Integrated Phytochemical Analysis Based on UPLC–MS/MS and Network Pharmacology Approaches to Explore the Effect of Odontites vulgaris Moench on Rheumatoid Arthritis

Odontites vulgaris Moench has the effect of clearing away heat, detoxification, dispelling wind, and clearing dampness. In this study, the potential anti-inflammatory compounds of O. vulgaris were investigated using ultra-high-performance liquid chromatography–mass spectrometry (UPLC–MS/MS) combined with the network pharmacology approach and further confirmed on an LPS-activated RAW 264.7 macrophage model. Monomer compounds were prepared from the active fraction using modern advanced separation and purification methods. UPLC–Q-Exactive HRMS was used to identify the chemical compounds in the active fractions of O. vulgaris. D-mannitol, geniposidic acid, salidroside, shanzhiside methyl ester, eleutheroside B, geniposide, 7,8-dihydroxycoumarin, gardoside methyl ester, arenarioside, vanillic acid, p-hydroxy-cinnamic acid, melampyroside, syringaresinol, tricin, and diosmetin were isolated from O. vulgaris for the first time. A compound database of O. vulgaris was established based on the existing literature to predict the mechanism of O. vulgaris in the treatment of rheumatoid arthritis. The results suggest that the PI3K-Akt pathway mediates O. vulgaris and deserves more attention in the treatment of RA. Finally, the anti-rheumatoid arthritis effects of the four target compounds were validated with the decreased levels of NO, TNF-α, IL-6 and IL-1β in RAW 264.7 macrophage cells treated with LPS. The present study explored the potential targets and signaling pathways of O. vulgaris in the treatment of RA, which may help to illustrate the mechanisms involved in the action of O. vulgaris and may provide a better understanding of the relationship between O. vulgaris and RA. This study provides novel insights into the development of new drugs and utilization of Mongolian traditional Chinese medicine resources.


INTRODUCTION
Rheumatoid arthritis (RA) is a chronic autoimmune disease characterized by persistent synovial hyperplasia, inflammatory infiltration, pannus formation, and bone erosion, which may lead to joint deformity, disability, and death (Li et al., 2019a;Li et al., 2019b;Guo et al., 2020). Thus far, pain relief and reduction of inflammation have been considered the main therapeutic strategies for RA. With a high disability rate, RA poses a serious threat to human health and is a prevalent, globally distributed incurable diseases. Without timely treatment, the incidence of disability in patients with RA is reported to be approximately 70% (Li and Gu, 2020). Nonsteroidal anti-inflammatory drugs are commonly used for the treatment of such patients. However, long-term use of these drugs can cause considerable side effects; for example, gastrointestinal mucosal injury and bleeding occur more frequently in patients with long-term use of these drugs than in those with gastrointestinal diseases (Vetal et al., 2013;Li et al., 2018). Biological blockers that suppress inflammation, such as that mediated by anti-tumor necrosis factor-α, may lead to a higher risk of infection owing to their role in weakening the immune system (Wang et al., 2020a). The interaction of a drug with an unintended target may be a source of adverse drug reactions and side effects. Therefore, there is an urgent need to find safe and effective drugs for the treatment of RA with fewer side effects than those currently used. In recent years, owing to the low level of ethnopharmacological adverse reactions, Ethnopharmacological Mongolian medicine has attracted attention as an alternative treatment for inflammation (Wang et al., 2020b). These Mongolian medicinal treatments exert anti-inflammatory effects through the NF-κB, MAPK, PI3K-Akt, arachidonic acid metabolism, inflammasome, and other pathways (Mao et al., 2019).
In Mongolian medicine, RA is often called "Xieri Wusu" or "Tuolai disease"  and is a "Taolai" disease (Bai et al., 2015). "Basaga," with its name derived from the transliteration of a Sanskrit word, is often used in Mongolian medicine with remarkable curative effects, such as cooling blood, reducing itching, detoxification, and diuresis (Zhao et al., 2017). According to traditional Chinese medicine, RA is a disease caused by wind, cold, dampness, heat, and other evil gases that close and obstruct meridians and collateralization, leading to pain in the limbs, bones, and joints caused by poor flexion and extension, stiffness, and deformation (Liu et al., 2015). Odontites vulgaris Moench ( Figure 1) is a commonly used Mongolian medicine variety of "Basaga." It helps in clearing away heat, detoxification, dispelling wind, and clearing dampness. The primary compounds present in O. vulgaris are phenylethanol glycosides, iridoids, and flavonoids. Some compounds have been shown to have certain anti-inflammatory effects (Gong et al., 2020;Nigro et al., 2020;Wang et al., 2020d;Cicek et al., 2021;Tian et al., 2021). Through interviews with the local populace and hospital personnel in Inner Mongolia, we found that O. vulgaris is often used to treat RA; therefore, we conducted an in-depth study on the effects of O. vulgaris on RA. A flowchart of this process is shown in Figure 2.
At present, multi-compound, multi-target therapy has been reported to be more effective and less toxic than the use of traditional single-target drugs (Zhang et al., 2013a). Network pharmacology is a new, promising, and low-cost drug development method based on bioinformatics, systems biology, and multi-pharmacology. For Mongolian traditional Chinese medicines with complex compositions and unclear mechanisms of action, especially medicine varieties with weak foundations in research, network pharmacology not only serves as an effective tool for studying the mechanism of action, but also provides a basis for designing in vivo and in vitro experiments (Zhang et al., 2013a;Zheng et al., 2014;Zhou et al., 2016;Jiang et al., 2019).
To the best of our knowledge, this study is the first to use UPLC-MS/MS technology to study the chemical compounds produced by O. vulgaris. We isolated 17 compounds, of which 15 compounds were newly reported. We used network pharmacology combined with computer assisted techniquesmolecular docking to predict the target and action pathways of O. vulgaris in RA treatment and detected its target compounds from the level of inflammatory factors in vitro. This study explored potential targets and signaling pathways for the treatment of RA with O. vulgaris, which may help to illustrate the mechanisms of action involved in RA. Moreover, our results can promote the development and utilization of Mongolian traditional Chinese medicine resources.

