Uncovering the underlying mechanisms of Compound Yuxingcao Mixture in the treatment of COVID-19 based on network pharmacology


 Background and objective: The novel coronavirus named COVID-19 emerged in Wuhan, China in December, 2019 and has spread rapidly in China and around the world. The traditional Chinese medicine Compound Yuxingcao Mixture (CYM) has been recommended in recent editions of the national guideline while the underlying mechanisms are still unclear. In this study, we analyzed the effectiveness and potential mechanisms of CYM on COVID-19 based on network pharmacology and molecular docking approach. Methods: The active ingredients and potential targets of CYM were screened using TCMSP and STITCH databases. Genes related severe acute respiratory syndromes (SARS) and Middle East respiratory syndrome (MERS) were queried on the DisGeNET and MalaCards databases. CYM-COVID-19 common target protein interaction network was established by STRING database. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were conducted to generate the relative pathways based on KOBAS databases. In addition, the possible binding site of screened compounds were also predicted by Autodock vina software. Results: A total of 103 active ingredients and 205 putative targets were screened from CYM, of which 32 overlapped with the targets of COVID-19 and were considered therapeutic targets. The analysis of the network diagram demonstrated that the CYM activity of ingredients of quercetin, luteolin, β-sitosterol and kaempferol may play a crucial role in treating COVID-19 by regulating TNF, IL-6, IL-1β, etc. The analysis of molecular binding energy showed that β-sitosterol had the lowest binding energy with COVID-19 3CLpro (-8.63 kJ/mol). GO and KEGG enrichment analysis revealed that these targets were closely associated with inflammatory responses and immune defense processes. Conclusion: In summary, our study identified the potential mechanisms and targets of CYM for the prevention of COVID-19, providing directions for further clinical research.


Introduction
Since December 2019, the large-scale spread of the new coronavirus disease (COVID- 19), o cially known as severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), is a major epidemiological event(1) [1]. As of April 14, 2020, COVID-19 has affected more than 1,948,617 patients in 210 countries and regions around the world, resulting in approximately 121,846 deaths worldwide (2,3) [2,3]. Since there are no effective control measures, the World Health Organization (WHO) eventually requested member countries to expand their emergency response mechanisms to nd new vaccines and effective drugs(4) [4].
Traditional Chinese medicine (TCM), as a core component of the national healthcare system, has been recommended in recent editions of the national guideline for the treatment of COVID-19(5) [5]. COVID-19 belongs to the category of "pestilence" in TCM and is named "pulmonary pestilence". Clinically, it has radiological characteristics of pneumonia, and presenting as fever and respiratory symptoms (6) [6]. Subsequently, the Sichuan Provincial Health Commission recommended the Chinese patent medicine of Compound Yuxingcao Mixture (CYM) for clinical treatment(7) [7]. This prescription consists of Houttuyniae Herba, Isatidis Radix, Scutellariae Radix, Lonicerae Japonicae Flos, Forsythiae Fructus. Among them, Houttuyniae Herba is regarded as sovereign medicine which has the functions of clearing heat and detoxi cation, diuresis and dehumidi cantion (8,9) [8,9]. The CYM has been used in the clinical treatment of COVID-19 and achieved expected results, but its underlying mechanisms remains to be clari ed.
Network pharmacology is an interactive network based on the theory of system biology, including cheminformatics, bioinformatics, network biology and pharmacology. It clearly reveals the underlying mechanisms of active ingredients acting on human body by constructing drug-active ingredient-targetdisease network, which is consistent with the general view of TCM. The purpose of this study is to explore the underlying mechanism of CYM on COVID-19 disease through network pharmacology, drug targeting interaction database and biological analysis methods. Our owchart is shown in Figure 1.

Collection of active ingredients
The active ingredients of CYM were collected from the Traditional Chinese Medicine Systems Pharmacology (TCMSP) platform (http://tcmspw.com/tcmsp.php). Oral bioavailability (OB) refers to the percentage of unmodi ed drugs that enter the circulatory system after oral administration. Drug likeness (DL) is a vague concept that refers to the similarity between compounds and known drugs. According to the relevant parameters of the pharmacokinetic properties, the active ingredients of CYM were screened using the OB% ≥ 30% and DL ≥ 0.18 as parameters (10,11) [10,11].

