Metabolomics Combined with Network Pharmacology to Explore the Active Components and Mechanism of Tetrastigma planicaule (Hook.) Gagnep. in the Treatment of Rheumatoid Arthritis

Abstract

Context. Rheumatoid arthritis (RA) is a chronic autoimmune inflammatory disease. It is characterized by persistent joint damage. Traditional Chinese medicine (TCM) has demonstrated notable efficacy in managing RA. Prior investigations have indicated that the ethyl acetate extract of Tetrastigma planicaule (Hook.) Gagnep. (TP) possesses substantial anti-inflammatory properties, suggesting the potential for screening TCM drugs with antirheumatoid arthritis attributes. However, the precise mechanism underlying its pharmacological effects and material basis remains unclear, impeding the advancement of TCM innovation. Objective. This study is to elucidate the active components and mechanism of TP in the treatment of RA. Materials and Methods. A rat model of collagen-induced arthritis (CIA) was established for conducting pharmacological experiments to assess the effectiveness of ethyl acetate extract from TP in treating RA. In addition, nuclear magnetic resonance (NMR) metabolomics technology was employed to identify potential endogenous biomarkers for further metabolic pathway analysis. The active compounds and key targets were investigated using ultraperformance liquid chromatography-quadrupole time-of-flight mass spectrometry (UPLC-Q-TOF-MS/MS) technology in combination with network pharmacology. Moreover, common gene targets discovered by metabolomics and network pharmacology were validated by molecular docking with the relevant active components. Result. 12 endogenous biomarkers were screened by 1H-NMR metabolomics technology. The metabolite pathways are primarily implicated in lipid metabolism and amino acid metabolism. 49 compounds were identified in TP, of which 10 were considered active ingredients through network pharmacological analysis. In a comprehensive analysis, it was found that TP exhibited a strong association with the PLA2G4A and lipid metabolite pathways in RA. Molecular docking studies further demonstrated a high affinity between the active compounds of TP and PLA2G4A. Discussion and Conclusions. TP may play a therapeutic role in RA by regulating PLA2G4A and participating in back-regulating the glycerophospholipid metabolic pathway. The study elucidated the multicomponent, multitarget, and multipathway mechanism of TP in the treatment of RA, laying the groundwork for a deeper understanding of its therapeutic mechanisms.

This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

1. Introduction

Rheumatoid arthritis is an autoimmune inflammatory disease. It is characterized not only by swelling and pain in the joints, but in severe cases, it can lead to disability and affect multiple organs throughout the body, greatly affecting the physical and mental health of the patient [1]. According to statistical data, 18 million people worldwide are living with rheumatoid arthritis. About 70% of people living with rheumatoid arthritis are women, and 55% are older than 55 years [2]. Due to the challenges it poses for affected individuals, RA is increasingly becoming one of the globally recognized public health issues. Currently, treating RA in clinical practice mainly relies on drugs, such as glucocorticoids and nonsteroidal anti-inflammatory drugs (NSAIDs). However, these medications often come with significant adverse effects, such as gastrointestinal damage, leukopenia, and thrombocytopenia [3]. TCM has been used for the treatment of RA for thousands of years [4–6]. Because of its stable therapeutic efficacy, lower side effects, and high safety, TCM has become an excellent choice for the development of drugs to treat RA [7–9].

Tetrastigma planicaule (Hook.f.) Gagnep. (TP) is a woody plant that can be mainly discovered in the western and eastern parts of China. At the same time, it is also a relatively common medicinal herb for the Zhuang people in China. It is known for its anti-inflammatory and antitumor effects and is often used to treat diseases, such as RA [10]. The constituents of the ethyl acetate fraction from TP possess anti-inflammatory activity and exhibit notable anti-inflammatory effects [11]. However, there is a lack of research on its active compounds and the underlying mechanisms by which they exert their pharmacological effects.

Metabolomics, or metabolomic profiling, is a technique that allows for the quantitative and characterization analysis of all low-molecular-weight metabolites present in a particular organism or cell. By utilizing metabolomics, it becomes possible to analyze and compare abnormal changes in the composition and levels of metabolites, thereby revealing the final response of the organism to perturbations [12, 13]. This technology provides a powerful tool for integrating traditional Chinese medicine with modern medicine, facilitating the advancement of traditional Chinese medicine toward modernization. Currently, commonly used metabolomics techniques include gas chromatography-mass spectrometry (GC-MS), liquid chromatography-mass spectrometry (LC-MS), and nuclear magnetic resonance (NMR) spectroscopy, each with its advantages and limitations. NMR metabolomics [14] has good objectivity and reproducibility and a large dynamic range for the simultaneous detection of multiple metabolites and is used for targeted analysis and absolute quantification of major metabolites. Network pharmacology is a systematic analysis method that explores the interactions among biological systems, compounds, and diseases at the network level, providing a holistic perspective to elucidate the occurrence of disease [15]. Network pharmacology is used to obtain “compound-target-disease” network information to show the relationship between the therapeutic targets and active compounds [16]. In the field of TCM, network pharmacology is widely utilized to uncover the mechanisms of TCM in treating diseases. Molecular docking is used to evaluate the binding activity of a receptor to an active molecule.

In this study, we constructed a metabolite-response-enzyme-gene network through which to combine metabolomics and network pharmacology. It can reveal in depth the complex connections between metabolites and multiple targets in TCM [17–19]. Finally, molecular docking was utilized to further validate the key components of the screen with the key targets to assess the binding potential. In conclusion, we elucidated the mechanism of TP’s antirheumatoid arthritis effects from a holistic perspective, providing support for the development of innovative TCM for the treatment of RA.

2. Material and Methods

2.1. Regents and Material

The TP was purchased from Guangxi Xianzhu Traditional Chinese Medicine Technology Co., Ltd. (batch No: 20190801). They were identified by Vice Chief Pharmacist Mali Fei from Guangxi Yixin Pharmaceutical Company (Guangxi, China) as the dried stems of Tetrastigma planicaule (Hook.f.) Gagnep. The HPLC-grade acetonitrile was purchased from Merck Group in Germany (Darmstadt, German). Methanol, 95% ethanol, and analytical grade ethyl acetate were purchased from Chengdu Kolon Chemical Co., Ltd. (Chengdu, China). The complete Freund’s adjuvant was purchased from SIGMA Co., Ltd. (Missouri, USA); bovine type II collagen was purchased from Chondrex Co., Ltd. (Parker-Hannifin, USA). Methotrexate tablets were purchased from Shanghai Xinyi Co., Ltd. (Shanghai, China). Hydroxypropyl methylcellulose sodium was purchased from Shanghai Kailin Chemical Co., Ltd. (Shanghai, China).

