Uncovering anti-influenza mechanism of Ophiocordyceps sinensis using network pharmacology, molecular pharmacology, and metabolomics

Abstract

Ophiocordyceps sinensis is a precious Chinese traditional herb with a long medicinal history. This study used UPLC-MS metabolomics to explore and compare the metabolic profiles of the stroma (OSBSz), sclerotium (OSBSh), and mycelium (OSBS) of O sinensis to analyze their differential metabolites and identified potential active components. Then combined with network pharmacology and molecular docking to explore the mechanism of differential metabolites with anti-influenza properties. The results indicate that the stroma, sclerotium, and mycelium showed significant differences in metabolites. The key pathways for differential metabolites were butanoate metabolism, thiamin metabolism, alanine, aspartate and glutamate metabolism, citrate cycle, and arginine biosynthesis. Protein-protein interaction analysis identified potential targets, including SRC, RHOA, HSP90AA1, VEGFA, ITGB1, PRKCA, and ITGA1, and the key protective pathways in-volved PI3K-Akt, HIF-1, influenza A, and Coronavirus disease 2019. The molecular docking results showed that the core metabolite D-(−)-glutamine has high binding affinity with SRC, RHOA, and EGFR, re-flecting the multi-component and multi-target network system of O sinensis. In short, the combination of metabonomics, network pharmacology and macromolecular docking technology provides a new way to explore the anti-influenza research of O sinensis. This is undoubtedly an important theoretical support for the clinical application of O sinensis in the future.

1. Introduction

Ophiocordyceps sinensis (Berk.) Sung et al, is an important traditional Chinese medicine (TCM) from the Qinghai–Tibet Plateau and its surrounding regions, including Gansu, Qinghai, Sichuan, Tibet, Yunnan, Nepal, Bhutan, and India.^[1] Many pharmacological constituents have been isolated from O sinensis; these include ergosterol, adenosine, and cordyceps acid. However, no previous study has provided a comprehensive metabolic profile of the stroma, sclerotia, and cultured mycelium of O sinensis.

O sinensis has multiple bioactivities, including those relating to antitumor, immune regulation, cardiovascular protection, treatment of diabetes, and anti-oxidation.^[2–7] The cultured mycelium of O sinensis is regarded as a substitute for O sinensis and appears to have significant biological activity. Indeed, Wu et al found that sphingolipids isolated from the cultured mycelium of O sinensis demonstrated immunological activity by inhibiting the viability of mouse splenocytes, as well as that of splenocyte-derived B and T cells.^[8] Other studies also revealed that preparations of mycelium could alleviate renal injury in diabetic nephropathy mice by regulating metabolic disorders. Further metabolomic analysis of metabolites in the plasma, urine, and kidneys of mice revealed that the mycelium modulated glucose and lipid metabolism and disturbed amino acid turnover.^[9] Zhong et al demonstrated the anti-inflammatory effect of water extract from fresh O sinensis on single stimulation of cigarette extract and combined influenza virus infection in vitro, suggesting that O sinensis has the potential to protect against influenza viruses.^[10] Metabonomics is a new discipline that emerged after genomics, transcriptomics, and proteomics, which focuses on small-molecule metabolites present in biological systems.^[11] Although genomics and proteomics explore life activities at the gene and protein levels, many life activities in cells occur at the metabolite level, including cell signal release, energy delivery, and intercellular communication regulation by metabolites. Due to its high-throughput and high-resolution advantages, metabolomics has become one of the hot spots in life science research. Xu et al performed a metabolomics study based on UPLC-MS to discriminate the potential biomarkers of diabetic nephropathy.^[12] Metabonomics, combined with pattern recognition, expert system, and other analysis methods, represents one of the best methods to discover potential active ingredients at present.^[13]

The development of network pharmacology is based on the deepening understanding of the interaction between molecules and proteins. Hence, network pharmacology has great potential for understanding the occurrence and development of diseases, TCM syndromes, and the mechanisms underlying treatment of diseases with TCM.^[14] Network pharmacology integrates disciplines such as systematic biology, multi-directional pharmacology, computational biology, and network analysis to explore the relationship between TCM and diseases from a holistic perspective.^[15] Molecular docking technology is a mature technology for the direct design of chemical drugs using computer aids.^[16] Molecular docking uses computer technology to simulate the geometric structure and intermolecular forces of molecules through chemometric methods, with the aim to elucidate small molecules (or ligands) with known molecules and construct an active site of a macromolecule (or receptor) in a low-energy binding mode.^[17] Network pharmacology and molecular docking technologies are used to predict the treatment of diseases and explore the underlying mechanisms.^[18,19]

Therefore, in this study, we employed UPLC-MS metabolomics to explore and compare the metabolic profiles of the substrates, sclerotia, and cultured mycelia of O sinensis to analyze their differential metabolites and identify potential active components. Combining network pharmacology and molecular docking technology allowed us to explore the mechanism of differential metabolites with anti-influenza properties. The goal of this study was to provide support for pharmacological activity research of O sinensis.