Screening of target genes for active ingredients
The names and numbers of active ingredients obtained from the TCMSP database were gathered and retrieved on the PubChem database (https://pubchem.ncbi.nlm.nih.gov/), and the corresponding SMILE numbers were recorded as well. Subsequently, the following two databases DrugBank (http://www.drugbank.ca) and STITCH (http://stitch.embl.de/) were used to predict the target information of active ingredients. Finally, all the target information was standardized using UniProt (http://www.UniProt.org/).

Screening of potential targets for COVID-19
No data on COVID-19-related genes were available in the DisGeNET (https://www.disgenet.org/) and MalaCards (https://www.malacards.org/) databases in the present study. Because the new coronavirus is very similar to SARS-CoV and MERS-CoV, Severe Acute Respiratory Syndrome (SARS) and Middle East Respiratory Syndrome (MERS) was used in our research to gather the target genes that may be related to the new coronavirus.

PPI network construction
To explain the interaction between target genes, the selected genes were uploaded to STRING database (http://string-db.org) to obtain the information of protein-protein interaction (PPI). In the present study, the species was set as Homo sapiens, the minimum required interaction score was set as medium con dence 0.400, then the remaining parameters remained the default settings. Subsequently, the information of PPIs was downloaded and visualized using Cytoscape 3.7.1. In all nodes, the size and color of the nodes were used as the criteria for screening hub genes.
Drug-active ingredient-target network construction A Venn diagram was used to visualize the amount of overlap between the genes related to the CYM ingredients and COVID-19-related genes. After removing the redundant genes, the drug-active ingredienttarget network was constructed by cytoscape 3.7.1 software. In the network diagram, the nodes represent ve traditional Chinese medicines, active ingredients, and therapeutic targets respectively. The edges in the network were used to connect drugs-active ingredients-therapeutic targets and the amount of edges were measured in degrees.

Gene ontology and KEGG enrichment analysis of target genes
To elucidate the potential biological function of CYM in the treatment of COVID-19, GO (Gene Ontology) functions and KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway enrichment analysis were performed. The gene list of targets were inputted into STRING (http://string-db.org) and KOBAS (http://kobas.cbi.pku.edu.cn/) to perform GO and pathway analysis. The P-value<0.05 were considered to be statistically signi cant, then the bubble plot of pathways were drawn via imageGP (http://www.ehbio.com/) platform.

Molecular docking
The protein COVID-19 3CL pro (PDB ID:6LU7) required for docking was obtained from PDB (https://www.rcsb.org/) database. The AutoDockTools 1.5.6 software was used to remove water molecules, perform hydrogenation and calculate the charge of the protein. Finally, it was saved as pdbqt format le. The structure of small molecule ligands of the TCM that need to be docked were obtained from Pubmed (https://www.ncbi.nlm.nih.gov/Pubmed) and ZINC (https://zinc.docking.org/). Subsequently, the Autodock vina software was used for molecular docking (the parameters were set as num_modes = 10, energy_range = 4, exhaustiveness = 100).

Results
The herb-property-avor-meridian tropism network for CYM prescription Based on TCM theory, the ve herbs in CYM were classi ed according to their property, avor and meridian tropism (Table 1 and Figure 2). A network was constructed among elements of the TCM classi cation system (Table 1) and the ve herbs of the CYM prescription via R software. The connection degree of lung, heart, stomach and small intestine is the highest in the meridian tropism group, and the corresponding connection values are four, three, two and two, respectively. Four herbs (Houttuyniae Herba, Scutellariae Radix, Lonicerae Japonicae Flos, Forsythiae Fructus) in CYM are associated with the lung meridian. Among the avor group, the greatest degree of connection is bitter, with three degrees of association.