2.2. Preparation of TP

An appropriate amount of powdered TP was taken and underwent an 8-fold reflux extraction using 70% ethanol until a color change was observed. Upon completion, the extraction was halted, and the extracted liquids were merged. Subsequently, the combined extract was introduced into a rotary evaporator for vacuum concentration. Then, it was extracted three times using equal volumes of ethyl acetate and each extract was collected to again concentrate until dryness was attained.

2.3. Animal Experiment

The study was approved by the Ethics Committee of Guangxi University of Chinese Medicine of China (DW20211216-206). The 24 male Sprague Dawley (SD) rats of specific pathogen-free (SPF) grade, weighing 180–220 g, were purchased from the Animal Center of Guangxi Medical University (certificate number: SCXK Yue 2016-0041, Guangxi, China). All rats were housed in a standard environment (55 ± 5% relative humidity, 12/12 h light/dark cycle, 24 ± 2°C) and had free access to standard rat chow and water. After adaptive feeding, they were randomly divided into the following four groups (n = 6 per group): the control group, the CIA model group, the methotrexate group, and the TP group. At the first immunization, all rats were injected with 2 μl of a mixture of emulsified bovine type II collagen and complete Fuchs’ adjuvant at the right hind paw toe and tail root, except for the control group, which was injected with an equal amount of saline. After Day 7, a second booster immunization was administered to establish the CIA rat model in the same manner as the first injection at 100 microliters per rat. On the 1st day of modeling, the treatment groups were treated with TP (20 g/kg), the methotrexate group was treated with methotrexate (1.05 mg/kg), and the model and control groups had the same volume of 0.1% CMC-Na for 22 days.

At the end of the experiment, 24 rats were anesthetized by intraperitoneal injection (0.2 ml/100 g) using 3% pentobarbital sodium, and blood was collected from the abdominal aorta. The collected blood was collected in 5 ml of anticoagulant-free vacuum blood collection tubes and allowed to stand for 30 minutes, and then centrifuged for 15 min at 4000 rpm at 4°C, and the upper serum layer was collected in EP tubes and stored in a −80°C refrigerator for further metabolomics analysis.

2.4. Evaluation of the CIA Model and TP Efficacy

During the experiment, the body weight of 24 rats was measured on days 0, 4, 8, 12, 16, 20, 24, and 28. The swelling thickness of the right hind toe and arthritis index (AI) scores of all rats were measured before secondary immunization and every 7 days after secondary immunization. The standards for AI scoring are as follows: normal = 0; erythema and mild swelling = 1; erythema and moderate swelling extending from the ankle to the midfoot = 2; erythema and moderate swelling from the ankle extending to the metatarsophalangeal joints = 3; and erythema and severe swelling encompassing the knee, foot, and entire paw = 4. Scores were performed independently by three experimenters, and the mean value was used as the AI score.

2.5. Metabolomics Analysis

2.5.1. Preparation of Serum Samples

100 μl of buffer solution (0.2 mol/l Na₂HPO₄ and 0.2 mol/l NaH₂PO₄, pH 7.4) and 150 μl of D₂O were added to serum samples (300 μl), which were vortexed to homogeneity and centrifuged for 5 mins at 4000 rpm at 4°C. Next, 500 μl of the supernatant that had passed through the 0.22 μm microporous filter membrane was transferred to a 5 mm NMR tube and stored in a 4° refrigerator.

2.5.2. Data Acquisition

NMR spectra data were obtained on Bruker AVANCE 500 MHz NMR instruments (Bruker, Karlsruhe, Germany). The data were obtained using a Carr-Purcell-Meiboom-Gill (CPMG) pulse sequence with parameters of 10 kHz spectral width, 32 k sampling points, 128 accumulations, 3 s relaxation time, and 100 ms spin-echo time. Lactate served as a chemical shift reference (δ1.336). The baseline of the plots was aligned manually, and the chemical shifts of the plots were calibrated using MestReNova (Mestrelab Research, Spain) software. The spectral regions of each sample were differentiated at equal intervals of 0.02 units, and the residual water peaks (δ 4.80 to δ 5.06) were removed to perform segmental integration of the plots. The data were exported as text files for further statistical analysis.

2.5.3. Data Analysis

Principal component analysis (PCA) and orthogonal partial least squares discriminant analysis (OPLS-DA) were performed by SIMCA-P 14.1 software (Umetrics, Sweden). Compounds that met variable importance of projection (VIP) >1 and $P<0.05$ from the OPLS-DA model were considered potential biomarkers with significant differences [19]. Then, MetaboAnalyst 5.0 was employed for pathway analysis to insight into the mechanisms.

2.6. UPLC-Q-TOF-MS/MS Combined with Network Pharmacology Analysis

2.6.1. TP UPLC-Q-TOF-MS/MS Study

The samples were prepared as follows: ethyl acetate extract of TP was taken in appropriate amounts, weighed accurately, added methanol to ultrasonic (550 W, 53 kHz) for 5 min, and transferred to a 2 mL measuring flask to be fixed volume and shaking, through 0.22 μm microporous filtration membrane. Standards were weighed precisely, dissolved in methanol, and volume-dispersed into a 10 ml volumetric flask to obtain a mixed standard solution.

The chemical components were identified by a Sciex X500R Q-TOF liquid mass spectrometer (Sciex, USA) equipped with an ESI source and Sciex OS data processing software. The chromatographic column was an Accucore C18 (100 × 2.1 mm, 2.6 μm; Thermo Fisher Scientific, USA). The mobile phase consisted of acetonitrile (A) and 0.1% aqueous formic acid (B). The chromatographic separation conditions of the serum were as follows: 0–1 min, A: 5%, B: 95%; 1–5 min, A: 5–27%, B: 95–73%; 5–15 min, A: 27–70%, B: 73–30%; 15–18 min, A: 70–95%, B: 30–5%; 18–20 min, A: 95–5%, B: 5–95%. The sample injection volume was 3 μl, the column temperature was maintained at 40°C, and the flow rate was 0.4 mL/min. The data were simultaneously collected in positive and negative ion modes (electrospray ionization ESI⁺ and ESI⁻ modes, respectively) with a scan range (m/z) of 100–2000. The optimized parameters were as follows: gas temperature, 600°C; nebulizer gas pressure, 55 psi; ion spray voltage floating, 5.5 kV; and collision energy, 35 V.

2.6.2. Prediction of Potential Targets

Compounds identified by UPLC-Q-TOF-MS/MS were imported into the PubChem database (https://pubchem.ncbi.nlm.nih.gov) for obtaining 3D structural information. Then, the compounds were screened for active ingredients through the SwissADME database (http://www.swissadme.ch) by using the criteria of not satisfying at least three conditions of Lipinski’s rule [20, 21] and GI absorption of “High” as the screening criteria. Its structure information was also uploaded to the SwissTargetPrediction platform (http://www.swisstargetprediction.ch) to screen the targets of the active ingredients of TP. Finally, the GeneCards database (https://www.genecards.org) was searched for disease targets with the keyword “Rheumatoid arthritis,” and Venn diagrams were utilized to search for potential anti-RA targets.