2. Materials and methods

2.1. Metabolomics

2.1.1. Sample and culture.

O sinensis was collected from the Baima Snow Mountains, and was identified by Professor Yu Hong of Yunnan University. Cultured mycelium was prepared in the Yunnan herbals Laboratory of Yunnan University. Three biological replicates were performed for each experiment.

2.1.2. Metabolite extraction.

After slowly thawing the sample at 4°C, 25 mg of the sample was weighed and placed in an Eppendorf tube. Next, 800 μL of extract solution (methanol: acetonitrile: water = 2:2:1, precooling at −20°C), 10 μL of internal standard, and 2 small steel balls were added and placed in a tissue grinding machine for grinding (50 Hz, 5 minutes). After ultrasonic treatment in a 4°C water bath for 10 minutes, the samples were placed at −20°C for 1 hour, before centrifuging at 25,000 rpm at 4°C for 15 minutes. After centrifugation, 600 μL of supernatant was removed and drained by a vacuum concentrator. Subsequently, 200 μL of complex solution (methanol: H₂O = 1:9) was added for redissolution, followed by vortexing for 1 minute and ultrasonic treatment in a 4°C water bath for 10 minutes. Finally, the sample was centrifuged at 25,000 rpm for 15 minutes at 4°C, before removing the supernatant and placing in a sample bottle.

2.1.3. UPLC-MS conditions.

The samples were analyzed on a Waters 2D UPLC (Waters, MA), coupled to a Q-Exactive mass spectrometer (Thermo Scientific, MA) with a heated electrospray ionization source and controlled by Xcalibur 2.3 (Thermo Scientific, MA). Chromatographic separation was performed on a Waters ACQUITY UPLC BEH C₁₈ column (1.7 μm, 2.1 × 100 mm, Waters, MA), and the column temperature was maintained at 45°C. The mobile phase consisted of 0.1% formic acid (A) and acetonitrile (B) in the positive mode, and in the negative mode, the mobile phase consisted of 10 mM ammonium formate (A) and acetonitrile (B). The gradient conditions were as follows: 0 to 1 minute, 2% B; 1 to 9 minutes, 2% to 98% B; 9 to 12 minutes, 98% B; 12 to 12.1 minute, 98% B; 12.1 to 15 minutes, 2% B. The flow rate was 0.35 mL/minutes and the injection volume was 5 μL.

The mass spectrometric settings were as follows: spray voltage, 3.8/−3.2 kV; sheath gas flow rate, 40 arbitrary units (arb); aux gas flow rate, 10 arb; aux gas heater temperature, 350°C; capillary temperature, 320°C. The full scan range was 70 to 1050 m/z, with a resolution of 70000. The automatic gain control target for MS acquisitions was set to 3e6, with a maximum ion injection time of 100 ms. The top 3 precursors were selected for subsequent MSMS fragmentation, with a maximum ion injection time of 50 ms, a resolution of 30000, and automatic gain control of 1e5. The stepped normalized collision energy was set to 20, 40, and 60 eV.

2.1.4. Data processing and statistical analysis of metabolomics.

The mass spectrometry raw data collected by UPLC-MS was imported into Compound Discoverer 3.1 (Thermo Fisher Scientific, Waltham, MA) for data processing, including peak extraction, retention time correction within and between groups, additive ion pooling, missing value filling, background peak labeling, and metabolite identification. Finally, the compound molecular weight, retention time, peak area, and identification results were exported.^[20]

2.1.5. Metabolite identification and analysis of differentially expressed metabolites.

Metabolites were identified by combining the mzCloud (https://www.mzcloud.org/), ChemSpider (https://www.chemspider.com/), HMDB (https://hmdb.ca/), and Kyoto encyclopedia of genes and genomes (KEGG) (https://www.genome.jp/kegg/compound/), and LipiMaps (https://www.lipidmaps.org/) databases. The main parameters for metabolite identification were as follows: precursor mass tolerance, <5 ppm; fragment mass tolerance, <10 ppm, RT tolerance, <0.2 minutes. To compare differences between groups, the SIMCA 13.0 software package was used for multivariate statistical analysis. Principal component analysis (PCA) was used to explain the population distribution of all samples. Partial least squares discriminant analysis (PLS-DA) was used to identify the general separation of groups. Linear discriminant analysis effect size (LEfSe) analysis, hierarchical cluster analysis, and content percentage analysis were completed using Bioincloud (https://bioincloud.tech/). Metabolomics pathway analysis based on differential metabolites was performed with Metaboanalyst (https://www.metaboanalyst.ca/), a website tool that includes the KEGG (www.genome.jp/kegg/) and HMDB (www.hmdb.ca/) databases.

