Elucidating the material basis and potential mechanisms of Ershiwuwei Lvxue Pill acting on rheumatoid arthritis by UPLC-Q-TOF/MS and network pharmacology

Ershiwuwei Lvxue Pill (ELP, མགྲིན་མཚལ་ཉེར་ལྔ།), a traditional Tibetan medicine preparation, has been used hundreds of years for the clinical treatment of rheumatoid arthritis (RA) in the highland region of Tibet, China. Nevertheless, its chemical composition and therapeutic mechanism are unclear. This study aimed to uncover the potentially effective components of ELP and the pharmacological mechanisms against RA by combing UPLC-Q-TOF/MS and network pharmacology. In this study, 96 compounds of ELP were identified or tentatively characterized based on UPLC-Q-TOF/MS analysis. Then, a total of 22 potential bioactive compounds were screened by TCMSP with oral bioavailability and drug-likeness. Preliminarily, 10 crucial targets may be associated with RA through protein-protein interaction network analysis. The functional enrichment analysis indicated that ELP exerted anti-RA effects probably by synergistically regulating many biological pathways, such as PI3K-Akt, Cytokine-cytokine receptor interaction, JAK-STAT, MAPK, TNF, and Toll-like receptor signaling pathway. In addition, good molecular docking scores were highlighted between five promising bioactive compounds (ellagic acid, quercetin, kaempferol, galangin, coptisine) and five core targets (PTGS2, STAT3, VEGFA, MAPK3, TNF). Overall, ELP can exert its anti-RA activity via multicomponent, multitarget, and multichannel mechanisms of action. However, further studies are needed to validate the biological processes and effect pathways of ELP.


Sample preparation
Accurately-weighed ELP sample (0.5 g) was placed in a conical flask for ultrasound-assisted extraction with 70% methanol (30 mL) at 25˚C for 30 min. The extracted solution was adjusted to the original weight by adding 70% methanol, and then the extracted solution was centrifuged at 14 000 rpm for 5 min. The supernatant was filtered through a 0.22 μm microporous membrane filter prior to injection into the UPLC-Q-TOF/MS system. Standard stock solutions of the reference standards were prepared by dissolving appropriate amounts of the pure substances in methanol.
Xevo G2-XS Q-TOF (Waters, Manchester, UK) equipped with an electrospray ionization ion (ESI) source was applied for mass spectrometry data's acquisition in negative and positive ionization mode ranging from m/z 100 to m/z 1200. The parameters of the source were set as follows: the ion source temperature was set at 120˚C, cone voltage 25 V, capillary voltage 2.5 kV, collision energy 10 V, desolvation temperature 350˚C, and desolvation gas flow rate, 1000 L/h.

Acquisition of active compounds and their targets
All compounds obtained by UPLC-Q-TOF/MS analysis were retrieved based on the absorption, distribution, metabolism, and excretion (ADME) parameters in the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP, http://lsp.nwu. edu.cn/tcmsp.php), the following screening criteria were used: oral bioavailability (OB) �30%, drug-likeness (DL) �0.18 [20]. Then, the selected bioactive compounds were converted to a standard Canonical Simplified molecular input line entry system (SMILES) format in Pub-Chem database (https://pubchem.ncbi.nlm.nih.gov/), at the same time, SMILES format file was imported to Swiss Target Prediction database with properties for "Homo sapiens", and eventually ELP potential targets were built in the bioactive compounds information database [21].

Protein-Protein Interaction Data
Protein-Protein Interaction (PPI) Data were extracted using STRING (https://string-db.org/) with a reasonable confidence range for PPI data scores (low: <0.4, medium: 0.4 to 0.7, and high: >0.7) [25]. Common targets of bioactive compounds associated with RA were inputted into the STRING database, with the species to "Homo sapiens," and a confidence score higher than 0.7. The visualization of the PPI networks was conducted in Cytoscape 3.8.0 software (http://www.cytoscape.org/) [26].

Functional enrichment analysis
Based on potential common targets for this study, GeneOntology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses were conducted to predict the action mechanism of ELP in treating RA by using the Database for Annotation, Visualization, and Integrated Discovery (DAVID v6.8, https://david.ncifcrf.gov/) [27].

