Fresh flowers of Nymphaea “Eldorado” were sourced from the Liuzhou Shixian Planting and Breeding Cooperative (Liuzhou, China). Fresh flowers on the first day of flowering were collected and washed with water (Supplementary Figure S1A). The flowers were placed in a container filled with water and exposed to daylight for 24 h. The flowers were opened, and stalks and sepals were removed. The ovary was cut by 1/3, and the flowers were dried at 70°C–80°C until the moisture content was less than 10% (Supplementary Figure S1B). The dried flowers were crushed in a grinder, and the powder was sieved through a 20-mesh sieve and stored at −20°C for later use.

The extract (5,120 mg) was placed in a 1 L flask and mixed with 400 mL of water (yielding a solid-to-solvent ratio of 1:78, w/v) at 100°C. After standing for 30 min at room temperature, the mixture was centrifuged. The supernatant was collected, and the pellet was resuspended in 400 mL of water at 100 C. After standing for 30 min, the sample was centrifuged as described above. The supernatant was collected and combined with the first supernatant. The stock solution (6,400 mg/L) was stored at −20°C.

The stock solution was diluted to obtain working concentrations (100, 200, 400, 800, 1600, 3200, and 6,400 mg/L) (Supplementary Figure S1). At 72 h post-fertilization (hpf), AB strain wild-type zebrafish embryos were placed in 24-well plates with 10 embryos per well and three replicates per concentration (Supplementary Figure S1C). The culture medium was removed, and 2 mL of NEWE was added to each well. Based on the results of these assays, NEWE concentrations that were not toxic to zebrafish embryos were used in subsequent experiments (Supplementary Figure S1). All procedures involving animals were approved by the Animal Care and Use Committee of the Guangxi Academy of Agricultural Sciences and complied with Chinese government regulations.

This study used the transgenic zebrafish line Tg(mpx:GFP), which expresses GFP under the neutrophil-specific myeloperoxidase promoter, as described previously (Renshaw et al., 2006). Zebrafish embryos were placed in 90 mm dishes containing 1×E3 medium (pH was adjusted to 7.0–7.2 using 5 mM NaCl, 0.33 mM CaCl₂, 0.17 mM KCl, and 0.33 mM MgSO₄) with 100 embryos per dish. At 24 hpf, dead embryos were removed, and healthy embryos were added. At 72 hpf, normal embryos were transferred to 24-well culture plates with 10 embryos per well and incubated with 20 μM CuSO₄ + NEWE (0, 25, 50, 75, 100 μg/mL) for 2 h at 28°C ± 1°C to induce neutrophil migration. GFP-expressing neutrophils in the anterior yolk sac were observed under a fluorescence microscope (Axio Imager M2, Zeiss) at 2 h after CuSO₄ + NEWE treament to evaluate the degree of inflammation.

WT zebrafish embryos were incubated with NEWE (0, 25, 50, 75, 100, 1255 μg/mL) or 100 μM quercetin (positive control) for 1 h at 28°C ± 1°C, followed by treatment with 10 μM CuSO₄ for 20 min as described in Section 2.3. After washing with culture medium, embryos were incubated with 20 μg/mL of the fluorescent indicator 2′,7′-dichlorodihydrofluorescin diacetate (DCFH-DA) for 1 h at 28°C ± 1°C. CuSO₄-induced ROS in embryos was detected by the formation of the oxidized compound dichlorofluorescein by fluorescence microscopy (Axio Imager M2, Zeiss) (Andoh et al., 2006; Girard-Lalancette et al., 2009; Zhang et al., 2023a; Zhang et al., 2023b).

Tg(mpx:GFP) zebrafish embryos were incubated with 20 μM 20 μM CuSO₄+NEWE 2 h at 28°C ± 1°C as described in Section 2.3. RNA from each sample (50 μg/mL) was extracted using TRIzol and reverse transcribed using the PrimeScript RT reagent kit. qRT-PCR was performed using the TB Green Premix Ex Taq II reagent kit (TaKaRa, Dalian, China) on the LightCycler 480 II System (Roche, Indianapolis, IN, United States). The cycling conditions were as follows: initial denaturation at 95°C for 30 s, followed by 40 cycles of 95°C for 5 s and 60°C for 20 s. Melting curves were generated by holding at 95°C for 5 s and 65°C for 1 min and increasing to 95°C at a rate of 0.1°C per second. The primers used in this study are listed in Supplementary Table S1. Relative gene expression levels were quantified using the 2^−ΔΔCT method (Livak and Schmittgen, 2001). The data were visualized using Origin 2021 (OriginLab, Northampton, MA, United States).

