Arundina graminifolia (D. Don) Hochr samples used in the present study were collected from the Yunnan Branch of the Institute of Medicinal Plant Development (Chinese Academy of Medical Sciences, Jinghong, China). It was identified as Arundina graminifolia (D. Don) Hochr. by Research Professor Lixia Zhang (Yunnan Branch of the Institute of Medicinal Plant Development, Chinese Academy of Medical Sciences, Jinghong, China). The appraisal report is shown in Fig. S1.

The following reagents were used in the present study: Resveratrol (cat. no. 111535; National Institutes for Food and Drug Control), pterostilbene (cat. no. 455753; J&K Chemical Ltd.), acetonitrile [American Chemical Society/high-performance liquid chromatography (HPLC) grade; cat. no. AH015-4HC; Honeywell International Inc.], methanol (HPLC grade; cat. no. 01104351; Adamas-Beta^® Ltd.), formic acid (HPLC grade; cat. no. F112034; Shanghai Aladdin Biochemical Technology Co., Ltd.), distilled water [Watsons Water; AS Watson Group (HK) Ltd.], Cis (cat. no. 275688; J&K Chemical Ltd.), H&E (cat. no. G1120) and Masson staining kit (cat. no. G1346; both from Beijing Solarbio Science & Technology Co., Ltd.), kidney injury molecule 1 (KIM-1)/hepatitis A virus cellular receptor 1 rabbit polyclonal antibody (cat. no. 30948-1-AP) and lipocalin-2/neutrophil gelatinase-associated lipocalin (NGAL) rabbit polyclonal antibody (cat. no. 30700-1-AP; discontinued; both from ProteinTech Group, Inc.), mouse IL-6 (cat. no. YJ063159), IL-1β (cat. no. YJ35346) and TNF-α (cat. no. YJ002095) ELISA kits (all from Shanghai Yuanju Bio-Technology Co., Ltd.) migration inhibitory factor (MIF) recombinant rabbit monoclonal antibody (cat. no. ET1703-89; HUABIO), MIF rabbit monoclonal antibody (cat. no. A22623; ABclonal, Inc.), phosphorylated (p)-NF-κB rabbit polyclonal antibody (cat. no. WL02169), IL-6 (cat. no. WL02841), IL-1β (cat. no. WL02257) and TNF-α rabbit polyclonal antibody (cat. no. WL01581; all from Wanleibio Co., Ltd.), S-vision polymer secondary antibody for immunohistochemistry (IHC; goat anti-rabbit; cat. no. G1302; Wuhan Servicebio Technology Co., Ltd.), sodium citrate antigen retrieval buffer (cat. no. G1219-1L; Wuhan Servicebio Technology Co., Ltd.), anti-NF-κB p65 rabbit polyclonal antibody (cat. no. GB11997-100; Wuhan Servicebio Technology Co., Ltd.), HRP-conjugated β-actin recombinant rabbit monoclonal antibody (cat. no. PSH03-63; HUABIO) and HRP-conjugated goat anti-rabbit IgG polyclonal antibody (cat. no. HA1001; HUABIO).

A total of 100 g BaiYangJie powder was added to 1,000 ml methanol and treated ultrasonically (800 W; 40 kHz; 25±3˚C) for 30 min, filtered and concentrated under reduced pressure. The concentrated solution was stored overnight at -80˚C, then freeze-dried to a powder (pro-4055/4085 freeze-dryer; Nanjing Genscience Instrument and Equipment Co., Ltd.) and stored at 4˚C. A total of 20 mg freeze-dried methanol extract powder was added to a 100 ml volumetric flask and methanol was added to make up the volume to 100 ml. Then, the mixture was sonicated and finally filtered through a 0.22 µm microporous membrane for further analyses.

Preparation of the reference substance. A total of ~1 mg resveratrol and pterostilbene were added to 100 ml volumetric flasks. Methanol was added to dilute to scale and make up a total volume of 100 ml to obtain a 10 µg/ml standard solution. This solution was filtered through a 0.22 µm microporous membrane for further analysis.