Plant Materials
O. vulgaris was collected from the Xilinguole League of Inner Mongolia Autonomous Region in 2018 and identified as Odontites vulgaris Moench by a specialist. The specimen (specimen number: 152500180629025LY) was kept in the herbarium of Baotou Medical College.

Extraction and Processing
Whole dried O. vulgaris (6.04 kg) powder was successively extracted with 70% ethanol; approximately 1.2 kg of the extract was obtained. The extract was dissolved in water and  Logarithmic phase RAW 264.7 cells seeded in 96-well plates at a density of 1 × 10 5 cells/ml were cultured at 37°C in a 5% CO 2 incubator for 24 h. After adherence of the cells to the plates, a series of different concentrations (25,50,100,200, and 400 μg/ ml) of the O. vulgaris extracts (water, n-butanol, ethyl acetate, and petroleum ether) was added. The control group (without the extract) was set at the same time. Five parallel wells were used in each group. After 48 h of culture, 10 μl of the freshly prepared 5 mg/ml MTT solution was added to each well, and the culture was incubated at 37°C for 4 h in the dark. The supernatant was discarded, and 150 µl of dimethyl sulfoxide (DMSO) was added to each well. After oscillation for 10 min, the absorbance (OD 570 nm) of each well was measured using a microplate analyzer (Thermo Scientific Multiskan FC, Thermo Fisher Scientific).

Determination of Indomethacin Concentration
Logarithmic phase RAW 264.7 cells seeded in 24-well plates at a density of 5 × 10 5 cells/ml were placed in a 37°C 5% CO 2 incubator for 24 h. After adherence to the plate, the cells were randomly divided into the control group, model group (LPS), and positive drug group (indomethacin group). Each group was placed in parallel holes. The cells were cultured with LPS (1 μg/ml) for 4 h, with the positive drug group being administered indomethacin solutions of different concentrations to co-interact with LPS cells for 20 h at the following concentrations: 6.25, 12.5, 25, 50, and 100 µM. The cell supernatant of each group was collected after culture, and NO release in the cell supernatant was detected according to the manufacturer's instructions. The concentration of NO was determined using sodium nitrite as the control substance, and the 50% inhibition rate (i.e., the IC 50 value) was calculated according to the concentration of NO. Absorbance at 550 nm was determined using a microplate analyzer.

Cell Viability Assay and Protective Effects of the Extracts on LPS-Treated RAW 264.7 Cells
Single-cell suspension RAW 264.7 cells were seeded at a density of 5 × 10 5 cells/ml in 24-well plates and placed at 37°C in a 5% CO 2 incubator for 24 h until the logarithmic phase was reached. The group demarcation based on treatment in this study was as follows: control group with normal cell culture for 24 h without any treatment; inflammation model group with cells stimulated using 1 μg/ml LPS for 24 h; negative control group with cells treated using different extracts (alcohol, ethyl acetate, n-butanol, water, and petroleum ether fractions) of O. vulgaris at 400 μg/ml for 20 h; and the experimental group with cells stimulated with 1 μg/ml LPS for 4 h and different concentrations (12.5, 25, 50, 100, and 200 μg/ml) of O. vulgaris extracts added to the cells for 20 h. The cell supernatant of each group was collected after culture, and NO release in the cell supernatant was detected according to the Griess test described in the NO kit. Absorbance was measured at 550 nm (Gong et al., 2020).