Compound-Target-Pathway (C-T-P) network
Based on the results in the DAVID database, the C-T-P network was created in Cytoscape 3.8.0 software. In a complex associative network, nodes and edges represent compounds/entities and their direct interactions, respectively. A high degree value equates to a prominent node status.

Molecular docking simulation
The molecular docking simulation was performed on the selected targets and corresponding compounds by using Maestro version 11.5 from the Schrodinger software suite. The lowest/ minimum energy conformation was used for molecular docking via the default parameters. The docking score is the negative logarithm of the experimental dissociation/inhibition constant (pKd/pKi) and usually ranges from 0 to 10 (weak to strong combination force) [28]. The human protein structures with the highest resolution related to RA were selected from the UniProt database (https://www.uniprot.org/). Meanwhile, the X-ray crystal structures of these proteins were obtained from the RCSB PDB database (https://www.rcsb.org/). which contain or lose -C 7 H 5 O 5 (m/z 169), -C 7 H 4 O 4 (m/z 152), -C 6 H 5 O 3 (m/z 125) ion fragments in the MS/MS spectrum. Ellagitannins, a hydrolyzable tannin class of polyphenols, hydrolyzed with acids or bases to produce hexahydroxy diphenolic acid (HHDP), which were further spontaneously esterified into ellagic acid [29]. ESI-MS analysis of flavonoids exhibited some diagnostic features such as loss of methyl, methoxyl, H 2 O (18 Da), CO (28 Da), CO 2 (44 Da), C 2 H 2 O (42 Da) [30]. Meanwhile, for the identification of flavonoid aglycones, the retro-Diels-Alder (RDA) reaction provides structurally relevant information on the number and types of substituents in the A-and B-rings of the two C-C bonds in Cleavage of the C-ring [31] [32].

Network pharmacology analysis
Potential bioactive compounds and targets of ELP in the treatment of RA. After ADME screening, 22 potential bioactive compounds (OB � 30%, DL � 0.18) in ELP were identified (Tables 2 and S2). Firstly, 145 potential targets of the 22 potential bioactive compounds were obtained from the Swiss Target Prediction platform (Figs 2 and 3) (S2 Table). Secondly, 361 RA-related targets were retrieved from GeneCards, CTD, and OMIM databases (S3 Table). Finally, 46 targets directly and indirectly associated with RA were obtained by precisely matching the potential targets of the above two steps through the online tool Venny 2.1 (http://bioinfogp.cnb.csic.es/tools/venny/) (Fig 4).
Protein-Protein Interaction (PPI) network. The PPI relationship of 46 target genes was obtained by the STRING tool, and the visualization was realized by Cytoscape 3.8.0 software. The network of PPI relationships contained 46 nodes and 563 edges when a combined score of > 0.4 was used (Fig 5 and S4 Table). The 10 target genes with the highest connectivity degree were selected as the hub genes for RA. Thus, the hub genes, which might play a crucial role in RA progression, were IL6, TNF, TP53, AKT1, JUN, VEGFA, MAPK3, STAT3, IL1B, and PTGS2.
Gene Ontology (GO) function and KEGG pathway enrichment analysis. GO function and KEGG pathway enrichment analysis of the 46 candidate target genes were conducted by DAVID v6.8 to explore the molecular mechanism of ELP in treating RA. GO evaluations were illustrated using biological process (BP), cell component (CC), and molecular function (MF) terms (Fig 6A-6C). A total of 79 enrichment results in the related items of BP, involving regulation of apoptosis, protein amino acid phosphorylation, defense response, and intracellular signaling cascade; 20 enrichment results are related to CC, which includes a membraneenclosed lumen, organelle lumen, cytosol, and cell fraction; 39 enrichment processes are related to the MF which cover the nucleoside binding, ATP binding, protein kinase activity, cytokine activity and so on. Enrichment results of each p-value were calculated (p < 0.01 was considered as significant enrichment). Subsequently, a total of 29 signaling pathways were obtained (p < 0.05), and 12 signaling pathways related to RA were identified (Table 3 and Fig  6D). Meanwhile, the top 6 signaling pathways closely associated with RA were screened according to the number of targets. Hence, ELP probably produces anti-RA effects by synergistically regulating many biological pathways, such as PI3K-Akt signaling pathway, Cytokinecytokine receptor interaction, JAK-STAT signaling pathway, MAPK signaling pathway, TNF signaling pathway, and Toll-like receptor signaling pathway, and so on. Network construction and analysis. A C-T-P Network of ELP for treating RA was constructed using Cytoscape 3.8.0 software (Fig 7). The network map showed that the relationships among 22 potential bioactive compounds, 46 protein targets, and 12 signaling pathways. With the excavation of the C-T-P network, the mechanism of ELP in treating RA was preliminarily understood.

