First, mRNA expression profiles and clinical information of GSE166722 (n=53), GSE31210 (n=226), GSE30219 (n=182), GSE50081 (n=125), and GSE42127 (n=284) datasets were obtained from the Gene Expression Omnibus (GEO) database (GEO, https://www.ncbi.nlm.nih.gov/GEO/). Additionally, mRNA expression data and clinical information of the TCGA-LUAD cohort were acquired from the UCSC Xena browser (https://xenabrowser.net/). In the GSE166722 dataset, 53 cases of early-stage lung adenocarcinoma were classified into indolent tumors (ais and mia) and invasive tumors (iac) based on pathological grading. The GSE31210 dataset provided complete information on recurrence status and disease-free survival time for stage I&II patients. Therefore, the primary analysis was mainly based on GSE166722 and GSE31210, while the GSE30219, GSE50081, GSE42127, and TCGA-LUAD datasets were used as validation sets. Additionally, mutation information of the LUAD cohort from TCGA database was downloaded.

We utilized the WGCNA R package (v1.68) to identify the gene co-expression networks associated with invasion and recurrence in the GSE166722 and GSE31210 datasets. Hub genes within specific modules were defined based on gene significance (GS), where the Pearson correlation of each gene with invasion and recurrence > 0.2, and module membership (MM) > 0.8. Subsequently, the clusterprofiler R package (v3.14.3) was employed to perform Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses on the hub genes within the co-expression modules.

The random survival forest and random forest are machine learning algorithms commonly used for dimensionality reduction analysis. The identified gene set associated with invasion and recurrence was included in the random survival forest analysis, using the “randomForest” and “randomSurvivalForest” packages in R, selecting the top 30 genes. Key genes were obtained by analyzing the overlapping genes using a Venn diagram. Subsequently, the effectiveness of these key genes was assessed through survival analysis and the construction of ROC curves.

The prognostic role of key genes in several independent LUAD cohorts was analyzed using Kaplan-Meier curve analysis. The correlation between key genes and clinical pathological features, including pathological stage, TNM stage, age, and gender, was assessed using the chi-square test. The prognostic independence of key genes in predicting Overall Survival (OS) and Progression-Free Interval (PFI) was determined using the univariate and multivariate Cox regression analysis.

Cell lines derived from Anhui University of Science and Technology School of Medicine were used in this study. All cell lines were authenticated using short tandem repeat (STR) analysis. Human lung epithelial cells BEAS-2B,the lung cancer cell lines A549, H1975, and H1299, were cultured in DMEM medium containing 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin (Biological Industries, Israel) at 37°C in a 5% CO₂ incubator. A549, H1975, and H1299 cells were seeded in plates and transfected with siRNA using Lipofectamine 2000 (Invitrogen, USA) when the cells reached 70% confluency. The siRNA reagents were purchased from GenePharma.The siRNA sequences were as follows: si_CDC20#1 (S: 5’-CGCCUGAAAUCCGAAAUGATT-3’, AS: 5’-UCAUUUCGGAUUUCAGGCGTT-3’); si_CDC20#2 (S: 5’-CACAGAACCAGCUAGUUAUTT-3’, AS: 5’-AUAACUAGCUGGUUCUGUGTT-3’); si_CDC20#3 (S: 5’-CCACCAAGAAGGAACAUCATT-3’, AS: 5’-UGAUGUUCCUUCUUGGUGGTT-3’); at the same time, we used a scramble sequence that did not specifically target any gene transfected cells as the NC group, si_CDC20#NC(S: 5’-UUCUCCGAACGUGUCACGUTT-3’, AS: 5’-ACGUGACACGUUCGGAGAATT-3’), we used this as a siRNA control targeting CDC20.

