Prognostic role of CDCA3 in lung adenocarcinoma with immune infiltration

Abstract

Lung cancer is a leading cause of cancer-related deaths, often resistant to conventional therapies like platinum-based chemotherapy. Identifying new biomarkers and therapeutic targets is essential for improving outcomes. In this study, we explored the role of CDCA3 in lung adenocarcinoma (LUAD) as a prognostic marker and its association with immune cell infiltration and immunotherapy. Using data from The Cancer Genome Atlas (TCGA), we evaluated CDCA3 expression and its prognostic significance through Kaplan–Meier survival analysis. Our results showed that CDCA3 was significantly upregulated in LUAD tumor tissues compared to normal tissues, with higher CDCA3 expression linked to poorer overall survival. Additionally, CDCA3 expression correlated with clinical features such as age, gender, cancer stage, and smoking status. Immune infiltration analysis revealed significant associations with CD4-activated memory T cells and macrophages. Single‑cell transcriptomics revealed that the proportions of CD8⁺ exhausted T cells, CD8⁺ cytotoxic T cells, M1 macrophages and M2 macrophages varied in association with CDCA3 expression levels in LUAD. CDCA3 expression also correlated with immune scores and immune checkpoint gene expression. Drug sensitivity analysis using the “oncoPredict” R package suggested that CDCA3 expression may influence chemotherapy responses. In vitro experiments in A549 cells showed that CDCA3 knockdown significantly reduced CDCA3 expression at both the mRNA and protein levels. While flow cytometry indicated increased apoptosis in CDCA3-knockdown cells, suggesting CDCA3’s role in promoting LUAD cell proliferation and survival. These findings indicated that CDCA3 could serve as a valuable prognostic biomarker in LUAD, influencing prognosis, immune response, and drug sensitivity, potentially guiding therapeutic and immunotherapy decisions.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-34254-2.

Introduction

Lung cancer ranks among the most prevalent malignancies, exhibiting the highest occurrence and morbidity rates¹,with a poor 5 year survival rate of 19%². It accounts for 29% of cancer-related deaths in males and 26% of cancer-related deaths in females³. Undoubtedly, smoking is the primary cause of lung cancer. In addition, it is worth noting that until recently, the incidence and mortality rates of lung cancer were higher in men than in women³. These facts demonstrate that lung cancer poses a significant threat to human health. According to pathological classification, lung cancer is categorized into two main types: small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC)⁴. Despite early surgical resection being the best cure for NSCLC, the majority of patients present with advanced or metastatic disease. Despite advances in targeted gene therapy, biomarkers, and immuno-oncology for advanced NSCLC, the 5-year survival rate for lung cancer patients is still below 20%, and even lower after metastasis⁵. Therefore, it is imperative to investigate the pathogenesis of lung cancer and identify therapeutic targets that possess high specificity and sensitivity.

In our quest to discover new biomarkers or therapeutic targets, we have studied the protein Cell Division Cycle-Associated Protein 3 (CDCA3), also referred to as Tom-1, which is located on the human chromosome 12p13. CDCA3 is a component of the SKP1-Cullin-RING-F-box (SCF) ubiquitin ligase (E3) complex that targets the endogenous cell cycle inhibitor WEE1 G2 checkpoint kinase (WEE1) for degradation, thereby regulating the cell cycle^6,7. Aberrations in the cell cycle have been linked to uncontrolled cell proliferation, which can promote the development of malignant tumors. CDCA3 has been reported to be upregulated in various cancers, including hepatocellular carcinoma (HCC)⁸, oral squamous cell carcinoma (OSCC)⁹, prostate cancer (PC)¹⁰, gastric cancer (GC)¹¹, and bladder cancer (BC)¹², where elevated malignant expression of CDCA3 strongly correlates with poor patient outcomes. Studies show that CDCA3 expression in HCC correlates with patient prognosis and immune infiltration in the tumor microenvironment, making it a potential diagnostic biomarker and therapeutic target for HCC management¹³. However, comprehensive investigations of CDCA3 in lung cancer, particularly LUAD, remains limited. Yang et al. noted that its expression and clinical significance in LUAD are still unclear¹⁴, and although miR-144-5p–mediated downregulation of CDCA3 suppresses LUAD cell proliferation and invasion¹⁵, the intrinsic biological functions of CDCA3 in this subtype have yet to be systematically elucidated. Moreover, elevated CDCA3 levels have been linked to platinum-based chemotherapy sensitivity and EGFR-TKI resistance in EGFR-mutant NSCLC¹⁶, underscoring the need for a comprehensive analysis of its expression patterns, prognostic value, and mechanistic roles in LUAD. Therefore, exploring the role of CDCA3 in LUAD is essential.

In this study, we compared the expression levels of CDCA3 in LUAD and normal tissues and performed independent prognostic analysis, correlation analysis of CDCA3 expression with clinicopathological parameters, and functional exploration. We also investigated the relationship between high and low CDCA3 subgroups and immune infiltrating cells, immune checkpoint inhibitors (ICIs), and drug sensitivity. By examining CDCA3 and LUAD from multiple perspectives, our study provides valuable theoretical guidance for the clinical management of this disease.