Ultra-high-performance Liquid Chromatography-Mass Spectrometry (UPLC-MS/MS) System
The UPLC high-resolution MS Dionex Thermo Scientific UItimate 3,000 system connected with the Thermo Q Exactive focus machine was used in this study. For the analysis, the extracts Frontiers in Pharmacology | www.frontiersin.org August 2021 | Volume 12 | Article 707687 of n-butanol and ethyl acetate were redissolved (1 mg/ml) in methanol and the sample was filtered through a 0.22 µm microporous membrane to obtain the test solution, and 10 µl of the filtered solution was injected into the instrument.

Data Processing
After the sample solution was analyzed and detected according to the above mass spectrum conditions, the Qualbrowser feature of the Xcalibur 3.0 software was used to complete data acquisition and processing, including the extraction of the ion chromatogram and fragmentation behavior information. The mass error of each main fragment ion was within 5 ppm. Full-scan mode highresolution mass spectrometry data were analyzed using the Mass Frontier 7.0 software, and the possible structure formulae were deduced. The actual molecular weights of the detected compounds were then matched with the theoretical values. The Discoverer 2.1 (including mzCloud, Chinese medicine compound database, Chemspider, and Sci finder) database was used to search for the compounds, match the first and second mass spectra of the compounds, analyze the pyrolysis law, and identify the compounds.

Prediction of O. vulgaris Compound Information
The compound database was established through literature retrieval and compound identification in Chinese and English databases (CNKI, Wanfang, PubMed). Compound information was collated in the PubMed database (Hao et al., 2013). The PubChem database constantly adds new data, and the SMILES or chemical structure formulae can be uploaded to the Swiss target prediction database (Daina et al., 2019).

Prediction of O. vulgaris Targets for RA Treatment
Using "rheumatoid arthritis" as the keyword, RA-related targets were identified in the OMIM, GeneCards, and DisgeNet databases. The TBtools software was used to match the targets related to the chemical compounds of O. vulgaris with the targets of RA. Additionally, a Venn diagram was drawn to obtain the potential therapeutic targets of the compounds of O. vulgaris for RA.

Construction and Analysis of the Protein-Protein Interaction Networks of the Target Proteins
To study the interaction between the target proteins in the treatment of RA by O. vulgaris, a PPI network was constructed using the String Version 11.0 platform. We set a score greater than 0.7 to indicate high confidence (Szklarczyk et al., 2017;Deng et al., 2019;Zhou et al., 2019), set the species as Homo sapiens (human), and hid the nodes that were disconnected from the network, leaving the rest of the parameters as default . The network analyzer function in the Cytoscape 3.7.1 software was used to study the PPI network topology attributes and draw the PPI network diagram.

Enrichment Analysis of the Target Using Genome Ontology and the Kyoto Encyclopedia of Genes and Genomes
The KEGG pathway and GO enrichment analyses of 198 targets of O. vulgaris in the Clusterprofiler software package were carried out using R Version 4.0.2, and the results were visualized.

Molecular Docking
To further investigate the possibility of interaction between the four selected compounds (verbascoside, isoacteoside, melampyroside, and tricin) and the ten key targets (AKT1, MAPK1, EGFR, HRAS, APP, HSP90AA1, PIK3CA, SRC, PIK3R1, and AR), the structure of the small molecule compounds downloaded from PubChem (https://pubchem.ncbi.nlm.nih.gov/) was saved in SDF format. The Chem 3D software was used to convert the SDF file into a * Mol2 file, and the 3D structure of the target protein (PDB format) was downloaded from the PDB database (https://www.rcsb.org/) (Burley et al., 2017). The PyMol software was used to dehydrate the proteolytic enzyme and remove the active center of the original ligand. At the same time, hydrogenation, charge adding, and nonpolar hydrogen merging were carried out using the AutoDock Tools 1.5.6 software. We used AutoDock Vina 1.1.2 for docking operations (Tanchuk et al., 2016). The receptor-ligand pairs were sorted and screened according to the binding free energy [Affinity (kcal/mol)]. Binding energy less than 0 indicated that the compound and protein could spontaneously bind and interact with each other. The lower the energy, the more stable the molecular conformation. Generally, a binding energy of ≤5.0 kcal/mol indicated a good binding effect. The PyMol 2.3.2 software was used for visualization.

Effects of the Target Compounds of O. vulgaris on Cell Viability
Logarithmic phase RAW 264.7 cells distributed as single-cells seeded in 96-well plates at a density of 1 × 10 5 cells/ml were placed in a 37°C 5% CO 2 incubator for 24 h. After the cells adhered to the plate, different concentrations (6.25, 12.5, 25, 50, 100, and 200 μM) of the target compounds of O. vulgaris (isoacteoside, verbascoside, tricin, and melampyroside) were added to the culture. The normal control group (without the extract) was established at the same time. Five parallel wells were used for each group. After 48 h of culture, 10 μl of the freshly prepared 5 mg/ml MTT solution was added to each well and the culture was incubated at 37°C for 4 h in the dark. The supernatant Frontiers in Pharmacology | www.frontiersin.org August 2021 | Volume 12 | Article 707687 was discarded, and 150 µl of DMSO was added to each well. After oscillation for 10 min, the absorbance (OD 570 nm) of each well was measured using a microplate analyzer.