Tg(mpx:GFP) zebrafish embryos were cultured as described in Section 2.3. Three treatment groups were established: control (untreated), model (treated with CuSO₄), and treatment (treated with 20 µM CuSO₄ + 50 μg/mL NEWE).

The construction of sequencing libraries and RNA-seq were performed using the Illumina HiSeq platform (Biomarker Biotechnology Corporation Co., Beijing, China). High-quality reads were assembled using StringTie. Genes with log2 fold-change >2.0 or < −2.0 and false discovery rate (FDR) < 0.01 were considered differentially expressed genes (DEGs). DEGs were identified using edgeR (https://bioconductor.org/packages/release/bioc/html/edgeR.html). Relative transcript abundance was measured as fragments per kilobase of exon per million fragments mapped using Cuffdiff, with an FPKM threshold set at a minimum of 1.0 to ensure the reliability of expression levels. Histograms, scatter plots, volcano plots, Venn diagrams, and heatmaps were generated on the Biomarker Cloud Platform (Biomarker Biotechnology Corporation Co., Beijing, China) or https://hiplot.com.cn/cloud-tool/drawing-tool/list.

The stock solution was diluted to 50 mg/L in culture medium. Metabolomics analyses were performed by MetWare Biotechnology (Wuhan, China). Metabolite extraction, identification, and quantification were performed as described previously (Yuan et al., 2018). Samples were freeze-dried, ground into powder, and dissolved in 70% methanol as internal standard (1:30 v/v). The samples were centrifuged at 12,000 rpm for 3 min at 4 C, and the supernatant was filtered through a 0.22 μm membrane for analysis.

Metabolites were analyzed by ultra-performance liquid chromatography (Exion LC AD System, AB Sciex, Darmstadt, Germany) coupled with electrospray tandem mass spectrometry (ESI-Q TRAP-MS/MS (AB SCIEX Q TRAP 4000, Applied Biosystems, Foster City, CA, United States). Chromatographic separation was performed on an SB-C18 column (1.8 µm, 2.1 mm × 100 mm, Agilent). Elution was carried out using a system consisting of solvent A (0.1% formic acid in water) and solvent B (0.1% formic acid in acetonitrile) and a linear gradient from 95% A to 95% B in 9 min, 95% B for 1 min, 95% B to 95% A in 1.1 min, and 95% A in 2.9 min. The flow rate was 0.35 mL/min at 40°C, and the injection volume was 2 μL. The ESI conditions were as follows: source temperature, 550 C; ion spray voltage (IS), 5500 V (positive mode)/-4500 V (negative mode); gas pressure of 50, 60, and 25 psi for GS1, GS2, and CUR gases, respectively. The collision-activated dissociation was set to high. Multiple reaction monitoring mode was utilized for metabolite detection, with declustering potential and collision energy individually optimized for each transition.

Metabolites abundance was categorized into two groups (classes I and II) based on peak areas.

Genes potentially targeted by metabolites were identified using the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (https://old.tcmsp-e.com/tcmsp.php) and SwissTargetsPrediction (http://www.swisstargetprediction.ch/). Drug-target networks integrating transcriptomic and metabolomics data were constructed using Cytoscape version 3.9.1 (Li et al., 2022).

