PTL was purchased from GLPBIO (Montclair, USA) and the human LUAD cell line, H1975, was purchased from the Shanghai Institutes for Biological Sciences, China. The RPMI 1640 medium was purchased from Thermo Fisher Scientific (Waltham, MA, USA). Fetal bovine serum was purchased from Oumashi Biotechnology Co., Ltd. (Shanghai, China). All amino acids and stable isotope-labelled standards were obtained from Sigma-Aldrich (St. Louis, MO, USA). Analytical grade ammonium acetate was obtained from Sigma-Aldrich (St. Louis, MO, USA). Methanol (Optima LC-MS), acetonitrile (Optima LC-MS), and formic acid (Optima LC-MS) were purchased from Thermo Fisher Scientific (FairLawn, NJ, USA). Ultrapure water was purchased from Millipore (Billerica, MA). TCEP (tris (2-carboxyethyl) phosphine) and IAA (iodoacetamide) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Sequence-grade trypsin (sequence grade) was purchased from Promega (Madison, WI, USA). NP-40 (Nonidet P 40) and acetone were purchased from Sangon Biotech (Shanghai, China). The protease inhibitor cocktail was obtained from Kangchen Biotech (Shanghai, China). RNAkey Total RNA Extraction Kit was purchased from Beijing Saiwen Innovation Biotechnology Co., Ltd. The qPCR quantitative analysis and reverse transcription kits were purchased from Novozan Biotechnology(Nanjing,China). The poly (vinylidene fluoride) membrane and ECL reagent were purchased from Sigma (USA). The relevant antibodies were provided by Abclonal (Beijing, China). An amino acid extraction kit was purchased from Solarbio (Beijing, China). The NADPH/NADP detection kit was obtained from Wuhan Saiyu Biotechnology Co., Ltd. The GSH/GSSG detection kit was provided by Nanjing Jiancheng Bioengineering Research Institute.

H1975 cells were cultured in complete RPMI 1640 medium containing 10% fetal bovine serum, 100 U/mL streptomycin, and 100 U/mL penicillin. H1975 cells were incubated in an environment of 5% CO₂ and 37°C. According to the appropriate concentration (inhibitory concentration 50 [IC50]) standardized in the initial experiment, PTL was dissolved in 0.05% DMSO, diluted to 10 μM with PBS, added to the treatment group, and an equal amount of complete culture medium or culture medium lacking amino acids was added to the control group.

The process was divided into nine steps as follows: 1) sample lysis; 2) BCA(Bicinchoninic Acid Assay) quantification; 3) acetone precipitation; 4) protein reconstitution, reduction, alkylation, and enzymatic hydrolysis; 5) removal of SDC(sodium deoxycholate); 6) peptide desalination; 7) Nano HPLC(High Performance Liquid Chromatography) separation; 8) LC-MS/MS(Liquid Chromatography-Tandem Mass Spectrometry); and 9) MaxQuant analysis and LFQ(label-free quantification).

We constructed a PPI network of differentially expressed proteins using the String data platform (https://cn.string-db.org/), and the final graph was implemented using the R package”igraph” (1.2.6). The biological processes, molecular functions, and related signaling pathways of the differentially expressed proteins were analyzed using MetaScape(http://metascape.org/gp/index.html#/main/step1) software. The parameters related to the enrichment analysis were set to a minimum overlap of 3, p-value cutoff of 0.01, and minimum enrichment of 1.5. Using R language - ggplot2 [3.4.4] package was used for GSEA (Gene Set Enrichment Analysis) Analysis.

Stock solutions of the individual amino acids were mixed and prepared in an amino acid-free matrix to obtain a series of amino acid calibrators. Certain concentrations of L-Alanine-d4 and Phenylanine-d2 were mixed as internal standards (IS). The stock solution of all of these and the working solution were stored in -20°C. The samples were added to 450 μL water and 2 mL of acetonitrile/methanol (1:1), which contained mixed internal standards, and vortexed for mixing. Next, the samples were placed in liquid nitrogen and thawed at room temperature. Subsequently, the samples underwent ultrasonic treatment for 4 min, was placed in liquid nitrogen again for 5 min, and thawed at room temperature. The samples were centrifuged at 12,000 rpm for 20 min at 4°C. Finally, the supernatant was injected into a LC-MS/MS system for analysis.

