Extraction, Purification of C‑Glycosides from Arisaema decipiens for Studying its Anti-Inflammatory and Bioactivities

Abstract

Arisaema decipiens­(A. decipiens), a traditional Chinese folk medicine, is recognized for its anti-inflammatory and analgesic properties. In this study, we report the extraction, purification, and anti-inflammatory mechanisms of C-glycosides from A. decipiens (ADCG). We developed an integrated strategy for the efficient enrichment and identification of polar C-glycosides from A. decipiens. Sample pretreatment via freeze-drying preserved thermolabile constituents, and reflux extraction with 60% aqueous ethanol optimized polar compound recovery. A key innovation was the implementation of a repeated-loading strategy using polyamide resin chromatography, which significantly enhanced the dynamic adsorption capacity and prevented leakage of target analytes. This approach achieved a 15.8-fold enrichment of target C-glycosides and enabled the identification of seven compounds, including chrysin- and apigenin-based derivatives. It effectively provides an efficient and cost-effective approach for the large-scale enrichment of this fraction. ADCG significantly suppressed granuloma formation in a cotton ball-induced model. In vitro, ADCG demonstrated potent anti-inflammatory effects in LPS-stimulated RAW264.7 macrophages by dose-dependently inhibiting the production of NO, TNF-α, and IL-6, without inducing cytotoxicity. In TNF-α-induced human rheumatoid arthritis synovial MH7A cells, ADCG inhibited proliferation, migration, invasion, and the secretion of pro-inflammatory mediators IL-6, IL-8, IL-1β, and MMP-1. Mechanistic investigations combining molecular docking and Western blot analysis revealed that ADCG directly inhibits NF-κB activation by suppressing IκBα degradation and p65 phosphorylation, thereby blocking nuclear translocation and the expression of downstream pro-inflammatory genes. Furthermore, constituents of ADCG exhibited high binding affinity for JAK2 and EGFR, suggesting a potential multitarget mechanism. These findings underscore the potential of ADCG as a complementary therapeutic agent for inflammatory diseases, particularly rheumatoid arthritis, and provide important insights into its molecular mechanisms of action.

1. Introduction

Arisaema decipiens Schott (A. decipiens), a plant from the Araceae family indigenous to southern China, has been historically prescribed for its efficacy in dispelling wind-dampness and alleviating pain. Modern pharmacological investigations have corroborated its antiarthritic and analgesic properties. The plant’s ability to modulate multiple inflammatory pathways, including reduction of pro-inflammatory cytokines, and rheumatoid factor (RF) expression, makes it particularly valuable for managing chronic inflammatory conditions. Additionally, it enhances the pain threshold in mice, reduces writhing responses, and inhibits ear edema in a dose-dependent manner. Clinical data on TCM treatment of rheumatoid arthritis (RA) indicated that capsules prepared from A. decipiens achieved an efficacy rate exceeding 80% in RA management.

While previous phytochemical studies have identified various bioactive compounds in A. decipiens, our research specifically focuses on its unique C-glycoside profile, which includes a class of C-glycosides identified in our prior work. C-glycosides are naturally occurring in foods and medicinal plants, are characterized by a carbon–carbon (C–C) bond linking the sugar moiety to the aglycone. − Mori et al. summarized the literature and reported that approximately 300 C-glycosides have been identified to date, among which compounds containing a single C-glycosidic bond are the most common. These compounds exhibit diverse pharmacological activities, including anti-inflammatory, − antioxidant, − and anticancer properties. Puerarin, a well-known C-glycosides, alleviating renal inflammatory damage, protects against sepsis-induced myocardial. − It has been formulated into various pharmaceutical preparations. Our previous studies revealed that A. decipiens contains a large number of C-glycosides derived from apigenin and chrysin. Unlike conventional C-glycosides that are linked by a single C–C bond, C-glycosides from ADCG uniquely possess dual C-glycosides bonds. This distinctive structural feature enhances their water solubility and preserves bioavailability through the addition of glycogroups. SS exerts inhibitory effects on inflammation in BV2 microglial cells stimulated by oxygen-glucose deprivation, LPS-induced pulmonary inflammation, and osteoporotic osteoarthritis. ISS demonstrates antitumor and anti-inflammatory activities, while chrysin and its derivatives exhibit anti-inflammatory and antioxidant properties. , However, the therapeutic effects and underlying mechanisms of ADCG on RA remain unclear.

RA is a chronic autoimmune disease characterized by synovial inflammation and synovial tissue hyperplasia. The RA fibroblast-like synovial cells are a key pathological cell population in RA, characterized by aberrant hyperplasia, migration, invasion and inadequate apoptosis. − These cells promote synovial hyperplasia, secrete matrix metalloproteinases (MMPs) that erode articular cartilage and underlying bone, and release inflammatory factors and chemokines, thereby sustaining persistent inflammation. Studies have shown that inhibiting the proliferation of synovial fibroblasts is one of the key objectives in modern RA treatment. The management of RA necessitates the suppression of synovial cell proliferation and facilitation of apoptosis. At present, the treatment strategy of RA patients is based on a treatment pyramid consisting of resting nonsteroidal anti-inflammatory drugs (NSAIDs), and disease-modifying antirheumatic drugs (DMARD). They include conventional synthetic drugs, biologics, and targeted JAK inhibitors. These agents work by suppressing the proliferation and activation of lymphocytes to reduce the production of pro-inflammatory cytokines, or by directly inhibiting cytokines such as TNF-α and IL-6. However, the therapeutic efficacy of these treatment strategies remains limited.

In recent years, certain pharmaceutical agents have been shown to exert therapeutic effects by suppressing fibroblast-like synoviocyte (FLS) proliferation through inhibition of the intracellular JAK-STAT signaling pathway. , Meanwhile, herbal medicine constitutes an important component of adjuvant therapy and is commonly employed as an alternative treatment for RA. − A growing number of natural compounds derived from traditional Chinese medicine have been found to effectively inhibit synovial fibroblast proliferation, reduce the production of inflammatory factors, and promote apoptosis by targeting the NF-κB pathway. − For example, triptolide have been reported to potently suppress NF-κB activation through blockade of IKK activity, thereby preventing IκB phosphorylation and degradation. , This mechanism contributes to the attenuation of FLS proliferation and inflammatory cytokine secretion.

This study aims to extract, purify, and isolate ADCG, and to investigate its anti-inflammatory activity and mechanism of action through both in vitro and in vivo experiments. The extraction and purification processes will be optimized using Box-Behnken Design (BBD) and repeated loading experiments. C-glycosides present in ADCG will be isolated and analyzed. Potential therapeutic targets will be predicted through network pharmacology analysis and molecular docking. The anti-inflammatory effects will be evaluated using the cotton pellet-induced granuloma model in vivo and LPS-stimulated RAW264.7 macrophage cells in vitro. Furthermore, the mechanism underlying the inhibition of fibroblast-like synovial cells will be investigated using TNF-α-induced MH7A cells. This research will provide a new reference for the development of effective and safe therapeutic agents for RA.

2. Experimental Section

2.1. Experimental Drugs

The A. decipiens plants were collected from Gaopo Township, Guiyang City, Guizhou Province. It was certified by Prof. Sun Qinwen from the School of Pharmacy of Guizhou University of Traditional Chinese Medicine. Fresh A. decipiens specimens were thoroughly washed, subsequently freeze-dried, and finally pulverized.