The molecular structure of L-pyroglutamic acid, monomyristin, and protocatechuic acid were obtained from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/). The 3D structure of the target protein, Serpine1, also known as plasminogen activator inhibitor-1, was obtained from the Protein Data Bank (https://www.rcsb.org/structure/4DTE). Protein-metabolite interactions were predicted using AutoDock Tools version 1.5.7, which applies a semi-flexible docking approach to predict binding affinities by exploring ligand and receptor flexibility. The docking grid was centered on the protein’s active site, with grid spacing and box size parameters optimized to encapsulate potential binding pockets. For each ligand, binding modes were generated based on the lowest binding energy utilizing the Lamarckian Genetic Algorithm as the primary search strategy, which has been validated in several studies for its efficacy in docking simulations. The AutoDock results were visualized and analyzed using PyMOL version 4.6.0 (Morris et al., 2009; Seeliger and de Groot, 2010), allowing detailed observation of ligand binding conformations, hydrogen bonding, and hydrophobic interactions, providing insights into the stability and specificity of ligand binding to Serpine1.

Data normality was assessed using the Kolmogorov–Smirnov test. Normally distributed continuous variables were expressed as mean ± standard deviations, and the significance of differences between means was determined using paired Student’s t-test or one-way analysis of variance followed by Tukey’s post hoc test. Differences between groups were compared using the Kruskal–Wallis test and Dunn–Sidak multiple comparison test. Box plots were drawn using Origin 2021 (OriginLab, Northampton, MA, United States). P-values of less than 0.05 were considered statistically significant.

Acute toxicity assessment revealed that the 24-h LC₅₀ of NEWE for 72 h post-fertilization (hpf) zebrafish embryos was 3200 μg/mL (Supplementary Figure S1D). Based on this safety profile, we evaluated the anti-inflammatory potential of safety concentrations of NEWE using a copper sulfate (CuSO₄) induced inflammation model in 3 days post-fertilization (dpf) larvae, this model utilizes the chemotactic properties of CuSO₄ to disrupt neutrophil homeostasis, enabling real-time tracking of GFP-labeled neutrophil migration toward lateral line neuromasts - a validated method for quantifying inflammatory responses in zebrafish (Renshaw et al., 2006). As show in Figure 1A, neutrophils primarily localized to the caudal hematopoietic tissue in the control group, consistent with normal distribution patterns (Zhang et al., 2023b), but CuSO₄ treatment prompted neutrophil migration towards the horizontal midline, where they formed clusters near the lateral line neuromasts (Figure 1B), indicating a pronounced inflammatory response. However, treatment with varying concentrations of NEWE effectively inhibited CuSO₄-induced neutrophil migration in a does dependent manner, suggesting a potential anti-inflammatory activity (Figure 1B). Further, we investigated the effect of NEWE on expression of inflammatory genes using qRT-PCR, and found CuSO₄ treatment significantly upregulated the expression of inflammatory genes, interleukin 1 beta (il1b), nterleukin 11a (il11a), C-X-C motif chemokine ligand 8a (cxcl8a), and nuclear factor kappa B inhibitor alpha a (nfkbiaa). Specifically, il1b expression increased by 37.22-fold, il11a by 9.67-fold, cxcl8a by 4.59-fold, and nfkbiaa by 3.91-fold compared to controls, respectively. However, treatment with NEWE could restore the CuSO₄-indecued inflammatory gene expression. For example, in the presence of 100 μg/mL NEWE CuSO₄ induced il1b expression was reduced to 8.78-fold from 37.22-fold (Figures 1C–F). Moreover, NEWE could restore the CuSO₄-induced expression of il11a, cxcl8a and nfkbiaa to control level (Figures 1C–F). Collectively, our findings demonstrate that NEWE could effectively mitigate CuSO₄-induced inflammation by reducing neutrophil recruitment to inflammatory site and downregulating inflammatory gene expression, highlighting its potential anti-inflammatory agent.