Cell Viability Assay and the Protective Effects of Target Compounds of O. vulgaris on LPS-Treated RAW 264.7 Cells
We performed this analysis using the methods described in Chemical Materials and Extraction and Processing. The group demarcation based on treatment in this study was as follows: control group normal cell culture for 24 h without any treatment; inflammation model group with cells stimulated using 1 μg/ml LPS for 24 h; negative control group with cells treated using different target compounds (isoacteoside, verbascoside, tricin, and melampyroside) of O. vulgaris at 100 μM for 20 h; experimental group with cells stimulated using 1 μg/ml LPS for 4 h and different concentrations (6.25, 12.5, 25, 50, and 100 μM) of different target compounds (isoacteoside, verbascoside, tricin, and melampyroside) of O. vulgaris for 20 h. Three parallel wells were set in each group. The cell supernatant of each group was collected after culture, and NO release in the cell supernatant was detected according to the Griess test described in the NO kit. Then, the absorbance was measured at 550 nm (Gong et al., 2020).

Enzyme-Linked Immunosorbent Assay
Logarithmic phase RAW 264.7 cells made into single-cell suspension cells under the scratch seeded in a 6-well plate at a density of 5 × 10 5 cells/ml were placed at 37°C in a 5% CO 2 incubator for 24 h. The group demarcation based on treatment in this study was as follows: control group with normal culture without any treatment; inflammation model group with cells stimulated with 1 μg/ml LPS for 24 h; negative control group with cells treated with different target compounds (isoacteoside, verbascoside, tricin, and melampyroside) of O. vulgaris at 100 μM for 20 h; experimental group with cells stimulated using 1 μg/ml LPS for 4 h and different concentrations (25, 50, 100 μM) with different target compounds (isoacteoside, verbascoside, tricin, and melampyroside) of O. vulgaris for 20 h. Three parallel wells were set in each group. The supernatant of each group was collected, centrifuged for 5 min, and the supernatant was carefully removed for storage. The expression levels of TNF-α, IL-6, and IL-1β were detected using an ELISA kit.

Statistical Analysis
All data were analyzed using SPSS 25.0, and the results were expressed as mean ± standard deviation (SD) of the three independent experiments. Differences were considered statistically significant if p < 0.05.

Determination of Indomethacin Concentration
The results showed that indomethacin could significantly inhibit the content of NO (p < 0.0001), and the inhibitory effect was proportional to the concentration of indomethacin (i.e., the higher the concentration, the stronger the inhibitory effect).
The IC 50 value of indomethacin was 58.82 μM in linear regression analysis. Therefore, 60 μM indomethacin was selected as the positive drug concentration for subsequent experiments (Figure 3).

Effects of the Crude Extracts of O. vulgaris on Cell Viability
The effects of different polar fractions (alcohol, ethyl acetate, petroleum ether, water, and n-butanol) on the proliferation activity of mouse macrophage RAW 264.7 cells were detected. There were no significant differences in cell proliferation between the control group and the RAW 264.7 cells subjected to the different polar fractions of O. vulgaris (25, 50, 100, and 200 μg/ml) ( Figure 4A; p > 0.05). When the concentration of the different polar fractions of O. vulgaris was greater than 200 μg/ ml, the viability of the ethyl acetate-and n-butanol extracttreated cells decreased significantly (p < 0.01). The experimental results showed that the different polar fractions of O. vulgaris had no effect on cell proliferation and cytotoxicity at 25, 50, 100, and 200 μg/ml concentrations; thus, they were selected for subsequent experiments.

Cell Viability Assay and Protective Effects of the Extracts on LPS-Treated RAW 264.7 Cells
Analysis of the protective effects of the different crude extracts of O. vulgaris on LPS-induced inflammation in RAW 264.7 cells showed that the release of NO in the LPS model group was significantly higher than that in the control group (p < 0.0001).
Compared with the LPS model group, the ethyl acetate ( Figure  4B-b) and n-butanol fractions ( Figure 4B-c) at concentrations of 25, 50, 100, and 200 μg/ml significantly inhibited the release of NO (p < 0.05) in a concentration-dependent manner. This indicated that the ethyl acetate and n-butanol fractions could improve the inflammatory response in LPS-induced mouse macrophage RAW 264.7 cells in the experimental FIGURE 3 | Effect of indomethacin on NO production in RAW 264.7 induced by LPS. Data are presented as means ± SD, n 5. ** means p < 0.01, *** means p < 0.001, **** means p < 0.0001 compared with LPS group. △△△ means p < 0.001 compared with control group.
Frontiers in Pharmacology | www.frontiersin.org August 2021 | Volume 12 | Article 707687 6 concentration range. IC 50 values were calculated according to the concentrations of the ethyl acetate and n-butanol fractions of O. vulgaris. The n-butanol fraction (135.9 μg/ml) showed a higher anti-inflammatory effect than the ethyl acetate fraction (166.8 μg/ ml). The alcohol, water, and petroleum ether fractions d,e) had an inhibitory effect on NO secretion by RAW 264.7 cells stimulated with LPS at 100 and 200 μg/ml concentrations that showed a statistically significant difference to that on NO secretion in the LPS model group (p < 0.05). There was no statistically significant inhibitory effect at other concentrations (p > 0.05).