2.2. Instruments

The analysis was performed using a high-performance liquid chromatography (HPLC) system (LC-2030C, Shimadzu Corporation, Kyoto, Japan) equipped with a Waters XBridge C18 column (5.0 μm, 4.6 mm × 150 mm; Waters Corporation, Milford, MA, USA). Sample preparation utilized an electronic analytical balance (MS105DU, Mettler Toledo, Greifensee, Switzerland). Cell morphology was observed under an inverted microscope (ECLIPSE Ts2, Nikon, Tokyo, Japan). Absorbance and fluorescence measurements were conducted using a multifunctional microplate reader (FlexStation 3, Molecular Devices, San Jose, CA, USA). All cell culture procedures were carried out under aseptic conditions in a laminar flow cabinet (SW-CJ-1FD, Suzhou Antai Air Technology Co., Ltd., Suzhou, China), and cells were maintained in a CO₂ incubator (MCO-15AC, SANYO, Osaka, Japan).

2.3. Reagents

Chromatography-grade acetonitrile (Batch: 20230514) were obtained from TEDIA (Ohio, USA). SYBR Green Master Mix was procured from Vazyme Biotech Co. Ltd. (Nan jing, China). Polyamide resin (30–60 mesh, Sinopharm Chemical Reagent Co., Ltd., Shanghai, China). The reference compounds, including Apigenin-6-C- d -glucosyl-8-C–D-xyloside (AGX), Apigenin-6-C–D-xylosyl-8-C- d -glucoside (AXG), Schaftoside (SS), Chrysoeriol-6-C-xylopyrano-8-C-galactoside (CXG), Apigenin-6-C-galactosyl-8-C-arabinoside (AGA), Isoschaftoside (ISS), and Chrysoeriol-6-C-glucose-8-C-arabopyranoside (CGA) were made in the laboratory. Their structures were verified by 1H NMR, 13C NMR, and their purity (>98%) was confirmed by HPLC using the peak area normalization method. Phosphate-buffered saline (PBS, Batch:10091–148), Dulbecco’s modified Eagle’s medium (DMEM, Batch: PM150210), cell counting kit-8 (CCK-8, Batch/HY-K0301), cell lysis buffer were acquired from Hyclone Biotechnology (Shanghai, China), Rapid Blocking Buffer (CWBIO, Beijing, China), ELISA kits for Interleukin-1β­(IL-1β, Batch:H211230–007b), Interleukin-6­(IL-6, Batch: H211230–008b), Interleukin-8­(IL-8, Batch: H211230–002b), and Matrix metalloproteinase1­(MMP1, Batch: H211230–134b) were purchased from Neobioscience (Shenzheng, China).

2.4. Animals and Cells

Specific pathogen-free (SPF) KM mice (8 weeks old, weighing 18–22 g) were obtained from Tianqin Biotechnology Co. Ltd. (Changsha, China, Animal License No. SCXK­(Xiang) 2019–0014). All animal procedures were approved by the Animal Ethics Committee (Approval No. 20250930002). Animals were maintained under controlled conditions: 12 h light/dark cycle, temperature 25.0 ± 2.0 °C, relative humidity 45.0 ± 15.0%, with ad libitum access to standard laboratory chow and autoclaved water. RAW264.7 (Procell, Wuhan, China) and MH7A (SynthBio, Hefei, China) cells were cultured in DMEM supplemented with 10% (v/v) fetal bovine serum FB.

2.5. Methods

2.5.1. Extraction of ADCG

2.5.1.1. Single-Factor Experiment Design

A quantity (2.0 g) of A. decipiens powder was accurately weighed, mixed with 30 mL 70% ethanol. It was then extracted by reflux twice at 60 °C for 1 h each time. Filter the sample extraction solution and combine the filtrate. Then, filter it again with a 0.45 μm microporous filter membrane. SS, CXG, ISS, and CGA were quantitatively analyzed by LC-2030C system (Figure S1). The response value (R) was used to evaluate the efficiency of C-glycosides extraction methods. R was calculated from four chromatographic peaks according to the following formula as an evaluation index.

$$R=\sum _{i=1}^4{X_i}^{*}\frac{{\mathrm{X̅}}_i}{X}$$

Where: R, the adjustment number of response value; Xi, the peak area corresponding to peaks 1–4; ${\mathrm{X̅}}_{\mathrm{i}}$ , the average value of peaks 1–4.

According to the above test conditions, the Single-Factor were carried out by changing one of the factors: ethanol concentration (from 0 to 100%), reflux temperature (from 30 to 90 °C), liquid–material ratio (from 12:1 to 24:1 mL/g).

2.5.1.2. Box–Behnken Design

Based on the single-factor experiments, ethanol concentration (A), reflux temperature­(B), liquid–material ratio­(C) were selected as factors to optimize the extraction process of ADCG. A three-factor, three-level BBD was employed using Design-Expert­(v8.0.6). Model adequacy was assessed using statistical parameters such as R², adjusted R², and lack-of-fit tests.

2.5.2. Purification of ADCG

2.5.2.1. Pretreatment of Polyamide Resin

Polyamide resin was immersed in the anhydrous ethanol solution for 12 h to ensure complete swelling. The resin was then loaded into a chromatographic column and washed sequentially with anhydrous ethanol, 80%, 60%, 40%, and 20% ethanol with two column volumes (CVs) each. Finally, the resin was rinsed with distilled water until no residual ethanol was detected.

2.5.2.2. Static Adsorption and Desorption Test

A quantity (1.0 g) of polyamide resin (30–60 mesh) and solution of ADCG extract (concentration, 2 mg/mL; 30 mL) was added to a conical flask (50 mL) placed in a thermostatic water bath shaker for oscillation adsorption (25 °C, 120 rpm) for 12 h. Aliquots of 100 μL supernatant were collected every hour, and the adsorption rates (%) of four C-glycosides were determined. Static adsorption kinetics curves were plotted to evaluate the adsorption efficiency and equilibrium time.

2.5.2.3. Dynamic Adsorption Test

Treated polyamide resin (5 mL) was packed into a chromatographic column with a diameter of 1.5 cm. A constant flow pump was utilized to introduce 200 mL of ADCG extract into the column at flow rates of 2 mL/min, 3 mL/min and 4 mL/min. Effluent samples corresponding to one column volume were collected, and the adsorption rates of four C-glycosides were measured. The adsorption rate of the four C-glycosides in the effluent was measured, and a dynamic desorption curve was constructed.

2.5.2.4. Repeated Loading Experiment

Sample loading was performed using a repeated protocol, wherein the column effluent was reback to the inlet. Under conditions where the target compound concentration in the effluent remained below a critical thresholdspecifically, when the amount of target compound per unit column volume was lower than its instantaneous adsorption ratetheoretical models predicted no breakthrough would occur. The maximum theoretical loading capacity of the polyamide column was computationally determined. This predicted loading capacity was then empirically validated through multiple replicate loading experiments.

2.5.2.5. Eluting Solvent Selection

The elution efficiency of the target compound was evaluated across a gradient of ethanol concentrations (20%, 40%, 60%, 80%, and 95% v/v). For each concentration, effluent corresponding to 3 CVs was collected, pooled, and quantitatively analyzed.

2.5.3. Component Separation and Analysis

2.5.3.1. Components Separation and Identification

The crude extract was fractionated on macroporous resin (HPD-100) using stepwise ethanol–water gradients (0, 30%, 50%, 70%, 95%). The active 50% ethanol eluate was then subjected to repeated silica gel column chromatography (ethyl acetate–ethanol–water system) for further separation. The resulting fractions were further purified through repeated silica gel column chromatography, polyamide resin chromatography, and gel filtration. Finally, seven individual compounds were isolated by preparative HPLC. Their structures were unequivocally confirmed by comparison of ¹H NMR, ¹³C NMR, and MS data with those reported in the literature.