Materials and methods

2.1 Data collection

mRNA expression profiles and associated clinical information for 541 lung adenocarcinoma (LUAD) samples and 59 normal controls were retrieved from the Genomic Data Commons Data Portal (https://portal.gdc.cancer.gov/). To validate prognostic associations, we obtained an independent LUAD cohort (GSE31210, n = 226) from the Gene Expression Omnibus via the KM Plotter web tool (https://www.kmplot.com) for Kaplan–Meier survival analysis. Furthermore, three additional GEO datasets served as external validation: GSE43458 (bulk RNA‑seq, 80 tumor and 30 normal samples), GSE32863 (bulk RNA‑seq, 58 tumor and 58 normal samples) and GSE207422 (single‑cell RNA‑seq, eight LUAD samples).

Data preprocessing

All downstream analyses were performed in R (v4.1.2). For the TCGA‑LUAD cohort, raw counts were converted to transcripts per million (TPM) and, when gene symbols were duplicated, only the highest TPM value for each gene was retained. The bulk RNA‑seq validation sets (GSE43458 and GSE32863) were merged and then quantile normalized using the normalize Between Arrays function in the limma package (v3.50.3)¹⁷. Batch effects were removed with ComBat from the sva package (v3.42.0)¹⁸, including sample group as a covariate in the model. 2.3 Single-cell RNAseq data analysis.

Single-cell RNA-seq data (GSE207422) were processed in Seurat (v5.1.0) by filtering out cells with fewer than 500 UMIs, fewer than 200 detected features, and > 10% mitochondrial reads. Doublets were identified and excluded using DoubletFinder (v2.0.6)¹⁹. Remaining batch‑associated variation across sample was corrected via canonical correlation analysis (CCA)–based integration. This standardized workflow ensured that all datasets were harmonized and of high quality before downstream analyses. CellChat (v2.2.0)²⁰was employed to infer all cell–cell communications across all cell types.

The relationship between CDCA3 and LUAD

The TCGA data were separated into two groups based on the median value of CDCA3, and the prognostic impact of CDCA3 was analyzed using the KM test. The mRNA gene chip module for lung cancer in the KM plotter website was then utilized to confirm the role of CDCA3 (223307_at) in the GEO dataset.

Differential expression analysis

We categorized LUAD samples into two subgroups based on the median expression level of CDCA3. Because CDCA3 and other gene expression values exhibited non-normal, right-skewed distributions (see Supplementary Fig. S1 for normality diagnostics), differential expression was assessed using the Wilcoxon rank‐sum test, with fold change defined as log₂(mean_tumor/mean_normal). P‐values were adjusted by the Benjamini–Hochberg method, and genes with |log₂FC|> 1 and FDR < 0.05 were considered differentially expressed.

Functional enrichment analysis

To investigate the biological processes and pathways associated with differentially expressed genes (DEGs), we conducted Gene Ontology (GO)²¹ and Kyoto Encyclopedia of Genes and Genomes (KEGG)²² pathway enrichment analyses performed by a package called clusterProfiler²³. The GO terms included biological process (BP), cellular component (CC) and molecular function (MF). Significant enrichment was observed with both p and q values below 0.05.

Tumor microenvironment assessment and immune infiltration analysis

We applied the ESTIMATE algorithm ²⁴ to TCGA-LUAD RNA-seq profiles to infer the composition of the tumor microenvironment by computing three scores per sample—StromalScore, which reflects the abundance of stromal cells; ImmuneScore, which reflects the extent of immune cell infiltration; and EstimateScore, the sum of StromalScore and ImmuneScore. Because both StromalScore and ImmuneScore correlate negatively with tumor purity, a lower EstimateScore indicates higher tumor purity. In addition, we utilized the CIBERSORT²⁵ algorithm to investigate the variations in immune cell composition across different groups. Moreover, we assessed the association between CDCA3 expression levels and immune infiltration in R.

Immunotherapy evaluation

The immunophenoscore(IPS) is a composite score based on immune checkpoint molecules and immune cell infiltration, among other factors, and has been shown to be a reliable predictor of response to immune checkpoint inhibitors (such as anti-programmed cell death protein 1 (anti-PD-1) and anti-cytotoxic T lymphocyte antigen-4 (anti-CTLA-4))²⁶. We obtained the IPS data for LUAD patients from the TCIA database (https://tcia.at/)²⁷. Based on the expression of CDCA3, we divided the samples into low and high expression subgroups and compared the IPS between these subgroups under different immunotherapy response conditions.

Drug susceptibility analysis

The tumor group was divided into high and low risk groups based on the median expression level of CDCA3, and oncoPredict²⁸ package was used to calculate the drug imputed sensitivity score for drugs from Sanger’s Genomics of Drug Sensitivity in Cancer (GDSC) v2, aiming to explore possible variations in therapeutic drug effects among LUAD patients.

Statistical analysis

All statistical analyses and data visualizations were performed using the R programming environment (version 4.1.2). To compare two groups, we conducted a single comparison using the Student’s t-test. Survival analysis was performed according to Kaplan–Meier analysis and log-rank test. For the survival analysis was conducted exclusively for CDCA3 (i.e. a single hypothesis), no multiple‐testing correction was performed. In settings where multiple genes are tested simultaneously, raw p‐values would be adjusted by the Benjamini–Hochberg false discovery rate procedure to control for false positives. The correlation between gene expressions was evaluated using Person correlation. The R package “ggplot2” was used to create all boxplots in this study. Statistical significance was considered as a P-value less than 0.05, with significance levels denoted by * for p < 0.05, ** for p < 0.01, and *** for p < 0.001.

