores (Fig.8B).

Changes of pathological
Histopathological analysis indicated that rats in the normal group, the colon glands were arranged neatly, the mucosa was intact, and no in ammatory cells in ltrated into the submucosa. However, the majority of the colon glands in rats induced by DSS were destroyed, the colonic epithelial cells were detached, and numerous in ammatory cells were observed in the submucosal tissues. The pathological condition of rats treated with baicalein (20 mg/kg) or SASP showed signi cant improvement (Fig.8C).

Changes of pro-in ammatory cytokines
Pro-in ammatory cytokines play a signi cant role in the process of DSS-induced acute enteritis rat [4].To investigate whether baicalein is associated with proin ammatory cytokines in the repair of DSS-induced acute enteritis, the levels of TNF-α, IL-1β, IL-6 and IL-8 in the serum were tested by enzyme linked immunosorbent assay. Baicalein (15, 20 mg/kg) and SASP could remarkably reduce the levels of IL-1β in the serum which signi cantly increased after DSS challenge. Baicalein (20 mg/kg) and SASP could signi cantly reduce the TNF-α, IL-6 and IL-8 in the serum after DSS challenge (Fig.9.). The results indicated that the protective effect of baicalein on DSS-induced colitis is closely related to the down-regulation of the expression of these pro-in ammatory cytokines.
Changes of the PI3K/Akt/HIF-1 signaling pathway According to the previous pathway analysis studies, we proposed that baicalein may excert its anti-AC effects by regulating the PI3K/Akt/HIF-1 signaling pathway. Therefore, western blot analysis were used to study the expressions of some key proteins in the pathway. As shown in Fig.10, Compared to the DSS group, the protein expressions of phospho-PI3K (p-PI3K) and HIF-1α were signi cantly decreased in Baicalein (20 mg/kg) group and SASP group. Similarly, the protein expressions of phospho-AKT(p-AKT ) were signi cantly decreased in Baicalein (15, 20 mg/kg) group and SASP group, and the p-AKT expressions were down-regulated by baicalein dose-dependently. In addition,compared to the DSS group, the expression of total-PI3K, total-AKT were hardly affected by baicalein and SASP.