SiRNA transfected H1975 and H1299 cells were collected 24h after transfection, and RIPA lysis buffer was added to extract cell proteins on ice. The protein samples were subjected to 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred onto a PVDF membrane. The membrane was blocked with 5% skim milk for 1 hour and incubated overnight at 4°C with specific primary antibodies (diluted 1:1000). The specific primary antibodies used were as follows: CDC20 (PTG, 10252-AP), ERK (Servicebio, GB11004-100), JNK (Servicebio, GB114321-100), P38 (Servicebio, GB113380-100), and GAPDH (ABclonal, AC001). After washing the PVDF membrane three times with TBS-T solution for 10 minutes each, chemiluminescent detection was performed, and the protein band intensity was analyzed using Image J software.

Total RNA was extracted from cultured cells using Trizol reagent (Invitrogen) according to the manufacturer’s instructions. RNA extracted from each group was used for reverse transcription through the Revert Aid First Strand cDNA Synthesis Kit (Thermo Scientific). The obtained cDNAs were used for real-time quantitative PCR (RT-qPCR) through 2X Universal SYBR Green Fast QPCR Mix system. The results were analyzed in a QuantStudio 3 Real-Time PCR System (quantstudio design & analysis software, Thermo Scientific).

The H1975 and H1299 cell suspensions were added to a 6-well plate and placed in a cell incubator. Once the cells formed a confluent monolayer, the cells were transfected with siRNA. After transfection for 6h,a sterile pipette tip (200μL) was used to create scratch wounds. Subsequently, the wells were washed three times with PBS. Images of the scratch wounds were captured at 0h and 24h using an inverted microscope.Use serum-free medium after scratch.

SiRNA transfected H1975, H1299 cells were collected 24h after transfection and cultured in 6-well plates at a density of 2000 cells per well for 7-10 days. The medium was changed every three days. After colonies formed, the colonies were fixed, washed, and stained to observe clone formation. The colonies were counted using Image J software, and statistical analysis was performed.

The upper compartment was filled with 100μL of 1%BSA DMEM cell suspension containing 1 × 10⁴ H1975 or H1299 cells collected 24 hours after transfection of siRNA.Additionally, 600μL of complete culture medium was added to the lower chamber. The chamber was incubated at 37°C with 5% CO₂ for 24 hours. Subsequently, the chamber was removed and the cells were fixed in 4% paraformaldehyde for 20 minutes. Excess cells in the upper chamber were gently wiped with a cotton swab, washed twice with PBS, and stained with crystal violet for 20 minutes. Finally, the cells were observed under an inverted microscope, and random images were captured.

The cell cycle was determined using propidium iodide (PI) DNA staining. H1975 and H1299 cells were seeded in cell culture dishes and incubated at 37°C with 5% CO₂ for 24 hours. The cells were then treated with siRNA and incubated again at 37°C with 5% CO₂ for an additional 24 hours. After isolating the cells, they were centrifuged, washed with PBS, fixed with 75% cold ethanol, and stored overnight at 4°C. The cells were washed with PBS, treated with RNase and PI, and incubated at 37°C for 60 minutes. Flow cytometry analysis was performed using a flow cytometer, with an excitation wavelength of 488 nm and an emission wavelength of 610 nm. The collected data from 10,000 cells were analyzed using Modfit 5.0 software.

One day prior to transfection, HEK-293 (Pricella) cells were seeded to reach a density of 60% to 80%. Six hours before transfection, the cells were switched to a non-complete culture medium. 2μg of sh-CDC20 lentiviral plasmid were diluted in 400 μl of Opti-MEM (ThermoFisher Scientific, #31985070), and simultaneously, 8 μl of Lipofectamine 2000 were diluted in another 400 μl of Opti-MEM. Both solutions were gently mixed and allowed to stand for 5 minutes. Then, the Lipofectamine 2000 mixture was slowly added to the plasmid solution and allowed to stand for an additional 10 minutes. Subsequently, the combined solution was added to the cell culture plate and gently mixed. The cells were cultured for 6 hours at 37°C with 5% CO2, followed by replacement with fresh DMEM medium. After 24 hours, the expression of ZsGreen protein in HEK-293 cells was observed using a fluorescence microscope to assess the efficiency of plasmid transfection.