Cell culture and knockdown of CDCA3

Human lung carcinoma cells (A549, AW—CCH011, Abiowell, China) were maintained in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% fetal bovine serum, 100 U/mL penicillin, and 100 µg/mL streptomycin at 37 °C in a 5% CO₂ incubator. For CDCA3 knockdown, cells were transfected with siRNA1 targeting CDCA3 (sense: 5′-GCAAUAGAUGGAAACCAAATT-3′; antisense: 5′-UUUGGUUUCCAUCUAUUGCTT-3′) or a non-targeting siRNA control (siNC) using Lipofectamine® 3000 (Invitrogen), according to the manufacturer’s protocol. Gene silencing efficiency was assessed 48 h post-transfection.

Western blot

Western blot analysis was conducted as previously reported with minor modifications. Briefly, tumor tissues were homogenized in ice‑cold modified RIPA buffer, and A549 cells (control and CDCA3‑knockdown lines) were lysed in the same buffer supplemented with protease inhibitors. Protein concentrations were determined using the BCA assay (Pierce). Equal amounts of protein (40 µg) were separated on 10% SDS–polyacrylamide gels and transferred to PVDF membranes (Millipore). Membranes were blocked in 5% non‑fat milk for 2–4 h at room temperature and then incubated overnight at 4 °C with rabbit anti‑CDCA3 antibody (1 : 1,000, Bioss) and mouse anti‑β‑actin antibody (1 : 5,000, Sigma Aldrich). After three washes in TBST, membranes were incubated for 2 h at room temperature with HRP‑conjugated goat anti‑rabbit or goat anti‑mouse IgG secondary antibodies (1 : 5,000; Biyuntian). Chemiluminescent signals were visualized using the ECL detection system (UVP BioDoc‑It Imaging System), and band intensities were quantified by densitometric analysis in ImageJ, with β‑actin as the internal loading control. Tumor tissue samples were obtained from lung cancer patients at Tangdu Hospital, Fourth Military Medical University, under IEC approval (202,102‑25), and all participants provided informed consent. All methods were performed in accordance with the relevant guidelines and regulations.

Quantitative real-time PCR

Total RNA was isolated from A549 cells (control and CDCA3 knockdown) using TRIzol reagent (Invitrogen) according to the manufacturer’s instructions. cDNA synthesis was performed with the PrimeScript RT reagent kit (Takara). Quantitative PCR was conducted on an ABI 7500 real-time PCR system using SYBR Premix Ex Taq II (Takara). The thermal profile included an initial denaturation at 95 °C for 30 s, followed by 40 amplification cycles of 95 °C for 5 s and 60 °C for 34 s. GAPDH was used as the internal control, and relative mRNA expression levels were determined using the 2^^− ΔΔCt method. All reactions were performed in triplicate.

Apoptosis assay by flow cytometry

A549 cells were collected 48 h after siRNA transfection, washed twice with cold PBS, and suspended in the appropriate binding buffer. Cell apoptosis was assessed using an Annexin V-FITC/PI Apoptosis Detection Kit (BD Biosciences), following the manufacturer’s protocol. After a 15-min incubation in the dark at room temperature, fluorescence signals were acquired using a BD FACSCalibur flow cytometer. The percentage of apoptotic cells was quantified using FlowJo software.

Results

The levels of CDCA3 in lung cancer and other cancers

In order to enhance the comprehension of the differential expression of CDCA3 in tumor and normal samples, the expression level of CDCA3 in various tumors and adjacent normal tissues was assessed using the TIMER2.0 database. CDCA3 expression was higher in bladder urothelial carcinoma (BLCA), breast invasive carcinoma (BRCA), cholangiocarcinoma (CHOL), colon adenocarcinoma (COAD), esophageal carcinoma (ESCA), head and neck squamous cell carcinoma (HNSC), HPV-positive head and neck squamous cell carcinoma (HNSC-HPV +), kidney renal clear cell carcinoma (KIRC), kidney renal papillary cell carcinoma (KIRP), liver hepatocellular carcinoma (LIHC), lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), prostate adenocarcinoma (PRAD), rectum adenocarcinoma (READ), stomach adenocarcinoma (STAD), thyroid carcinoma (THCA), and uterine corpus endometrial carcinoma (UCEC) tumors than in normal samples(Fig. 1A). Collectively, these results underscore the pervasive upregulation of CDCA3 across multiple malignancies and provide the rationale for our subsequent in-depth analysis of its expression and clinical significance specifically in lung cancer. In the TCGA LUAD cohort, we first compared CDCA3 expression in 541 tumor samples versus 59 adjacent normal tissues, and then examined 58 matched tumor–normal pairs. In both unpaired and paired analyses, CDCA3 levels were significantly higher in tumors than in normal controls (Figs. 1B). To validate these findings, we next analyzed two GEO datasets, GSE43458 and GSE32863, comprising a total of 138 tumor and 88 normal samples. Consistent with the TCGA results, CDCA3 expression remained markedly elevated in tumor samples in both overall and paired box‑and‑whisker plots (Fig. 1C).

Fig. 1.