2.5.3.2. Components Analysis

Chromatographic separation was performed on a Waters XBridge C18 column (5.0 μm, 4.6 mm × 150 mm), with the mobile phase consisting of solvent A (0.01 mol/L (NH₃)₂HPO₃ in water: 0.01 mol/L NH₄H₂PO₄ in water 4:1) and solvent B (acetonitrile). The gradient elution was 5% B for 1–10 min, 5–8% B for 10–40 min, 8–15% B for 40–60 min, 15%–5% B for 60–61 min, 5% B for 61–65 min. The column temperature was 30 °C, detection wavelength was 370 nm, flow rate was 0.8 mL/min and injection volume was 20 μL.

2.5.4. Animal Experiment Design

After a 3 day acclimatization period under standard conditions, KM mice were randomly divided into five groups (n = 12 per group) and treated daily via oral gavage as follows: the normal control (NC) group received 0.5% sodium carboxymethyl cellulose (CMC-Na); the model (M) group received 0.5% CMC-Na; the positive control group received prednisone acetate (PA, 7 mg/kg); and three experimental groups received ADCG at doses of 60, 120, and 180 mg/kg, respectively. According to the Guizhou Provincial Standard for Chinese Ethnic Medicine Herbal Pieces, the recommended clinical dose of A. decipiens for human use is 0.3–0.6 g/day. Based on an average adult body weight of 70 kg, the corresponding human dose is 4.29–8.57 mg/kg. Using the body surface area-based Km coefficient method (Km = 37 for humans, Km = 3 for mice), the equivalent mouse dose was calculated to be approximately 52.8–105.4 mg/kg. On this basis, and taking into account the dose range recommended for human use, as well as considerations of convenient administration and the available amount of test material, three doses-60, 120, and 180 mg/kg-were ultimately selected as the low, medium, and high dose groups, respectively. All test articles were prepared in 0.5% CMC-Na and administered by gavage at a volume of 10 mL/kg once daily for 7 consecutive days.

2.5.4.1. Cotton Pellet-Induced Granuloma Test

Mice were anesthetized with a subcutaneous injection of 3% chloral hydrate (0.15 mL/10 g). Following sterilization, a small incision was made on the dorsal surface, and two sterile cotton pellets (10 mg each) were aseptically implanted into the left and right axillae. The incision was sutured and disinfected. Treatments were administered orally once daily for 14 consecutive days. The control and model groups received 0.5% CMC-Na solution (0.1 mL/10 g), the PA group received prednisone acetate (7 mg/kg), and the ADCG groups received their respective doses. Two hours after the final administration on day 14, the mice were euthanized. The granulomatous tissue encapsulating each cotton pellet was carefully dissected, freed from extraneous adipose tissue, and dried at 70 °C for 4 h. The dry weight of the granuloma was determined after cooling.

The granuloma weight (mg) was calculated as.

Granuloma weight = Dry weight of (granuloma tissue + cotton pellet) - Dry weight of cotton pellet alone.

2.5.4.2. Hematological Analysis

Blood samples were collected and analyzed using an automatic hematology analyzer to determine the counts of white blood cells (WBC), lymphocytes (LYM), monocytes (MON).

2.5.5. Network Pharmacology Construction and Molecular Docking Analysis

2.5.5.1. Screening Targets of ADCG Relevant to RA

The canonical SMILES, CAS numbers, and 3D structures of seven C-glycosides were obtained from the PubChem database. The canonical SMILES or CAS of components were inputted to the TCMSP (https://old.tcmsp-e.com/tcmsp.php), accessed on 8 April 2025, CTD (http://ctdbase.org/), accessed on 8 April 2025, DrugBank (https://go.drugbank.com/), accessed on 8 April 2025. , In cases where compounds could not be retrieved using canonical SMILES or CAS numbers, their respective 3D structures were submitted to the Swiss Target Prediction platform (http://www.swisstargetprediction.ch/), accessed on 12 April 2025. Subsequently, potential targets associated with inflammatory were obtained from GeneCards (https://www.genecards.org/), accessed on 12 April 2025, OMIM (https://omim.org/), accessed on 12 April 2025, DrugBank accessed on 12 April 2025, and PharmGKB (https://www.pharmgkb.org/), accessed on 12 April 2025 using ‘RA’ as a keyword.

2.5.5.2. Protein–Protein Interaction Analysis

The intersection of ADCG targets and inflammatory-related targets were analyzed by the Draw Venn Diagram (http://bioinformatics.psb.ugent.be/webtools/Venn/, accessed on 12 April 2025). The ADCG-inflammatory intersected targets were imported into String database version 11.5 (https://cn.string-db.org/) to obtain PPI data. The organism selected for analysis was “Homo sapiens”, and a protein interaction parameter score value greater than 0.9 was set as the threshold (indicating higher confidence in protein interactions). Cytoscape (ver. 3.7.1) was used to establish the topological properties of the PPI network.

2.5.5.3. Molecular Docking Analysis

Molecular docking was carried out to evaluate the binding affinity between the compounds and the key target. The structure of the target proteins were obtained from the Protein Data Bank (http://www.rcsb.org/, accessed on 14 April 2025), while seven C-glycosides components were retrieved from PubChem databases. AutoDock Vina 1.1.2 software was utilized for molecular docking analysis. PyMol 2.5.5³ was employed to preprocess the receptor protein, including removal of water molecules, salt ions, and small molecules. Subsequently, a grid box was designed to define the search space within the protein pocket for ligand binding exploration. ADCG Rsuite 1.0⁴ facilitated conversion of receptor proteins and molecular formats as required by Vina for molecular docking simulations with a global search detail set at 20. Finally, the conformation with highest binding affinity score for each core protein was selected as optimal configuration using Discovery Studio Viewer for visual analysis.

2.5.6. Cell Culture and Treatment

RAW264.7 and MH7A cells were cultured in DMEM supplemented with 10% (v/v) fetal bovine serum­(FBS), penicillin­(100 units/mL) and streptomycin­(100 μg/mL). Cells were maintained at 37 °C in a humidified atmosphere containing 5% CO₂.

2.5.6.1. Cell Viability

Cell viability was assessed using CCK-8. Briefly, RAW264.7 murine macrophages­(2 × 10⁴ cells/well) and MH7A human synovial cells­(2 × 10⁵ cells/well) were inoculated into 96-well plates. After 24h (RAW264.7) or 48h (MH7A) incubation with graded concentrations of ADCG (1, 5, 10, 20, and 100 μg/mL), the treatment medium was aspirated and replaced with CCK-8 working solution (10% v/v in serum-free medium). The medium was then replaced with CCK-8 working solution (10% v/v in serum-free medium), followed by incubation at 37 °C/5% CO₂ for 4 h. Absorbance at 450 nm was measured using a FlexStation 3 multimode microplate reader. Viability was calculated as

$$\mathrm{V}\mathrm{i}\mathrm{a}\mathrm{b}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{y}(\%)=(\mathrm{A}1-\mathrm{A}0)/(\mathrm{A}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}-\mathrm{A}\mathrm{b}\mathrm{l}\mathrm{a}\mathrm{n}\mathrm{k})\times 100\%$$

where A0: absorbance of medium + CCK-8 (no cells). A1: absorbance of ADCG treatment, A2: absorbance of untreated cells.