All animal experiments were approved by the Institutional Ethics Review Committee of the Yunnan Branch of the Institute of Medicinal Plant Development, Chinese Academy of Medical Sciences (Jinghong, China; approval no. 20240213001). The present study was strictly performed in accordance with the guidelines of the National Research Council Guide for the Care and Use of Laboratory Animals (13). Male Kunming (KM) mice [weight: 20-22 g; aged 8 weeks; Sibefu (Beijing) Biotechnology Co., Ltd.] were housed in SPF environments with a controlled temperature (22±2˚C), relative humidity (50±5%), atmospheric pressure (101±2 kPa) and a 12-h light/dark cycle, before being allowed to acclimatise for 7 days, during which animals were provided with ad libitum access to water and standardised feed. A total of 30 KM mice were randomly allocated into three experimental cohorts (n=10/group): i) Nor [0.5% sodium carboxymethyl cellulose (CMC-Na), intragastrically], ii) Cis (Cis 13 mg/kg, intraperitoneally) and iii) MEAG (800 mg/kg, intragastrically). The MEAG cohort received daily oral gavage for 10 consecutive days, while the mice in the Nor group were given equivalent volumes of 0.5% CMC-Na solution (0.1 ml/10 g). On day 7, Cis-induced nephropathy was established in the Cis through a single intraperitoneal dose (13 mg/kg). Biological specimens (blood and kidney tissues) were collected on day 10. Mice were anaesthetised with 4% isoflurane in an induction chamber until loss of consciousness and then maintained at a concentration of 1-3% using a mask. Once the toe pinch reflex disappeared for >60 sec, breathing became deep and slow and there was no limb contraction or respiratory change upon strong pinching, the head was fixed and the jugular vein was gently compressed to engorge. A capillary tube was inserted into the inner canthus of the eye to collect blood, with the volume reaching 20-30% of the total blood volume (0.5-0.8 ml for 25 g mice). Immediately after blood collection, euthanasia was performed (under deep anaesthesia, rapid cervical dislocation), To clarify, cervical dislocation rather than exsanguination (blood loss) was used for euthanasia as waiting for natural mortality due to blood loss was prohibited. Mortality was determined when breathing and heartbeat stopped for ≥2 min and the pupils dilated and fixed. The left kidney was collected and fixed in 4% paraformaldehyde at 24˚C for 24 h for histopathological examination, while the right kidney was rapidly frozen in liquid nitrogen and stored at -80˚C for molecular biological experiments.

Serum aliquots (80 µl) were mixed with 320 µl methanol/acetonitrile (ACN) solution (1:1; v/v) followed by 30 sec vortex-mixing. The mixture underwent cryoprecipitation at -20˚C for 60 min. Subsequent centrifugation (4˚C; 16,100 x g; 15 min) yielded supernatant that was lyophilised using the Bionoon-VAC1 vacuum concentrator (Shanghai Bionoon Biotechnology Co., Ltd.) at 4˚C. The lyophilised products were reconstituted in 160 µl ACN/water (1:1; v/v) solution. Quality control (QC) samples were prepared by pooling 10 µl aliquots from each reconstituted sample. Prior to analysis, all specimens were filtered through 0.22 µm membrane filters to remove particulate matter.

Chromatographic separation was achieved using an Acquity HSS T3 column (2.1x100.0 mm; 1.8 µm; Waters China Ltd.) on a Shimadzu LC400 system (Shimadzu Scientific Instruments) coupled with AB SCIEX X500B Q-TOF mass spectrometer (SCIEX) operating in dual electrospray ionization mode. The eluent system comprised of (A) 0.1% aqueous formic acid and (B) 0.1% formic acid in acetonitrile. The following parameters were used: Flow rate, 0.3 ml/min; column oven, 35˚C; injection volume, 5 µl; nitrogen gas temperature, 350˚C; and nitrogen nebuliser flow rate, 10 l/min. Mass detection employed information-dependent acquisition mode with the following settings: Primary MS spectrometry scan, m/z 100-1,200; and secondary tandem MS scan, m/z 50-1,200. The top 15 intense ions were selected for fragmentation.

For the MEAG samples, the mobile phase gradient was 0-2 min, 5% B; 2-5 min, 5-20% B; 5-8 min, 20-40% B; 8-10 min, 40-60% B; 10-15 min, 60-100% B; 15-19 min, 100% B; 19-20 min, 100-5% B; and 20-23 min, 5% B. Nitrogen gas was used in each gas path and ionisation voltage was 5,500 V. Sprayer pressure was 60 psi, the auxiliary heating gas was 60 psi, the air curtain gas was 35 psi, the cone hole voltage was 100 V and the collision energy was 10 V.