An ultra-high performance liquid chromatography coupled to tandem mass spectrometry (UHPLC-MS/MS) system (ExionLC™ AD UHPLC-QTRAP 6500+, AB SCIEX Corp., Boston, MA, USA) was used to quantitate amino acids in Novogene Co., Ltd. (Beijing, China). Separation was performed on an ACQUITY UPLC BEH Amide column (2.1×100mm, 1.7μm), which was maintained at 50°C. The mobile phase, consisting of 0.1% formic acid in 5 mM ammonium acetate (solvent A) and 0.1% formic acid in acetonitrile (solvent B), was used at a flow rate of 0.30 mL/min. The solvent gradient was set as follows: initial 80% B, 0.5 min; 80_–70% B, 2 min; 70_–45% B, 4 min; 45_–80% B, 6.01min; 80% B, 9 min.

The mass spectrometer was operated in positive multiple reaction mode (MRM). Parameters were as follows: IonSpray Voltage (5500 V), Curtain Gas (35 psi), Ion Source Temp (550°C), and Ion Source Gas of 1 and 2 (50 and 60 psi).

To explore the relevant targets of the disease, the keyword “lung adenocarcinoma” was searched using OMIM (http://omim.org/), GeneCards (http://www.genecards.org/), and DisGeNET (https://disgenet.com/) databases. Relevant targets of parthenolide were extracted from the SEA search server (https://sea.bkslab.org/), PharmMapper server (https://lilab-ecust.cn/pharmmapper/index.html), and SwissTargetPrediction(http://swisstargetprediction.ch/). The intersection of disease targets, drug targets, and proteomic results was obtained. The intersection of these targets was considered as a potential target for the treatment of LUAD with parthenolide. Subsequently, the standard names of the targets were retrieved from UniProt KB (http://www.uniprot.org/). Relevant targets were obtained from differentially expressed amino acid metabolites using MetScape in Cytoscape 3.8.0. Finally, a compound–target–amino acid network diagram was constructed that included the relationships between parthenolide, related target genes, and amino acids.

The PDB files of the targets were downloaded from the PDB database (RCSB PDB, http://www.rcsb.org/pdb/). They were preprocessed using Pymol software, and protein hydrogenation and charge calculations were performed using Autodock software and converted into pdbqt files. The SDF file for the compound was downloaded from PubChem (https://pubchem.ncbi.nlm.nih.gov/). Chem3Dpro software was used to calculate its minimum energy, convert it into a mol file, and use autodock to convert it into pdbqt; then Vina molecule docking was used, with the ligand position in the PDB file as the docking interface; the CONF file of each folder was the docking location; Vs_desult was the docking result; The binding energy between components and targets, as well as the number of hydrogen bonds formed, were important criteria for evaluating molecular docking results. The lower the binding energy and the more hydrogen bonds there are, the more stable the binding becomes, and greater the possibility of target molecule interactions. Pymol (open-source) was used to draw 3D and 2D graphs, and Discovery Studio 4.5 Client was used to display docking sites.

A 100 ns MD simulation was performed on the composite using Gromacs 2023. The protein adopted CHARMM 36 force field parameters, and the ligand topology was constructed using GAFF2 force field parameters. The protein-ligand complex was placed in a cubic box under periodic boundary conditions. The box was filled with water molecules using the TIP3P water model. The particle grid Ewald (PME) and Verlet algorithms were used to separate the electrostatic interactions. Subsequently, 100000 steps of the isothermal isostatic ensemble equilibrium and isothermal isobaric ensemble equilibrium were performed using a coupling constant of 0.1 ps and a duration of 100 ps. The van der Waals and Coulomb interactions were calculated using a cut-off value of 1.0 nm. Finally, the system was subjected to molecular dynamics simulations using Gromacs 2023 at a constant temperature (300 K) and pressure (1 bar) for a total duration of 100 ns.