Expression of CDCA3 in pan-cancer samples and Kaplan‑Meier survival analysis. (A)The expression of CDCA3 was analyzed and compared in various tumor and normal tissues using the TIMER database, with significance levels of *p < 0.05, **p < 0.01, and ***p < 0.001. (B) CDCA3 expression levels and CDCA3 expression levels in paired samples in LUAD as retrieved from the TCGA database. (C) CDCA3 expression levels and CDCA3 expression levels in paired samples in LUAD as retrieved from the GEO database (GSE43458 and GSE32863). (D) Kaplan–Meier analysis was conducted to assess the prognostic impact of CDCA3 expression on lung cancer using data from TCGA. (E) Kaplan–Meier analysis was conducted to assess the prognostic impact of CDCA3 expression on lung cancer using data from GSE31210. (F) Quantification of CDCA3 protein levels by Western blot in adjacent normal (NC) and LUAD tumor tissues; β-actin was used as a loading control. The uncropped original WB images are provided in Supplementary Fig. 1.

To further evaluate the prognostic potential of CDCA3 for lung cancer, especially LUAD, Kaplan–Meier survival analysis was conducted on both the TCGA database and the GEO datasets. patients were divided into high and low expression groups based on their CDCA3 gene expression levels, using the median as a cutoff point. We found a significant negative correlation between CDCA3 expression level and OS in both TCGA (p < 0.01; Fig. 1D) and GEO dataset GSE31210(p < 0.010; Fig. 1E). It indicates that CDCA3 overexpression is associated with poor prognosis in patients with LUAD.

In order to verify whether CDCA3 changes in lung cancer, we detected the expression level of CDCA3 by WB. We found that the expression level of CDCA3 was high in tumor sample than adjacent tissue (Fig. 1 F).

The association of CDCA3 expression with clinical characteristics and biomarker genes in LUAD patients.

To further assess the significance of CDCA3 expression in LUAD, we initially examined the relationship between CDCA3 and different clinical characteristics. Two groups were formed in TCGA samples based on CDCA3 expression, and the clinical characteristics were compared between them. As shown in Fig. 2, CDCA3 was significantly correlated with age, gender, stage, advanced TNM stage and N stage. However, no correlation was found between CDCA3 expression and M stage (Fig. 2, Table 1). Furthermore, we assessed the expression correlation between CDCA3 and several biomarker genes of lung cancer. We observed a significant positive correlation between CDCA3 expression and the expression of TROAP, FOXM1, AUPKB, and CENAP (Fig. 3A–D). TROAP expression in lung adenocarcinoma correlates with clinical features and may serve as an independent prognostic factor for poor survival²⁹. Upregulation of FOXM1 is not only linked to poor clinical outcomes in lung cancer patients, but it also contributes to resistance against anticancer treatments. This suggests that CDCA3 may be involved in the development and progression of lung cancer, potentially influencing these biomarker genes. While the observed correlations suggest an association between CDCA3 and these genes, it is important to note that correlation does not imply causation. This analysis focuses on expression levels, and further research, including experimental validation, is needed to confirm whether CDCA3 directly regulates the expression of TROAP, FOXM1, AUPKB, and CENAP. A limitation of this study is that it does not establish a direct regulatory mechanism between CDCA3 and these genes, and future studies should explore the underlying molecular interactions to clarify this potential relationship.

Fig. 2.

Heatmap of clinicopathologic features among high- and low-expression CDCA3 subgroups. Student’s t test was used to assess significant differences among groups, with significance levels of *p < 0.05, **p < 0.01, and ***p < 0.001.

Table 1.

Table of clinicopathologic features among high- and low-expression CDCA3 subgroups. Student’s t test was used to assess significant differences among groups, with significance levels of *p < 0.05, **p < 0.01, and ***p < 0.001.

Clinical_feature	Low_expression	high_expression	P_value	Significance
Age	<  = 65: 105 (42.3%); > 65: 143 (57.7%)	<  = 65: 134 (53.8%); > 65: 115 (46.2%)	0.0135	*
Gender	FEMALE: 156 (61.4%); MALE: 98 (38.6%)	FEMALE: 122 (46.6%); MALE: 140 (53.4%)	0.001	***
T	T1: 103 (40.9%); T2: 117 (46.4%); T3: 24 (9.5%); T4: 8 (3.2%)	T1: 66 (25.3%); T2: 161 (61.7%); T3: 23 (8.8%); T4: 11 (4.2%)	0.0015	**
N	N: 1 (0.4%); N0: 176 (71.5%); N1: 41 (16.7%); N2: 28 (11.4%); N3: 0 (0%)	N: 0 (0%); N0: 156 (60.2%); N1: 55 (21.2%); N2: 46 (17.8%); N3: 2 (0.8%)	0.0357	*
M	M: 3 (1.6%); M0: 173 (94%); M1: 5 (2.7%); M1a: 1 (0.5%); M1b: 2 (1.1%)	M: 1 (0.5%); M0: 174 (90.6%); M1: 13 (6.8%); M1a: 1 (0.5%); M1b: 3 (1.6%)	0.332	ns
Stage	Stage I: 152 (61.3%); Stage II: 56 (22.6%); Stage III: 31 (12.5%); Stage IV: 9 (3.6%)	Stage I: 124 (47.7%); Stage II: 66 (25.4%); Stage III: 53 (20.4%); Stage IV: 17 (6.5%)	0.0089	**
Subdivision	L-Lower: 47 (19.1%); L-Upper: 49 (19.9%); R-Lower: 50 (20.3%); R-Middle: 9 (3.7%); R-Upper: 91 (37%)	L-Lower: 31 (12.2%); L-Upper: 74 (29%); R-Lower: 47 (18.4%); R-Middle: 12 (4.7%); R-Upper: 91 (35.7%)	0.0683	ns

Fig. 3.