UPLC-MS Analysis
Using UPLC-MS, 35 compounds were identified from the n-butanol and ethyl acetate fractions, of which 16 were identified by the reference substance. The chemical compounds of the n-butanol and ethyl acetate fractions were preliminarily determined and were mainly divided into four categories: phenylethanoid glycosides (verbascoside, isoacteoside, arenarioside, salidroside), flavonoids (kaempferol, cynaroside, luteolin, luteolin-7-O-glucuronide, apigenin, apigenin-7-O-glucoside, tricin, chrysoeriol, quercetin, diosmetin, hydroxygenkwanin), iridoids (melampyroside, shanzhiside methyl ester, aucubin, geniposide, geniposidic acid, shanzhiside methyl ester), phenolic acids (caffeic acid, trans-cinnamic acid, 2-hydroxycinnamic acid, salicylic acid, ferulic acid, 7,8-dihydroxycoumarin), and others (adenosine, syringaresinol, D-Mannitol, esculetin). The total ion chromatograms of the ethyl acetate and n-butanol extracts of O. vulgaris, both in the positive ion mode, are presented in Figure 5B. The retention time, chemical formula, measured mass, and theoretical mass of the 35 identified compounds are presented in Table 1. The characteristics of the representative compounds in the ethyl acetate and n-butanol fractions of the O. vulgaris were as follows: given its molecular ion peak of m/z 449.1458 in negative ion mode, the molecular formula of peak 22 (melampyroside) was determined to be C 22 H 26 O 10 . Accordingly, a second-order spectrum was found, and fragment ion peaks at m/z 213.1622, 173.0118, and 121.0283 were recorded. Caffeic acid (peak 8) was detected in the ethyl acetate fraction of the herb, and its fragment ion peaks were at m/z 135.0441 and 107.0350. Peak 20 of n-butanol showed a molecular ion peak at m/z 623.1974, which was identified as that of verbascoside. There were fragment ions peaks at m/z 313.8787, 161.0232, and 133.0282 in the secondary spectrum, which was consistent with the fragmentation mode of the reference substance and that in     previous reports. The molecular ion peak at m/z 623.1971 of peak 21, which was speculated to be that of the isomer of verbascoside, also showed a similar fragmentation mode. On comparison with the fragmentation mode of the reference substance, peak 21 was identified as that of isoacteoside. Peak 27 at m/z 301.0355 in the ethyl acetate fraction was identified as that of quercetin. The fragment ions with peaks at m/z 151.0029, 121.0282, 107.0122, and 65.0017 were found in the secondary spectrum, which was consistent with the fragmentation mode of the reference substance.

Chemical Composition of O. vulgaris
The known and predicted targets of the 36 compounds, after eliminating duplicates, were examined ( Table 2). In this study, we only included results with probability >0.1 as the prediction target, and there were 294 targets in total.

Targets Relevant to O. vulgaris Treatment of RA
A total of 5,387 RA-related genes were compared with the target genes of 36 compounds to determine the potential therapeutic effect of each compound on RA target genes. The potential processing capacity of O. vulgaris to affect RA target genes was comprehensively obtained, and a total of 198 target genes were reported ( Figure 6A).

Construction and Analysis of the PPI Network of the Target Protein
A PPI network of the candidate target protein of O. vulgaris was constructed. A visual PPI network diagram was generated to evaluate the topology parameters of the network nodes. A total of 198 nodes and 552 edges were obtained from these interactions. Nodes that were larger and darker in blue were those with higher degrees of freedom ( Figure 6C). In PPI networks, nodes with a higher degree may play a more important role; thus, the top 30 proteins here may be key molecular targets for the treatment of RA. Further screening showed that 28 compounds were linked to the top 30 proteins, of which 14 were flavonoids, two were iridoids, three were phenylethanolglycosides, six were phenolic acids, and three were other compounds (Table 3). Akt1, MAPK1, EGFR, and HRAS were found to be associated with 51, 48, 42, and 40 proteins, respectively ( Figure 6B; Table 4).