2.5.6.2. Inflammatory Biomarkers

RAW264.7 cells and MH7A cells (2 × 10⁵ cells/well) were seeded in 96-well plates, and incubated for 24 h. RAW264.7 cells were randomly divided into blank control group, LPS (1 μg/mL) model group, LPS + ADCG (5, 10, and 20 μg/mL) group, and LPS + dexamethasone (DEX, 50 μM) at positive group. MH7A cells were divided into blank control group, TNF-α model group­(10 ng/mL), TNF-α+ ADCG (5, 10, and 20 μg/mL) group, and TNF-α+ Methotrexate (MTX, 0.1 μM). Cells were pretreated with ADCG or DEX or MTX for 2 h followed by stimulation with LPS (RAW264.7) or TNF-α (MH7A) for 24 h. In LPS-induced RAW264.7 cells, NO production was measured via Griess reaction, TNF-α, and IL-6 levels were quantified by ELISA. In TNF-α-stimulated MH7A cells, IL-6, IL-8, IL-1β, and MMP1 were assessed by ELISA.

2.5.6.3. Wound Healing Assay

MH7A cells (5 × 10⁵ cells/well) were seeded in 6-well plate cultured to 100% confluence. After serum starvation for 12 h to minimize proliferation effects, cells were pretreated for 16 h with blank control, TNF-α model, TNF-α+ ADCG (5, 10, 20 μg/mL), and TNF-α+ MTX (0.1 Μm). A uniform scratch wound was generated using a sterile 200 μL pipet tip, followed by TNF-α stimulation (20 ng/mL) for 24 h. Wound images were captured at 0 and 24 h using an Olympus CKX-41 inverted microscope (10× magnification). Scratch areas were quantified with Image-Pro Plus (v. 6.0), and migration rate was calculated as.

Migration distance = D 24 h – D 0 h where D 0 h = initial distance, D 24 h = residual distance at 24 h.

2.5.6.4. Transwell Assay

The lower chamber of Transwell was supplemented with DMEM medium containing 10% FBS. MH7A Cells (5 × 10⁴ cells/well) in 100 μL serum-free medium were seeded in the upper chamber. Cells were pretreated with ADCG (5, 10, and 20 μg/mL) or MTX (0.1 μM; positive control) for 16 h, then stimulated with TNF-α (20 ng/mL) for 24 h. Inserts were inverted onto absorbent paper, rinsed 3× with PBS, and fixed with 4% paraformaldehyde (1 h). After 3× PBS washes, membranes were stained with 0.1% crystal violet (20 min), washed 3× with PBS, and nonmigrated cells on the upper surface gently removed by cotton swab. Migrated cells in five random fields per membrane were imaged at 100× magnification using ImageJ software.

2.5.6.5. Western Blotting

MH7A Cells were seeded in 6-well plates at a density of 1 × 10⁵ cells/well. Following treatment with ADCG (5, 10, and 20 μg/mL) for 24 h, and TNF-α (20 ng/mL) was added for 10 min. The cells were collected and washed using RIPA buffer. The total protein content was determined utilizing the BCA Protein Assay Kit. The proteins were isolated through SDS-PAGE gel electrophoresis and transferred onto a PVDF membrane. After blocking with Rapid Blocking Buffer, the membranes were incubated overnight at 4 °C with primary antibodies. The membranes were cultured with secondary antibodies for 2 h at room temperature. The chemiluminescence was employed to reveal the antibody-reactive bands. Band intensities were quantified using Gel-Pro Analyzer 4 software (Media Cybernetics, USA). The expression level of each target protein was normalized to the corresponding GAPDH band as an internal loading control. The final quantitative data presented in the bar graphs represent the absorbance ratio (AR), calculated as the integrated optical density value of the target protein band divided by that of the GAPDH band.

2.6. Statistal Analysis

Data were analyzed using SPSS­(v. 24.0). For in vitro cell-based assays, experiments were performed in triplicate (n = 3 independent experiments). For in vivo animal studies, the sample size was n = 12 per group. Data are expressed as the mean ± standard deviation (SD). Statistical comparisons between groups were analyzed using one-way analysis of variance (ANOVA) followed by Dunnett’s post hoc test. A p-value < 0.05 was considered statistically significant.

3. Results

3.1. Effects of the Different Factors on the Extraction Yield of the ADCG

3.1.1. Single-Factor Analysis

As depicted in Figure , the extraction yield of ADCG was influenced by ethanol concentration, extraction temperature, and material–liquid ratio. With increasing ethanol concentration, the yield of C-glycosides first rose gradually, peaking at 50% ethanol (Figure a). This maximum extraction efficiency likely resulted from the optimal polarity matching between the solvent and target compounds, which enhanced solubility. Further increasing or decreasing the ethanol concentration beyond this level led to a clear decline in yield. Extraction temperature also played an important role in determining ADCG yield. As illustrated in Figure b, the yield increased steadily with temperature up to 70 °C, above which it stabilized. The plateau may be attributed to starch gelatinization at elevated temperatures, hindering the release of ADCG into the solvent. Varying the liquid-to-material ratio similarly affected extraction efficiency. According to Figure c, the yield improved as the ratio increased, reaching an optimum at 1:24 (w/v). Beyond 1:16 (w/v), however, further increases in solvent volume resulted in only marginal gains of less than 1% in yield.

1.

Effects of different extraction parameters on the ADCG extraction rate. (a) Ethanol concentration, (b) temperature, (c) material–liquid ratio.

3.1.2. Optimization Using a Box-Behnken Design

Extraction temperature (A), ethanol concentration (B), and material–liquid ratio (C) were chosen as independent variables, while the response values of four C-glycosides served to evaluate extraction efficiency. Guided by the single-factor results, a three-factor, three-level BBD was adopted for process optimization. The detailed experimental design and corresponding outcomes are summarized in Table .

1. Box-Behnken Design and Experimental Result .

run	(A)/°C	(B)/%	(C)/mL	response value
1	–1(30)	1(90)	0(32)	543.3
2	0(55)	0(60)	0(32)	769.0
3	0(55)	0(60)	0(32)	763.7
4	1(80)	1(90)	0(32)	778.4
5	1(80)	0(60)	1(48)	885.5
6	1(80)	0(60)	–1(16)	739.6
7	0(55)	1(90)	–1(16)	686.2
8	0(55)	–1(30)	–1(16)	261.8
9	0(55)	0(60)	0(32)	779.0
10	–1(30)	0(60)	1(48)	834.3
11	0(55)	1(90)	1(48)	642.4
12	0(55)	0(60)	0(32)	808.4
13	–1(30)	0(60)	–1(16)	884.2
14	1(80)	–1(30)	0(32)	835.5
15	–1(30)	–1(30)	0(32)	716.8
16	0(55)	–1(30)	1(48)	464.7
17	0(55)	0(60)	0(32)	767.6

^aThe fitted model was as follows: Y = 777.53–23.35A + 150.51B+31.88C+29.10AB+48.93AC-61.67BC+131.57A2–190.58B2–73.20C2–208.14A2B+111.77AB2.

As shown in Table , the model’s F-value was 103.556 (p < 0.0001), indicating that the model is highly significant for predicting the extraction rate of ADCG accurately. The lack-of-fit p-value (0.2869) was not significant relative to the pure error, and the determination coefficient (R²) was 0.9958, demonstrating that the model explains 99.58% of the total variability. These results confirm the high reliability of the model and its ability to accurately reflect experimental trends. The effects of extraction temperature, ethanol concentration, and material-liquid ratio on the yield of ADCG were further analyzed using three-dimensional response surface plots and contour maps (Figure ). All three factors exhibited significant impacts on ADCG extraction efficiency, consistent with the ANOVA results. Based on the model predictions, the optimal extraction parameters were determined as follows: extraction temperature of 30 °C, ethanol concentration of 55.86%, and material–liquid ratio of 1:15.62 (w/v). To facilitate practical implementation, the parameters were adjusted to extraction temperature of 30 °C, ethanol concentration of 55%, material–liquid ratio of 1:16 (w/v), and two extraction cycles of 60 min each. Under these conditions, the predicted extraction rate of ADCG was 938.9 μg/g. The actual average extraction rate obtained from triplicate experiments was 930.9 μg/g, with a RSD of 0.85%, which is in close agreement with the predicted value. These findings substantiate the reliability and efficacy of the optimized extraction procedure.