The antioxidant activity of NEWE was assessed by fluorescence microscopy and qRT-PCR analysis. The DCFH-DA fluorescence probe operates through a redox-sensitive mechanism: non-fluorescent DCFH-DA passively diffuses across cell membranes and is deacetylated by intracellular esterases to form DCFH, which becomes fluorescent DCF upon oxidation by reactive oxygen species (ROS) (Zhang et al., 2023b). This oxidation process generates fluorescence intensity proportional to ROS levels, allowing semi-quantitative assessment of oxidative stress in vivo (Nguyen et al., 2020). Fluorescence imaging showed that CuSO₄ remarkably increased oxidative stress in zebrafish larvae, whereas 100 μM quercetin (positive control) and different concentrations of NEWE reduced fluorescence intensity (Figures 2A,B), suggesting NEWE possesses antioxidant properties by reducing CuSO₄-induced oxidative stress levels. The qRT-PCR results showed that CuSO₄ exposure led to an upregulation of the oxidative stress-related genes, ptgs2a and ptgs2b, which encode prostaglandin synthase 2a and 2b, respectively (Ishikawa et al., 2007). Specifically, CuSO₄ induced expression of ptgs2a to 4.1-fold, and ptgs2b to 37.42, relative to control, respectively. In contrast, NEWE treatment at 25–125 μg/mL effectively reversed these effects, reducing ptgs2a and ptgs2b expression (p < 0.05, Figures 2C,D). These findings demonstrate that NEWE mitigates CuSO₄-induced oxidative stress in zebrafish larvae.

In order to examine the molecular mechanism of NEWE’s anti-inflammatory, we performed RNA-seq to profile global gene expression changes. Volcano plots showed that distinct gene expression patterns across the control, model group (treated with CuSO₄), and treatment groups (CuSO₄ + 50 μg/mL NEWE) (Figures 3A–C). A total of 339 differentially expressed genes (DEGs) was shared across the three groups and were selected for further functional annotation (Figure 3D), genes previously validated by qRT-PCR - including il1b (interleukin 1 beta), ptgs2a/b (prostaglandin-endoperoxide synthase 2), and cxcl8a (C-X-C motif chemokine ligand 8a) - emerged as core regulatory nodes (Figure 3H). Gene Ontology (GO) analysis showed that the selected DEGs were enriched in several biological processes, cellular components, and molecular functions (Figures 3E,F). Notably, genes associated with inflammatory response were significantly enriched, indicating their potential role in inflammatory pathways. Additionally, genes related to antioxidant activity were differentially expressed (Figures 2C,D), suggesting that antioxidant mechanisms may help mitigate inflammation by NEWE. Thus, future research should screen for DEGs related to both inflammatory and antioxidant activities to elucidate molecular mechanisms underlying NEWE’s anti-inflammatory effects. A heatmap showed that the expression profiles of these DEGs varied between different treatments (Figure 3G). A protein-protein interaction (PPI) network analysis indetified 30 hub DEGs as potential targets of NEWE, of particular significance, stat3, serpine1, and mmp9 formed a tightly interconnected subnetwork that bridges inflammatory signaling and extracellular matrix reorganization (Figure 3H). These hub genes exhibited strong co-expression patterns with qRT-PCR validated targets, confirming their central role in NEWE-mediated anti-inflammatory responses.

To further understand the molecular mechanism of NEWE in anti-inflammatory pathway, widely targeted metabolomic analysis was employed to identify potential compounds involved in NEWE-mediated anti-inflammatory (Supplementary Figure S2). A total of 891 compounds were identified and quantified, revealing a high degree of chemical diversity in the extract (Figures 4A,B). Flavonoids (172 compounds), alkaloids (85 compounds), lipids (97 compounds), and phenylpropanoids (142 compounds) were predominant. Subsequent network pharmacology analysis predicted potential biochemical interactions between these chemical compounds and target genes (Figures 4C–F; Supplementary Tables S2-S5). These interactions indicate the potential pharmacological relevance of these compounds and the importance of understanding these interactions to elucidate NEWE’s pharmacological effects.