The association between CDCA3 expression and biomarker genes of LUAD patients. (A) The expression of CDCA3 was significantly correlated with TROAP (P < 2.2e-16). (B) The expression of CDCA3 was significantly correlated with FOXM1 (P < 2.2e-16). (C) The expression of CDCA3 was significantly correlated with AURKB (P < 2.2e-16). (D) The expression of CDCA3 was significantly correlated with CENPA (P < 2.2e-16).

Identification and functional analysis of differentially expressed genes

To investigate the role of the CDCA3 gene, we classified TCGA samples into two subgroups based on the median CDCA3 expression level and detected differentially expressed genes (DEGs) between the two groups. We detected a total of 100 DEGs using the threshold (|log₂FC|> 1 and FDR < 0.05), consisting of 50 down-regulated genes and 50 up-regulated genes (Fig. 4A). To gain a deeper insight into the biological functions of the DEGs associated with CDCA3, we performed GO and KEGG enrichment analyses with the R package clusterProfiler. We identified several enriched gene sets, including nuclear division for GO biological process, chromosomal region for GO cellular component, and microtubule motor activity and single − stranded DNA helicase activity for GO molecular function (Fig. 4B). We further investigated the functional implications of these DEGs by KEGG pathway analysis. We found that a significant number of DEGs were enriched in KEGG pathways, with the top 10 pathways being: Cell cycle, Complement and coagulation cascades, Motor proteins, DNA replication, Cytoskeleton in muscle cells, Fanconi anemia pathway, ECM − receptor interaction, Oocyte meiosis, Protein digestion and absorption (Fig. 4B).

Fig. 4.

Differentially expressed gene identification and GO enrichment and KEGG pathway enrichment analyses. (A) Heatmap of differentially expressed genes between high and low CDCA3 expression subgroups. The color scale bar indicates the level of gene expression. (B) LEFT: GO enrichment analyses of differentially expressed genes. The color of the dots corresponds to the P-values, while the size of the dots indicates the number of enriched target genes in the pathway.RIGHT: KEGG pathway enrichment analysis of differentially expressed genes. The color of the dots corresponds to the P-values, while the size of the dots indicates the number of enriched target genes in the pathway. The pathway map is adapted from KEGG (www.kegg.jp), following the citation guidelines: Kanehisa, M., Furumichi, M., Sato, Y., Matsuura, Y. and Ishiguro—Watanabe, M.; KEGG: biological systems database as a model of the real world. Nucleic Acids Res. 53, D672—D677 (2025).

CDCA3 expression and its relationship with immunity

We present the ESTIMATE-derived StromalScore, ImmuneScore, and EstimateScore for CDCA3 high- versus low-expression groups. The CDCA3 high-expression group exhibited lower StromalScore, ImmuneScore, and EstimateScore, indicating reduced stromal and immune cell infiltration and, consequently, higher tumor purity in tumors with elevated CDCA3 expression (Fig. 5A). This suggests that high CDCA3 expression in LUAD patients is associated with increased tumor purity. To further explore how CDCA3 expression relates to the immune microenvironment, we applied the CIBERSORT algorithm to estimate the proportions of 22 tumor‑infiltrating immune cell subsets in TCGA‑LUAD specimens. Fourteen of these tumor infiltrating immune cells showed significant associations with CDCA3 levels (Fig. 5B). In particular, eight subsets were positively correlated with CDCA3 expression: CD8⁺ T cells, activated memory CD4⁺ T cells, follicular helper T (Tfh) cells, regulatory T (Treg) cells, resting NK cells, and M0, and M1 macrophages. We then validated these results in two GEO cohorts (GSE43458 and GSE32863), where two immune cell types remained significantly correlated with CDCA3; notably, activated memory CD4⁺ T cells again demonstrated a positive association, in agreement with the TCGA findings (Fig. 5C). In TCGA-LUAD, immune cells significantly related to the expression of CDCA3 include CD4-activated memory T cells, and M0 macrophages (Fig. 5C,E). In GEO-LUAD, immune cells significantly related to the expression of CDCA3 also include CD4-activated memory T cells and M0 macrophages (Fig. 5D, F). The findings suggest that CDCA3 could potentially function as an indicator of immune activity, and appears to be closely associated with T-cell nuclei and macrophages.

Fig. 5.

Immune cell infiltration in different CDCA3 subgroups. (A) The tumor microenvironment (TME) score in the low- and high-expression CDCA3 subgroups (B) Infiltrating abundance of distinct immunocytes in the low- vs. high-expression CDCA3 LUAD subgroups in TCGA and GEO databases. (C) Correlation analysis of CDCA3 and immune cell infiltration in TCGA-LUAD. The color of the dots corresponds to the P-values, The size of the dot represents the strength of the correlation. (D) Correlation analysis of CDCA3 and immune cell infiltration in GEO-LUAD (GSE43458 and GSE32863). (E) Correlation analysis between CDCA3 expression and three immune cell types in TCGA-LUAD. (F) Correlation analysis between CDCA3 expression and three immune cell types in GEO-LUAD (GSE43458 and GSE32863).