2.6.3. Enrichment Analysis and Network Construction

We imported potential targets into the STRING (http://string-db.org/) database for constructing a protein-protein interaction (PPI) network and screened the core targets in the network with the cytoHubb plug-in. Meanwhile, we uploaded the potential targets into the DAVID database (https://david.ncifcrf.gov/) and selected “Homo sapiens $P<0.05$” for further pathway enrichment analysis. Furthermore, the “compounds-targets-pathways” network was constructed by Cytoscape 3.6.1 (http://cytoscape.org/) [22].

2.6.4. Molecular Docking

First, the protein structures were obtained from the PDB (https://www.rcsb.org/) database and the 3D chemical structures of the compounds were obtained from the Chem3D software. Before molecular docking, proteins need to be removed from water molecules and polar hydrogen atoms by PyMOL version 2.5 software (https://pymol.org/). Next, the docking grid box was constructed by AutoDock version 1.5.6 (California, USA) at the active site of each target [23]. Then, AutoDock was utilized for further molecular docking to calculate the binding energy between the proteins and the compounds. Finally, the docking results were visualized by PyMOL.

2.6.5. Statistical Analysis

Data were presented as mean ± standard deviation and analyzed by the SPSS version 22.0 statistical software (IBM Corp., Armonk, NY, USA). Multisample data were compared by one-way ANOVA, with $P\le 0.05$ indicating a significant difference, and statistical plots were drawn using GraphPad Prism 8.0 software.

3. Result

3.1. Efficacy of TP in the Treatment of RA

To investigate the efficacy of TP on RA, we recorded the trend of body weight and swelling thickness of the right hind paw toe of all rats (Figures 1(a), 1(b), 1(c)). Before modeling, there were no significant differences in all groups ($P>0.05$). After modeling, compared with the control group, the body weight of the rats in the model group increased slowly and the thickness of the swelling of the right hind paw toes increased significantly. After the treatment, rats in the methotrexate and TP groups showed a significant increase in body weight ($P<0.05$) and a decrease in right hind paw toe thickness ($P<0.05$). Arthritis score is also an important indicator to for assessing the RA, as shown in Figure 1(d). At Day 19, the arthritis index scores showed the highest trend, and then, the scores of the rats in all groups began to decrease, and the difference was significant compared with the model group ($P<0.01$).

[figure(s) omitted; refer to PDF]

3.2. Results of Metabolomics

To further explore the potential mechanisms of TP for the treatment of RA, we performed untargeted metabolomics. PCA exhibited a separation tendency of groups. The PCA plot (Figure 2(a)) showed that the control group, the model group, the methotrexate group, and the TP group were separated from each other. The OPLS-DA method was utilized for further analysis to assist in differentiating the changes in the metabolites of each group. OPLS-DA plot showed significant separation between the control and model groups, as well as the model and TP groups. As shown in Figure 2(b), the OPLS-DA plots of the model and control groups were significantly different; $R^{2}X$ = 0.902, $R^{2}Y$ = 0.912, and $Q^2$ = 0.854, and the results of the permutation test (n = 200) are shown in Figure 2(c); the model group was clearly distinguished from the group of the TP, shown in Figure 2(d); $R^{2}X$ = 0.61, $R^{2}Y$ = 0.909, and $Q^2$ = 0.78, and the results of the permutation test (n = 200) are shown in Figure 2(e). The 200-permutation test showed the goodness of fit of the OPLS-DA model. Metabolites meeting VIP >1 and t-test $P<0.05$ were considered differential metabolites. Finally, 12 differential metabolites were selected after treatment, all of these metabolites were able to regress to normal levels in rats with significant regression ($P<0.05$) as shown in Table 1.

[figure(s) omitted; refer to PDF]

Table 1

Serum endogenous differential metabolites.

NO	Metabolites	VIP	KEGG	Model/control	TP/model
1	Arginine	1.67	C00062	↓##	↑${}^{\ast \ast }$
2	Leucine	1.41	C00123	↑##	↓${}^{\ast \ast }$
3	Valine	1.39	C00183	↑##	↓${}^{\ast }$
4	CholineGlycerophosphate	1.27	C00670	↑##	↓${}^{\ast }$
5	Ethanolamine	1.43	C00189	↑##	↓${}^{\ast }$
6	γ-Aminobutyric acid	1.04	C00334	↓##	↑${}^{\ast \ast }$
7	Lactate	1.55	C00186	↑##	↓${}^{\ast }$
8	Ethanol	1.20	C00469	↑##	↓${}^{\ast }$
9	Uridine	1.67	C00299	↑##	↓${}^{\ast }$
10	Arabinose	1.61	C02479	↑##	↓${}^{\ast \ast }$
11	Pantothenate	1.42	C00864	↑##	↓${}^{\ast }$
12	Mevalonic acid	1.41	C00418	↑##	↓${}^{\ast }$

^#$p<0.05$, ^##$p<0.01$ vs. control group; ${}^{\ast }p<0.05$, ${}^{\ast \ast }p<0.01$ vs. model group; ↑: increase, ↓: decrease.

Metabolism pathways of the differential metabolites were obtained and analyzed through the MetaboAnalyst 5.0 (https://www.metaboanalyst.ca) website. 12 differential metabolites screened were mainly related to arginine and proline metabolism; valine, leucine, and isoleucine biosynthesis; valine, leucine, and isoleucine degradation; and glycerophospholipid metabolism, as shown in Figure 3.

[figure(s) omitted; refer to PDF]

3.3. TP by UPLC-Q-TOF-MS/MS

Composition identifications of the ethyl acetate fraction of TP were performed using UPLC-Q-TOF-MS/MS, and the positive and negative total ion chromatogram (TIC) were obtained (Figures 4(a) and 4(b)). Combined with SCIEX OS 2.0 software and comparison with standards information and literature, we identified 49 chemical compounds, including 12 organic acids, 12 flavonoids, 4 alkaloids, 4 coumarins, and 17 other classes, as shown in Table 2.

[figure(s) omitted; refer to PDF]

Table 2

Compounds of ethyl acetate extract from TP identified by UPLC-Q-TOF-MS/MS.