We selected sh-CDC20 LA-4 stable cells to verify the effect of CDC20 on tumor formation in mice. Healthy male C57BL/6 mice, aged 4-6 weeks, were randomly divided into control and shCDC20 groups. Subcutaneously injected (100μl) into the armpit of the forelimb of mice, and the cell density was 1 × 10⁸/ml.The mice’s health condition and normal food intake were observed. Tumor size was measured using calipers, and on the 28th day, the mice were euthanized using cervical dislocation, and the tumors were completely dissected. Sh group and NC group, 5 animals in each group.Tumor size and volume were calculated using the formula: V = (L × W²)/2, where L and W represent the length and width of the subcutaneous tumor, respectively. The animal experiments were approved by the Medical Ethics Committee of Anhui University of Science and Technology.

All independent experiments were repeated three times, and the results were analyzed using GraphPad Prism software. Independent sample t-test and Wilcoxon signed-rank test were used between the two groups; one-way ANOVA (One-way ANOVA) and multiple comparisons were used between the three groups to compare the control and experimental groups. Statistical significance is denoted as follows: ns, not significant; * p < 0.05; ** p ≤ 0.01; *** p ≤ 0.001.

WGCNA was used to identify gene modules highly correlated with invasion and recurrence in the GSE166722 and GSE31210 datasets (Figures 1A, B). The blue (cor = 0.8, P < 1e-200), sky blue (cor = 0.63, P = 2.3e-10), and yellow-green (cor = 0.73, P = 1.5e-81) modules in the GSE166722 dataset were found to be associated with invasion. Similarly, the blue (cor = 0.46, P = 5e-123), sky blue (cor = 0.52, P = 5.5e-07), and yellow-green (cor = 0.38, P = 4.9e-18) modules showed correlation with the mRNA stemness index (mRNAsi) score ( Figure 1C ). In the GSE31210 dataset, the yellow module (cor = 0.44, P = 7.6e-45) was selected as key genes for further analysis based on MM > 0.8 and GS > 0.2 ( Figure 1D ). Intersection of genes from these modules resulted in the identification of 91 hub genes ( Figure 1E ). Functional enrichment analysis using Gene Ontology (GO) revealed their association with microtubule binding, spindle, chromosomal region, organelle fission, and nuclear division ( Figure 1F ).

To further reduce the dimensionality of genes, we employed random survival forests and intersected their results, yielding three hub genes: CDC45, CDC20, and TPX2 ( Figures 2A–C ). Concurrently, we analyzed the impact of CDC45, CDC20, and TPX2 expression on patient OS and 1-, 3-, and 5-year predictive abilities in the GSE31210 cohort. We found that high expression of these three genes led to poorer prognosis but better survival prediction capabilities ( Figure 2D ). Moreover, in the GSE166722 cohort, all three genes demonstrated strong predictive abilities for high invasion ( Figure 2E ).

Additionally, we assessed the correlation between CDC45, CDC20, and TPX2 with tumor stemness using Pearson correlation analysis ( Figure 2F ). We found that the correlation between CDC20 and stemness was 0.63 and 0.46 in both datasets, respectively, which was higher than that of CDC45 (0.62,0.41) and TPX2 (0.53,0.33).

Due to the relatively high correlation between CDC20 and tumor stemness, and limited research on the association between CDC20 and postoperative recurrence and invasion in lung adenocarcinoma, subsequent studies primarily focused on CDC20. The prognostic value of CDC20 mRNA expression was evaluated in the Kaplan-Meier Plotter database, and overall survival (OS) curves were plotted for LUAD patients from the TCGA, GSE30219, GSE50081, and GSE42127 cohorts. As shown in Figures 3A–D , higher mRNA expression of CDC20 was significantly associated with a poorer prognosis in LUAD patients. Additionally, progression-free interval (PFI) curves from TCGA data and progression-free survival (PFS) curves from GSE30219 and GSE50081 data were plotted ( Figures 3E–G ). The results revealed that patients with high CDC20 expression had shorter progression-free intervals and progression-free survival compared to those with low CDC20 expression. Furthermore, a comparison of CDC20 expression between normal and tumor tissues in TCGA and GSE31210 cohorts showed significantly higher expression in tumor tissues ( Figure 3H ). These findings suggest a significant association between CDC20 mRNA expression and the prognosis of LUAD patients, highlighting its potential as a useful biomarker for predicting patient survival.