2.2. Methods of network pharmacology and molecular docking technology

2.2.1. Collection of active ingredients.

Based on the analysis of the differential metabolites of the metabolome, 48 metabolites were screened out. The structure of the active ingredient was confirmed on the PubChem (https://pubchem.ncbi.nlm.nih.gov/) platform, and the 2D structure of the active ingredient was downloaded and saved as an SDF structure. Alternatively, ChemDraw18.0 software was used to draw the 3D structural formula of the active ingredient, before minimizing the energy and save it as an SDF structure. Next, the file was imported into the Swiss ADME database (www.swissadme.ch), the smiles number of each active ingredient was obtained, and “Run” was clicked to run. According to the level of GI absorption, indicating oral bioavailability, active ingredients with good pharmacokinetic properties were screened out using the drug-like index. In this database, the components whose gastrointestinal absorption was “High” were set as the conditions that can be absorbed, before subjecting to oral bioavailability screening. At least 2 of the 5 drug-like prediction results of lipinski, ghose, veber, egan, and muegge were collected as Compounds with “Yes.” The 2 collected results were combined and the duplicates were deleted. These synthesized results were regarded as the potential active components of O sinensis and as the potential material basis for predicting the target of action.

2.2.2. Screening of active ingredients with access to components and disease targets.

The SDF file in item 2.2.1 was imported into Swiss Target Prediction for target prediction, and the targets were screened with Probability >0 to obtain all of the targets of the active ingredient. Influenza-related disease genes were obtained by searching the Gene Cards (www.genecards.org) and OMIM (www.omim.org) databases with “influenza” as the keyword.

2.2.3. Construction of the TCM-ingredient-target-disease network graph.

The drug genes and influenza targets were entered into the Bioinformatics (bioinformatics.cn), and the intersection targets of TCM targets and disease targets were screened to construct a Venn diagram. The intersection target name and active ingredients of TCM were imported into Cytoscape 3.2.1 for visual analysis, and the network diagram of the TCM-component-target was constructed.

2.2.4. Construction of the protein-protein interaction (PPI) network.

Use the STRING 11.5 platform (https://cn.STRING-db.org/cgi/) to construct a PPI network of O sinensis target protein interaction, the multiple proteins tool was selected, and the protein type was set to “Homo sapiens.” Through the network analyzer plug-in of Cytoscape 3.2.1 software, the network Topology properties of O sinensis influenza “TCM component target disease” network were analyzed, and the active components with higher Betweenness, Closeness and Degree values were obtained as key components. Further analysis of the obtained targets was carried out using the CytoHubba plugin of Cytoscape 3.2.1 software, using the maximal clique centrality topology analysis method.

2.2.5. Gene ontology (GO) functional enrichment in KEGG biological pathway enrichment analysis.

The target genes of O.sinensis were imported into the Metscape database (https://metascape.org/gp/index.html#/main/step1) for GO biological process and KEGG metabolic pathway and enrichment analyses, and the species selection is “H.sapiens.” The data results were saved, and bioinformatics was used for visual analysis.

2.2.6. Molecular docking.

The SDF format of compounds was downloaded from the PubChem (nih.gov) platform and saved. The 3D crystal structures of key targets were downloaded from the RCSB database (http://www.rcsb.org/pdb/home/home.do) and saved in PDB format. PyMOL 2.5.1 was used to remove irrelevant ions such as water and phosphate. The target protein receptor and ligand were split, small ligand molecules were extracted, and the pretreated target protein and original ligand were saved as PDB format files. AutoDock Vina 1.5.6 was used to set the size of the Grid Box, before establishing the protein active site (active port) parameters and converting the active ingredients, preprocessed target proteins, and original ligands into pdbqt format files for molecular docking. Using the top 5 small molecules in 2.2.1 as ligands and the top 3 proteins in 2.2.4 as docking receptors. Finally, PyMOL was used to visualize the docking results.

3. Results

3.1. PCA

PCA is one of the most widely used unsupervised methods. The PCA score plots (R²X = 0.77, Q² = 0.51) of OSBS, OSBSz, and OSBSh are shown in Figure 1A. All of the samples were in the 95% confidence and data points of OSBS, and OSBSz and OSBSh were significantly distinguished in the PCA score plot. The results showed significant differences in the metabolites of OSBS, OSBSz, and OSBSh.