Discussion
In this study, UPLC-LTQ-Orbitrap-MS was used to analyze and identify the active compounds in rat serum after oral administration GQD, so as to systematically clarify the effective substance of GQD; Then, network pharmacology strategy was used to study the mechanism of GQD for the treatment of acute colitis, including revealing its key components, targets, and regulated pathways; Finally, the therapeutic effects and mechanism of baicalein, which is an important component of GQD, were investigated in dextran sulfate sodium (DSS)-induced AC rats. This article hopes to lay a foundation for the study of quality control and mechanism of GQD, and provide evidence for the investigative modernization of TCM.
There are 23 components were identi ed from rat serum after oral administration GQD by UPLC-LTQ-Orbitrap-MS analysis system, and the results indicated that avonoids and alkaloids were the main active ingredients of GQD. The results of molecular docking showed that the constituent may target protein such as ICAM1, JAK2, AKT1, MTOR and PPARG. Here, three representative proteins(ICAM-1, AKT1 and JAK2) were selected to futher discuss their function.
The main cause of acute exacerbation of chronic colitis is the in ux of neutrophils in the lining of the epithelium, which leads to the formation of crypt abscesses. After the crypts rupture, the pus ows to the intestine or neutrophils spread into the mucosa or connective tissue. Therefore, the focus of controlling this disease seems to be to limit the damage of neutrophils. The study found that ICAM-1 is key to the ow of neutrophils into the colon [41]. From the link between AC and ICAM-1 genes, ICAM-1 was found to be a clear target for the treatment of AC.
Protein kinase B (AKT) consisted of three isoforms AKT1, AKT2 and AKT3, regulates many processes including cell survival and apoptosis [42].
Studies con rm that the activation of AKT would regulate the entry of nuclear transcription factor (RELA, NF-κB ) into the nucleus, and promote the release of various in ammatory cytokines and pro-apoptotic proteins, initiate in ammatory responses and cell apoptosis [26 , 43] , So AKT and NF-κB play key roles in the pathogenesis and treatment of AC.
At present, it is believed that the pathogenesis of colitis is closely related to the disorder of immune regulation function. JAK2 is an essential signaling events in both innate and adaptive immunity, Yang et al. analyzed the genomic differences between colitis patients and normal people, and found that JAK2 gene was signi cantly associated with colitis in Korea [30].
Subsequently, compound-target network analysis suggested that baicalin, baicalein, wogonoside , liquiritin and daidzin which with the highest degree were the core components of GQD. The therapeutic effect and mechanisms of baicalin and wogonoside on AC rats were con rmed by many consulting published literatures. For example, Sun et al [44] reported that wogonoside could exert repair and therapeutic effects on AC rats by inhibition of NF-κB and NLRP3 in ammasome pathway. Baicalin could effectively treat DSS-induced ulcerative colitis by inhibiting IL-33 expression and its subsequent activation of NF-κB [45]..In addition, Hong et al [46] demonstrated the therapeutic effect of baicalein on AC, but its mechanism of action was not deciphered.
KEGG pathway enrichment analysis predicted that GQD could suppress AC by regulating the HIF-1 signaling pathway and PI3K-Akt signaling pathway and Toll-like receptor signaling pathway.
Studiesshowed that in the pathogenesis of AC, intestinal mucosal microcirculation disorder would make the colonic tissue in the environment of hypoxia , and then the colonic tissue will turn on oxidative stress and cause the expression of HIF-1α, so as to realize the hypoxia tolerance and in ammatory repair of colon tissue. Therefore, HIF-1 signaling pathway plays a key role in maintaining intestinal barrier function and treatment of AC [47][48][49]. The HIF-1 signaling pathway is derived from KEGG database, as shown in Figure 11, It is found that the expression of HIF-1α is regulated by PI3K-Akt signaling pathway, which is consistent with the literatures [50 , 51]. PI3K-AKT signaling pathway can regulate a wide range of cellular, is involved in the regulation and release of multiple pro-in ammatory cytokines, and these pro-in ammatory cytokines play an important role in the development of AC [43]. So we speculated that the active components of GQD may achieve the therapeutic effect on AC by regulating HIF-1 signaling pathway and PI3K-Akt signaling pathway. Toll-like receptor signaling pathway is involved in the activation of innate and adaptive immune responses [52]. Although the pathogenesis of AC is not clear, it is thought to be closely related to the immune function barrier, intestinal barrier function and environmental factors [53]. Therefore, according to the research results, GQD may improve intestinal in ammation and restore intestinal function through immune regulation to achieve the treatment of AC.
Finally, animal experiments were performed to determine the e cacy of baicalein on AC rats and its potential mechanism. The results showed that after the intervention of baicalein, the weight of rats increased, the DAI scores decreased, and the injury of colon were improved compared with the model group. in addition, the levels of TNF-α, IL-1β, IL-6 and IL-8 were decreased compared with model rats, the protein expressions of phospho-PI3K(p-PI3K), phospho-Akt(p-Akt ) and HIF-1α were down-regulated compared with model rats, which demonstrated that baicalein may affects the phosphorylation process of PI3K and AKT, thereby inhibiting the PI3K/Akt/HIF-1 signaling pathway and signi cantly reduce pro-in ammatory cytokines, such as IL-1β, IL-6, IL-8 and TNF-α,to reduce intestinal mucosal damage and achieve a therapeutic effect on acute colitis.
According to the researches [19,[54][55][56], on one side, DSS-induced colitis has been considered to be driven by activated intestinal macrophages, which release proin-ammatory cytokines and chemokines to up-regulate the expression of HIF-1α, cause tissue damage. In other hand, oxidative stress can cause intestinal epithelial barrier dysfunction and up-regulate the expression of HIF-1α. The mechanism of action of HIF-1 alpha need to be further studied in the following work.
Taken together, our study identi ed 23 active ingredients in rat serum after oral administration of GQD, and the 23 components closely interacted with 41 protein targets associated with in ammation, immunity and enteric mucos, the core components of GQD were baicalin, baicalein and wogonoside etc, the most promising candidate targets of GQD were ICAM1, MAPK14, AKT1 and PPARG etc.The KEGG pathway enrichment analysis predicted that GQD could suppress AC by regulating the HIF-1 signaling pathway, PI3K-Akt signaling pathway and Toll-like receptor signaling pathway. And the animal experiments con rmed that baicalein could ameliorate intestinal mucosal damage by regulate PI3K/Akt/HIF-1 signaling pathway. These results demonstrate that multi-component synergistic system are the advantage of TCM when play a therapeutic role in the treatment of AC, which deserves further exploration.

Conclusion
In this study, we used a UPLC-LTQ-Orbitrap-MS method combined with an integrated network pharmacology strategy were used to identify differential blood components, predict potential targets and identify critical compounds of GQD that were e cacious in the treatment of AC. In addition, animal experiments were performed to determine the e cacy of the predicted components on AC. The network pharmacology method was used to establish a biological network of chemical components interacting with target proteins at the molecular and system-wide level.
This approach helped in identifying the molecular mechanisms of GQD action in reducing AC, and should be valuable for understanding the molecular mechanisms of other traditional Chinese medicines.  Network of compound-protein interaction induced by GQD. The 23 representative compounds were associated with 41 potential protein targets determined by high-throughput molecular docking analysis. The yellow ellipse nodes represent compounds, while the 41 protein targets identi ed were associated with in ammation (green ellipse nodes), enteric mucosa (blue ellipse nodes), immunity(purple ellipse nodes).   The PPI network of GQD anti-ACwas established in the String database     Expression levels of TNF-α, IL-1β, IL-6 and IL-8 in the serum were determined by Elisa. Experimental data were expressed as mean ± SD.*P<0.05 compared to DSS-induce AC rats.**P<0.01 compared to DSS-induce AC rats.***P<0.001 compared to DSS-induce AC rats.

Figure 11
The HIF-1 signaling pathway was downloaded from the KEGG database

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. ReportonExperimentalAnimalWelfareandEthics.pdf