Additionally, the combined analysis of clinical and pathological features demonstrated that patients with high CDC20 expression tended to exhibit more advanced pathological stages ( Figure 3I ). CDC20 may serve as a potential therapeutic target for LUAD treatment. Multivariable Cox regression analysis further conducted on the TCGA-LUAD cohort indicated that CDC20 could be an independent prognostic factor in LUAD ( Figure 3J ). Overall, these results strongly suggest a high correlation between CDC20 and tumor malignancy.

CDC20’s association with stemness was assessed in the GSE31210 and GSE166722 datasets. Positive correlations were observed between CDC20 and 26 stemness gene sets ( Figure 4A ). Subsequently, oncogenic pathways in these datasets were screened, revealing a positive correlation between CDC20 and the Hypoxia pathway in both datasets ( Figure 4B ). Differential analysis was further performed on six datasets (GSE30219, GSE31210, GSE42127, GSE50081, GSE166722, TCGA) to explore 33 key genes, including CDC20 ( Figures 4C, D ). Subsequently, a protein-protein interaction (PPI) network analysis was conducted on these genes to assess their protein-level interactions, as shown in Figure 4E . Additionally, the mutation landscape of these key genes was visualized ( Figure 4F ). Collectively, these results partially elucidate the molecular characteristics of tumors with high CDC20 expression.

To identify potential drugs highly sensitive to CDC20 expression, this study utilized predictive algorithms to estimate the chemotherapeutic sensitivity of chemotherapy drugs and compared them between high and low CDC20 expression groups. As shown in Figure 5A , the estimated IC50 values of Docetaxel, Gemcitabine, Cisplatin, Vinorelbine, Paclitaxel, Afatinib, Erlotinib, and Gefitinib were significantly lower in the high CDC20 expression group across each LUAD dataset, suggesting that individuals with high CDC20 expression might exhibit higher sensitivity to these drugs. Subsequently, the study employed CAMP visualization to screen and prioritize the top 20 drugs targeting CDC20 scores, including JNJ-26854165, masitinib, methotrexate, pyrvinium-pamoate, 7b-cis, serdemetan, CD-437, pyrimethamine, NSC-3852, angiogenesis inhibitor, aminopurvalanol-a, purvalanol-a, L-168049, palbociclib, vidarabine, floxuridine, fludarabine, antimycin-a, SA-792728, and calyculin ( Figure 5B ). Additional drug screening was performed for the gene clusters obtained from ( Figures 4E ), revealing potential drug categories identified by CAMP that could serve as a foundation for targeted therapies (Figure 5C).

To clarify the promotional effect of CDC20 on lung adenocarcinoma, our study initially verified the expression levels of CDC20 protein and mRNA in human epithelial cell line BEAS-2B and three lung cancer cell lines, H1299, H1975, and A549. We found that the protein and mRNA expression levels of CDC20 were significantly higher in all three lung cancer cells compared to the normal epithelial cells, with H1299 and H1975 cells showing relatively higher expression ( Figures 6A, B ). As the level of CDC20 protein expressed in A549 cells was relatively low, it was considered futile to knock down CDC20 in these cells. Consequently, to more effectively elucidate the impact of CDC20 reduction on migration and invasion behavior, we selected the H1299 and H1975 cell lines, which exhibited higher levels of CDC20 expression, for knockdown experiments.