Figure 1.

(A) PCA model score of OSBSz, OSBSh, and OSBS. (B) PLS-DA model score of OSBSz, OSBSh, and OSBS. (C) Model random permutation test for PLS-DA (n = 200). PCA = principal component analysis.

3.2. PLS-DA

PLS-DA can effectively discriminate observations between groups and identify the influencing variables that lead to differences between groups. As shown in the PLS-DA score plot in Figure 1B (R²X = 0.77, R²Y = 0.94, Q² = 0.85), the samples of OSBS, OSBSz, and OSBSh were distributed in different quadrants with good sample repetition, which proved that the metabolites of the 3 were significantly different (Fig. 1B). To test whether the PLS-DA model was over-fitting, a random group of 200 replacement tests was performed. The results showed that R² and Q² in the model were very close in Figure 1 (R² = 0.35, Q² = −0.2), confirming the reliability of the established model (Fig. 1C).

3.3. Differential metabolite screening and LEfSe analysis

PLS-DA is used to screen differential metabolites from a large number of signals. The variable important in projection (VIP) in the model is an important variable weight value, which can be used to measure the impact strength of the differences in the accumulation of each metabolite on the classification and discrimination of each group of samples. Metabolites with a VIP > 1 in the PLS-DA model and P < .05 by one-way ANOVA were considered as differential metabolites of OSBS, OSBSz, and OSBSh. Differential metabolites include amino acids, nucleosides, flavonoids, and alkaloids.

LEfSe is a software for discovering high-dimensional biomarkers and revealing genomic features, including genes, metabolism, and taxonomy, which is used to distinguish 2 or more biological conditions.^[19] LEfSe analysis has a powerful recognition function through biological statistical differences. At present, LEfSe analysis is mainly used in the study of microbial diversity to explain the species with significant differences in abundance between groups. During LEfSe analysis, linear discriminant analysis (LDA) was used to estimate the effect of the abundance of each component on the differential effect, and the abundance of each group was significantly higher than that of the other groups. The metabolites with LDA scores of ≥3.0 are shown in clad ults demonstrated 59 significantly different metabolites between the OSBSz, OSBSh, and OSBS groups. Among these differential metabolites, 48 (font bold in Fig. 2A) were also screened by VIP > 1 and P < .05; among them, all 30 differential metabolites in OSBS satisfied VIP > 1 and P < .05.

Figure 2.

(A) LDA effect size (LEfSe) analysis of OSBSz, OSBSh, and OSBS (LDA score >3.0) characterizing the differences in metabolites among samples. Different group bars are colored uniquely. (B) Hierarchical cluster analysis for OSBSz, OSBSh, and OSBS. LDA = linear discriminant analysis, LEfSe = linear discriminant analysis effect size.

3.4. Hierarchical cluster analysis

Next, the samples were clustered to study the similarity between different samples. The differential metabolites obtained by the above analysis were selected to realize sample clustering. The results showed that the metabolites of OSBSz, OSBSh, and OSBS were significantly different (Fig. 2B), and compared to the difference between OSBSz and OSBSh, OSBS was more different from OSBSz and OSBSh. The relative expression of adenosine in OSBS was higher than that of OSBSz and OSBSh. Tryptamine, styrene, and pipemidic acid were more abundant in OSBSz, while sesipramine, N-propyl gallate, and tropine were more abundant in OSBSh. Most metabolites, including thiamin, caffeine, pipecolate, adenosine, and maltose, were more abundant in OSBS, suggesting that they play an important role in the growth of the stroma and sclerotium, and in the process of mycelial propagation.

3.5. Column chart of percent content

The significantly different metabolite content of each sample is shown by the stacked column chart of percentage content. The content of metabolites such as l-arginine, pyroglutamate, citrate, and 2,3-dihydroxypropyl hexadecanoate in OSBS was significantly higher than that in OSBSz and OSBSh. The content of metabolites such as oleamide, gamma-aminobutyric acid, dl-tryptophan, and (+)-eudesmin was significantly higher than that in OSBS. We also identified significant differences in the metabolite content between OSBSz and OSBSh. Indeed, the content of l-arginine in OSBSh is higher than that in OSBSz, while the opposite is true for tryptamine (Fig. 3A).

Figure 3.

(A) Column chart of 48 metabolites in percentage stacking. (B) Differential metabolite metabolic enrichment analysis. (C) Differential metabolite metabolic pathway map.