2. ANOVA of Box-Behnken Design .

ANOVA source	sum of squares	df	mean square	F-value	p-value
model	413740	11	37612.72	103.556	<0.0001
A	2181.654	1	2181.654	6.006565	0.0579
B	90614.05	1	90614.05	249.4801	<0.0001
C	8132.432	1	8132.432	22.39035	0.0052
AB	3387.942	1	3387.942	9.327738	0.0283
AC	9576.095	1	9576.095	26.36507	0.0037
BC	15210.98	1	15210.98	41.87912	0.0013
A2	72891.19	1	72891.19	200.6853	<0.0001
B2	152937.4	1	152937.4	421.0699	<0.0001
C2	22564.09	1	22564.09	62.12383	0.0005
A2B	86643.03	1	86643.03	238.5471	<0.0001
AB2	24983.42	1	24983.42	68.78477	0.0004
residual	1816.057	5	363.2115	/	/
lack of fit	496.94	1	496.94	1.506886	0.2869
pure error	1319.117	4	329.7793	/	/
cor total	415556	16	/	/	/

^a R ² = 0.9958, R ² _Adj = 0.9873, R ² _Pred = 0.8748.

2.

Optimization of the extraction process of C-glycosides from Arisaema decipiens (ADCG) by Box-Behnken design. (a) Temperature and ethanol concentration; (b) Temperature and material–liquid ratio; (c) Ethanol concentration and material–liquid ratio.

3.2. Purification of ADCG

3.2.1. Static Adsorption and Dynamic Adsorption

The polyamide resin exhibited a rapid adsorption rate toward the target C-glycosides, as illustrated in Figure a. In static adsorption tests, the resin achieved 48.49% of its total adsorption capacity within 4 h, rising to 57.81% by 7 h, after which the process gradually approached equilibrium. In dynamic adsorption experiments (Figure b), noticeable leakage of the target compounds occurred when 3 CVs of sample were loaded. This leakage diminished progressively as the loading volume increased to 15 CVs. At a flow rate of 2 mL/min, leakage remained relatively low, whereas higher flow rates led to accelerated leakage and consequently reduced purification efficiency. Therefore, a flow rate of 2 mL/min was chosen to maximize the dynamic adsorption capacity during sample loading.

3.

Leakage kinetic curves of ADCG. (a) Static adsorption kinetics; (b) dynamic adsorption curves at different flow rates.

3.2.2. Repeated Loading

To enhance the adsorption efficiency of the polyamide resin and minimize the loss of valuable components, the loading volume was systematically optimized. According to the theoretical adsorption capacity of the resin (Table S1), loading 6 column CVs of the sample solution allowed the resin to retain partial adsorption capacity. Maximum adsorption efficiency was achieved with 7 CVs (Table S2), whereas loading 8 CVs led to detectable leakage of the target compounds, resulting in component waste (Table S3). Therefore, 7 CVs was determined as the optimal loading volume to maximize adsorption while reducing loss. The robustness of using 7 CVs was confirmed through three independent replicate experiments, all of which yielded consistent results with no leakage, thus supporting its adoption in subsequent purification steps.

3.2.3. Eluting Solvent Selection

As the ethanol concentration increased, the content of C-glycosides in the effluent rose continuously and reached equilibrium at 60% (v/v, Figure a). Although higher ethanol concentrations could further improve elution efficiency, a concentration of 60% was chosen to minimize organic solvent usage in consideration of industrial cost-effectiveness and environmental impact. The elution process was further monitored by measuring residual sample content in the solvent. As depicted in Figure b, after elution with 5 CVs of 60% ethanol, the residual content fell below the detection limit, indicating near-complete desorption (>96%) of the target compounds from the resin.

4.

Elution conditions of C-glycosides extraction from A. decipiens. (a) Ethanol concentration; (b) Elution solvent volume.

After repeated loading with 7 CVs and elution using 60% ethanol (5 CVs), the target C-glycosides were successfully purified. The final concentration of the target compounds increased by 15.8-fold compared to the crude extract, with a recovery rate of 53.79%. These results demonstrate that the polyamide resin exhibits high purification efficiency for C-glycosides, achieving both significant enrichment and acceptable recovery.

3.3. Component Separation and Analysis of ADCG

Seven components were isolated from ADCG. The isolated compounds were subjected to structural elucidation using 1H NMR and 13C NMR, which provided critical information on the C-glycosidic linkages and substitution patterns. The results of the identification are shown in Table S4. As shown in Figure , seven C-glycosides compounds were successfully identified within ADCG: AGX, AXG, SS, CXG, AGA, ISS, and CGA. These identifications were confirmed by their characteristic spectral data, ensuring the reliability of the results.

5.

Analysis of C-glycosides in ADCG.

3.4. Anti-inflammatory Effects of ADCG in the Cotton Pellet Granuloma Model

The cotton pellet-induced granuloma in mice is a classic model of chronic proliferative inflammation. Administration of ADCG at various doses and prednisone acetate significantly suppressed granuloma formation to different extents. Compared to the model group, granuloma weight was significantly reduced in the positive control group (PA, 7 mg/kg) and the ADCG-120 and ADCG-180 mg/kg groups (p < 0.05, Figure a,b). These findings demonstrate the potent inhibitory effect of ADCG on chronic inflammation in this model. Analysis of hematological inflammatory markers revealed that the model group exhibited a significant increase in WBC, LYM, and MON counts compared to the control group (p < 0.01, Figure c,d,e). Treatment with PA (7 mg/kg), ADCG-120, and ADCG-180 mg/kg significantly reversed these elevations, effectively lowering the WBC, LYM, and MON counts relative to the model group (p < 0.05). Notably, the efficacy of ADCG-180 mg/kg in reducing MON count was comparable to that of PA. This finding provides crucial evidence to support further investigation into the underlying mechanisms of ADCG.

6.

ADCG attenuates inflammation in a cotton pellet granuloma model, (a) Cotton pellet-induced granuloma, (b) Granuloma weight, (c) White blood cell (WBC), (d) Monocyte (MON), (e) Lymphocyte (LYM). NC: negative control group, M: cotton pellet granuloma model group, PA: prednisone acetate positive control group, ADCG: ADCG administration groups; Note: data are presented as mean ± SD (n = 12);^# p < 0.05, ^## p < 0.01 vs NC group, *p < 0.05, **p < 0.01 vs M group.

3.5. Network Pharmacology Analysis

3.5.1. Predicted Component and Disease Targets

Potential targets of the seven identified C-glycosides were retrieved from the TCMSP, CTD, DrugBank, and SwissTargetPrediction platform. Targets were filtered using TCMSP’s pharmacokinetic criteria and SwissTargetPrediction’s probability threshold. After removing duplicates, 128 unique compound targets were retained. RA-related targets were obtained from GeneCards (relevance score ≥1), OMIM, DrugBank, and PharmGKB. Duplicate entries and nonhuman genes were excluded, resulting in 4402 high-confidence inflammatory targets. The overlapping targets between ADCG components and inflammatory were identified using Venny 2.1 (https://bioinfogp.cnb.csic.es/tools/venny/), yielding 73 shared targets (Figure a). These targets were prioritized for subsequent pathway analysis.