Cells were divided into three groups: control, PTL, and PTL + OE-GCTG (also known as GGCT). The concentration of PTL was 10 μM. qPCR experiments were conducted 48 h after cell transfection. Plasmids were prepared using GenePharma (Shanghai, China). qPCR of extracted mRNA was performed according to the manufacturer’s instructions. The primer sequence was as follows: GCTG (upstream): 5’- TCGGCGGCGTTCTTCTGTG-3’; GCTG (downstream):

In addition, RT-qPCR experiments involving human tissues were performed after obtaining informed consent from the patients and were approved by the Ethics Committee of Shengjing Hospital (Ethics Code: 2023PS1232K). For basic experimental analysis. 32 pairs of cancer and adjacent normal tissue samples were collected from patients who underwent the same surgery at Shengjing Hospital between January 2019 and December 2021.

Proteins were extracted from the three groups of cells mentioned earlier, quantified, and denatured. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was used to separate the same amount of protein, which was then transferred to a polyvinylidene fluoride membrane with a pore diameter of 0.45 μM. Subsequently, the membrane was sealed with a rapid blocking solution and incubated overnight with primary antibody at 4°C. After washing, the membrane was incubated with a secondary antibody at room temperature for 1 h. Finally, ECL reagent was used to visualize the protein bands on a gel imaging system (GE Healthcare). In addition, the grayscale values of each band were analyzed using ImageJ software.

In each group, 800 cells/well were inoculated in a 6-well culture plate, which consisted of complete culture medium of 2 ml per well, with three wells in each group. The inoculated cells were shaken well and incubated on a clean bench for 20 minutes, then cultured in the incubator. The medium was changed every 3 days, the cell status observed, the supernatant discarded, and the cells washed once with PBS until the number of cells in most individual clones in the well exceeded 50. Next, 1 mL of 4% paraformaldehyde was added to each well and the cells were fixed at 4°C for 60 min and washed once with PBS. Crystal violet staining solution (1 mL) was added to each well and the cells were stained for 15 min. The cells were then washed several times with PBS, air-dried, photographed, and counted.

Cells were incubated using RPMI 1640 medium without amino acids. Cells were collected into centrifuge tubes and the supernatant was discarded after centrifugation. According to the instructions, 1 mL of the corresponding reagent was added to every 5 million cells, and the cells were sonicated (power 20%, sonication for 3 seconds, interval of 10 seconds, repeated 30 times). Subsequently, the liquid was centrifuged at 10000 rpm and 4°C for 10 minutes, and the supernatant was collected. The enzyme-linked immunosorbent assay (ELISA) reader was preheated for more than 30 min, and the wavelength was adjusted to 570 nm. Amino acid content in the supernatant was measured and calculated according to the manufacturer’s instructions.

Eight ml NADPH probe buffer (component B-II) and 2 ml NADPH probe (component B-I) was mixed together. Two μl of storage solution was diluted with PBS to obtain 2 μM working solution. The cells were seeded in a 6-well plate, counted to 1x10^6, and treated. The cells were resuspended in 50 μl of component D and repeatedly freeze thawed thrice in liquid nitrogen. A 96-well plate was prepared and 50 μl of PBS solution was added to blank and standard wells 1–6, respectively; 50 μl of NADPH working solution was added to standard well 7. Fifty microliters of the processed sample suspension was added to each experimental well; next, 50 μl of the mix was added to all wells and incubated at room temperature in the dark for 1 h. A ELISA reader was used to detect the OD value at 460 nm.