3.5 Single-cell transcriptomic analysis of CDCA3 expression and immune heterogeneity

To further validate this finding, we analyzed the LUAD single-cell transcriptome data GSE207422, a dataset containing eight LUAD tumor samples. Totally, eleven cell types were annotated by unsupervised uniform manifold approximation and projection (UMAP) analysis, including B cells (with expression of MS4A1), Epithelial cells (EPCAM), Fibroblasts (COL1A1). Mast cells (KIT), Myeloid cells (LYZ), Neutrophil (CSF3R), NK cells (FGFBP2), Plasma cells (IGHG1), Plasmacytoid dendritic cells (LILRA4), and T cells (CD3E). However, Cancer cells were annotated by CopyKAT analysis, which calculates CNV to assess epithelial cell benignity and malignancy (Fig. 6A). Based on the average expression of CDCA3 in the cells of different samples, the samples were categorized into CDCA3 high-expression group and CDCA3 low-expression group according to the median, each containing four samples (Fig. 6B). We found a relatively large difference in the cell percentage of T cells and myeloid cells in the CDCA3 group, which consists to our previous results (Fig. 6C, D). Further, we performed subpopulation analysis of T cells and myeloid cells. A total of six T cell subpopulations were annotated, including CD4 + Exhausted T cells, CD4 + Naïve T cells, CD8 + Cytotoxic T cells, CD8 + Exhausted T cells, Proliferating T cells, Regulatory T cells, among others, CD8 + Cytotoxic and Exhausted T cells which differed the most between CDCA3 groups (Fig. 6E, F). In myeloid cells, a total of five subpopulations were annotated, including Alveolar Macrophages, M1 Macrophages, M2 Macrophages, Monocyte, and Proliferating Macrophages, among others, M1 and M2 macrophages that differed significantly between CDCA3 groups (Fig. 6G, H). Cell–cell communication analysis using CellChat revealed that the interaction of cancer cells with CD8 + cytotoxic T cells, CD8 + Exhasted T cells, and M1/M2 macrophages was enhanced in the CDCA3 high-expression group compared to the CDCA3 low-expression group, and the number of cellular communications of CD8 + Exausted T cells was also increased in the CDCA3 high-expression group. (Fig. 6I). In high CDCA3 group cancer cells, both outgoing and incoming autocrine signals via PPIA–BSG and MDK–SDC1/4 are markedly reduced, indicating a shift away from Cyclophilin A–Basigin and Midkine–Syndecan self‑stimulation. At the same time, MDK → (ITGA4 + ITGB1) efferent signals from CD8⁺ cytotoxic T cells and PPIA → BSG incoming signals into tumor cells are attenuated, suggesting impaired T cell–tumor crosstalk and enhanced immune escape. In contrast, cancer cell → M1/M2 macrophage signaling through SCGB3A2–MARCO, PPIA–BSG, MT‑RNR2–FPRL2, MDK–SDC1/2/4 and MDK–NCL is significantly strengthened, consistent with increased macrophage recruitment, survival and immunosuppressive polarization to support tumor progression (Fig. 6J).

Fig. 6.

Single-cell transcriptomic profiling of LUAD tumors stratified by CDCA3 expression. (A) UMAP embedding of cells from eight LUAD samples, colored by major cell types (B cells, epithelial cells, fibroblasts, mast cells, myeloid cells, neutrophils, NK cells, plasma cells, plasmacytoid dendritic cells, T cells) and malignant epithelial cells identified by CopyKAT CNV analysis. (B) Bar plot shows the classification of samples into CDCA3-high and CDCA3-low groups based on median CDCA3 expression. (C)Comparison of all cell proportions between CDCA3-high and CDCA3-low samples in stacked histogram. (D)Comparison of all cell proportions between CDCA3-high and CDCA3-low samples in radar chart. (E) UMAP embedding of six T cell subtypes. (F) Stacked histogram shows the proportional composition of T cell subtypes. (G) UMAP embedding of five myeloid cell subtypes. (H) Stacked histogram shows the proportional composition of myeloid cell subtypes. (I) CellChat-inferred interaction network among cancer cells, CD8⁺ cytotoxic T cells, CD8⁺ exhausted T cells, and M1/M2 macrophages. (J) Bubble map of ligand–receptor pairs involving cancer cells.

Immune checkpoint gene expression and prediction of immunotherapy response

The development of immune checkpoint inhibitors has transformed the landscape of cancer treatment. We performed correlation analysis between the expression of CDCA3 and 21 immune checkpoint genes by calculating the Pearson correlation coefficient for each gene pair. The results showed that CDCA3 was negatively correlated with most of the immune checkpoints, such as CD40LG, CD27, TNFSF15, CD28, CD200R1, among others (Fig. 7A). These findings may indicate that CDCA3 plays a certain role in the process of tumor immune escape.

Fig. 7.

CDCA3 Predicts the response to immunotherapy in LUAD (A) Correlation between CDCA3 and immune checkpoints. The color of the point represents the strength of the correlation. (B) Differences in levels of IPS in the high and low CDCA3 groups in LUAD.