7.

Network pharmacology analysis of compound–target interactions. (a) Common targets. (b) PPI of common targets (larger nodes and deeper color indicate higher degree value).

3.5.2. Construction and Evaluation of PPI Network

To explore functional associations among the 73 common targets, a protein–protein interaction (PPI) network was constructed via the STRING database. A high-confidence interaction threshold (combined score >0.9) was applied to minimize false positives. Disconnected nodes were excluded from the final network. As shown in Figure b, the resultant network contained 61 interconnected nodes and 344 edges, with a network density of 18.8% (344 edges/1830 possible edges) and an average node degree of 5.6, indicating moderate topological complexity. Ten hub targetsRelA (p65), TP53, LCK, JAK2, EGFR, AKT1, MAPK8, PRKACA, ITGAV, and HIF1Awere prioritized based on the composite ranking of three centrality metrics: degree (connectivity), betweenness (bridging role), and closeness (information flow efficiency).

3.6. Molecular Dockin

Molecular docking studies were performed using AutoDock Vina 1.1.2 software. From the 10 core genes, prioritizing their direct relevance to inflammatory pathways, five inflammation-related targets (RelA [PDB ID: 1NFI], JAK2 [PDB ID: 8EX2], EGFR [PDB ID: 2RGP], AKT1 [PDB ID: 4GV1], and MAPK8 [PDB ID: 4AWI]) were selected for docking with seven C-glycoside compounds. A binding energy threshold of ≤ – 7.9 kcal/mol was applied to indicate high binding affinity. The binding affinities of the compounds for the core targets are summarized in Table , and representative molecular docking interaction diagrams are shown in Figure S2. Molecular docking revealed that ADCG constituents exhibit high-affinity binding to core inflammatory targets, with CXG showing the strongest interaction with RELA (NF-κB p65) at −10.2 kcal/mol, followed by AGX and SS binding JAK2 at −10.1 and −10.0 kcal/mol, respectively. Notably, SS demonstrated multitarget engagement (RELA/JAK2/EGFR ≤ −9.5 kcal/mol), while CGA bound both EGFR and JAK2 at −9.9 kcal/mol.

3. Binding Affinities of Compounds to Core Targets.

ingredient	binding affinity (kcal/mol)
	RelA(p65)	JAK2	EGFR	AKT1	MAPK8
AGX	–9.2	–10.1	–9.2	–8.2	–8.9
AXG	–9.2	–8.7	–8.8	–8.7	–9.2
SS	–9.5	–10.0	–9.6	–8.9	–8.8
CXG	–10.2	–8.8	–9.2	–8.4	–8.8
AGA	–9.1	–10.0	–8.6	–7.9	–9.2
ISS	–9.2	–8.9	–8.8	–8.5	–8.9
CGA	–9.0	–9.9	–9.9	–8.8	–9.0

3.7. Effects of ADCG on RAW264.7 Cells

ADCG displayed minimal cytotoxicity across the tested concentration range (1–100 μg/mL), with cell viability remaining above 98% after 24 h of treatment (Figure a). The half-maximal inhibitory concentration (IC₅₀) was beyond the highest concentration tested (100 μg/mL), indicating a favorable biosafety profile suitable for further anti-inflammatory investigation.

8.

Inhibitive effects of ADCG on LPS-induced RAW264.7 cells, (a) Cell viability, (b) NO content, (c) IL-6, (d) TNF-α. NC: normal control group, LPS: lipopolysaccharide induced model group, DEX: dexamethasone positive control group, ADCG: ADCG administration groups; Note: data are presented as mean ± SD (n = 3); ^## p < 0.01 vs NC group, **p < 0.01 vs LPS group.

As a key mediator in immunopathological processes and inflammatory tissue injury, NO production was markedly increased upon LPS stimulation compared to the control group (p < 0.01). However, pretreatment with ADCG (10–20 μg/mL) for 24 h significantly suppressed LPS-induced NO release in a concentration-dependent manner (p < 0.01, Figure b), suggesting its potential to mitigate macrophage-driven inflammatory activation.

Further assessment by ELISA revealed that LPS stimulation significantly upregulated the secretion of TNF-α and IL-6 (p < 0.01 vs control), while ADCG (10–20 μg/mL) dose-dependently reduced their release, restoring cytokine levels close to baseline (p < 0.01 vs LPS group; Figure c,d). Notably, the inhibitory effect of ADCG on IL-6 was comparable to that of 50 μM DEX­(Figure c). Together, these results demonstrate that ADCG effectively attenuates inflammatory cytokine production in activated macrophages.

3.8. Effects of ADCG on TNF-α-Induced MH7A Cells

3.8.1. Anti-Migratory in TNF-α-Induced MH7A Cells

Consistent with the observations in RAW264.7 cells, ADCG exhibited no significant cytotoxicity toward MH7A cells, with an IC₅₀ value exceeding 100 μg/mL in both cellular models (Figure a). This concordance across different cell types reinforces the favorable safety profile of ADCG and provides a reliable foundation for its application in anti-inflammatory studies.

9.

ADCG significantly inhibited the proliferation, migration and invasion of TNF-α-induced MH7A cells, (a) Cell viability. (b) Migration distance. (c) Number of invasive cells. (d) The healing area of the wound at 0 and 24 h were photographed by microscopy­(The red dashed box represents the area counted after migration). (e) The migration of TNF-α-induced MH7A cells in Transwell migration assays. NC: negative control group, TNF-α: TNF-α induced model group, MTX: Methotrexate positive control group, ADCG: ADCG administration groups; Note: data are presented as mean ± SD (n = 3); ^## p < 0.01 vs NC group, **p < 0.01 vs TNF-α group.

The antimigratory effect of ADCG was evaluated using wound healing and Transwell assays. In the wound healing assay (Figure b, d), TNF-α stimulation significantly promoted MH7A cell migration relative to the control (p < 0.01), whereas ADCG treatment (5–20 μg/mL) effectively suppressed this TNF-α-induced migration in a concentration-dependent manner (p < 0.01). Correspondingly, the Transwell assay (Figure c,e) confirmed that ADCG (5–20 μg/mL) significantly inhibited TNF-α-driven cell migration after 24 h (p < 0.01). Together, these findings indicate that ADCG possesses notable antimigratory activity in TNF-α-stimulated MH7A cells.

3.8.2. Suppression of Pro-inflammatory Mediators in TNF-α-Induced MH7A Cells

Consistent with their established roles in driving inflammatory responses and immune activation, IL-1β, IL-6, IL-8, and MMP1 were quantified in TNF-α-induced MH7A cells using ELISA. As illustrated in Figure a–d, TNF-α stimulation markedly increased the secretion of these key inflammatory mediatorsIL-1β, IL-6, IL-8, and the tissue-degrading protease MMP1compared to the NC group (p < 0.01). Treatment with ADCG (5–20 μg/mL) significantly and dose-dependently reduced their release (p < 0.01 vs TNF-α group), exhibiting an inhibitory effect comparable to that of the clinical reference MTX (0.1 μM). These results indicate that ADCG not only attenuates pro-inflammatory signaling but also mitigates extracellular matrix degradation in activated synoviocytes.

10.