3.6. Analysis of partial differential metabolite pathways and enrichment analysis

Differential metabolites were functionally annotated through the KEGG database to determine their functional properties and identify their major biochemical metabolic pathways. Pathway enrichment analysis showed that the enriched pathways were mainly subordinate to metabolic functions, followed by biosynthesis, and were mainly enriched in 21 metabolic pathways (Fig. 3B). Pathways significantly enriched for differential metabolites were further bubble-plotted (Fig. 3C), and the most enriched important pathways were butanoate metabolism, thiamin metabolism, alanine, aspartate and glutamate metabolism, citrate cycle (TCA cycle), arginine biosynthesis, porphyrin and chlorophyl metabolism, glyoxylate and dicarboxylate metabolism, and purine metabolism. Adenosine and sulfuric acid were highly expressed in OSBS compared to in OSBSz and OSBSh, and both are mainly involved in the purine metabolism pathway. Gamma-aminobutyric acid, mainly involved in butanoate metabolism, is highly expressed in OSBSz; thiamin, mainly involved in thiamin metabolism, is highly expressed in OSBS; alpha-ketoglutaric acid is mainly involved in alanine, aspartate, and glutamate metabolism pathways. Alpha-ketoglutaric acid and citrate are major components of the citrate cycle (TCA cycle) pathway, and alpha-ketoglutaric acid and l-arginine are mainly involved in the arginine biosynthesis pathway. Porphobilinogen of OSBS is mainly involved in the porphyrin and chlorophyl metabolism pathway. The abscissa where the bubble is located and the size of the bubble represent the influence value, where the larger the bubble, the greater the importance of the pathway. The ordinate where the bubble is located and the color of the bubble represent the P value of the enrichment analysis, where the redder the color, the closer the P value is to 0, and the more significant the enrichment.

3.7. Screening of active ingredients and construction of the TCM-ingredient-target-disease network graph

After screening in item 2.2.1, 37 compounds were screened from 48 compounds (Table 1). The 37 active ingredients after screening were used to predict potential targets through the Swiss Target Prediction database, and a total of 527 targets were obtained after deleting duplicate values. Moreover, 1383 influenza-related targets were retrieved from the Gene Cards (www.genecards.org/) database, and 1534 related targets were retrieved from the OMIM (www.omim.org) database. A total of 1417 influenza-related targets were obtained after combining the results and removing the duplicate values. The obtained targets were intersected with the potential targets of O sinensis, and 107 potentially related potential targets were obtained (Fig. 4A). The O sinensis-influenza “TCM-ingredient-target-disease” network diagram was drawn by Cytoscape 3.2.1 software (Fig. 4B), which contained a total of 133 nodes and 203 edges. The network topology properties were analyzed by the Network Analyzer plug-in. The active ingredients with larger betweenness, closeness, and degree were d-(−)-glutamine, Asp-leu, caffeine, 4-(2-aminopropyl)-2-methoxyphenol, 2,3 dihydroxypropyl hexadecanoate, corresponding to 29, 16, 14, 13, and 12 targets, respectively. These active ingredients are the center of the network, indicating that the interaction between these components and the target plays a key role in the network and how different active components act on multiple targets. The same target was found to correspond to different components, reflecting the multi-component and multi-target network system of O sinensis.

Table 1.

Screening of active ingredients.

ID	PubChem CID	Compound	ID	PubChem CID	Compound
OS1	119	Gamma-aminobutyric acid	OS20	535	1-Aminocyclopropanecarboxylic acid
OS2	1150	Tryptamine	OS21	12035	n-acetyl-l-cysteine
OS3	73117	(+)-eudesmin	OS22	7361	Furfuranol
OS4	197139	4-(2-aminopropyl)-2-methoxyphenol	OS23	6992367	Asp-leu
OS5	235498	17beta-Hydroxy-4,17-dimethyl-4-azaandrost-5-en-3-one	OS24	849	Pipecolate
OS6	2724760	D-(−)-glutamine	OS25	13591	1-nitrosopyrrolidine
OS7	288	Dl-carnitine	OS26	5375048	Trans-3-indoleacrylic acid
OS8	525	Dl-malic acid	OS27	440014	Cis-4-hydroxy-d-proline
OS9	4831	Pipemidic acid	OS28	1021	Porphobilinogen
OS10	5283387	Oleamide	OS29	169148	N3, n4-dimethyl-l-arginine
OS11	124688	Hydroxycarteolol	OS30	107689	L-(+)-lactic acid
OS12	13591	1-nitrosopyrrolidine (npyr)	OS31	4947	N-propyl galiate
OS13	442424	Genipin	OS32	2995	Desipramine
OS14	8424	Tropine	OS33	1130	Thiamine
OS15	1148	Dl-tryptophan	OS34	51	Alpha-ketoglutaric acid
OS16	319074378	Α-l-fucopyranose	OS35	2519	Caffeine
OS17	22035687	1-stearoylglycerol	OS36	602	Dl-alanine
OS18	7405	Pyroglutamate	OS37	10341	Butenolide
OS19	14900	2,3-dihydroxypropyl hexadecanoate

Figure 4.