Molecular docking simulation
Molecular docking simulation was performed using Maestro11.5 (Schrodinger Suite) between the 22 potential bioactive compounds and 10 key targets. The three-dimensional (3D) structures of the 10 selected targets were obtained from the PDB database (https://www.rcsb.org/), which is an archive that includes experimentally determined atomic-level 3D structures of biological macromolecules (DNA, RNA, and proteins) [33]. The docking scores were depicted in   Table. The hydrogen bonding and π-π stacking were involved between the targets and the potential compounds. Finally, according to the heat map analysis, good molecular docking scores were highlighted between five promising bioactive compounds (ellagic acid, quercetin, kaempferol, galangin, coptisine) and five core targets (PTGS2, STAT3, VEGFA, MAPK3, TNF). The typical schematic representation of 3D and 2D molecular docking patterns of target proteins and compounds are shown in Fig 8B, including PTGS2 with coptisine, STAT3 with ellagic acid, MAPK3 with quercetin, and VEGFA with kaempferol. For instance, the binding mode of coptisine in the active site of PTGS2 has been represented in its 3D and 2D modes. Coptisine showed an H-bond interaction and three π-π stacking. The oxygen of coptisine forms a hydrogen bond with ASN382, and three π-π stacking were formed by binding the six-membered ring to HIE388 and HIS207, respectively.

Discussion
As a common systemic inflammatory autoimmune disease, with high disability and incidence, RA can severely impair physical function and quality of life [34]. Recently, discoveries have improved the understanding of rheumatoid inflammation and its consequences [35]. The pathophysiology of RA involves chronic inflammation of the synovial membrane, which can destroy articular cartilage and juxta-articular bone [36]. RA is characterized by infiltration of the synovial membrane in joints with T cells, B cells, and monocytes, and the common symptoms are musculoskeletal pain, swelling, and stiffness in clinical practice [37]. As reported that the innate immune response produces some cytokines, among those, IL-1, IL-6, and TNF-α played important parts in the progress of RA [38]. In addition, the process of RA involves the activation of multiple inflammatory signaling pathways and interactions with inflammatory cytokines [39]. Nevertheless, given the uncertainty and complexity of its pathogenesis, there's no specific medicine that can effectively treat RA. Modern therapies such as nonsteroidal antiinflammatory drugs or pain medications only improve symptoms but do not prevent damage progression and irreversible disability [37]. The mainly therapeutic strategy is applicated disease-modifying antirheumatic drugs for patients with RA at present. For example, methotrexate, one of the therapeutic drugs, is the most effective and commonly used first-line drug [40]. Unsatisfactory, it has large side effects. Consequently, it is urgent to discover and develop a safe and effective therapeutic. The P-value of each biological process was less than 0.01. In the bar plots, each bar represents a GO term on the vertical axis (Fig 6A-6C). The number of genes enriched in each term is represented on the vertical axis. The color of each bar represents each GO term. Similarly, in the bubble graphs, each bubble represents a KEGG path on the vertical axis (Fig 6D). The number of the genes is represented on the horizontal axis. The size of each bubble indicates the number of genes enriched in each KEGG pathway. The larger the bubble, the greater is the number of genes involved in the pathway. The color of each bubble represents the adjusted P-value for each KEGG path. The redder the bubble, the smaller the adjusted P-value is. https://doi.org/10.1371/journal.pone.0262469.g006