ADCG significantly inhibited the TNF-α-induced production of IL-6, IL-8, IL-1β, and MMP1 in MH7A cells, (a) IL-6, (b) IL-8, (c) IL-1β, (d) MMP1. NC: negative control group, TNF-α: TNF-α induced model group, MTX: Methotrexate positive control group, ADCG: ADCG administration groups. Note: data are presented as mean ± SD (n = 3); ^## p < 0.01 vs NC group, **p < 0.01 vs TNF-α group.

3.9. Effects of ADCG on Protein Expressions of p65, p-p65, IκBα in TNF-α-Induced MH7A Cells

To validate the predictions from network pharmacology and molecular docking, we examined the effect of ADCG on the NF-κB signaling pathway using Western blotting. The NF-κB pathway, activated by TNF-α binding to its receptors, plays a pivotal role in inflammatory responses by triggering the expression of pro-inflammatory factors. As shown in Figure , stimulation of MH7A cells with TNF-α led to a substantial increase in the phosphorylation of p65 (p-p65) and a marked decrease in the total protein level of IκBα, consistent with canonical pathway activation. The expression of total p65 was also moderately upregulated.

11.

Regulation of p65, p-p65 and IκBα proteins levels by ADCG. (a) P65, p-P65 and IκBα protein bands. (b) The relative expression level of IκBα/GAPDH. (c) The relative expression level of p65/GAPDH. (d) The relative expression level of p-p65/GAPDH. respectively. Note: data are presented as mean ± SD (n = 3);^## p < 0.01 vs NC group, **p < 0.01 vs TNF-α group.

Treatment with ADCG (5–20 μg/mL) for 24 h significantly attenuated the TNF-α-induced activation of NF-κB signaling. Specifically, compared to the TNF-α group, ADCG treatment resulted in a concentration-dependent reduction in the protein levels of both total p65 and p-p65 (Figure a–d). Conversely, the degradation of IκBα was effectively suppressed, as evidenced by a significant increase in its protein level compared to the TNF-α-induced model group.

4. Discussion

This study successfully established and optimized an integrated process for the efficient enrichment of C-glycosides from the Araceae plant A. decipiens. In contrast to previous research on this plant family, which has predominantly focused on alkaloids and nucleosides, this work systematically shifts the focus to its highly polar and challenging-to-isolate C-glycoside constituents. , Previous studies on the fingerprint of A. decipiens conducted by our research group revealed that C-glycosides might be potential quality markers. Reported evidence indicates that C-glycosides possess certain bioactivities. , Therefore, this study focuses on the extraction, isolation, and activity validation of C-glycosides from A. decipiens.

The compound reported in this study possesses a unique dual C-glycosidic bond structure, in which both sugar units are directly linked to the aglycone via carbon–carbon (C–C) bonds. To highlight this structural feature, we have juxtaposed the structures of a representative O-glycoside and a C-glycoside in Figure . Based on the glycosidic bond type, glycosides can be classified into O-, C-, S-, and N-glycosides, among which O-glycosides represent the largest class in nature. Flavonoids typically occur as O-glycosides, with sugar moieties attached to the hydroxyl groups at positions 3, 5, or 7. In contrast, flavonoid C-glycosides are characterized by C–C linkages, usually at the C-6 or C-8 positions. From the perspective of chemical bond stability, the C–C bond is stronger than the C–O–C bond, making it more resistant to hydrolysis during normal in vivo metabolism. As a result, flavonoid C-glycosides maintain high water solubility while also acquiring partial lipophilicitythe latter arising from their larger molecular architecture. The reported biological activities of C-glycosides have largely focused on cardiovascular diseases, with examples such as vitexin, and breviscapine, whose activities are closely related to the chemical properties determined by their structural characteristics. −

12.

Chemical structures of O-glycoside and C-glycoside.

C-Glycosidic bonds occur much less frequently than the common O-glycosidic bonds, and dual C-glycosides bearing two independent C–C glycosidic linkages are even rarer. From a chemical stability perspective, C-glycosidic bonds are significantly more stable than O-glycosidic bonds. The acetal oxygen in O-glycosidic bonds is prone to protonation, leading to heterolytic cleavage under acidic or enzymatic conditions. In contrast, the C–C bond in C-glycosides lacks lone-pair electrons and cannot be activated by protonation, thus exhibiting strong resistance to both acid hydrolysis and conventional glycosidase activity. Studies have shown that C-glycosidic bonds typically resist acidic and enzymatic hydrolysis, remaining stable in simulated saliva, gastric fluid, and ileal fluid, and are only slowly metabolized in the colonic fluid by specific gut microbiota. For instance, C-glycosides such as puerarin and mangiferin remain stable during intestinal transit and absorption, whereas O-glycosides are readily hydrolyzed by gastric acid or gut microbiota. − Furthermore, C-glycosidic bonds confer unique pharmacokinetic advantages. It has been reported that the biological half-life of C-glycosides can be extended compared to conventional O-glycosides. These characteristics make the dual C-glycoside structure an ideal scaffold for drug design and natural product lead optimization, particularly for chronic inflammatory diseases requiring long-lasting effects. Therefore, the dual C-glycosidic bonds revealed in this study not only represent a structural novelty but also directly relate to the compound’s sustained efficacy and clinical translational potential, bearing clear scientific significance and practical value.

A critical focus of this study was the extraction and enrichment methodology. Methodologically, this study first optimized the extraction process via Box-Behnken design-response surface methodology, identifying parameters that yielded a highly predictive and reproducible protocol. More crucially, to address the technical challenges posed by the strong polarity and increase the sample utilization rate to avoid sample waste, we innovatively implemented a repeated-loading strategy during the dynamic adsorption on polyamide resin. This approach effectively utilized the resin’s residual adsorption capacity, concentrating the target compounds by 15.8-fold with a recovery rate of 53.79%. This integrated strategy (freeze-drying, optimized extraction, repeated-loading chromatography) not only overcomes the bottleneck in enriching such components but also provides an efficient and scalable platform for the systematic study of highly polar metabolites in Araceae plants and beyond, representing an advancement not achieved in prior studies. Subsequent chemical characterization of the enriched fraction led to the identification of seven C-glycosides, all derived from chrysin or apigenin aglycone scaffolds, including AGX, AXG, SS, CXG, AGA, ISS, and CGA. The method not only offers a practical solution for studying under-explored polar metabolites in this plant family but may also serve as a valuable reference for similar investigations in other challenging matrix systems.

Although anti-inflammatory mechanisms of individual C-glycosides have been sporadically reported, the integrated mechanism of ADCG as a complex multicomponent mixture remained unexplored. , Our study, employing a combined strategy of animal experiment, network pharmacology prediction, molecular docking, and experimental validation, preliminarily elucidated its multitarget characteristics. We demonstrated ADCG obtained via this optimized method significant anti-inflammatory activity in a chronic cotton pellet-induced granuloma model, providing crucial in vivo pharmacodynamic evidence for further investigation.

Follow-up cellular experiments further revealed that ADCG demonstrated an excellent biosafety profile, with no significant cytotoxicity observed at concentrations up to 100 μg/mL. The present study reveals that ADCG exhibits notable anti-inflammatory properties in LPS-stimulated RAW264.7 murine macrophages. The anti-inflammatory efficacy of ADCG was primarily reflected in its potent, concentration-dependent suppression of key inflammatory mediators. Specifically, ADCG significantly inhibited the LPS-induced overproduction of NO, a critical gaseous mediator involved in inflammatory processes and tissue injury. , Moreover, ADCG markedly reduced the secretion of the major pro-inflammatory cytokines TNF-α and IL-6, with its inhibitory effect on IL-6 comparable to that of the potent synthetic glucocorticoid DEX.