No	$T_R$ (min)	Molecular formula	Ion mode	MS/MS	Error (ppm)	Identification
1	0.54	C4H6O5	[M − H]−	59.0140, 71.0136, 89.0248, 115.0038, 133.0142	−3.5	L-malic acid
2	0.79	C4H6O4	[M − H]−	73.0401, 117.0192	−1.7	Succinic acid
3	2.36	C7H6O3	[M − H]−	108.0187, 137.0244	−1.1	Protocatechuic aldehyde
4	1.73	C9H8O4	[M − H]−	90.9707, 135.0422, 179.0348	−0.7	Caffeic acid${}^{\ast }$
5	3.74	C21H20O9	[M − H]−	257.0839, 415.1034	−0.7	Daidzin
6	0.54	C7H12O6	[M − H]−	85.0292, 87.0085, 111.0213, 191.0561	−0.6	Quinic acid
7	0.73	C9H12N2O6	[M − H]−	66.0352, 82.0301, 200.0567, 243.0621	−0.2	Uridine
8	6.47	C16H12O6	[M − H]−	157.0274, 271.0252, 299.0566	−0.2	Kaempferol
9	18.8	C30H48O4	[M − H]−	471.3511	−0.2	Hederagenin
10	0.54	C6H8O7	[M − H]−	147.0135, 191.0199	0.2	Citric acid
11	5.69	C21H20O12	[M − H]−	151.0015, 300.0366, 301.0349	0.2	Quercetin-3′-O-glucoside
12	3.22	C9H6O4	[M − H]−	93.0347, 121.0306, 177.0372	0.7	Daphnetin
13	18.8	C30H48O4	[M − H]−	471.3511	0.7	Corosolic acid
14	3.22	C9H6O4	[M − H]−	121.0306, 149.0227, 177.0194	0.8	Esculetin
15	10.2	C20H30O5.HCOOH	[M − H]−	286.994, 395.2081	0.8	Andrographolide
16	6.47	C16H12O6	[M − H]−	227.0364, 299.0566	1	Hydroxygenkwanin
17	1.63	C8H10O3	[M − H]−	123.0185, 153.0558	1.4	Hydroxytyrosol
18	1.52	C13H18O7.HCOOH	[M − H]−	124.0176, 331.1029	1.6	Orcinol glucoside
19	3.74	C21H20O9	[M − H]−	397.0713, 415.1034	1.6	Puerarin
20	15.3	C38H44N2O6	[M − H]−	623.024	2.1	Dauricine
21	6.46	C27H34O11.HCOOH	[M − H]−	533.2228, 579.2089	2.1	Forsythin
22	5.69	C21H20O12	[M − H]−	301.0349, 301.1084, 343.2146, 417.2497, 463.1178	2.1	Hyperin${}^{\ast }$
23	6.47	C16H12O6	[M − H]−	226.0268, 227.0364, 278.9915, 299.0430	3.8	Diosmetin
24	5.40	C21H20O11	[M − H]−	285.0406, 163.0015, 447.0934	0.3	Isoorientin
25	4.97	C7H6O3	[M − H]−	93.0345, 93.0202, 137.0242	−1.3	Salicylic acid${}^{\ast }$
26	3.18	C9H10O5	[M − H]−	197.0437, 181.0148, 153.0536, 123.0090	−0.3	Ethyl gallate
27	5.17	C21H20O11	[M − H]−	447.092, 301.0361, 300.0291	0.3	Quercitrin
28	4.55	C21H20O12	[M − H]−	300.0366, 301.0349, 463.0885	0.7	Isoquercitrin
29	6.51	C15H10O7	[M − H]−	301.0348, 151.0032, 107.0132,273.0395	−0.1	Quercetin${}^{\ast }$
30	3.13	C15H12O6	[M − H]−	287.0558, 201.0530, 150.0317, 125.0225	−1.1	Aromadendrin
31	17.9	C16H32O2	[M − H]−	255.2332, 241.0966, 191.1098	1	Palmitic acid
32	4.35	C14H12O3	[M − H]−	142.0950, 185.0907, 227.0712	−0.6	Resveratrol
33	3.11	C9H10O5	[M − H]−	197.037, 123.0090	−0.3	Syringic acid
34	4.97	C7H6O3	[M − H]−	93.0345, 137.0238	−1.3	P-hydroxybenzoic acid
35	2.86	C8H8O4	[M − H]−	167.0352, 153.9237, 108.0212	−0.6	Vanillic acid${}^{\ast }$
36	4.31	C10H8O4	[M − H]−	191.0562, 133.0135	−0.8	Isoscopoletin
37	0.6	C4H6O4	[M − H]−	73.0295, 117.0183,	0.7	Succinic acid
38	3.48	C8H8O3	[M − H]−	151.0398, 109.0297, 108.0216	−0.3	Vanillin
39	1.65	C7H6O4	[M − H]−	108.0216, 109.0292, 153.0189	−0.6	Protocatechuic acid${}^{\ast }$
40	16.2	C35H60O6	[M + H]+	577.557	0.9	Sitogluside
41	7.21	C22H26N2O3	[M + H]+	81.0763, 105.0730, 275.2165, 367.2526	−5.5	Hirsutine
42	5.68	C27H32O15	[M + H]+	227.1859, 597.3298	0.8	Neoeriocitrin
43	0.55	C17H18O6	[M + H]+	225.0943, 301.1621, 319.1341	2.1	Agarotetrol
44	0.55	C16H16O6	[M + H]+	131.0678, 147.0645, 203.0949, 305.1206	1.3	Oxypeucedanin hydrate
45	7.21	C22H26N2O3	[M + H]+	105.0730, 367.2526	2.9	Geissoschizine methyl ether
46	0.53	C20H18NO4	[M + H]+	319.1784, 337.0621	4.1	Epiberberine
47	3.18	C15H10O6	[M − H]−	175.0407, 151.0401, 107.0507	4.3	Luteolin${}^{\ast }$
48	3.15	C15H14O6	[M − H]−	245.0815, 203.0710	−1.9	Epicatechin${}^{\ast }$
49	0.97	C7H6O5	[M − H]−	125.0240, 107.0139	0.4	Gallic acid${}^{\ast }$

${}^{\ast }$The compound has passed the standards comparison.

3.4. Network Pharmacology Analysis

3.4.1. Prediction of Potential Targets of TP Acting on RA

25 chemical components meeting the conditions under 2.8.1 were screened as active substances. 1575 component targets were predicted with the help of the Swiss Target Prediction platform. Then, 317 anti-RA-related targets were obtained through the GeneCards platform, and 156 common targets were obtained after the intersection of the two. The rheumatoid arthritis-component-target network has been constructed and the core targets in this network have been screened by the cytohub plugin, as shown in Figure 5(a). Of them, STAT3, SRC, MAPK3, PIK3R1, and MAPK1 were ranked high in the network and were considered to be important targets against RA.

[figure(s) omitted; refer to PDF]

Then, we used the DAVID database to perform GO and KEGG pathway enrichment analysis, the results suggested that the antirheumatoid arthritis effect of TP might be closely related to 116 pathways ($P<0.05$). The enrichment pathways include the cancer pathway, the hepatitis B pathway, the Kaposi sarcoma-associated herpesvirus infection, etc. Figure 5(b) showed the top 15 pathways, and the larger the node, the smaller the $P$ value, the higher the number of targets enriched on this pathway, and the color was from purple to darker. GO analysis mainly includes three parts: 106 biological process entries (BP), 14 cell cellular component entries (CC), and 33 molecular function entries (MF). The top ten results with the smallest $P$ values are shown in Figure 5(c). We also constructed a component-target-pathway network by Cytoscape 3.6.1 for visualization and analysis, Figure 5(d). Component nodes with higher degrees were selected for analysis, the higher the number of edges, the nodes may have a significant impact on the network, and the constituents in them can be considered as active components. The result showed that 10 compounds, including resveratrol, kaempferide, hydroxygenkwanin, hirsuteine, hederagenin, geissoschizine methyl ether, esculetin, epiberberine, daphnetin, and corosolic acid, are the important active compounds, as shown in Figure 4(c).