Cells were plated in a 6-well plate and treated before counting. The number of cells in the sample were adjusted to 1x10⁶, and the cells were resuspended using 110 μl of reagent D. Configuration of 50 μM GSH standard and 50 μM GSSG standard was performed. Next, 10 μl PBS was added to the blank well of the 96-well plate, 10 μl GSH standard to the standard well, and 10 μl of the test cell suspension to the experimental well. Then, 100 μl of reagent A and 10 μl of reagent B was added to each well in sequence, and incubated at room temperature for 2 minutes. Then 50 μl of reagent C was added to each well. After 30 s, the OD value at 450 nm was measured using the ELISA reader and the reading was recorded (A1); after 10 minutes, OD value was measured again under the same conditions and the reading was recorded (A2). Result analysis: GSH (μM)=[Measurement well (A2-A1)/standard (A2-A1)] x standard concentration (50 μM). GSSG was determined as described above.

5×10⁵ cells were seeded in a 6-well plate and subjected to the corresponding treatment. Next, 10 μM ROS probe reagent was added to each well (leaving one blank well). After incubation at 37°C for 30 minutes, the treated cells were digested with trypsin and resuspended. After adjusting the parameters, the average fluorescence intensity of the blank well cells was set to 10 and the average fluorescence intensity of the other cells was sequentially detected by adding ROS probes.

Biotinylated drug (10 μM) was incubated with streptavidin beads (10 μL) in binding buffer (200 μL total volume) for 1 h at 4°C with rotation.Beads were washed 3× with ice-cold binding buffer to remove unbound drug.Target protein lysate (100 μg) was added to the drug-bound beads and incubated for 2 h at 4°C with rotation.Beads were pelleted using a magnetic rack and washed 3× with binding buffer.

All experimental procedures involving animals were approved by the Animal Ethics Committee of the Shengjing Hospital (Ethics Code: 2023PS1263K). Twelve female nude mice (BALB/C, 4–5 weeks old, weight 13–15 g) were purchased from Beijing Huafukang Biotechnology Co., Ltd. All nude mice were kept under SPF conditions (22 ± 0.5°C, 50 ± 10% relative humidity, free access to food and water). Nude mice were housed in a SPF-grade animal breeding center for 1 week. Subsequently, 12 nude mice were randomly divided into control, PTL, and PTL+OE-GCTG groups (n=4). Lentivirus preparation was performed by Genepharma. Normal H1975 or OE-GCTG H1975 cells were digested, centrifuged, resuspended in serum-free medium, and counted to a final concentration of 1x10⁷/ml. Cells were randomly inoculated into the posterior lower part of the right forelimb of nude mice using a sterile syringe, with an injection volume of 120 μl. Normal H1975 cells were inoculated in the control and PTL groups, whereas OE-GCTG H1975 cells were inoculated in the PTL + OE-GCTG group. PTL was dissolved in 0.05% DMSO and diluted in PBS. The PTL and PTL+OE-GCTG groups were intraperitoneally injected with PTL [(20 mg/kg, dose selection based on the results of our previous study (unpublished)]; results showed that 20 mg/kg PTL significantly inhibited the volume of LUAD tumors in nude mice. Therefore, this dose was selected for subsequent analyses. The control group was administered an equal volume of DMSO+PBS without any other treatment. The drugs were injected three times a week for four consecutive weeks, and the body weight and tumor volume of the nude mice were measured twice a week. After 4 weeks, the mice were euthanized using the cervical dislocation method after anesthesia. Photographs of the tumor were taken, and relevant experiments were conducted.

The extracted proteomic and amino acid metabolomic data were analyzed using univariate statistical analysis, multivariate statistical analysis, hierarchical clustering, and correlation analysis. Univariate statistical analysis was conducted using the Student’s t-test and multiple variation analysis. When the p-value was less than 0.05, the distribution of the measured data showed a decrease of less than 0.67 times or an increase of more than 1.5 times, which was considered a significant difference between cells treated with estradiol. The multivariate statistical analysis included unsupervised principal component analysis (PCA), supervised partial least squares discriminant analysis (PLSDA), and orthogonal partial least squares discriminant analysis (OPLSDA). The discrimination of the tested substances was determined by the variable importance (VIP>1) and p-value (p<0.05) of the projection (VIP) parameter. Gel Pro Analyzer and Image J software were used for grayscale analysis, while SPSS 22.0 and GraphPad Prism7.0 software were used for statistical analysis. Quantitative data were presented in the form of mean ± standard deviation. After testing whether the data in each group followed a normal distribution, a t-test or one-way ANOVA was used to detect intergroup differences. P<0.05 indicated statistical significance.