For the metabolomics samples, the mobile phase gradient was 0-2 min, 5% B; 2-8 min, 5-50% B; 8-12 min, 50-80% B; 12-14 min, 80-95% B; 14-17 min, 95% B; 17-18 min, 95-5% B; and 18-21 min, 5% B. Nitrogen gas was used in each gas path. The ionisation voltage was 4,500 V, the sprayer pressure was 60 psi, the auxiliary heating gas was 60 psi, the air curtain gas was 35 psi, the cone hole voltage was 80 V and the collision energy was 20 V.

Active components were initially screened using SwissADME (http://www.swissadme.ch) with dual criteria: Oral bioavailability ≥30% and drug-likeness. Potential therapeutic targets were identified through SwissTargetPrediction (swisstargetprediction.ch). Cis-AKI targets were systematically retrieved from five major databases: Online Mendelian Inheritance in Man (OMIM; https://omim.org/), DrugBank (version 5.1.10; https://go.drugbank.com/), GeneCards (version 5.22; https://www.genecards.org/), Therapeutic Target Database (TTD; https://db.idrblab.net/ttd/; accessed on October 31, 2023) and DisGeNET (version 24.0; https://www.disgenet.org/). The targets shared by MEAG components and Cis-AKI were identified using the Venn analysis module in SRplot (bioinformatics.com.cn).

Protein-protein interaction (PPI) networks were constructed using Search Tool for the Retrieval of Interacting Genes/Proteins (STRING; version 11.5; https://cn.string-db.org/), with species restricted to Homo sapiens and a confidence cutoff >0.4. Following network visualisation in Cytoscape (14) (version 3.7.1), the ‘CytoNCA’ plugin (version 2.4.6) was used to prioritise nodes based on multidimensional centrality metrics: Degree, betweenness and closeness. A network of 12 metabolites was constructed using the Search Tool For Interactions Of Chemicals (https://stitch-db.org/) and Cytoscape to represent the metabolic module regulated by MEAG. The consensus targets underwent comprehensive functional annotation using Kyoto Encyclopedia of Genes and Genomes (KEGG) (15,16) pathways and Gene Ontology (GO) enrichment analyses in Metascape (metascape.org), with organism specification restricted to Homo sapiens. Based on the number of gene enrichments, diagrams were drawn using SRplot.

Original data for each sample were processed to obtain peak alignment, peak extraction, peak comparison and standardisation using MS DIAL software (version 4.9.221218) as previously described (17). The online MetaboAnalyst (version 6.0; https://www.metaboanalyst.ca/) database was used for principal component analysis (PCA) and orthogonal partial least squares discriminant analysis (OPLS-DA). By analysing differences between the Nor group and Cis group and between the Cis group and MEAG group, differential metabolites [variable importance in the projection (VIP)>1; P<0.05] were identified and imported into MetaboAnalyst for enrichment and metabolic pathway analyses.

Serum creatinine (Scr) and blood urea nitrogen (BUN) concentrations were quantified using the SMT-120VP clinical chemistry system (Chengdu Seamaty Technology Co., Ltd.) with a seven-parameter renal function assay kit (cat. no. AW03076; Chengdu Seamaty Technology Co., Ltd.), according to the manufacturer's protocol.

Kidney specimens were fixed in 4% paraformaldehyde at 24˚C for 24 h, followed by sequential dehydration at room temperature and paraffin embedding at 60˚C for 2 h. Subsequently, 4 µm microtome sections were prepared. A total of three tissue slices per sample were deparaffinized with xylene at room temperature for 15 min twice, rehydrated through a graded ethanol series at room temperature and rinsed with distilled water. For H&E staining, the sections were stained with haematoxylin at room temperature for 5 min, rinsed with running tap water, differentiated, blued and then counterstained with eosin at room temperature for 2 min. For Masson's trichrome staining, the sections were stained according to the manufacturer's protocol. Briefly, the sections were stained with haematoxylin at room temperature for 5 min, stained with Masson's ponceau acid fuchsin solution at room temperature for 5 min, differentiated with phosphomolybdic acid solution at room temperature for 2 min and counterstained with aniline blue solution at room temperature for 2 min. After staining, all sections were dehydrated, cleared and mounted with neutral resin. Renal morphological alterations were visualised using an Olympus BX53 bright-field microscope at x200 and x400 magnification (Olympus Corporation). The glomerular architecture, proximal convoluted tubule morphology, basement membrane characteristics of glomeruli and collagen deposition patterns were observed.