3.4.2. Integrating Metabolomics Analysis with Network Pharmacology

To comprehensively investigate the mechanism of TP for RA, a “differential metabolite-reaction-enzyme-gene” network was built by the MetScape plug-in function of Cytoscape [24], as shown in Figure 6. TP regulated the corresponding metabolic pathways by affecting this network, in which ethanolamine and glycerophosphocholine were involved in lipid metabolism; valine and leucine were involved in valine, leucine, and isoleucine degradation; pantothenate was involved in pantothenate and CoA biosynthesis; and arginine and aminobutyric acid were involved in the metabolism of arginine and proline. Comprehensively analyzing network pharmacology and metabolomics networks, we identified PLA2G4A as a common key target and glycerophosphocholine as a key metabolite, which both are involved in lipid metabolism.

[figure(s) omitted; refer to PDF]

3.4.3. Molecular Docking

For further validation, molecular docking was utilized to test the key target (PLA2G4A) and 10 potentially active compounds for binding. The docking information is shown in Table 3. The binding energy of −5.0 kcal/mol is usually considered standard [25]. The molecular docking results are shown in Figure 7. Our study results showed that the binding energies of the active compounds were all <−5.0 kcal/mol, suggesting that they have a strong binding ability to PLA2G4A. It further suggested the potential of these components to inhibit PLA2G4A activity.

Table 3

The docking scores of PLA2G4A and 10 related active compounds.

No	Compounds	Binding energy (kcal mol−1)
1	Corosolic acid	−11.44
2	Daphnetin	−5.90
3	Epiberberine	−8.16
4	Esculetin	−5.40
5	Geissoschizine methyl ether	−7.47
6	Hirsuteine	−7.78
7	Hederagenin	−11.88
8	Hydroxygenkwanin	−5.67
9	Resveratrol	−5.88
10	Kaempferide	−6.09

[figure(s) omitted; refer to PDF]

4. Discussion

In our previous experiments, we performed a screening of the efficacy of TP at different acetates for the treatment of RA. In our study, we found that the ethyl acetate extract of TP had the best efficacy. Therefore, we chose the ethyl acetate extract of TP for this experiment.

At present, due to the complexity of the causes of RA, the mechanisms for treating its disease have not yet been elucidated. In recent years, metabolomics has been utilized to screen for biomarkers associated with disease and has shown advantages in the early diagnosis and treatment of a variety of diseases, including RA [26–28].

In our metabolomics results, we compared and analyzed the metabolism pathways and found that most of them were associated with amino acid metabolism and lipid metabolism. The glycerophosphocholine levels were increased in CIA rats in comparison with the control group, and the corresponding glycerophospholipid metabolism pathway was abnormal, which was downregulated and returned to the normal level after intervention by TP. Leucine and valine were found to be jointly involved in metabolism pathways, such as valine, leucine, and isoleucine of biosynthesis and degradation. The leucine and valine content of rats in the model group was lower than that of rats in the normal group, and the corresponding metabolic pathways were disorganized, which is consistent with the pattern of amino acid metabolism changes in RA [29]; after treatment, the levels of leucine and valine in the rats in TP group increased and were regulated back to the normal level. Moreover, it has been shown that arginine levels can be used as an indirect evaluation of the degree of inflammatory response in RA, and that body arginine levels can decrease with increasing synovial inflammatory response [30]. After treatment with TP, the arginine level in CIA rats increased significantly, effectively alleviating RA. The study suggested that TP holistically corrected the disordered metabolic profiles, and exerted the pharmacological effects of RA treatment through multiple pathways.

Studies have shown that metabolic pathway abnormalities such as lipids and amino acids can lead to abnormal activation of fibroblast-like synoviocytes (FLS), which can invade and destroy the joints and mediate inflammation, causing the development of RA [31, 32]. Glycerophosphocholine is mainly involved in the lipid metabolism pathway and is a product of the hydrolysis of two fatty acid chains on the lecithin molecule in the human body. It contributes to the synthesis of phospholipids in cell membranes, thereby enhancing cell membrane fluidity and positively affecting lipid metabolism, as well as having anti-inflammatory effects [33–35]. Amino acids serve an essential physiologic function in the progression of the disease, and synovial hyperplasia and inflammation are closely associated with changes in amino acid levels in the body. Leucine and valine are branched-chain amino acids (BCAAs) that function in trophic signaling and metabolic regulation and are involved in immune regulation, affecting the synthesis and release of cytokines, such as IL-1, IL-2, and TNF-α, and the levels of BCAAs can be altered with changes in synovial inflammation [36, 37]. A normal level of leucine in the body regulates the release of proinflammatory cytokines and increases the secretion of anti-inflammatory cytokines, and abnormal leucine metabolism may lead to an imbalance in the body’s immunity in RA patients [38]. Arginine mainly exerts its immunomodulatory effects through the NO pathway as well as regulating the body’s endocrine hormones to promote immunoglobulin production [39].

We constructed a “compound-target-pathway” network through network pharmacology and screened 10 active compounds. The binding activity of 10 active compounds to key targets was further validated by molecular docking. In the meanwhile, it also suggested that the 10 active substances may be the pharmacodynamic substances of TP for RA. Studies have shown that resveratrol can play a role in the alleviation of RA by inhibiting the secretion of inflammatory factors IL-6, IL-1β, and TNF-α [40]. Daphnetin has anti-RA effects, mainly involving the protein kinase RNA-like endoplasmic reticulum kinase, activating transcription factor 4, and (C/EBP-homologous protein) pathways [41]. Esculetin acts as an anti-inflammatory by suppressing platelet-derived growth factor-induced phenotypic transformation of airway smooth muscle cells by inhibiting the phosphatidylinositol 3-kinase and protein kinase B pathway [41]. Hederagenin can block the nuclear factor kappa-B signaling pathway and reduce the release of inflammatory factors, such as IL-6, IFN-γ, TNF-α, NO, and PGE₂, which prevents and reduces joint rupture [42]. In summary, these active compounds mostly produce anti-inflammatory effects by regulating inflammatory proteins.

The targets of the antirheumatoid arthritis of TP in network pharmacology are mainly proteases, which are involved in apoptotic activities, such as apoptosis of cartilage and synovial inflammatory cells [43–45].