GO Analysis of Target Proteins
To verify the biological characteristics of 198 targets of O. vulgaris, GO enrichment analysis was performed on the presumed targets to clarify the related biological processes (Figure 7; p < 0.01). The top 10 significantly enriched terms in the BP (Biological Process), MF (Molecular Function), and CC (Cellular compound) categories were selected. Among them, BP (2,175 records), MF (167 records), and CC (92 records) accounted for 89.36, 6.86, and 3.78%, respectively. In the BP category, the target proteins were associated with cellular response to chemical stress (GO:0062197), response to oxidative stress (GO:0006979), and rhythmic processes (GO:0048511). In the MF category, the target proteins were associated with protein tyrosine kinase activity (GO:0004713), insulin-like growth factor In the CC category, the target proteins were associated with membrane rafts (GO:0045121), membrane microdomains (GO: 0098857), and vesicle lumen (GO:0031983).

KEGG Classification of Target Proteins
To determine the potential pathway underlying the effects of O. vulgaris on RA treatment, KEGG pathway enrichment analysis was performed on 198 genes. Figure 7 shows the 20 most significant signaling pathways. According to these results, the most important pathways associated with the effects of O. vulgaris on RA treatment were the PI3K-Akt, HIF-1, VEGF, estrogen, MAPK, Fc epsilon RI, Rap1, FoxO, and Prolactir signaling pathways, suggesting that these pathways mediated the effect of O. vulgaris on RA.

Construction of Compound-Target and Target-Pathway Network
Based on the target prediction results and KEGG analysis results of the chemical compounds of O. vulgaris, the Cytoscape 3.7.1 software was used to construct the network of the "compound-target" and "target and pathway" of O. vulgaris in the treatment of RA for visual analysis. In this network, the orange nodes represented compounds, blue and pink nodes represented targets, green nodes represented pathways, and red and purple nodes represented O. vulgaris and RA, respectively. The potential target of a compound was linked by an edge ( Figure 8A). In this study, 30 core genes and their corresponding 36 compounds were used to establish an "compound-target" map, and 20 pathways from KEGG enrichment results and their corresponding targets were used to establish a "target-pathway" map. Figure 8B shows the PI3K-Akt signaling pathway. The enrichment results showed that 17 common targets, namely, IGF1R, KDR, AKT1, PRKCA, PIK3R1, INSR, IL2, PTK2, CDK6, CDK2, GSK3B, FLT3, PIK3CG, SYK, KIT, MET, PKN1, and EGFR, were enriched in the PI3K-Akt signaling pathway.

Compound Analysis
Verbascoside, isoacteoside, melampyroside, and tricin were isolated from the ethyl acetate and n-butanol fractions and were representative of the effective compounds. According to the area normalization method, the relative peak areas of verbascoside, isoacteoside, tricin, and melampyroside in the n-butanol and ethyl acetate fractions of O. vulgaris were up to 47.97% and 21.89%, respectively. Tricin is a flavonoid that presents anti-inflammatory activity. It acts mainly by inhibiting cyclooxygenase (COX), TNF and prostaglandin (PG), IL, NO, and other inflammatory mediators. It has potential medicinal value for the treatment of RA (Zhan et al., 2018;Yin et al., 2020). Verbascoside, and isoacteoside are phenylethanolglycosides that have been shown to have significant anti-inflammatory activity and potential value in the development of new drugs for RA . Melampyroside is an iridoid compound, and its relative peak areas in the ethyl acetate and n-butanol extracts reached 11.39% and 13.31%, respectively. Iridoids have been studied for their anti-inflammatory activity (Gui and Zheng, 2018). Taken together, the four selected compounds were representative and formed a basis for the verification experiment.

Molecular Docking
We selected the top 10 targets (AKT1, MAPK1, EGFR, HRAS, APP, HSP90AA1, PIK3CA, SRC, PIK3R1, and AR) in the PPI network and four compounds (verbascoside, isoacteoside, melampyroside, and tricin) for analysis of molecular docking. The docking result is depicted as a heat map display in Figure 9A. The docking visualization results are shown in Figure 9B. The results show that the binding energy of each compound to protein was less than −5 kcal/mol, indicating that each compound could bind well to protein. Of them, the binding energies of isoacteoside and tricin to SRC were −9.2 and −8.6 kcal/mol, respectively. The binding energies of verbascoside and melampyroside were −9.6 and −8.5 kcal/mol, respectively.

Anti-Inflammatory Assay
Protective Effects of Target Compounds on LPS-Treated RAW 264.7 Cells The effects of different target compounds (melampyroside, isoacteoside, verbascoside, and tricin) on the proliferation activity of mouse macrophage RAW 264.7 cells were examined. The results showed that there was no significant difference between the proliferation of control cells and that of RAW 264.7 cells upon treatment with different target compounds of O. vulgaris at 6.25, 12.5, 25, 50, and 100 μM (p > 0.05). When the concentrations of melampyroside and tricin were more than 100 μM, cell viability decreased significantly (p < 0.0001). The different target compounds of O. vulgaris had no effect on cell proliferation and cytotoxicity at 25, 50, and 100 μM concentrations; therefore, the 25, 50, and 100 μM concentrations were selected for subsequent experiments ( Figure 10A).