Screening potential targets of CYM
To identify the active compounds of CYM, two classical ADME parameters, OB%≥30% and DL≥0.18, were used for screening. Through mining the TCMSP, a total of 594 potential compounds were screened from ve traditional Chinese medicines (Supplemental Tabel 1). After that, we inputted the molecular name and SMILE number of the active ingredients into the Drugbank and STITCH databases to obtain 151 and 80 putative targets of CYM, respectively. The putative targets included PTGS1, IL-6, JUN, TNF, IFNG, IL-2 and so on. Detailed information about the potential targets of CYM was shown in ( Figure 3A, Supplemental Table 2).

The acquisition of known therapeutic targets for COVID-19
Genes related to COVID-19 were retrieved from the DisGeNET database and MalaCards database using the keywords "SARS" and "MERS". In the study, a total of 260 known therapeutic targets were collected from DisGeNET database and 144 known therapeutic targets were collected from MalaCards database. After eliminating redundant targets, a total of 351 known therapeutic targets were collected ( Figure 3B).
Then the potential targets in CYM was mapped to the COVID-19 targets to obtain 32 therapeutic genes using the ImageGP platform, and a Venn diagram was drawn ( Figure 3C and Table 2). PPI networks construction and core genes screening PPI network was used to explore the function of diverse targets in CYM and COVID-19. A acquisition of 32 targets were inputted into the STRING database, the medium con dence score was set as 0.400 for further analysis, and then visualized by Cytoscape 3.7.1. There were 32 protein nodes and a total of 225 interactive connecting lines in the PPI network. The average node degree of freedom is 14.1, and the avg. local clustering coe cient is 0.758. Among all core targets, the bigger the deeper, the more important it is.
Further analysis from cytoHubba revealed that the degree of tumor necrosis factor (TNF), interleukin 6 (IL-6) and interleukin 1β (IL-1β) is 30, 28 and 25 respectively, which were closely related to the in ammatory response ( Figure 4).

Drugs-Compounds-Targets network analysis
To elucidate the potential mechanism of CYM in the treatment of COVID-19, we used Cytoscape 3.7.1 software to build the Drugs-Active compounds-Targets network. The hexagon nodes represent ve traditional Chinese medicines, the ellipse nodes represent 103 active ingredients and the diamond nodes represent 32 overlapping targets for disease. The edges indicate that nodes can interact with each other. Further analysis of the network topology shows that the centralization and heterogeneity are 0.339 and 1.602 respectively, which indicates that the compound-target space has a tendency for certain compounds and targets. Therefore, the network contains some core ingredients with multi-targets, such as Quercetin (degree = 75), Luteolin (degree = 24), Beta-sitosterol (degree = 28), Wogonin (degree = 18), Acacetin (degree = 10), Kaempferol (degree = 24), which may play a crucial role in the treatment of COVID-19 ( Figure 5).

Gene ontology enrichment analysis
To verify whether the 32 target genes are related to COVID-19, we entered the candidate targets into the STRING platform for GO functional enrichment and annotation. In the study, a total of 750 biological processes, 52 molecular functions, and 25 cellular components were obtained. The top 10 entries were selected based on false discovery rate (FDR) <0.05 and the number of enriched genes, and then visualized by Cytoscape ( Figure 6). The X-axis indicates the number of enriched genes for the term, and the Y-axis indicates the GO term. The results demonstrated that the biological processes mainly involved response to stimulus (32 targets), cellular process (32 targets), biological regulation (30 targets) ( Figure 6A

Analysis of molecular docking and binding modes results
It is worth noting that we have made several milestone discoveries, including the crystal structure of 3CL pro , which can serve as candidate molecular target for SARS therapy(12) [12]. To evaluate the targeting of several active ingredients in CYM to COVID-19 3CL pro , we selected the top six ingredients with larger degrees from the network diagram for molecular docking and binding energy analysis (Table 3 and Figure  8). The binding energy of the core active ingredients of CYM are all less than -5 kJ/mol, of which the lowest binding energy of active compounds with COVID-19 3CL pro is beta-sitosterol (-8.63 kJ/mol).