Intracellular life activities involve genes, proteins, and micromolecular metabolites. Metabolites are downstream of the gene regulatory network, and changes in the function of upstream macromolecules are ultimately reflected at the metabolites level. Therefore, we combined metabolomics with network pharmacology to analyze the correlation between the differential metabolites and key targets to explore the mechanism of ethyl acetate extract of TP for the treatment of RA. We found PLA2G4A and related metabolites, glycerophosphocholine, in the glycerophospholipid metabolism pathways. PLA2G4A is a member of the phospholipase A2 (PLA2) family. PLA2 is an enzyme that catalyzes the hydrolysis of membrane phospholipids and is also used as a diagnostic indicator for RA [46]. Besides, it has been reported that inflammatory mediators, such as TNF, can exert proactivation through PLA2 and participate in the inflammatory response [47]. Evidence also suggested that PLA2G4A could be an important target for tripterygium glycoside tablets in the treatment of RA [48]. These further suggest that PLA2G4A is an essential target of TP for the treatment of RA.

5. Conclusion

In conclusion, the treatment of RA by TP is the result of coregulation through multicomponents, multitargets, and multipathways. With a comprehensive analysis of metabolomics and network pharmacology, we further found that the treatment of RA by TP is closely related to PLA2G4A and metabolic pathways (lipid metabolism). Furthermore, 10 active compounds were discovered using UPLC-Q-TOF-MS/MS in combination with network pharmacology. These compounds were verified through molecular docking to have a strong binding affinity to key targets. However, there are still some restrictions in this study, which require further in-depth research and validation of the key targets and active compounds. In our research, we will investigate the mechanism of action and clarify the pharmacopoeial substances in the follow-up study to provide a reference for the innovative drugs of traditional Chinese medicine for treating RA.

Authors’ Contributions

Qin Qiu and Chunying Huang contributed to the design of this study. Chunping Qin and Xueyan Meng drafted the main manuscript text. All the authors participated in the revised manuscript. GraphPad Prism 8.0 SPSS software was used for statistical analysis, GraphPad Prism 8.0 was used for graphing, and Simca 14.1 was used for metabolomics data analysis. DeepL Translator was used for some of the paper content translation.

Acknowledgments

This work was financially supported by the National Natural Science Foundation of China (No. 82060713), Guangxi Zhuangyao Medicine Key Laboratory (GuiKeJiZi[2014]32), Zhuangyao Medicine Collaborative Innovation Center (GuiJiaoKeYan[2013]20), Guangxi Research Center for Ethnic Medicine Resources and Application Engineering (Guifa Reform High-tech Letter [2020]2605), Guangxi Science and Technology Base and Talent Special Project (GuiKe AD21238031), Guangxi Key Research and Development Projects (GuiKe AB21196016), Guangxi Traditional Chinese Medicine Key Discipline Zhuang Pharmacy (GZXK-Z-20-64), Guangxi First Class Discipline of Chinese Materia Medica (Ethnic Pharmacy) (GuiJiaoKeYan [2018]12), the Training Program for Thousands of Young and Middle-Aged Backbone Teachers in Guangxi Colleges and Universities (GuiJiaoJiaoShi [2022]60), High-level Key Disciplines of Traditional Chinese Medicine (Zhuang Pharmacy) of the State Administration of Traditional Chinese Medicine (Letter of Human Education of State Administration of Traditional Chinese Medicine [2022]226), Guangxi University of Traditional Chinese Medicine “Guipai Xinglin Youth Talents” Training Project (2022C032), and Guangxi University of Traditional Chinese Medicine transversal project (2021020).

References

[1] O. Al-Jobory, J. Varghese, A. Dweik, M. Al-Bayati, N. Davey, "Organizing pneumonia as first manifestation of seronegative rheumatoid arthritis," Cureus, vol. 14 no. 7, 2022.

[2] N. Abdissa, M. Induli, P. Fitzpatrick, J. P. Alao, P. Sunnerhagen, G. Landberg, A. Yenesew, M. Erdélyi, "Cytotoxic quinones from the roots of Aloe dawei," Molecules, vol. 19 no. 3, pp. 3264-3273, DOI: 10.3390/molecules19033264, 2014.

[3] J. M. Kremer, D. A. Lawrence, R. Hamilton, I. B. McInnes, "Long-term study of the impact of methotrexate on serum cytokines and lymphocyte subsets in patients with active rheumatoid arthritis: correlation with pharmacokinetic measures," RMD Open, vol. 2 no. 1, 2016.

[4] J. Pang, S. Ma, X. Xu, B. Zhang, Q. Cai, "Effects of rhizome of Atractylodes koreana (Nakai) kitam on intestinal flora and metabolites in rats with rheumatoid arthritis," Journal of Ethnopharmacology, vol. 281, 2021.

[5] Y. Xie, C. T. Mai, D. C. Zheng, Y. F. He, S. L. Feng, Y. Z. Li, C. X. Liu, H. Zhou, L. Liu, "Wutou decoction ameliorates experimental rheumatoid arthritis via regulating NF-kB and Nrf2: integrating efficacy-oriented compatibility of traditional Chinese medicine," Phytomedicine, vol. 85, 2021.

[6] K. Zheng, Z. Chen, W. Sun, B. Liu, D. Fan, Q. Guo, H. Luo, J. Shen, L. Li, X. He, S. Kou, X. Li, G. Pang, H. Zhao, C. Lu, "Hei-gu-teng zhuifenghuoluo granule modulates IL-12 signal pathway to inhibit the inflammatory response in rheumatoid arthritis," Journal of Immunology Research, vol. 2018,DOI: 10.1155/2018/8474867, 2018.

[7] X. Zhang, H. Yang, X. Zuo, L. Wu, J. Peng, Z. Li, H. Li, W. Ji, L. Zhang, X. Li, L. Dai, L. Lu, N. Yang, W. Wei, Z. Shuai, Y. Jiang, Y. Liu, P. E. Lipsky, H. Chen, Yistar study group, "Efficacy and safety of tripterygium wilfordii Hook F plus TNF inhibitor for active rheumatoid arthritis: a multicentre, randomized, double-blind, triple-dummy controlled trial," Clinical Immunology, vol. 255, 2023.

[8] N. Zhang, L. B. Zhang, Z. Wang, T. Y. Lan, J. P. Wang, C. Xiao, Q. W. Tao, Y. Xu, "Efficacy and safety of the modified Zhiwang decoction combined with methotrexate in early rheumatoid arthritis: study protocol for a randomised controlled trial," BMJ Open, vol. 14 no. 1, 2024.

[9] J. Wang, Z. Wang, T. Lan, L. Zhang, Z. Li, X. Wang, Q. Zou, Y. Wang, Y. Li, R. Luo, N. Zhang, Y. Xu, M. Li, Q. Tao, "Wangbi granule as a combination therapy to achieve clinical deep remission in rheumatoid arthritis: protocol for a multicenter, triple-blind, randomised, placebo-controlled trial," Chinese Medicine, vol. 18 no. 1,DOI: 10.1186/s13020-023-00728-6, 2023.