Circulating concentrations of TNF-α, IL-1β and IL-6 in mouse serum were quantified using commercially available ELISA kits for mice according to the manufacturer's protocols.

Following dewaxing, the sections were rehydrated through a graded ethanol series and rinsed with PBS. Antigen retrieval was performed by boiling samples in sodium citrate antigen retrieval buffer at 100˚C for 10 min. Subsequently, tissues were treated with a quenching reagent (3% hydrogen peroxide solution) at room temperature for 5 min to block endogenous peroxidase activity, followed by blocking with 10% goat serum (Wuhan Servicebio Technology Co., Ltd.) for 60 min at room temperature. The goat serum was discarded and KIM-1 rabbit polyclonal antibody (1:300), NGAL rabbit polyclonal antibody (1:200) and MIF rabbit monoclonal antibody (1:800) were applied to the sections and incubated overnight (~12 h) at 4˚C. Next, secondary antibody conjugation was performed using S-vision polymer secondary antibody for IHC at room temperature for 60 min before the sections were washed three times with PBS. Signals were visualised using diaminobenzidine at room temperature for 3 min, counterstained with haematoxylin at room temperature for 1 min, dehydrated through a graded ethanol series, cleared with xylene and mounted. Tissues were viewed using standard bright-field illumination.

Renal tissues were homogenised in lysis buffer (cat. no. CW2333; CoWin Biosciences, Inc.) supplemented with 1% protease/phosphatase inhibitors, followed by a 1 h incubation to ensure complete lysis and subsequent centrifugation (16,100 x g; 20 min; 4˚C). Protein quantification was performed using a BCA protein detection kit (cat. no. CW0014; CoWin Biosciences, Inc.). To resolve proteins, a 10% SDS-PAGE gel was used and 30 µg of protein was loaded per lane before electrophoresis; the resolved proteins were transferred to PVDF membranes. Membranes were blocked with 5% non-fat milk/TBST (at 25˚C for 2 h; 0.05% Tween-20) prior to overnight incubation (4˚C) with anti-MIF (1:1,000), anti-p-NF-κB (1:2,000), NF-κB (1:1,000) anti-IL-6 (1:2,000), anti-IL-1β (1:2,000) and anti-TNF-α (1:2,000). After three TBST washes, membranes were incubated with an HRP-conjugated secondary antibody (25˚C for 2 h; 1:50,000). with β-actin as the internal reference. Signals were visualised with ECL reagents (New Cell & Molecular Biotech Co., Ltd.) and band intensities were quantified using an imaging system from Tanon Science and Technology Co., Ltd. with Image ProPlus (version 5.0; Media Cybernetics, Inc.; https://www.mediacy.com/).

Statistical analysis was performed utilising GraphPad Prism (version 9.0; Dotmatics) software. Data were compared using a one-way ANOVA followed by Tukey's HSD post hoc test. A total of three independent biological replicates were performed for each experiment. Results are presented as the mean ± SD. P<0.05 was considered to indicate a statistically significant difference.

Based on UPLC-Q-TOF-MS analysis, combined with database searches, reference compounds and related literature, 33 chemical components were identified in MEAG using multilevel MS information (18-24). The chemical components of the MEAG are listed in Table I and the base peak chromatograms of the positive and negative ion modes are shown in Fig. 1A and B.

A total of 14 active ingredients (Table SI) and 348 active ingredient targets were screened using the Swiss ADME and Swiss Target Prediction databases. A total of 2,514 non-redundant disease targets were retrieved from OMIM, DrugBank, GeneCards, TTD and DisGeNET. A total of 210 intersecting targets were obtained by intersecting the drug and disease targets (Fig. 2A).