PLOS ONE
ELP, a traditional Tibetan patented prescription medicine, is one of the commonly used drugs to cure RA in clinical trials [2]. More importantly, according to the Pharmacopoeia of the People's Republic of China (2015), only Pterocephalus hookeri (C.B.Clarke) Hoeck has minor toxicity in ELP, and the clinical dosage is 1-3 g/d. In contrast, according to the proportion of 25 herbs in ELP, the dosage of Pterocephalus hookeri (C.B.Clarke)Hoeck is only 0.11 g/ d, which is a safe dose. Therefore, ELP can be considered nontoxic. Hence, for the first time, a system pharmacological approach using UPLC-Q-TOF/MS and network pharmacology with molecular docking simulation was applied in this study to the prediction of promising bioactive compounds and mechanisms for the treatment of RA. In this work, it is found that ELP probably performed anti-RA effects via synergistically regulating many biological pathways, important six such as PI3K-Akt signaling pathway, Cytokine-cytokine receptor interaction, JAK-STAT signaling pathway, MAPK signaling pathway, TNF signaling pathway, and Tolllike receptor signaling pathway. Concurrently, good molecular docking scores were highlighted between five promising bioactive compounds (ellagic acid, quercetin, kaempferol, galangin, coptisine) and five core targets (PTGS2, STAT3, VEGFA, MAPK3, TNF).
Previous studies have suggested that the PI3K/Akt signaling pathway contributes to excessive cell proliferation, migration, and invasion in RA fibroblast-like synoviocytes (RA-FLSs) [41,42]. Excessive proliferation of FLSs is one of the critical features of RA, leading to cartilage and bone destruction [43]. Bone marrow MSCs, which play a key role in the healing of bone defects, have been applied for the treatment of RA via activation of the PI3K/AKT signaling pathway [44]. Consequently, according to the results of network analysis, one of the ELP mechanisms in anti-RA effects may be associated with regulating the PI3K/AKT signaling pathway. Cytokine-cytokine receptor interaction plays a vital role in both innate and adaptive inflammatory host defenses and development and repair processes aimed at the restoration of homeostasis, which is involved in the pathogenesis of inflammatory and autoimmune diseases including RA [45]. Notably, there are implicated in plenty of studies also suggesting cytokinecytokine interactions involved in the pathogenesis of RA [46]. In addition, large postinjury increases in RA-associated markers and differential upregulation of the cytokine-cytokine receptor interaction pathway that is closely associated with inflammation [47]. JAK-STAT signaling pathway is a key player in RA progression [48]. In previous studies, the levels of serum cytokines TNF-α and IL-1β are increased in RA patients. Meanwhile, activation of the JAK2-STAT3 pathway is regulated by inflammatory cytokine stimulation during the progression of RA [49,50]. Accumulating studies indicated that the MAPK signal transduction pathway can regulate the inflammatory cytokine and downstream cell transduction pathways, thereby affecting the inflammation and destruction of joints [51]. The MAPK family includes p38MAPK, extracellular signal-regulated kinase (ERK), and c-Jun N-terminal kinase (JNK), which were activated in the synovium of RA patients [52,53]. It was reported that CCR5 silencing suppresses the inflammatory response, inhibits viability, and promotes apoptosis of . It is demonstrated that functional suppression of RGS1 inhibits the inflammatory response and angiogenesis by inactivating the TLR signaling pathway in rats with CIA [58]. These investigations improved the prediction of ELP against RA via inflammation associated with this pathway. However, it is necessary to verify them through further experimental research. Literature for understanding the development of inflammatory arthritis revealed that LL-37 and IL17A can significantly enhance PTGS2 and TNF gene expression, then release its downstream pro-inflammatory cytokines, PGE2 and TNF, contributing to the enhancement of the pathogenesis mechanisms of inflammatory arthritis [59]. Meanwhile, PTGS2, IL1β, IL6, TNFA, and CCL20 have been shown that mediators transformed RA-FLS to be a major source Genes in the TNF family have been associated with RA and may be a potential therapeutic target [67]. Plenty of shreds of evidence illustrated that TNF perpetuates synovial inflammation via activating RA-FLS inducing a constellation of genes [55]. TNF is a key upstream target of the TNF signaling pathway, and the combination of the core compounds in ELP with TNF can block the RA immune response induced by TNF signaling Pathway (S3D Fig). According to the relationship between the above targets and signal pathways, it can be known that ELP exerts anti-RA effects through multiple components, multiple targets, and multiple pathways at the molecular level. Those provide strong evidence for the prediction of ELP to treat RA via intervention between these key targets, but it is necessary to further validate the claims using molecular biological methods.
It is necessary to clarify the activity ingredients of ELP revealing the scientific connotation of its on RA. Molecular docking analysis showed that five promising bioactive compounds, ellagic acid, quercetin, kaempferol, galangin, and coptisine, docking own stable. The above compounds mainly belong to polyphenols, flavonoids, and alkaloids, and a large investigation explicated that all these compounds are efficient in RA [68][69][70]. In addition, the contents of five bioactive compounds were studied. By comparing the UPLC-Q-TOF-MS total ion chromatogram of ELP, it was found that ellagic acid, Quercetin, Kaempferol, Galangin and Coptisine were the compounds with high content in 96 identified components. In particular, Ellagic acid, Quercetin, galangin. Specifically, Ellagic acid is a phenolic acid compound, which comes from Phyllanthus emblica, Terminalia billerica and Terminalia chebula in ELP and is the main compound of the above three herbs [71][72][73]. Quercetin, Kaempferol and Galangin are flavonoids, which widely exist in nature. The above three bioactive ingredients are contained in a large amount in the herbal medicine of ELP, such as Cassia obtusifolia, Gossampinus malabarica, Abelmoschus manihot, Pterocephalus hookeri, Gentiana manshurica and Fraxinus rhynchophylla [74]. Coptisine is an alkaloid component and an important compound of Adhatoda vasica. Previous studies have shown that ellagic acid alleviated the adjuvant-induced arthritis model in mice by modulation of pro-and anti-inflammatory cytokines (IL-1β, TNF-α, IL-17, IL-10, and IFN-γ) [75]. Quercetin could diminish myeloperoxidase activity and ROS levels to aid the control of autoimmune inflammation in patients with RA [5]. Kaempferol inhibits the migration and invasion of fibroblast-like synovial cells in RA by blocking the activation of the MAPK pathway [76]. Experiments showed that galangin improved human RA FLS by inhibition of the NF-κB/NLRP3 pathway activation [77]. Coptisine has been reported to possess anti-inflammatory activity that significantly inhibited the IL-1β-induced NF-kB activation in human RA chondrocytes [78]. All the above shreds of evidence proved that the prediction of bioactive compounds in ELP anti-RA is reasonable and reliable.