As shown in Figure 1 , the flowchart illustrates the study process. A total of 157 differentially expressed proteins were screened by mass spectrometry-based parallel reaction monitoring (PRM), including 60 upregulated and 97 downregulated proteins ( Figure 2A , Supplementary Tables S1, S2 ). As shown in Figure 2B , the significantly altered proteins formed a cluster, indicating that they have similar expression patterns and may be closely related to metabolic processes. As shown in Figure 2C , the PPI network revealed significant differential protein interactions in this group, which may indicate that it plays a special role in certain molecular processes or cellular functions.

As shown in Figure 3A and Table 1 , the differentially expressed proteins mainly focus on “oxidoreductase activity,” “protein homodimerization activity,” and “nucleotide transferase activity” in terms of molecular function. As shown in Figure 3B and Table 2 , the differential proteins mainly focus on “response to extracellular stimuli,” “glutamine family amino acid metabolism process,” and “amide metabolism process” in terms of cellular biology processes. Subsequently, these proteins were found to play biological roles in pathways such as “glutathione metabolism,” “alanine, aspartate, and glutamate metabolism,” and “biosynthesis of amino acids” ( Figure 3C , Table 3 ). Similarly, GSEA indicated that the functions of the differentially expressed proteins were mainly focused on amino acid metabolism and antioxidative stress ( Figure 3D , Table 4 ). As shown in Figure 3E , the three most distinct biological effects were demonstrated by GSEA enrichment analysis. Therefore, the results of the GO and KEGG analyses indicate that the proteins affected by PTL may alter the amino acid metabolism and oxidative stress status in LUAD.

As shown in Figure 4A , 23 amino acids were detected in this study; we found significant changes in the content of 12 amino acids after PTL administration ( Supplementary Tables S3 ). Interestingly, all 12 amino acids significantly decreased after PTL administration. As shown in Figure 4B , these significantly altered amino acids formed a cluster, indicating that they had similar metabolic patterns and may be closely related to energy metabolism. Correlation analysis was performed to determine the degree of association between significant amino acids, with the aim of gaining a deeper understanding of the relationships between amino acid types in biological processes. Correlation analysis showed that each amino acid was significantly positively correlated with the other 11 amino acids (P<0.05,r>0.7) ( Figure 4C ). These results indicate that PTL reduces amino acid production in LUAD, thereby affecting energy metabolism.

From the PCA scoring chart, it was concluded that there was good repeatability within the groups, the sample data were very similar, and there was good discrimination between the groups ( Figure 5A ). Similarly, the PLS-DA and OPLS-DA models ( Figures 5B, C ) showed significant differences between the two groups. External permutation test was performed to avoid overfitting, with a R2 value of 0.429, a Q2 value of -0.279 and with a R2 value of 0.227, a Q2 value of -0.786, respectively ( Figures 5D, E ). The Variable Importance in Projection (VIP) scores showed that 11 significantly differentially expressed amino acid metabolites had scores greater than 1, indicating that these metabolites enhanced the explanatory power of the model ( Figure 5F , Table 5 ). The volcano map consistently showed similar results ( Figure 5G ).

Based on the SwissTarget Prediction (STP), SEA search server, and PharmMapper databases, 148 target genes were associated with PTL ( Figure 6A ). Based on DisGeNET, GeneCards Search, and OMIM databases, 70 target proteins were associated with LUAD ( Figure 6B ). When these selected genes were crossed with differentially expressed genes using quantitative proteomics, 29 potential targets of PTL against LUAD were identified ( Figure 6C ).