(A) Venn diagram of O sinensis and influenza targets. (B) O sinensis-influenza “Chinese medicine-ingredients-targets-disease” network diagram, where OS represents O sinensis, the purple circle represents the active ingredient of O sinensis, and the orange arrow represents the target. (C) PPI network of potential targets in influenza. (D) PPI key sub-network of O sinensis. PPI = protein-protein interaction.

3.8. PPI network construction

The PPI network was constructed after analyzing the target proteins of O sinensis using the STRING database. To obtain high confidence, the highest confidence of the PPI network was set to 0.9, and the disconnected nodes in the network were hidden. After screening, 82 targets were found to have protein interactions, and 229 edges represent the protein interaction relationship. The mean degree of freedom was 4.28 (Fig. 4C). Using cytoHubba to rank nodes according to the attributes of the nodes in the network, the MCC topology analysis method was selected to extract key sub-networks. The proteins with high degree value were calculated to be more likely to be key proteins. The top 10 key target proteins were SRC, RHOA, EGFA, HSP90AA1, VEGFA, MAPK1, ITGB1, JUN, PRKCA, and ITGA1 (Fig. 4D).

3.9. GO functional enrichment

GO functional enrichment analysis includes 3 parts, biological process, molecular function, and cellular component. The O sinensis-influenza intersection target was imported into the Metascape database, the target gene column was input, the species was limited to “Homo Sapiens,” and the P value cutoff was set to 0.01 for enrichment analysis. Biological processes involved the positive regulation of cell migration, positive regulation of the response to external stimulus, regulation of cell adhesion, cellular response to chemical stress, and response to lipopolysaccharide. Molecular functions in kinase binding, protein kinase activity, endopeptidase activity, heme binding, and protease binding were concentrated. Cell components associated with the membrane raft, side of the membrane, vesicle lumen, ficolin-1-rich granule, and focal adhesion were the greatest (Fig. 5A–C).

Figure 5.

(A) Gene ontology (GO) functional enrichment analysis of biological process (BP); (B) GO functional enrichment analysis of molecular function (MF); (C) GO functional enrichment analysis of cellular component (CC); and (D) Kyoto encyclopedia of genes and genomes (KEGG) enrichment analysis of anti-influenza activity.

3.10. KEGG enrichment analysis

A total of 170 pathways were obtained from the KEGG enrichment analysis of the Metascape database. The pathways were sorted according to the P value from small to large, irrelevant pathways were deleted, and the top 20 pathways were selected for visual analysis (Fig. 5D). The key pathways included PI3K-AKT, HIF-1, IL-7, influenza A, Coronavirus disease 2019 (COVID-19), and other pathways. Among them, the enriched targets of the influenza A pathway comprised 14 targets, including AKT1, CASP1, CASP3, CASP8, and CCND3. These results demonstrate the anti-influenza activity of O sinensis with multiple targets and multiple pathways.

3.11. Molecular docking

According to the results outlined in sections 3.8 and 3.9, the chemical components with the top 5 degrees and the key targets with the top 3 degrees were selected for molecular docking. The top 5 components were d-(−)-glutamine, asp-leu, caffeine, 4-(2-aminopropyl) − 2-methoxyphenol, 2,3-dihydroxypropyl hexadecanoate, and the top 3 targets in terms of degree value were SRC, RHOA, and EGFA. Generally, when the binding energy is <0, the receptor and the ligand can spontaneously bind, and when the binding energy is <5 kcal mol-1, the small molecule can be considered to have better binding activity with the protein. The lower the binding energy, the greater the possibility of the interaction between the small molecule and the protein, and the stronger the binding ability (Table 2). The binding energy between each chemical component and the target protein was <5 kcal mol⁻¹, indicating that the binding conformation between the ligand and the receptor is stable and has good binding activity. Among them, d-(−)-glutamine showed the most stable docking conformation with SRC, RHOA, and EGFR, and had the best binding ability. Visual analysis of the top 3 docking results (Fig. 6) suggested that the active ingredient spontaneously combined with key target proteins and formed a relatively stable docking conformation through hydrogen bonding and other forces.

Table 2.

Docking results of the main active components of O sinensis with key targets.

Main active compounds	Key target docking score
	SRC	RHOA	EGFR
2,3-dihydroxypropyl hexadecanoate	−6.8	−5.6	−5.9
4-(2-aminopropyl) − 2-methoxyphenol	−6.4	−6	−6.3
Asp-leu	−6.6	−6.1	−6.9
Caffeine	−5.5	−5.6	−6.1
D-(−)-glutamine	−8.5	−7.3	−7.5

Figure 6.

