Network Analysis, and Experimental Validation to Uncover the Mechanism of the Four Compounds in Artemisia annua (Qing Hao) Antimalarial

Background: Artemisinin is widely used to treat malaria, but the antimalarial mechanism and coordinative interactions governing the actions of artemisinin, scopoletin, arteannuin B and artemisinic acid have not been elucidated. Methods: Based on the existence of antimalarial drugs, the antimalaria targets of artemisinin, scopoletin, arteannuin B and artemisinic acid were investigated by molecular docking using the similarity theory of chemical structure, and the antimalaria mechanism of scopoletin and its coordinative antimalaria interactions with the other three ingredients of the mixture were subsequently examined. Results: Using the text information excavation method, the relevant proteins involved in the antimalarial effect of artemisinin were IL-6, ACHE, PC3, IPOB, CYC, TNF-α, UGT1A9, CASP3, XDH, IL-1β, VEGF, CAT, CREB, AMPK, UGT1A6, ADR, MAPK, COX2, LB24AB and CYP450. The relevant proteins involved in the antimalarial effect of scopoletin were TNF-α, PI3K, IL-8, IL- 6, VEGF, IL-1β, MAPK, CD4, SP2, CTNNB, CASP3, PRO1400, IgE, IL-4, ICAM1, p38, STAT3, TLR4 and API4. However, arteannuin B and artemisinic acid had little relevance to the abovementioned proteins. The interaction property between TNF-α and Artemisia annua was that the effect of the mixture of artemisinin, scopoletin, arteannuin B and artemisinic acid was greater than that of artemisinin alone, and the synergistic effect of the four elements was considered benecial to the progress of antimalarial treatment. Conclusion: The antimalarial targets of Artemisia annua ingredients were examined using data mining methods, and the antimalarial effect of scopoletin may be related to TNF. The combined application of the four elements achieved the same antimalarial effect and reduced the clinical use of artemisinin and deciency β2 killer cell-mediated cytotoxicity pathway. macromolecule verify binding the target binding The nding also showed that of antimalarial effect different. The combination of the four with TNF was very high, that the four have a synergistic effect. The combined use of the four artemisinin used the same antimalarial effect.


Background
Malaria is a major threat to human life. The World Health Organization (WHO) lists malaria, AIDS and cancer as the world's three major deadly diseases. Before the emergence and promotion of artemisinin, approximately 400 million people worldwide were infected with malaria, and at least 1 million people died from malaria annually The morbidity and mortality of malaria are especially high in sub-Saharan Africa.
According to WHO data for 2016, governments and international organizations spent 2.7 billion dollars on malaria control and elimination worldwide [1]. Antimalarial drugs primarily include quinolines, such as chloroquine, me oquine, and quinine, and antifolates, such as pyrimethamine and sulfadoxine. The use of these drugs effectively controlled the global spread of malaria. However, Plasmodium falciparum developed resistance to almost all antimalarial drugs [2]. China was once most strongly affected by malaria and undertook large-scale efforts to eliminate the disease [3]. The unique chemical compositions of the traditional Chinese materia medica have signi cant biological activity in major diseases. Chinese medicines with clear active ingredients are rare, and Artemisia annua (Qing Hao) is a typical chemical composition obtained by the modern, scienti cally veri ed traditional Chinese materia medica. Victory over the disease was nally achieved in 2014, when the number of malaria patients was controlled to 56 individuals in China [4]. Artemisinin was widely used in Thailand and other countries in Southeast Asia in the 1990s. Artemisinin was widely adopted in Africa and the Americas [5]. However, neither artemisinin monotherapy nor artemisinin-based combination therapy was e cacious, and delayed parasite clearance often confused clinicians, similar to other antimalarial drugs [6][7][8]. Plasmodium falciparum developed resistance to artemisinin in the Greater Mekong region, including Cambodia, Laos, Myanmar, Thailand and Vietnam. The WHO 2011 global plan attempted to tackle artemisinin resistance caused by the artemisinin partner drug [9]. The primary reason for this resistance may be the artemisinin partner drug [10]. The combining of artemisinin with another drug with low drug resistance may not delay parasite clearance. It remains entirely possible to rely on artemisinin and its new partner drugs to end the prevalence of malaria [11]. Drugs with low drug resistance must be novel because existing antimalarial drugs developed strong resistance. Scopoletin, arteannuin B and artemisinic acid were considered because their chemical structures are more similar to artemisinin. Compounds that selectively act on two or more targets of interest would theoretically have more pharmacological actions than single-target agents [12]. 'Polypharmacology' is a new methodology in drug discovery [13]. The present study used the method of network pharmacology to perform text mining and target prediction of four components of artemisinin, artemisinin, artemisinic acid and sorghum lactone, which may have antimalarial effects in Artemisia annua L.
Molecular docking is an essential procedure to verify network pharmacology in structural molecular biology and computer-assisted drug design. Molecular docking may be used to perform virtual screening of chemical compounds, rank the results, and propose structural hypotheses on how the ligands inhibit the target, which is highly valuable in lead optimization [14].