RA is an autoimmune disorder characterized by persistent synovial inflammation and hyperplasia. Among the key pathological drivers are RA-FLS, which exhibit abnormal proliferation, enhanced migratory and invasive behavior, and impaired apoptosis. These cells contribute to joint destruction through the excessive release of inflammatory cytokines, MMPs, and chemokine. Controlling RA progression therefore requires effective strategies to suppress synovial cell proliferation and promote apoptosis. The MH7A cell line, an immortalized human RA-FLS model generated by SV40 T antigen transfection, retains the inflammatory and hyperproliferative features of primary RA-FLS, making TNF-α-stimulated MH7A cells a widely accepted model for RA research. TNF-α potently enhances MH7A proliferation, migration, and invasion, and triggers the release of inflammatory mediators that collectively promote immune dysregulation and tissue injury.

In this study, ADCG significantly suppressed MH7A cell proliferation in a dose-dependent manner, as determined by CCK-8 assay. It also effectively attenuated the migratory and invasive capacities of these cells. Furthermore, ADCG reduced the secretion of inflammatory cytokines IL-6, IL-8, and IL-1β, as well as the tissue-degrading enzyme MMP-1, in a concentration-dependent fashion.

Integrating computational molecular docking with experimental Western blot analysis, we found that ADCG exerts its anti-inflammatory effects largely through multitarget suppression of key signaling nodes, with pronounced inhibition of the canonical NF-κB pathway.

Molecular docking predictions indicated that constituents of ADCG, such as CXG, could bind with high affinity to the RELA/p65 subunit of NF-κB. This computational insight was subsequently validated by Western blot results. This computational insight was subsequently validated by Western blot results. The experimental data confirmed that ADCG treatment effectively blocks TNF-α-induced NF-κB activation, as evidenced by the concentration-dependent suppression of p65 phosphorylation and the concurrent prevention of IκBα degradation. By stabilizing the NF-κB–IκBα complex in the cytoplasm, ADCG prevents the nuclear translocation of the p65-p50 dimer, thereby directly inhibiting the transcription of downstream pro-inflammatory genes.

This inhibition of the NF-κB pathway provides a mechanistic explanation for ADCG’s efficacy in reducing the secretion of IL-1β, IL-6, IL-8, and MMP-1 in MH7A cells, as well as its potent inhibition of synoviocyte migration2014all key pathological processes in RA driven by NF-κB. Furthermore, the docking results suggest this primary effect is potentiated by a polypharmacological strategy, with other constituents engaging additional critical targets like JAK2 and EGFR. Inhibition of JAK2 could potentially suppress the JAK-STAT pathway, further disrupting cytokine signaling and immune cell communication, while EGFR inhibition might interfere with growth factor-mediated cell proliferation and inflammation.

In summary, this study makes significant contributions on three fronts: it develops a novel and efficient process for the specific extraction and enrichment of C-glycosides, offering a valuable methodological reference; it reveals, for the first time, the marked anti-inflammatory and antirheumatoid arthritis activity of the resulting ADCG fraction in relevant experimental models; and it provides preliminary mechanistic elucidation by demonstrating its inhibition of the NF-κB pathway. Together, these findings not only furnish new scientific evidence supporting the development of A. decipiens as a potential anti-RA resource, but also establish a transferable technical framework for studying highly polar bioactive constituents in other plant species. Looking forward, further work should focus on validating the in vivo anti-RA efficacy of ADCG, and their synergistic relationships, and investigating its effects on JAK2 and EGFR phosphorylation and downstream signaling.

5. Conclusions

In conclusion, this study successfully developed and validated an innovative repeated-loading polyamide chromatography strategy, effectively overcoming the long-standing challenge of enriching polar C-glycosides from Araceae plants, and identified seven C-glycoside constituents from A. decipiens. We further demonstrated that ADCG exhibits potent anti-inflammatory and antiarthritic effects in vitro by directly inhibiting the NF-κB signaling pathway and suppressing the proliferative and migratory phenotype of rheumatoid synoviocytes, thereby positioning it as a promising multitarget therapeutic candidate for rheumatoid arthritis and establishing a robust platform for polar natural product discovery.

Glossary

Abbreviations

A. decipiens

Arisaema decipiens

ADCG

C-glycosides extract from Arisaema decipiens

RF

rheumatoid factor

RA

rheumatoid arthritis

C–C

carbon–carbon

MMPs

matrix metalloproteinases

NSAIDs

nonsteroidal anti-inflammatory drugs

DMARD

disease-modifying antirheumatic drugs

FLS

fibroblast-like synoviocyte

BBD

Box-Behnken Design

HPLC

high-performance liquid chromatography

AGX

Apigenin-6-C- d-glucosyl-8-C–D-xyloside

AXG

Apigenin-6-C–D-xylosyl-8-C- d -glucoside

SS

Schaftoside

CXG

Chrysin-6-C-xylopyrano-8-C-galactoside

AGA

Apigenin-6-C-galactosyl-8-C-arabinoside

ISS

Isoschaftoside

CGA

Chrysin-6-C-glucose-8-C-arabopyranoside

Specific pathogen-free

Specific pathogen-free

PBS

Phosphate-buffered saline

DMEM

Dulbecco’s modified Eagle’s medium

CCK-8

cell counting kit-8

NO

Nitric oxide

IL-1&beta

Interleukin-1β

IL-6

Interleukin-6

IL-8

Interleukin-8

MMP-1

Matrix metalloproteinase-1

R

response value

CV

column volume

NC

Normal Control

M

Model

PA

Prednisone acetate

TCMSP

Traditional Chinese Medicine Systems Pharmacology Database

CTD

Comparative Toxicogenomics Database

David

Database for Annotation, Visualization, and Integrated Discovery

PPI

protein–protein interaction

FBS

fetal bovine serum

DEX

dexamethasone

MTX

Methotrexate

WBC

white blood cells

LYM

lymphocytes

MON

monocytes

p-p65

phosphorylation of p65

IC50

half-maximal inhibitory concentration

Institutional Review Board Statement Not applicable.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.5c13026.

• Figure S1: Determination of Four C-Glycosides in ADCG by High-Performance Liquid Chromatography: 1. SS, 2. CXG, 3. ISS, and 4. CGA; Figure S2: The 3D (left and center) and 2D (right) binding conformations between the C-glycosides compounds and the core target­(a) AGX-JAK2, (b) SS-AKT1, (c) CGA-EGFR, (d) SS-EGFR, (e) AGA- JAK2, (f) SS - JAK2, (g) AXG-MAPK8, (h) CXG-RELA, yellow dashed lines represent hydrogen bonding interactions. Table S1. Theoretical leakage of 6 column volumes with repeat loading samples; Table S2. Theoretical leakage of 7 column volumes with repeat loading samples; Table S3. Theoretical leakage of 8 column volumes with repeat loading samples; Table S4 Identification information on seven C-glycosides (PDF)

X.H.: conceptualization, methodology, formal analysis, investigation, writingoriginal draft. X.Z.: project research, data original draft, and analysis. S.W.: methodology, writingreview and editing, funding acquisition, resources.W.D.: software, visualization, supervision. L.H.: validation, visualization. J.D.: validation, investigation.

This research was funded by Guizhou Provincial Basic Research Program­(Natural Science) (QianKeHe Foundation [2020]­1Y375), Guizhou Province Traditional Chinese Medicine Processing Technology Inheritance Base Construction Project (Qian ZhongYao Han [2024] No. 22), Research Center for Processing and Processing Techniques of Chinese Medicinal Materials, Guizhou University of Traditional Chinese Medicine (Gui ZhongYi ZX Hezi[2024]­047).

The authors declare no competing financial interest.