[10] Q. Qiu, J. Shi, C. Huang, Y. Liu, D. Zhen, X. Liu, B. J. Gu, L. Jiang, Y. Y. Liu, "Spectrum-effect relationship between hplc fingerprint and anti-inflammatory activity of n-butanol parts of Tetrastigma planicaule (Hook) Gagnep," Pakistan journal of pharmaceutical sciences, vol. 35 no. 4, pp. 1055-1066, 2022.

[11] Q. Qiu, X. X. Lu, B. G. Wan, Y. W. Liu, Y. L. Guo, Q. Y. Huang, D. D. Zhen, "Study on the anti-inflammatory spectrum-effect relationship of Tetrastigma planicaule rattan based on grey relational degree and partial least squares regression analysis," Journal of Chinese Medicinal Materials, vol. 44 no. 02, pp. 369-373, 2021.

[12] N. S. Ding, J. A. K. McDonald, A. Perdones-Montero, D. N. Rees, S. O. Adegbola, R. Misra, P. Hendy, L. Penez, J. R. Marchesi, E. Holmes, M. H. Sarafian, A. L. Hart, "Metabonomics and the gut microbiome associated with primary response to anti-TNF therapy in crohn's disease," Journal of Crohn’s and Colitis, vol. 14 no. 8, pp. 1090-1102, DOI: 10.1093/ecco-jcc/jjaa039, 2020.

[13] J. T. Bjerrum, Y. L. Wang, J. B. Seidelin, O. H. Nielsen, "IBD metabonomics predicts phenotype, disease course, and treatment response," EBioMedicine, vol. 71, 2021.

[14] Y. Pu, J. Zhang, N. Peng, R. Li, D. J. Xie, R. X. Zhang, L. T. Sun, J. L. Wang, H. Ye, "1 H NMR-based metabolomic study of striatal injury in rats with copper-loaded Wilson’s disease by Chinese and Western medicine intervention," Journal of Pharmaceutical and Biomedical Analysis, vol. 229, 2023.

[15] Z. Zhou, B. Chen, S. Chen, M. Lin, Y. Chen, S. Jin, W. Chen, Y. Zhang, "Applications of network pharmacology in traditional Chinese medicine research," Evidence-based Complementary and Alternative Medicine, vol. 2020,DOI: 10.1155/2020/1646905, 2020.

[16] J. Xu, J. Li, R. Zhang, J. He, Y. Chen, N. Bi, Y. Song, L. Wang, Q. Zhan, Z. Abliz, "Development of a metabolic pathway-based pseudo-targeted metabolomics method using liquid chromatography coupled with mass spectrometry," Talanta, vol. 192, pp. 160-168, DOI: 10.1016/j.talanta.2018.09.021, 2019.

[17] Y. S. Qi, J. B. Xie, P. Xie, Y. Duan, Y. Q. Ling, Y. L. Gu, X. L. Piao, "Uncovering the anti-NSCLC effects and mechanisms of gypenosides by metabolomics and network pharmacology analysis," Journal of Ethnopharmacology, vol. 281, 2021.

[18] M. Yang, S. Sun, X. Jia, X. Wen, X. Tian, Y. Niu, J. Wei, Y. Jin, Y. Du, "Study on mechanism of hepatoprotective effect of Chrysanthemum morifolium Ramat. based on metabolomics with network analysis and network pharmacology," Journal of Chromatography B, vol. 1222, 2023.

[19] Z. Zou, B. Han, M. Gong, S. Wang, S. Liang, "NMR-based metabonomic studies on stomach heat and cold syndromes and intervention effects of the corresponding formulas," Evidence-based Complementary and Alternative Medicine, vol. 2014,DOI: 10.1155/2014/528396, 2014.

[20] B. G. Giménez, M. S. Santos, M. Ferrarini, J. P. Fernandes, "Evaluation of blockbuster drugs under the rule-of-five," Pharmazie, vol. 65 no. 2, pp. 148-152, 2010.

[21] C. A. Lipinski, "Rule of five in 2015 and beyond: target and ligand structural limitations, ligand chemistry structure and drug discovery project decisions," Advanced Drug Delivery Reviews, vol. 101, pp. 34-41, DOI: 10.1016/j.addr.2016.04.029, 2016.

[22] P. Shannon, A. Markiel, O. Ozier, N. S. Baliga, J. T. Wang, D. Ramage, N. Amin, B. Schwikowski, T. Ideker, "Cytoscape: a software environment for integrated models of biomolecular interaction networks," Genome Research, vol. 13 no. 11, pp. 2498-2504, DOI: 10.1101/gr.1239303, 2003.

[23] P. Xin, X. Xu, H. Zhang, Y. Hu, C. Deng, S. Sun, S. Liu, X. Zhou, H. Ma, X. Li, "Mechanism investigation of Duhuo Jisheng pill against rheumatoid arthritis based on a strategy for the integration of network pharmacology, molecular docking and in vivo experimental verification," Pharmaceutical Biology, vol. 61 no. 1, pp. 1431-1445, DOI: 10.1080/13880209.2023.2252854, 2023.

[24] A. Karnovsky, T. Weymouth, T. Hull, V. G. Tarcea, G. Scardoni, C. Laudanna, M. A. Sartor, K. A. Stringer, H. V. Jagadish, C. Burant, B. Athey, G. S. Omenn, "Metscape 2 bioinformatics tool for the analysis and visualization of metabolomics and gene expression data," Bioinformatics, vol. 28 no. 3, pp. 373-380, DOI: 10.1093/bioinformatics/btr661, 2012.

[25] X. J. Liu, Y. Z. Wang, F. X. Wei, M. Lv, P. Qu, S. J. Chen, S. Y. Li, X. Qin, "The synergistic anti-depression effects of different efficacy groups of Xiaoyaosan as demonstrated by the integration of network pharmacology and serum metabolomics," Journal of Pharmaceutical and Biomedical Analysis, vol. 197, 2021.

[26] A. M. Weljie, R. Dowlatabadi, B. J. Miller, H. J. Vogel, F. R. Jirik, "An inflammatory arthritis-associated metabolite biomarker pattern revealed by 1H NMR spectroscopy," Journal of Proteome Research, vol. 6 no. 9, pp. 3456-3464, DOI: 10.1021/pr070123j, 2007.

[27] M. B. Lauridsen, H. Bliddal, R. Christensen, B. Danneskiold-Samsøe, R. Bennett, H. Keun, J. C. Lindon, J. K. Nicholson, M. H. Dorff, J. W. Jaroszewski, S. H. Hansen, C. Cornett, "1H NMR spectroscopy-based interventional metabolic phenotyping: a cohort study of rheumatoid arthritis patients," Journal of Proteome Research, vol. 9 no. 9, pp. 4545-4553, DOI: 10.1021/pr1002774, 2010.