To predict the response to immunotherapy using the Immune Checkpoint Inhibitor score, we used four subtypes of IPS values as substitutes for the immunotherapy response of lung cancer patients. The IPS scores showed that the scores with CTLA4_pos (both PD1_neg and PD1_pos) and PD1_pos (both CTLA4_neg and CTLA4_pos) in the low CDCA3 expression subgroup were all higher than those in the high CDCA3 expression subgroup (CTLA4_pos_PD1_neg: p = 1e-08; CTLA4_pos_PD1_pos: p = 0.093; CTLA4_neg_PD1_pos: p = 0.024, as shown in Fig. 7B). These results suggest that CDCA3, especially in the low expression subgroup, may be associated with PD-1 and CTLA4-related immunotherapy. The significant difference in the CTLA4_pos_PD1_neg subgroup suggests that low CDCA3 expression may enhance immune responses by interacting with CTLA-4 inhibition. This could activate immune cells and improve the efficacy of immune checkpoint inhibitors. While CDCA3, as a cell cycle regulator, might play a role in immune evasion mechanisms, further experimental validation is needed to confirm its association with immunotherapy outcomes.

Identification of potential therapeutic drugs for treatment of the CDCA3 expression subgroups

We further explored the drug sensitivity of the different CDCA3 groups by utilizing the “oncoPredict” R package, anti-cancer drug sensitivity for each sample in the TCGA cohort was predicted with the magnitude of the IC50 value. The drug sensitivity analysis revealed a significant difference in drug response between the high and low CDCA3 groups. Most of the drugs, including 5-Fluorouracil, Dabrafenib, Gefitinib, and Olaparib, were found to be more effective for patients in the high CDCA3 expression subgroup, as they exhibited a relatively low IC50 value (Fig. 8A–D). Taken together, the subpopulation with high CDCA3 expression is more suitable for targeted therapy using drugs that target gene mutations or specific cancer types. The observed differences in treatment sensitivity between the two subtypes also highlight the heterogeneity among patients, indicating that CDCA3 can serve as a valuable reference for medication guidance.

Fig. 8.

The correlation between CDCA3 expression and drug sensitivity in LUAD. (A–D) Drug sensitivity of LUAD patients in high- vs. low-expression CDCA3 subgroups.

Effects of CDCA3 Knockdown on Proliferation and Apoptosis in LUAD Cells

To further investigate the role of CDCA3 in LUAD cells, we established CDCA3 knockdown cell lines using the A549 LUAD cell line. Initially, we validated the knockdown efficiency through quantitative PCR and Western blot analysis, which demonstrated a significant downregulation of CDCA3 at both the mRNA and protein levels (Fig. 9A–B). Furthermore, flow cytometry analysis revealed that reduced CDCA3 expression was associated with an increased apoptosis rate in LUAD cells (Fig. 9C–D). These comprehensive findings suggest that elevated CDCA3 expression promotes the proliferation of LUAD cells.

Fig. 9.

Effects of CDCA3 knockdown on A549 cell proliferation and apoptosis. (A) Western blot analysis of CDCA3 protein levels in control (NC) and CDCA3-knockdown cells; β-actin was used as a loading control. The uncropped original WB images are provided in Supplementary Fig. 1. (B) Quantitative Real time PCR analysis of CDCA3 mRNA levels in control (NC) and CDCA3-knockdown cells. (C) Representative Annexin V-FITC/PI flow cytometry plots for NC (left) and CDCA3-knockdown (right) cells. (D) Quantification of late apoptotic cells from (C); data are presented as mean ± SD of three independent experiments. *P < 0.05 by unpaired Student’s t-test.

Discussion

Lung cancer is associated with one of the highest mortality rates among all types of cancer in China^30,31. The study indicates that CDCA3 is an important oncogene in multiple tumors^32,33. CDCA3 plays a critical role in promoting efficient G2/M cell cycle progression and tumor cell proliferation, and its depletion can trigger tumor cell senescence³⁴. In addition, CDCA3 serves as a chemosensitivity biomarker in NSCLC, with high expression levels of CDCA3 associated with increased sensitivity to platinum chemotherapy^34,35. However, the existing literature on CDCA3 in lung cancer is limited, and a more detailed and comprehensive analysis of CDCA3 is yet to be conducted. Thus, it is imperative to delve further into the clinical implications and prognostic value of CDCA3 in lung cancer.

In this study, we compared the expression of CDCA3 between tumor and normal samples across all cancer types in TCGA. Our findings revealed that CDCA3 expression was higher in tumor samples compared to normal samples in 17 cancer types, including lung cancer. Higher CDCA3 expression in LUAD samples was found to be significantly associated with poorer overall survival (OS) through survival analysis on TCGA and GEO datasets. Furthermore, the correlation between CDCA3 expression and certain biomarkers in lung cancer indicates that CDCA3 plays a crucial role in the initiation and progression of the disease.