To investigate the potential mechanisms by which NEWE’s diverse chemical components exert their anti-inflammatory effects at the molecular level, we conducted an integrated analysis of our metabolomics and transcriptomics data, focusing on the identified hub DEGs and their predicted interactions with the various classes of compounds present in NEWE. Venn diagrams (Figures 5A–D) showed that among the 30 hub DEGs associated with NEWE treatment, several genes were consistently targeted by different classes of compounds within NEWE, including flavonoids, amino acid derivatives, lipids, and phenylpropanoids. Specifically, 6 key genes, including il1b, stat3, serpine1, mmp9, ptgs2a, and ptgs2b showed overlap, and the recurring presence across multiple compound-targeted groups suggests that these genes play pivotal roles in the response to NEWE treatment. Figure 5E depicts a network pharmacology map linking these six genes to various chemical constituents, underscoring their potential importance in mediating NEWE’s therapeutic effects. Additionally, RNA-seq expression profiling (Figure 5F) shows that the expression patterns of these core genes were distinct across the control, model, and treatment groups and that treatment normalized the gene expression disrupted by CuSO₄ exposure. These findings collectively suggest that the six hub genes (il1b, stat3, serpine1, mmp9, ptgs2a, and ptgs2b) form a coordinated regulatory network through which multiple bioactive compounds in NEWE exert polypharmacological effects, simultaneously targeting inflammatory signaling, oxidative stress responses, and extracellular matrix remodeling. This multi-target engagement mechanism provides a molecular rationale for the observed anti-inflammatory efficacy of NEWE and highlights these genes as potential biomarkers for evaluating plant-derived therapeutics.

To validate the transcriptional regulation of key inflammatory mediators identified through multi-omics integration, we quantified the expression dynamics of six core target genes using FPKM (Fragments Per Kilobase per Million mapped reads) values derived from RNA-seq data. This normalized metric accounts for both transcript length and sequencing depth, providing a reliable measure of relative gene expression levels across experimental groups. Upon treatment with CuSO₄, six inflammatory genes were significantly upregulated: ptgs2a increased from 9.16 to 72.62 FPKM (approximately 7.9-fold), serpine 1 from 6.81 to 43.64 FPKM (around 6.4-fold), il1b from 0.42 to 50.39 FPKM (about 120-fold), mmp9 from 11.31 to 245.92 FPKM (approximately 21.7-fold), ptgs2b from 2.35 to 76.95 FPKM (around 32.8-fold), and stat3 from 10.85 to 26.90 FPKM (approximately 2.5-fold). Notably, il1b exhibited the most dramatic induction, consistent with its role as a primary cytokine driver in copper-induced inflammation. Treatment with NEWE reversed these effects, significantly reducing gene expression in comparison to the CuSO₄ group. These results indicate that NEWE possesses a potential anti-inflammatory effect, mitigating the overexpression induced by CuSO₄.

To validate the network pharmacology prediction that Serpine1 (plasminogen activator inhibitor-1) acts as a critical mediator of NEWE’s bioactivity, we conducted molecular docking analyses on three representative compounds identified in the metabolomic profile: L-pyroglutamic acid, monomyristin, and protocatechuic acid (Supplementary Table S2). As illustrated in Figure 7, docking revealed specific interactions within the Serpine1 substrate-binding cleft: L-pyroglutamic acid formed hydrogen bonds with Thr-183, Arg-348, and Lys-235 (binding energy = −2.67 kcal/mol); monomyristin bound Asp-209 and Lys-311, positioning its acyl chain in the hydrophobic S4 pocket (binding energe = −0.67 kcal/mol); protocatechuic acid engaged Asp-209 and Asn-243 via hydrogen bonds (binding energy = −1.26 kcal/mol). These residues are critical for the structural integrity of the reactive center loop.

The functional relevance of these predicted interactions was assessed by measuring the compounds’ effects on serpine1 gene expression in the zebrafish inflammation model. Treatment with L-pyroglutamic acid (0.20 mg/mL) significantly downregulated serpine1 expression (Figure 7G). Similarly, protocatechuic acid induced a dose-dependent and significant downregulation at concentrations of 0.01, 0.06, and 0.10 mg/mL (Figure 7I). In contrast, monomyristin treatment did not significantly alter serpine1 expression levels (Figure 7H).

The significant downregulation of serpine 1 by L-pyroglutamic acid and protocatechuic acid provides functional biological evidence supporting their predicted interaction with Serpine1 through molecular docking, aligning with the 6.4-fold serpine1 downregulation observed with whole NEWE extract (Figure 6B). This suggests these compounds contribute to NEWE’s anti-inflammatory effect, potentially through direct or indirect interference with Serpine1 function or its regulation.