Discussion
From the perspective of TCM, the medication paradigm for the treatment of COVID-19 is based on a comprehensive and dialectical understanding of the pathological evolution of COVID-19. As the epidemic disease outbreak is urgent, serious and highly infectious, the pathogen is generally attributed to dampness toxin according to the main symptom characteristics of this disease(13) [13]. TCM deduces that COVID-19 is located in lung and closely related to stomach and spleen. Its pathological changes involve the heart, liver and kidney in the later stage(14) [14]. Dampness pathogen can slowly evolve into cold-dampness or dampness-heat, and it depends entirely on the patients' body constitution. Clinical observation reveals that patients will present some symptoms of Qi and Yin de ciency (15,16) [15,16]. Therefore, TCM recommends that the patients should gradually restore su cient health Qi to dispel pathogen from the body, so that they will enter the recovery period. The CYM has the functions of clearing heat and detoxi cation, enhancing the body's immunity, and stimulating the body's immune defense system to indirectly exert anti-viral effects. Among them, Houttuyniae Herba and Scutellariae Radix is regarded as sovereign and assistant medicine respectively. And the rest are adjuvants, which cooperates with together to exert the e cacy.
In this study, the network pharmacology method was used to screen active ingredients in CYM, we obtained their target genes and established a visual network analysis of drugs-active ingredients-target genes. Preliminary analysis indicated that 103 potential active ingredients in CYM might play a crucial role in regulating 32 targets through multiple cellular processes mainly involved response to stimulus (32 targets), cellular process (32 targets), biological regulation (30 targets). Further analysis demonstrated that these 32 targets were roughly divided into three categories: in ammatory cytokines, mitogen-activated protein kinases and others. Among them, the in ammatory cytokines are mainly TNF, IL-6, IL-1β, CCL2, IL-10, IL-2, which concentrate on in ammatory responses. As reported, IL-6 is an important cytokine that initially expressed through the immune system in response to injurious or infectious processes(17) [17]. In the early stage, it activates the JAK/STAT signal pathway, and initiates B cell differentiation, plasma cell production and a series of acute reactions (18,19) [18,19]. The increase of IL-6 occurs early in the initiation of cellular stress and has a long duration, so it can be used as an indicator for the early diagnosis of acute infection as well as the severity and prognosis of infection(20) [20]. The Yi JH team analyzed 69 severe COVID-19 patients and found that the level of IL-6 in severe patients was signi cantly higher than in non-severe patients, which was closely related to the patients' maximum temperature and CT results(21) [21].
According to the characteristics of network topology, the core nodes including Quercetin (degree = 75), Luteolin (degree = 24), Beta-sitosterol (degree = 28), Wogonin (degree = 18), Acacetin (degree = 10), Kaempferol (degree = 24) were screened in the network diagram, which might be the potential ingredients Using the KEGG pathway enrichment tool, we can obtain some important pathways that may be regulated by the drugs studied in the process of treating diseases, thus helping us to understand the mechanisms of drugs in treating diseases. In this study, the selected targets were analyzed by KOBAS platform, and the top 20 entries were screened out including the in ammatory bowel disease (IBD), Jak-STAT signaling pathway, TNF signaling pathway and IL-17 signaling pathway, which demonstrated that CYM could indirectly exert anti-viral effect by enhancing the body's immunity and stimulating the body's immune defense system.

Conclusion
In summary, the network pharmacology method was used in our study to predict the therapeutic targets and underlying mechanisms of CYM in the treatment of COVID-19. Although the predicted results were consistent with recent researches, more clinical trials are warranted for our ndings to be con rmed.    Table 3 Molecular docking results of key targets. Figure 1 The owchart of a systematic network pharmacology-based strategy to explore the pharmacological mechanisms of CYM for the treatment of COVID-19.  Common target PPI network between CYM and COVID-19. The nodes represent target genes, and the connections represent the interaction between two targets. The bigger the deeper, the more important it is.

Figure 5
The drug-active ingredient-target network. The hexagon nodes represent traditional Chinese medicines, the ellipse nodes represent active ingredients and the diamond nodes represent overlapping targets for disease.
Gene ontology (GO) term enrichment for hub genes.

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