Materials And Methods
Reagents and materials

Plasma sample preparation
Approximately 40 mg of EDC and 10 mg of NHS were weighed, and 1 mL of solution was prepared with distilled water. This solution was injected within 5 min into two channels that were thoroughly rinsed with PBS buffer.

Page 4/14
Fifty micrograms of TNF-a protein was dissolved in 100 μL of PBS, and 10 μL of this solution was taken in three portions and diluted with sodium acetate solutions with pH values of 5.5, 6.0, and 6.5, respectively, and the nal concentration was 50 μg/mL. The ow rate was reduced to 20 µL/min, and the left channel was rinsed for 10 min to determine the optimal pH of sodium acetate.
After determination of the optimal pH value, 1 M ethanolamine hydrochloride was injected into the two channels for 10 min to complete the sample xation.

Target shing
Known therapeutic targets for the treatment of malaria were obtained from the DrugBank database (http://www.drugbank.ca/, version 4.3) [15]. The prediction of drug targets based on ligand structural features primarily includes chemical similarity searches and reverse pharmacophore searches. The theoretical basis of the chemical similarity search is that small molecular compounds with similar structural or physicochemical properties act on targets with the same or similar properties: "antimalaria" was selected as the key word, and the drug-target interactions of drugs approved by the USA Food and Drug Administration (FDA) for the treatment of menstrual disorders. All target gene/protein identi ers (IDs) were converted into the corresponding gene symbol/UniProtKB-Swiss-Prot IDs to facilitate further data analyses. After removing redundant entries, 25 target genes corresponding to 15 known antimalarial drugs were retrieved.
Protein-protein interaction (PPI) data PPI data were imported from the Human Annotated and Predicted Protein Interaction Database (HAPPI, http://bio.informatics. iupui.edu/HAPPI/, Version 31.2) [16]. Based on this PPI network database, an interaction network of Artemisia annua candidate target groups and known antimalarial drug target groups was constructed, and the distribution of target nodes in metabolic pathways and the corresponding diseases was determined. A direct interaction network of key nodes was established and divided into different modules according to the functions of the nodes. According to the malaria pathway (ko05144: Malaria) in the Kyoto Encyclopedia of Genes and Genomes (KEGG), molecules closely related to the malaria pathway were selected as candidates to be veri ed from the key nodes. The MCODE algorithm is used to intersect the PPI network for module analysis. The score was 2.2-4.7, while the node was 3-21, and the edge was 2-58.
3. The unit of docking score is kJ/mol. And the unit was mentioned in the manuscript.

Network construction and topological analysis
Compound-target (C-T), target-pathway (T-P), and target-disease (T-D) networks of malaria were constructed using Cytoscape 3.2 software (https://cytoscape. org/download.html), which is a general bioinformatics software package for data integration and visualization of biological networks (Bindea et al., 2009;Smoot et al., 2011). An interaction network of Artemisia annua candidate target genes with known antimalarial drug target genes was established and consisted of 85 nodes and 298 pairs of interactions. The topological characteristic value of each node was calculated in the network, and the median of the topological characteristic value was used as the card value. A total of 32 key nodes were screened. A direct interaction network of key nodes was established and processed according to the node functions. The module was divided, the malaria pathway in KEGG (ko05144: Malaria) was compared, and molecules closely related to the malaria pathway were selected as candidates for veri cation from the key nodes.

Molecular docking
The molecular structures of CDK4, NFKB1, PIK3CG, MAPK1, TNF and ITGB2 human protein targets were searched in the database UniProt (http://www.uniprot.org/). The structures of scopolamine and artemisinic acid were downloaded from the PubChem database (https://pubchem.ncbi.nlm.nih.gov). Chemical compositions and protein structures were dehydrated and hydrotreated, respectively. Molecular docking and gures were generated using Discovery Studio Visualizer 2.5 software.