We did the Wound healing assay and Transwell migration assay.In the Wound healing assay, we analyzed the migration rate at 24 hours. The average migration rate of H1299 cells in the NC group at 24 hours was 42.7%, with a standard deviation (SD) of 8. After knocking down CDC20, the migration rates decreased significantly to 21.8%, 13.8%, and 13.1%, with SD values of 3, 4, and 1, respectively. Similarly, after the same knockdown treatment on H1975 cells, the average 24-hour migration rate in the NC group was 35.4%, with an SD of 4. In the knockdown group, the 24-hour migration rates were 14.9%, 21%, and 12.4%, with SD values of 4, 1, and 3, showing a significant decrease in migration. Therefore, we can conclude that knocking down CDC20 in both cell types inhibits and significantly reduces cell migration ability. Additionally, in the Transwell migration assay, the average number of cells passing through the chamber for H1299 cells in the NC group and knockdown group were 139, 65, 16, and 28, respectively. For H1975 cells, the numbers were 110, 39, 48, and 58.The results from the Wound healing assay and Transwell migration assay demonstrated a significant inhibition of migratory and invasive abilities in lung cancer cells upon CDC20 knockdown ( Figures 6C, D ). In the colony formation assay, we obtained that the mean colonies of single cell clone formation in the NC group and knockdown group of H1299 cells were 229,143,139,165, respectively, and that in H1975 cells were 331,242,227,208, respectively ( Figure 6E ). Additionally, the colony formation assay showed decreased proliferative ability in H1299 and H1975 cells following CDC20 knockdown.In conclusion, these findings provide strong evidence supporting the crucial role of CDC20 in promoting the malignant characteristics of lung adenocarcinoma.

To elucidate the potential mechanistic role of CDC20 in tumor malignancy, we transfected H1299 and H1975 cells with siRNA. Flow cytometry analysis revealed that the knockdown group of H1299 and H1975 cells exhibited arrest at the S phase, indicating an impact on the cell cycle ( Figures 7A, B ). To further investigate the potential mechanisms of CDC20 in LUAD, we discovered a significant enrichment of the MAPK signaling pathway upon CDC20 knockdown based on other literature references (18).

In addition, previous research has reported that the activation of the MAPK pathway promotes the malignant progression of LUAD (19). Thus, we hypothesize that CDC20 may play a role in promoting LUAD through the MAPK signaling pathway. To investigate this hypothesis, we conducted Western blotting analysis to determine the impact of CDC20 knockdown on the expression of MAPK nodes in both H1299 and H1975 cells. The results revealed a significant downregulation of ERK1/2 (p-ERK1/2), p38 (p-p38), and JNK (p-JNK1/2) protein levels following CDC20 knockdown (Figures 7C, D). These findings suggest that CDC20 has an influence on the cell cycle and may potentially contribute to the progression of LUAD through the ERK/MAPK signaling pathway.

To further investigate the impact of CDC20 on tumor progression in vivo, a mouse xenograft model was established ( Figures 8A, B ). Tumor growth was monitored, and the shCDC20 group exhibited a significantly lower average tumor volume of 503.96 mm3 compared to the control group (p ≤ 0.001). As anticipated, the average tumor weight in the shCDC20 group was 0.424g less than that in the shNC group (p ≤ 0.001) ( Figure 8C ).

Immunofluorescence analysis of CDC20 expression in tumors from the shCDC20 and NC groups revealed a noticeable decrease in CDC20 fluorescence levels and expression in the shCDC20 group ( Figure 8D ). In addition, we conducted immunofluorescence detection of CDC20 in normal mouse lungs. Compared to normal lung tissue, the expression of CDC20 in mouse lung cancer tissues was significantly increased. After knocking down CDC20, the tumor volume and weight were significantly reduced, suggesting that CDC20 may affect the progression of lung cancer in mice. In conclusion, the results obtained in vivo validate the role of CDC20 downregulation in tumor initiation and progression in lung cancer.