Conclusions
In this study, it was the first time that an integrative strategy based on UPLC-Q-TOF/MS coupled with the UNIFI informatics platform was applied for chemical profile analysis of ELP. A sum of 96 compounds was identified or tentatively characterized from 70% methanol extraction of ELP by comparing retention times, mass spectra with authentic standards, fragmentation behaviors, and data previously reported, including tannins, flavonoids, penylpropanoids, terpenoids, quinonoids, steroids, alkaloids, and other compounds. Furthermore, based on the traditional Tibetan clinical efficacy and identified compounds of ELP in the treatment of RA, a C-T-P network was constructed by relying on the network pharmacology method. In addition, good molecular docking scores were highlighted between 5 promising bioactive compounds (ellagic acid, quercetin, kaempferol, galangin, coptisine) and 5 core targets (PTGS2, STAT3, VEGFA, MAPK3, TNF). These targets were further related to the key signaling pathways such as PI3K/Akt, JAK-STAT, MAPK, TNF, Cytokine-cytokine receptor interaction, and Toll-like receptor, which could explain the anti-RA effects of ELP.
In summary, a system pharmacological approach, using UPLC-Q-TOF/MS and network pharmacology with molecular docking simulation, was applied to the prediction of promising bioactive compounds in ELP and mechanisms for the treatment of RA, which provides a new idea for the research of other Tibetan medicine prescriptions. Nevertheless, the bioactive compounds, biological targets, and signaling pathways predicted the need to be confirmed and validated using the CIA rat model in further studies.