Subsequently, in Cytoscape 3.9.1, 11 different amino acid metabolites regulated by PTL were introduced into MetScape and 41 amino acid metabolite targets were obtained ( Figure 7A ). Finally, a compound-target-metabolite network consisting of 82 nodes (PTL, 70 related target genes, and 11 amino acid metabolites) and 153 edges was constructed ( Figure 7B ). Based on parameters such as degree, betweenness centrality, proximity, and stress in the network, GCTG (also known as GGCT), DAO, and BAAT were identified as key targets ( Table 6 ).

We conducted molecular docking studies using AutoDock Vina to further elucidate the interactions between PTL and potential targets. Docking analysis of GCTG showed that PTL formed hydrogen bonds with the SER A: 24 and TRP A: 64 residues at the active site. The van der Waals interactions included TYR A: 22, ASN A:25, ALA A: 69, and TYR A:139 residues. The binding energy between GCTG and PTL was -10.5 ( Figure 8A ). However, in the molecular docking analysis of DAO and BAAT with PTL, there was no interaction with hydrogen bonds, and the binding energies of the two were -8.5 and -7, respectively ( Figures 8B, C ). In summary, GCTG had the lowest binding energy for molecular docking with PTL and hydrogen-bonding interactions. Therefore, GCTG was selected as a key target for subsequent experiments.

First, the balance of the simulation system was evaluated using root mean square deviation (RMSD). Generally speaking, an RMSD value less than 2 Å and reaching stability within 10–20 ns indicates high conformational stability of the system. As shown in Figure 9A , the composite system reached equilibrium at around 20ns and eventually fluctuated at approximately 1.8 Å. Therefore, PTLs exhibit high stability when bound to a target protein. Further analysis revealed that the radius of gyration (Rg) and solvent-accessible surface area (SASA) of the composite system fluctuated stably during motion, indicating that the composite system remained stable and compact throughout the simulation ( Figures 9B, C ). The number of hydrogen bonds between PTL and the target protein during the dynamic process is shown in Figure 9D , and the number of hydrogen bonds between the complex system ranged from 0 to 8, indicating that the complex system had good hydrogen bond interactions. As shown in Figure 9E , the root mean square fluctuation (RMSF) of the composite was relatively low (mostly below 2 Å), resulting in low flexibility and high stability. In summary, the binding of the complex was stable, and the complex exhibited good hydrogen bonding properties. Therefore, PTL bind well to its target proteins.

As shown in Figures 10A–C , bioinformatics analysis revealed that GCTG was significantly upregulated in tumor tissues and could serve as a potential biomarker for predicting prognosis. High expression was significantly associated with poor prognosis. RT-qPCR results of 32 clinical samples also showed that GCTG was significantly overexpressed in cancer tissues ( Figure 10D ). As shown in Table 7 , GCTG expression significantly correlated with tumor size and staging. Therefore, both dry and wet experiments indicated that GCTG was significantly upregulated and had a pro-cancer effect in LUAD. As shown in ( Figure 10E ), compared to that in the control group, PTL significantly inhibited the expression of GCTG protein, whereas the rescue experiment partially reversed the expression of this protein. As shown in Figures 10F, G , PTL significantly inhibited the proliferation and amino acid metabolism of LUAD cells, whereas rescue experiments partially reversed these biological processes. Correspondingly, as shown in Figures 10H–J , PTL targeting GCTG enhanced ROS expression while inhibiting the GSH/GSSG and NADPH/NADP ratios. To further confirm whether PTL has a direct mechanism of action on GCTG, a pull-down assay was performed, and it was found that PTL directly regulates GCTG ( Figure 10K )

As shown in Figure 11A , nude mice were euthanized and their tumors were removed for photography. PTL significantly decreased tumor weight and volume ( Figures 11B, C ). Immunohistochemical analysis revealed that PTL inhibited GCTG expression ( Figure 11D ). The amino acid content determination experiment also showed that PTL inhibited amino acid metabolic reprogramming in LUAD tissues ( Figure 11E ). The aforementioned rescue experiments partially reversed this phenomenon.