Three best docking results. (A) SRC and d-(−)-glutamine; (B) RHOA and d-(−)-glutamine; and (C) EGFR and d-(−)-glutamine.

4. Discussion

Untargeted metabolomics, as a form of metabolite fingerprint analysis, analyzes the whole endogenous small molecules, as opposed to isolating and identifying a single component.^[21] Untargeted metabolomics is expected to have wider applications, particularly in terms of discovering biomarkers or elucidating metabolic signatures because of its scanning mode and ability to collect large amounts of data in a short period.^[20] However, traditional univariate analysis methods cannot find stable markers representing differences between groups from the large amounts of metabolic profile information, which is an important challenge in the current development of metabolomics.^[22] This also means that more scientific and rigorous models are needed for screening differential metabolites in process of metabolomic data analysis. PLS-DA is based on partial first squares regression. Previous studies have used the PLS-DA model to find differential components between different sample groups and identify biomarkers, largely because it is easier to perform model interpretation, and the screening criteria are usually VIP and P- value.^[22,23] LDA was conducted to find projections to minimize within-class distance while maximizing between class distance. Notably, LDA can find more representative biomarkers by setting thresholds, although LDA is more commonly used in the analysis of biodiversity.^[24,25] In this study, we attempted to combine LDA analysis and the PLS-DA model to screen common differential metabolites of OSBS, OSBSz, and OSBSh. The obtained differential metabolites were more typical and have research value, which might provide new ideas and references for screening differential metabolites in metabolomics.

In this study, information on the O sinensis stroma, sclerotia, and mycelium was obtained based on UPLC-MS of the untargeted metabolic profile, and the 1697 metabolites were measured by a reliable quantitative array. The metabolites included amino acids, flavonoids, alkaloids, and fatty acids. Combining LDA and PLS-DA, the 48 metabolites were found to be potential biomarkers of O sinensis, which have great drug potential. In the comparison and cluster analysis of the relative abundance of the 48 differential metabolites between groups, the differences among the 3 were visually displayed. OSBS was most pronounced, implying that changes in cultural methods may play an important role in the significant differences. Moreover, there were significant differences among the 3 in terms of composition and relative abundance. In the analysis of the relative abundance of differential metabolites, some important differential metabolites, such as oleamide, gamma-aminobutyric acid, DL-tryptophan, L-arginine, A-l-fucopyranose, and hydroxycarteolol, were found in high relative abundance. Oleamide has a vasoconstrictor effect and can have an impact on metabolic pathways of fatty acids in the treatment of diabetic nephropathy.^[26] Among the broad variety of functions, the most well-known is its sleep-inducing effect.^[27]

In future pharmacological research, these metabolites should be studied to provide a theoretical basis for further exploring their pharmacological effects. The metabolic pathway enrichment analysis of these important differential metabolites showed that they were mainly enriched in 21 metabolic pathways and played an important role in the in vivo metabolism of O sinensis. The composition of O sinensis determines its function and the content determined its efficacy.^[28] In future studies, we plan to focus more on elucidating the mechanisms of these metabolites, especially their pharmacological effects. Indeed, our team found that O sinensis can effectively prevent acute lung injury induced by lipopolysaccharide in mice.^[29] In the study, mice were divided into normal group, model group, positive control group and O sinensis group. By observing the pathological section of lung tissue and transmission electron microscope, we can closely identify the structural differences caused by damage among each group. Hematoxylin and eosin staining results showed that compared with the normal group, the model group had alveolar collapse, and compared with the model group, the inflammatory cell infiltration in the alveolar cavity of the O sinensis group was significantly reduced. The transmission electron microscopy results showed significant edema of type II alveolar cells in the model group. The status of type II alveolar cells in the O sinensis group and the positive group was similar to that in the normal group, which reflected the positive effect of O sinensis on pulmonary inflammation. Studies have confirmed that O sinensis has good therapeutic effects on a variety of respiratory diseases.^[30,31] It can reduce the infiltration of inflammatory cells, inhibit the release of inflammatory factors, and alleviate inflammatory reactions.