The PPI network reflects protein interactions; 210 intersecting targets formed a network with 208 nodes and 2,861 edges. Subsequently, the PPI network was visualised using Cytoscape and the topology analysis was performed using the ‘CytoNCA’ plug-in. The top 15 nodes of the network were sorted by betweenness centrality, closeness centrality and degree centrality values and the hub targets were the intersection of these three values. These hub targets were AKT1, EGFR, ALB, BCL2, HIF1A, HSP90AA1, SRC, ESR1, PTGS2, MAPK3 and GSK3B (Fig. 2B and C; Table SII).

Based on GO functional enrichment, biological process enrichment involved ‘cellular response to lipid’ and ‘cellular response to chemical stress’, ‘regulation of inflammatory response’ and ‘positive regulation of programmed cell death’. Cellular component enrichment included ‘perinuclear region of cytoplasm’, ‘centrosome’ and ‘basal plasma membrane’. Molecular function enrichment involved ‘protein kinase activity’ and ‘protein tyrosine kinase activity’ (Fig. 2D). Collectively, MEAG may have contributed to the treatment of Cis-AKI through the regulation of numerous components, targets and pathways.

The target set was imported into the Metascape database for KEGG pathway and biological process enrichment analysis. There were 184 KEGG pathways associated with Cis-AKI therapy in mice. In addition to the cancer pathway with the largest number of annotation genes, the target set was also closely associated with ‘FoxO signalling pathway’, arachidonic acid metabolism’, ‘NF-κB signalling pathway’ and ‘TGF-β signalling pathway’ (Fig. 2E).

General distribution of samples was examined by PCA, with results showing the QC samples clustered together, indicating that the instrument was stable (Fig. 3A and B). The three groups of test samples were clustered, which indicated similarities within the data. There were notable metabolic differences among the three groups. The Nor group was clearly separated from the Cis group and the MEAG group exhibited a downward trend after treatment, indicating that MEAG exerted an ameliorative effect on metabolic disorders in Cis-AKI model mice (Fig. 3A and B).

To establish an association model between the expression of metabolites and sample categories and to screen the differential metabolites, OPLS-DA in positive and negative ion mode was used (Fig. 3C, E, G and I) and the samples in the Nor and Cis groups and Cis and MEAG groups were all significantly separated.

The S-plot diagram of positive and negative ions showed differences in metabolites between the Nor and Cis groups and between the Cis and MEAG groups (Fig. 3D, F, H and J). VIP, fold change (FC) and P-values were then used to identify characteristic differentially expressed metabolites among the three groups. Thus, substances with log₂ (FC) >1, VIP >1 and P<0.05 were potentially upregulated differential metabolites and substances with log₂ (FC)<-1, VIP>1 and P<0.05 were potential downregulated differential metabolites. In the positive ion mode, six endogenous differential metabolites were identified between Nor and Cis groups, and Cis and MEAG groups, including L-phenylalanine, L-carnitine, creatine, phosphocholine, lysophosphatidylethanolamine (LysoPE 16:0/0:0, the notation 16:0/0:0 indicates the fatty acyl chain composition, with a 16-carbon saturated fatty acyl chain and no second fatty acyl chain) and stearoyl-L-carnitine. In negative ion mode, eight endogenous differential metabolites were identified between the Nor and Cis groups and the Cis and MEAG groups, including L-tryptophan, citric acid, ascorbic acid, L-phenylalanine, hyocholic acid, docosahexaenoic acid, L-tyrosine and L-pyroglutamic acid. All differential metabolites in different groups were visualised as a heatmap, in which blue to red indicated an increase in the abundance of differential metabolites (Fig. 4). Table SII and Fig. 5 provide detailed information regarding each differential metabolite and the changes in the relative contents of potentially different metabolites.

Metabolite and gene regulatory network is shown in Fig. 6A. Through topological analysis of metabolic modules, phenylalanine was identified as the hub with the highest centrality. Then, MetaboAnalyst was used to analyse the metabolic pathway enrichment of the 12 metabolites. As shown in Fig. 6B, 14 pathways affecting the metabolism of mice with Cis-AKI were obtained from 13 different metabolites. Among them, the potential metabolic pathways with path influence >0.1 were ‘phenylalanine metabolism’ and ‘phenylalanine, tyrosine and tryptophan biosynthesis’. Based on the results of metabolite network and pathway analyses, phenylalanine metabolism may be a key metabolic pathway regulated by MEAG in the treatment of Cis-AKI.