[28] X. Li, S. B. Yang, Y. P. Qiu, T. Zhao, T. L. Chen, M. M. Su, L. X. Chu, A. P. Lv, P. Liu, W. Jia, "Urinary metabolomics as a potentially novel diagnostic and stratification tool for knee osteoarthritis," Metabolomics, vol. 6 no. 1, pp. 109-118, DOI: 10.1007/s11306-009-0184-0, 2010.

[29] Y. Liu, C. Peng, Y. F. Xie, "Exploration of the regularity of rheumatoid arthritis and changes in amino acid metabolism," Sichuan Medical, vol. 42 no. 08, pp. 794-799, 2021.

[30] A. A. Kayacelebi, J. Willers, V. V. Pham, A. Hahn, J. Y. Schneider, S. Rothmann, J. C. Frölich, D. Tsikas, "Plasma homoarginine, arginine, asymmetric dimethylarginine and total homocysteine interrelationships in rheumatoid arthritis, coronary artery disease and peripheral artery occlusion disease," Amino Acids, vol. 47 no. 9, pp. 1885-1891, DOI: 10.1007/s00726-015-1915-3, 2015.

[31] C. R. L. Machado, F. F. Dias, G. G. Resende, P. G. Oliveira, R. M. Xavier, M. V. M. Andrade, A. M. Kakehasi, "Morphofunctional analysis of fibroblast-like synoviocytes in human rheumatoid arthritis and mouse collagen-induced arthritis," Adv Rheumatol., vol. 63 no. 1,DOI: 10.1186/s42358-022-00281-0, 2023.

[32] C. Y. Jia, Y. Chang, W. Wei, Progress on Key Molecules of Abnormal Energy Metabolism in Fibroblast-like Synoviocytes in Rheumatoid Arthritis, 2021.

[33] O. Levillain, M. Schmolke, W. G. Guder, "Influence of dehydration on glycerophosphorylcholine and choline distribution along the rat nephron," Pflügers Archiv, vol. 442 no. 2, pp. 218-222, DOI: 10.1007/s004240100534, 2001.

[34] P. Hartmann, N. Fet, D. Garab, A. Szabó, J. Kaszaki, P. K. Srinivasan, R. H. Tolba, M. Boros, "rats," Journal of Surgical Research, vol. 189 no. 1, pp. 32-40, DOI: 10.1016/j.jss.2013.12.025, 2014.

[35] T. Tőkés, E. Tuboly, G. Varga, L. Major, M. Ghyczy, J. Kaszaki, M. Boros, "Protective effects of L-alpha-glycerylphosphorylcholine on ischaemia-reperfusion-induced inflammatory reactions," European Journal of Nutrition, vol. 54 no. 1, pp. 109-118, DOI: 10.1007/s00394-014-0691-2, 2015.

[36] Y. L. Ke, J. Luo, H. Y. Wang, "Branched-chain amino acid metabolism and its relation to disease," Chinese Journal of Biochemistry and Molecular Biology, vol. 39 no. 01, pp. 24-32, 2023.

[37] D. Dubey, S. Kumar, S. Chaurasia, A. Guleria, S. Ahmed, R. Singh, R. Kumari, D. R. Modi, R. Misra, D. Kumar, "NMR-based serum metabolomics revealed distinctive metabolic patterns in reactive arthritis compared with rheumatoid arthritis," Journal of Proteome Research, vol. 18 no. 1, pp. 130-146, DOI: 10.1021/acs.jproteome.8b00439, 2019.

[38] B. Cruz, A. Oliveira, M. C. C Gomes-Marcondes, B. Cruz, A. Oliveira, M. C. C. Gomes-Marcondes, "L-leucine dietary supplementation modulates muscle protein degradation and increases pro-inflammatory cytokines in tumour-bearing rats," Cytokine, vol. 96, pp. 253-260, DOI: 10.1016/j.cyto.2017.04.019, 2017.

[39] D. Evoy, M. D. Lieberman, T. J. Fahey, J. M. Daly, "Immunonutrition: the role of arginine," Nutrition, vol. 14 no. 7-8, pp. 611-617, DOI: 10.1016/s0899-9007(98)00005-7, 1998.

[40] Y. Jiang, H. C. Fang, X. C. Jin, P. Miao, "Mechanism of resveratrol improving rheumatoid arthritis by remodeling intestinal flora," New Chinese Medicine, vol. 54 no. 1, 2022.

[41] J. Hu, Y. Y. Yang, H. Ren, X. M. Cui, N. Li, T. Qu, Z. Y. Chen, "Discriminatory compounds and mechanism of Erycibe schmidtii and three potential substitutes against rheumatoid arthritis," Chinese Traditional and Herbal Drugs, vol. 53 no. 16, pp. 4958-4972, 2022.

[42] L. Li, Z. Pan, D. Ning, Y. Fu, "Rosmanol and carnosol synergistically alleviate rheumatoid arthritis through inhibiting TLR4/NF- κ B/MAPK pathway," Molecules, vol. 27 no. 1,DOI: 10.3390/molecules27010078, 2021.

[43] K. Yamaoka, "Janus kinase inhibitors for rheumatoid arthritis," Current Opinion in Chemical Biology, vol. 32, pp. 29-33, DOI: 10.1016/j.cbpa.2016.03.006, 2016.

[44] I. B. McInnes, G. Schett, "Pathogenetic insights from the treatment of rheumatoid arthritis," The Lancet, vol. 389 no. 10086, pp. 2328-2337, DOI: 10.1016/s0140-6736(17)31472-1, 2017.

[45] C. J. Malemud, "Matrix metalloproteinases and synovial joint pathology," Progress in Molecular Biology and Translational Science, vol. 148, pp. 305-325, DOI: 10.1016/bs.pmbts.2017.03.003, 2017.

[46] E. Kaiser, "Phospholipase A2: its usefulness in laboratory diagnostics," Critical Reviews in Clinical Laboratory Sciences, vol. 36 no. 2, pp. 65-163, DOI: 10.1080/10408369991239187, 1999.

[47] W. Pruzanski, P. Vadas, "Phospholipase A2-a mediator between proximal and distal effectors of inflammation," Immunology Today, vol. 12 no. 5, pp. 143-146, DOI: 10.1016/s0167-5699(05)80042-8, 1991.

[48] Y. Gao, Q. Qian, G. Xun, J. Zhang, S. Sun, X. Liu, F. Liu, J. Ge, H. Zhang, Y. Fu, S. Su, X. Wang, Q. Wang, "Integrated metabolomics and network analysis reveal changes in lipid metabolisms of tripterygium glycosides tablets in rats with collagen-induced arthritis," Computational and Structural Biotechnology Journal, vol. 21, pp. 1828-1842, DOI: 10.1016/j.csbj.2023.02.050, 2023.