Seasonal influenza virus infection is the most common cause of death from pneumonia annually.^[11] In addition to direct inhibition of pathogens, TCM treatment of infectious diseases can also improve the body natural immunity and specific immune function. Modern pharmacological studies have confirmed that O sinensis has a significant beneficial anti-inflammatory effect and improved immunity. The newly determined fatty acids isolated from O sinensis by Nanshan Zhong and others have been shown to have a significant inhibitory effect on influenza virus. To maximize the medicinal value of O sinensis, we used the UPLC-MS metabolomics method to discover other biological activity markers from O sinensis, before combining network pharmacology and molecular docking technology to explore the potential significance of O sinensis in the treatment of influenza. Network pharmacology demonstrated that d-(−)-glutamine, asp-leu, caffeine, 4-(2-aminopropyl) − 2-methoxyphenol, and 2,3-dihydroxypropyl hexadecanoatez played key roles in the treatment of influenza. Supplementation with d-(−)-glutamine could regulate the salivary cytokine profile of physically active elderly subjects and increase the total specific secretory immunoglobulin A (SlgA) level of the influenza virus vaccine.^[32] Caffeine has significant pharmacological effects, including anti-inflammatory, antioxidant, anti-apoptotic, diuretic, and antifibrotic effects, as well as the ability to regulate angiogenesis. Caffeine is not only used for the treatment of apnea in prematurity but also for the prevention of bronchopulmonary dysplasia.^[33] Moreover, caffeine has been shown to effectively ameliorate lung damage and inhibit viral replication, showing similar efficacy to oseltamivir and ribavirin.^[34] We speculate that the anti influenza virus activity of O sinensis is related to these important compounds.

According to the PPI network, 82 key target proteins were obtained, and the top 10 key proteins were SRC, RHOA, EGFA, HSP90AA1, VEGFA, MAPK1, ITGB1, JUN, PRKCA, and ITGA1. As an actin cytoskeleton regulator, RHOA plays an important role in cell biological activities, such as membrane trafficking, cytokinesis, cell-to-cell adhesion, and hanmucin-mediated cell-to-cell adhesion. SRC plays key roles in immune responses, cell adhesion, cell cycle progression, and transformation regulation, and is involved in antigen-antibody, cytokine, and integrin-mediated transmembrane signaling. HSP90α, encoded by HSP90AA1, can repair the damage induced by stress. KEGG enrichment analysis showed that O sinensis had anti-influenza effects, which were mediated through PI3K-AKT, HIF-1, IL-7, influenza A, COVID-19, and other pathways. The PI3K/Akt signaling pathway plays a key role in cell proliferation, apoptosis, cell cycle, inflammatory response, and other processes, and mediates the effects of various immune cytokines, which can activate downstream genes after external pathological stimuli.^[35] SRC can also transduce cytoplasmic signals to the nucleus through 3 signaling pathways, including PI3K/AKT, MAPK, and JAK/STAT. The PIK3CA protein in the PI3K/AKT pathway can lead to abnormal enhancement of the catalytic activity of PI3Ks, producing second messenger inositol-like substances to abnormally activate AKT, which in turn affect the apoptosis process. Combined with the results of GO analysis showing that the positive regulation of cell migration, positive regulation of response to external stimulus, regulation of cell adhesion, cellular response to chemical stress, and response to lipopolysaccharide play key roles in anti-influenza activity, suggesting that O sinensis has a role in cell stress and cell migration. GO and KEGG pathway analyses showed that O.sinensis had anti-influenza effects, which were mediated through multiple targets and pathways. The results of molecular docking showed that the important metabolites of O sinensis had a strong binding ability to key targets. This result implies that PI3K-Akt, HIF-1, IL-7, influenza A, COVID-19, and other key pathways were regulated and controlled by the important active ingredients in O sinensis, via the key targets of SRC, RHOA, EGFA, HSP90AA1, VEGFA, MAPK1, ITGB1, JUN, PRKCA, and ITGA1 to promote anti-influenza activity.

5. Conclusions

Metabolomic analysis highlighted significant differences in the metabolites of the stroma, sclerotium, and mycelium of O sinensis, as well as the important differential metabolites in all 3, which may represent active substances. Indeed, oleamide, gamma-aminobutyric acid, DL-tryptophan, L-arginine, A-l-fucopyranose, and hydroxycarteolol are important additions to pharmacological activity studies of O sinensis and provide important references for its clinical use. The use of network pharmacology and molecular docking techniques to predict the mechanism of action of O sinensis anti-influenza activity provides direction and basis for more in-depth in vivo and in vitro studies.

Acknowledgments

We thank BGI Co., Ltd. for the excellent experiments.

Author contributions

Conceptualization: Jinna Zhou, Mu Wang, Hong Yu.

Data curation: Jinna Zhou, Xiaorong Zhou, Hong Yu.

Formal analysis: Jinna Zhou, Xiaorong Zhou, Yao Wang.

Investigation: Yao Wang.

Methodology: Jinhu Wang, Yao Wang, Mu Wang.

Resources: Tao Sun, Run Luo.

Software: Tao Sun, Jinhu Wang, Run Luo.

Writing – original draft: Ran Zhang.