Cell Viability Assay and the Protective Effects of Target
Compounds on LPS-Treated RAW 264.7 Cells NO release in the LPS model group was significantly higher than that in the normal control group ( Figure 9B; p < 0.0001). The  effect of different target compounds on NO release was not significant in the negative control group, and there was no significant difference compared with that in the normal control group (p > 0.05). Compared with that in the LPS model group, treatment with 6.25, 12.5, 25, 50, and 100 μM of verbascoside ( Figure  vulgaris. The IC 50 values were as follows: melampyroside, 129.7 μM; verbascoside, 129.9 μM; isoacteoside, 114.5 μM; and tricin, 101.6. Thus, the anti-inflammatory effect of the four target compounds was in following order: tricin > isoacteoside > verbascoside > melampyroside.

Enzyme-Linked Immunosorbent Assay
Compared with those in the control group, the levels of TNF-α, IL-6, and IL-1β in the model group were significantly increased ( Figure 11; p < 0.0001), indicating that LPS caused an inflammatory response in RAW 264.7 cells. Compared with that in the model group, TNF-α content decreased significantly after treatment with different concentrations (25, 50, and 100 μM) of melampyroside, verbascoside, isoacteoside, and different concentrations (50 and 100 μM) of tricin. A concentration dependence was observed in the effective concentration analysis (p < 0.05). Compared with that in the model group, IL-1β content decreased significantly after treatment with different concentrations (25, 50, and 100 μM) of verbascoside, and different concentrations (50 and 100 μM) of tricin, isocalycoside, and melampyroside. A certain concentration dependence was observed (p < 0.05). Compared with that in the model group, IL-6 content decreased significantly after treatment with different concentrations (25, 50, and 100 μM) of melampyroside, verbascoside, and tricin, and different concentrations (50 and 100 μM) of isoacteoside. A concentration dependence was observed (p < 0.05). The results showed that melampyroside, verbascoside, isoacteoside, and tricin could downregulate the levels of inflammatory factors and exert anti-inflammatory effects. There was no significant difference seen between the verbascoside treated groups and the positive control group (indomethacin) in IL-1β content at 100 μM (p > 0.05). There was no significant difference in IL-6 content between the verbascoside, and tricin groups and the positive control group (indomethacin) at 100 μM (p > 0.05).