Probe Kd determination
Approximately 40 mg of EDC and 10 mg of NHS were weighed, and 1 mL of solution was prepared with distilled water. This solution was injected within 5 min into two channels that were thoroughly rinsed with PBS buffer. Next, 19.2 mg of scopolamine, 28.2 mg of artemisinin, 24.8 mg of artemisinin, and 23.4 mg of artemisinic acid were precisely weighed in 1 mL DMSO and mixed well. TNF-α protein was immobilized on grafted sensor chips. The compound monomers and combinations were divided into 12 groups (Table  1). Each group of samples was injected from low to high concentrations, and a control group (PBS) was set at each concentration. Regression analysis was performed when concentration curves were separated and approaching equilibrium. The dissociation constant (Kd) and its maximum value (Bmax) were calculated by tting the titration curve to the single-site saturation binding equation [Y = Bmax*X/(Kd+X)] using GraphPad Prism software (GraphPad Software Incorporated, La Jolla, CA, USA).
Fifteen known antimalarial drugs were retrieved from the DrugBank database (Table 2). There were 93 targets with structures similar to scopolide and 15 known antimalarial drugs (similar score greater than 0.7). Artemisinic acid had 32 targets with structures similar to 15 known antimalarial drugs (similar score greater than 0.7).

Clustering analysis
The key node interaction network was divided into three functional modules. The rst functional module primarily involved immune-related pathways, the second module primarily involved multiple infectious diseases, and the third module was related to drug metabolism and tumour pathways. The KEGG malaria pathway (ko05144: Malaria) comparison is shown in Figure 2. The four key nodes were involved in four pathways: the T cell receptor signalling pathway, the Toll-like receptor signalling pathway, the TNF signalling pathway, and natural killer cell-mediated cytotoxicity, which all have important links to the malaria pathway. Therefore, the joint nodes involved in these four pathways were regarded as candidates for veri cation.  Figure 3). Artemisinic acid ITGB2 -

Experimental validation of key targets
According to the binding curve of the immobilized protein, the highest binding e ciency of sodium acetate solution at pH 5.5 was 110 μRIU/s, but the binding was unstable. The lowest binding e ciency at pH 6.5 was 68 μRIU/s. The binding e ciency was better at pH 6.0, and the comparison of binding was stable at 108 μRIU/s. Therefore, the optimal approach for protein xation was sodium acetate solution at pH 6.0.
TNF had good binding with scopolide, which suggests that the antimalarial effect of scopolide may be related to TNF. The binding of TNF to artemisinin, artemisinin B, and artemisinic acid was weak, but the combination of the 4 components of artemisinin, artemisinin B, artemisinic acid, and stigmalactone had good binding to TNF, which suggests that the combined application of 4 ingredients may achieve antimalarial effects by acting on TNF (Table 4, Figure 4).