Associations between metabolic pathways and their targets were next assessed. To comprehensively explore the mechanism of action of MEAG in treating Cis nephrotoxicity, the targets of phenylalanine metabolism were assessed using KEGG (25) and intersected with drug targets and disease targets, integrated the results of serum metabolomics and network pharmacology analysis and subsequently identified an intersection target (MIF).

Among the aforementioned key signalling pathways, NF-κB serves a central role in cellular inflammatory and immune responses (26). A previous study suggested that, in a Cis-AKI mouse model, injured renal tubular epithelial cells released inflammatory cytokines, activated the MIF/NF-κB pathway and triggered a cascade of inflammatory responses, thereby exacerbating renal tubular injury (27). In addition, phenylalanine metabolism is also associated with inflammation. Therefore, the ability of MEAG to regulate the MIF/NF-κB signalling pathway and associated inflammatory factors was evaluated. A schematic diagram of the regulatory pathway is shown in Fig. 7.

After intraperitoneal injection of Cis, mice exhibited clear nephrotoxicity, including weight loss and a significantly increased renal index (P<0.0001; Fig. 8A and B). The levels of Scr (P<0.0001) and BUN (P<0.0001) also increased significantly (Fig. 8C and D). H&E staining of renal tissue showed that intraperitoneal injection of Cis caused serious damage to renal tubules and glomeruli, with renal tubular dilatation, renal tubular epithelial cell shedding, gelatinous casts, glomerular atrophy and basement membrane thickening (Fig. 8E). Quantitative analysis further showed that the tubular injury score was significantly increased in the Cis group compared with that in the Nor group (P<0.0001; Fig. 8F). Masson's trichrome staining revealed severe fibrosis in renal tissue (Fig. 8G). Consistently, the collagen volume fraction was significantly increased in the Cis group compared with that in the Nor group (P<0.0001; Fig. 8H). MEAG treatment significantly ameliorated the loss of renal function of mice and significantly reduced the kidney index (P<0.001; Fig. 8B) and levels of Scr (P<0.0001) and BUN (P<0.0001), which were increased by Cis injection (Fig. 8C and D). Simultaneously, MEAG ameliorated renal tubular damage and reduced the degree of renal tissue fibrosis (Fig. 8E-H).

KIM-1 and NGAL are biomarkers of AKI (28). The expression and distribution of KIM-1 and NGAL in renal tissue were examined. IHC staining showed that the expression of KIM-1 (P<0.0001) and NGAL (P<0.0001) in mouse renal tissue significantly increased after Cis injection and were significantly reduced by MEAG treatment (P<0.0001; Fig. 9A-D).

Proinflammatory cytokine expression in mice after MEAG treatment was analysed. ELISA data showed that MEAG significantly reduced the increase in IL-1β (P<0.0001), IL-6 (P<0.0001) and TNFα (P<0.0001) levels in the serum of mice initially induced by Cis (Fig. 10A-C). Furthermore, western blotting showed that Cis significantly induced the expression of the inflammatory factors Il-1β (P<0.0001), IL-6 (P<0.0001) and TNF-α (P<0.0001) and that this induction was attenuated by MEAG treatment (Fig. 10F and H-J). These findings indicated that MEAG reduced the inflammatory response of mice following Cis treatment.

MIF expression profiles were quantitatively assessed by IHC and western blotting. Histopathological examination demonstrated a significant elevation of MIF immunoreactivity in Cis-challenged murine models compared with the Nor group (P<0.0001), which was attenuated following MEAG intervention (P<0.0001; Fig. 10D and E). Western blotting quantification further demonstrated a significant reduction in renal MIF protein content following MEAG intervention compared with the Cis group (P<0.01; Fig. 10F and G), corroborating the histopathological observations. In addition, a comparative analysis between experimental groups revealed significantly enhanced p-NF-κB/NF-κB levels in the Cis-treated group in comparison with the Nor group (P<0.001), whereas MEAG administration significantly suppressed this phosphorylation event (P<0.01; Fig. 10K and L). These collective findings demonstrate that MEAG treatment exerts regulatory effects on the pathological activation of the MIF/NF-κB signalling cascade in Cis-induced AKI models.