DISCUSSION
Based on the symptoms of RA and the pharmacodynamic basis of O. vulgaris, although O. vulgaris has enormous therapeutic potential, its full chemical characterization remains incomplete. Therefore, the purpose of this study was to evaluate the biological activity of O. vulgaris and analyze its chemical compounds to lay a foundation for the identification of new drugs. In this study, anti-inflammatory activity was determined in terms of the NO content of RAW 264.7 mouse macrophages induced by LPS. The results showed that the ethyl acetate and n-butanol extracts exhibited better activity. We isolated 17 compounds from the n-butanol and ethyl acetate extracts of O. vulgaris with compounds 1-9 and 12-17 being isolated from O. vulgaris for the first time. The UPLC-Q-Exactive assay was used to identify 35 compounds from the ethyl acetate and n-butanol extracts of O. vulgaris.
Network pharmacological results showed that 36 compounds, including flavonoids, iridoids, phenolic acids, and phenylethanolglycosides, were directly related to the therapeutic effect of O. vulgaris on RA. In this study, the PI3K-Akt pathway was found to be particularly important for the treatment of RA by O. vulgaris. The PI3K-Akt signaling pathway is closely related to the occurrence and development of RA and plays an important role in the regulation of differentiation and formation of osteoblasts . Overproliferation of synovial cells and RA can be reversed via the use of P13K-Akt signaling inhibitors, or by the upregulation of endogenous negative regulatory proteins in the signaling pathway. This inhibits glycogen synthase kinase 3 (GSK3) activity and accelerates cell apoptosis. In RA, the earliest phenomenon is the formation of new blood vessels, which can develop into a pannus in severe cases. Vascular endothelial growth factors (VEGFs) are an important class of signaling proteins and ligands secreted by cells to stimulate vasculogenesis and angiogenesis (Da and Li, 2012). They are also as important inducers and direct pro-inflammatory factors. VEGF proteins play a vital role in activating angiogenesis in RA through the upregulation of the VEGF pathway (Yi et al., 2016;Meloa et al., 2020). Cheng et al. (2011) used the immunohistochemical SP method to detect VEGF protein expression in the synovial tissues of the knee joints of 27 patients with active RA. The results showed that the expression of VEGF in the synovial tissues of controls and patients with osteoarthritis was significantly lower than that in patients with RA.
According to the component-target interaction analysis, ten targets with higher degrees on the PPI network (AKT1, MAPK1, EGFR, HRAS, APP, HSP90AA1, PIK3CA, SRC, PIK3R1, and AR) were identified. The docking scores of all compound-target pairs were less than −5 kcal/mol, indicating that each of the four Frontiers in Pharmacology | www.frontiersin.org August 2021 | Volume 12 | Article 707687 20 selected compounds (verbascoside, isoacteoside, melampyroside, and tricin) had good binding affinity to all 10 targets. The binding energies of isoacteoside, tricin to SRC were −9.2, −9.9, and −8.6 kcal/mol, respectively. The binding energies of verbascoside and melampyroside were −9.6 and −8.5 kcal/mol, respectively. The molecular docking simulation strengthened all the above investigations to some extent.
As a major pro-inflammatory factor in the pathogenesis of RA, TNF-α is considered the leading trigger for inflammation and joint destruction. It can stimulate the synovium, fibroblasts, osteoclasts, and other cells to produce destructive substances called matrix metalloproteinases, which can damage cartilage and gradually destroy bone tissue in patients with RA. Therefore, TNF-α level can be used as an indicator of inflammatory activity in RA . IL-6 is produced by synovial cells, and its abnormal expression and dysregulation are typical characteristics of RA [39]. IL-6 can enhance the effects of IL-1 and TNF-α and is considered to be an amplifying factor of some biological effects of TNF-α (Shi et al., 2013). At non-toxic doses, the four target compounds in O. vulgaris significantly reduced the secretion of TNF-α, IL-1β, IL-6, and NO, suggesting that O. vulgaris is a potential therapeutic agent for RA. The anti-inflammatory activity of O. vulgaris can be attributed to its bioactive compounds. Phenylethanolglycosides, flavonoids, and iridoids have a wide range of pharmacological activities, including anti-inflammatory activityies (Hanfer et al., 2017;Spagnuolo et al., 2018;Zhang et al., 2018). Verbascoside has shown anti-inflammatory activities, such as reducing the presence of pro-inflammatory factors in intestinal inflammation (Danae et al., 2013). As an isomeride of verbascoside, the similarities and differences in the antiinflammatory activity of isoacteoside and the underlying mechanism of action require further exploration. Melampyroside is an iridoid terpene compound, and the Data are presented as means ± SD, n 5. * means p < 0.05, ** means p < 0.01, *** means p < 0.001, **** means p < 0.0001 compared with LPS group. #### means p < 0.0001 compared with LPS group. △△△△ means p < 0.0001 compared with control group.
Frontiers in Pharmacology | www.frontiersin.org August 2021 | Volume 12 | Article 707687 relative peak area of the ethyl acetate and n-butanol fractions of O. vulgaris is relatively large. Therefore, pharmacological studies on O. vulgaris cannot be ignored. Tricin is a flavonoid, and its anti-inflammatory properties have been studied (Zhan et al., 2018;Yin et al., 2020); it is noteworthy that tricin has shown outstanding anti-inflammatory activity in in vitro experiments. Based on studies of previous chemical substances, we used bioinformatics to predict the targets and potential action pathways of O. vulgaris in the treatment of RA and detected its target compounds from the levels of inflammatory factors in vitro. The results of network analysis and prediction showed that the PI3K-Akt pathway may play an important role in O. vulgaris-mediated treatment of RA. Therefore, the development of anti-inflammatory drugs using O. vulgaris has considerable advantages. However, bioinformatics methods still have some limitations, for example, it is found that it is too one-sided to explore the efficacy of medicinal plants only from the perspective of network pharmacology. It is necessary to consider the relative content of compounds in medicinal plants, so as to give better play to the pharmacological role of medicinal plants. In addition, the binding energy of compounds in molecular docking experiments is not necessarily at the same level as their therapeutic potential in vivo, and further systematic molecular biology experiments are needed to verify the accurate mechanisms.
In summary, the results of this study support the potential anti-inflammatory activity of O. vulgaris, and combined with computer pharmacology, it is speculated that it may have the activity of treating RA. In the development of Chinese and Mongolian medicine, O. vulgaris could be a potential source of novel drugs for the treatment of RA. Our study also provides a novel paradigm to identify the potential mechanisms of pharmacological effects derived from a natural compound.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.

AUTHOR CONTRIBUTIONS
ML conceived the research ideas. XW advised on data collection. MJ and CW designed the experiments in detail and provided valuable guidance on data analysis. ML and TY provided important insights and suggestions for this research. XM aided in data consolidation. MJ and CW performed the majority of the data processing and wrote the manuscript. All authors significantly contributed to the manuscript and have read and approved the final manuscript.