Discussion
The sesquiterpene compounds represented by artemisinin are the most heavily researched compounds in Artemisia annua. Nearly 61 sesquiterpenoids, primarily artemisinin compounds, were identi ed from Artemisia annua, including artemisinic acid, artemisinol, artemisinin ether and artemisinin. Artemisinin is a sesquiterpene lactone that contains an endoperoxide bridge structure, and it is the main component of several antimalarial treatments. Research data showed that artemisinin and artemisinin may be converted to artemisinin in the original plants [17,18]. The biosynthetic routes of artemisinin may be summarized as 9 total synthetic routes and 5 semi-synthetic routes. The synthesis method of artemisinin is a long process with high cost and low total output that produces a maximum yield of 10%. Artemisinic acid is one of the main components of sesquiterpenes in Artemisia annua plants and is an important precursor of artemisinin synthesis. Tu Yu showed that young plants of Artemisia annua contained a large amount of artemisinic acid but exhibited a shortage of artemisinin. It is speculated that sesquiterpenoids, such as artemisinin, are converted from artemisinic acid. Levesque F and other researchers used synthetic biology to successfully produce artemisinin using genetically engineered yeast [19]. Scopolide has strong water solubility and stability to artemisinin and has pharmacological activity that re ects the e cacy of traditional artemisinin. Previous studies showed that scopolide had certain antimalarial effects and certain synergistic effects with artemisinin. With the rapid development of chemical genomes and pharmacological technologies, a large number of potential targets and massive biological activity data emerged. However, with the accumulation of complex data, simple analysis methods no longer satisfy the analytical needs of high-throughput and large-scale data [20]. The rapid development of chemical informatics recently met the requirements of big data processing and information extraction tasks that are urgently needed in chemical genomics. Chemical informatics primarily studies how to properly select compounds from diverse compound libraries, how to describe drug molecular characteristics, how to measure the differences between different molecules, how to identify drug-like molecules, molecular structure and biological performance relationships, and how to develop corresponding computer software and hardware. This methodology includes the research tasks and content of chemometrics and computational chemistry [21]. One important application of the chemoinformatics method in the post-genomic era is predicting the potential targets of small-molecule compounds based on existing biological and chemical information and explaining their mechanism of action to accelerate the development of drugs. The prediction of drug targets is considerably important to the evaluation of early drug molecules and the new use of old drugs. However, due to the limitations of throughput, accuracy and cost, it is di cult to widely apply experimental methods. As a quick and lowcost method, the development of computer-aided target prediction algorithms is receiving increasing attention. According to different research strategies, the prediction of drug targets based on chemoinformatics may be divided into three categories: predictions based on ligand characteristics; predictions based on protein structural characteristics; and predictions based on data mining methods [22,23]. According to the prediction of targets, the potential targets of scopolide are CDK4, NFKB1, PIK3CG, MAPK1, and TNF, and the potential target of artemisinic acid is ITGB2. Cyclin-dependent kinase (CDK) is a serine/threonine (Thr) kinase that is an important signal transduction molecule in cells. CDKcyclin is formed by the cyclin complex and is involved in cell growth, proliferation, dormancy, or apoptosis. During the cell cycle, cyclins are periodically and continuously expressed and degraded and bound to CDKs, which are transiently activated by cyclins. Cyclins catalyse the phosphorylation of different substrates via CDK activity to promote and transform different phases of the cell cycle. The CDK family includes CDK1-13, and cyclins are divided into cyclins A-L. Different CDKs are connected to different cyclins. CDK4/6-speci c activation is closely related to the proliferation of some tumours. Rb is present in approximately 80% of human tumours, and abnormalities in the cyclin D-CDK4/6-INK4-Rb pathway are common [24]. It is characterized by (1) p16 INK4a gene deletion, point mutation, or DNA methylation, which leads to the inactivation of p16 INK4a , and (2) CDK4 gene ampli cation or pointmutated T cells induce other cells to activate or interfere with lysis. CD3, CD4, and CD8 cells are involved in the T cell transcription of activation signals. Toll-like receptors (TLRs) play an important role in the identi cation of invading pathogenic microorganisms in early congenital immunity. These evolutionarily preserved receptors are homologous to the Drosophila Toll protein family in structure, and they recognize highly conserved structural motif (motif)-pathogen-associated molecular patterns expressed only on pathogenic microorganism molecular patterns (PAMPs). PAMPs stimulate TLRs to initiate a signalling cascade that includes several proteins that lead to the activation of the transcription factor NF-kB, which induces the secretion of pro-in ammatory cytokines and effector cytokines that are directly involved in the adaptive immune response. Integrin β2 (CD18) is an important member of the integrin family of adhesion molecules. Integrin β2 binds to different integrin subunits to form the leukocyte adhesion receptor group. Integrin β2 is primarily expressed in white blood cells, and its ligands are TCAM, iC3b, and brinogen. The cytoplasmic region of integrin β2 is linked to a variety of cytoskeletal proteins and is involved in signal transduction. The genetic defects of integrin β2 lead to leukocyte adhesion de ciency syndrome. Integrin β2 primarily exists in the natural killer cell-mediated cytotoxicity pathway. The present study used the biological macromolecule interaction instrument to verify the binding of the target protein TNF with scopolide, which had the highest score in the docking experiment. The binding effect was good and con rmed that scopolide acted on TNF and participated in its corresponding functions. The results con rmed that scorolactone had an antimalarial effect. However, the binding rates of artemisinin, artemisinin B, artemisinic acid and TNF were very low, which may be observed because these three compounds did not act on TNF. This nding also showed that the mechanism of the antimalarial effect of artemisinin may be different. The combination of the four ingredients with TNF was very high, which indicates that the four ingredients have a synergistic effect. The combined use of the four ingredients may reduce the amount of artemisinin and scopolamine used to achieve the same antimalarial effect. The characteristics of multicomponent, multitarget, and synergistic effects are common in traditional Chinese medicine preparations. Many experiments indicated that TNF had a certain killing effect on Plasmodium. TNF must be assisted by a certain factor (or factors) in the body to exert its ability to damage Plasmodium, which means TNF is not a terminal effector that directly kills Plasmodium. The immune-protective mechanism of TNF may have the following effects. (1) Enhancing the phagocytosis function of phagocytic cells: one study treated neutrophils with different doses of TNF for 30 min followed by incubation with Plasmodium falciparum and found that the phagocytosis of each stage of Plasmodium was strengthened, and the extent of the increase positively correlated with the dose of TNF within a certain range. (2) Acting via reactive oxygen mediator: when TNF and macrophages were coincubated for 30 min, the release of reactive oxygen species (ROS) from macrophages was detected, and ROS kill Plasmodium.
The combination of the four components artemisinin, arteannuin B, artemisinic acid, and stigmalactone exhibited good binding to TNF.