Based on the GO functional analysis, CDCA3 has been found to be potentially involved in cell division, chromosome segregation, and other related processes. It may play a role in tumor progression by facilitating cell cycle G2M transition, as demonstrated in Fig. 4B. This finding aligns with a prior study³⁶. There have been studies showing that the expression level of CDCA3 is associated with the p53 signaling pathway³⁷. The KEGG pathway analysis indicated the involvement of CDCA3 in multiple pathways, such as the cell cycle, p53 signaling pathway, cAMP signaling pathway, and others. Negative regulation of the p53 signaling pathway was found to exacerbate the proliferation and migration of esophageal squamous cell carcinoma cells, and Hu et al. reported that inhibiting p53 expression promoted the growth of ovarian cancer cells^38,39.

In recent years, the tumor microenvironment has garnered significant attention and has emerged as a crucial factor in cancer treatment. The important role of the tumor microenvironment in tumor invasion, tumor proliferation, microangiogenesis, and even immune evasion has been demonstrated by numerous studies. The tumor microenvironment of lung adenocarcinoma is a key factor influencing the response to immunotherapy and disease prognosis, and mainly contains complex interactions among immune cells (e.g., CD8 + T cells, macrophages), stromal cells, and tumor cells. The tumor microenvironment has been found to have immunosuppressive properties that promote tumor heterogeneity and metastasis, leading to poor patient prognosis^40,41. For example, the tumor microenvironment in patients with lung adenocarcinoma often exhibits abnormalities in the infiltration status of immune cells, especially the distribution of CD8 + T cells, CD4 + T cells and myeloid cells, which has a direct impact on disease progression and immune checkpoint inhibitor response^42,43. Based on CIBERSORT bioinformatics analysis, our study found significant differences in 14 out of 22 types of immune cells between the CDCA3 high-expression group and the low-expression group in TCGA-LUAD. Among these, the cell abundance most significantly correlated with CDCA3 was CD4 + activated memory T cells, M0 macrophages, and M1 macrophages, which was confirmed by the analysis of the GEO-LUAD data. CDCA3 has been found to be correlated with immune cell infiltration in HCC, with involvement of various immune cells including monocytes, tumor-associated macrophages (TAMs), M1 and M2 macrophages, CD4 + T cells, CD8 + T cells, NK cells, and more, as reported by previous studies. Notably, our single-cell analysis of eight LUAD specimens (GSE207422) stratified by CDCA3 expression both confirmed and refined these findings: in the CDCA3-high group, the proportions of CD8⁺ cytotoxic and exhausted T cells, as well as M1 and M2 macrophages, were significantly elevated compared to the CDCA3-low group. Moreover, CellChat-based communication inference revealed strengthened crosstalk between malignant cells and these effector populations. Together, these single-cell results reinforce the CIBERSORT-based observation that CDCA3 upregulation is tightly linked to altered T cell and macrophage landscapes in LUAD, and echo prior reports in HCC where CDCA3 has been shown to modulate infiltration of monocytes, tumor-associated macrophages, CD4⁺/CD8⁺ T cells and NK cells. Statistical analysis of IPS scores suggests that patients in the high-expression subgroup may not benefit from immunotherapy. Given that the IPS has been proposed as a surrogate marker for response to CTLA-4 and PD-1 blockade⁴⁴, the IPS pattern observed in our cohort suggests that CDCA3 expression could be used to stratify LUAD patients for immunotherapy. CDCA3-low tumors, which exhibit higher IPS values, may represent an “inflamed” immune phenotype that is more likely to benefit from PD-1/CTLA-4 inhibitors, whereas CDCA3-high tumors, characterized by enrichment of exhausted CD8⁺ T cells and reprogrammed macrophages, may require rational combination strategies that simultaneously target CDCA3-driven cell-cycle dysregulation and immune checkpoints. In this context, CDCA3 emerges not only as a prognostic cell-cycle regulator but also as a potential therapeutic node linking tumor-intrinsic proliferation programs to the composition and functional state of the tumor immune microenvironment. Prospective mechanistic and clinical studies will be needed to determine whether CDCA3-guided stratification or CDCA3 inhibition can enhance the efficacy of immune checkpoint blockade in LUAD. Consistent with our findings, a recent pan-renal cell carcinoma study reported that high CDCA3 expression was associated with lower IPS values, higher TIDE scores and an inferior predicted response to PD-1/CTLA-4 blockade, further supporting CDCA3 as a potential biomarker for immunotherapy stratification. Therefore, we further assessed the differences in drug susceptibility between the high- and low-expression groups. In vitro experiments in A549 cells further validated the functional role of CDCA3 in tumor proliferation and survival. CDCA3 knockdown significantly reduced both mRNA and protein levels of CDCA3, as confirmed by quantitative PCR and WB. Flow cytometry analysis demonstrated a significant increase in apoptosis in the CDCA3-knockdown cells. These findings suggest that CDCA3 contributes to LUAD cell proliferation and survival, further supporting its potential as a therapeutic target. Our findings not only facilitate risk stratification but also provide a foundation for individualized treatment selection. This study comprehensively investigated the association between CDCA3 and lung cancer from various perspectives, in order to identify potential therapeutic targets for preventing and treating the disease.

However, this study still has several limitations. Most of our analyses were based on retrospectively collected public datasets, and the experimental validation was restricted to in vitro knockdown of CDCA3 in a single LUAD cell line, so additional mechanistic work in other models and independent patient cohorts is needed to substantiate our conclusions. The cohorts we used are essentially cross sectional, which prevents us from following dynamic changes in CDCA3 expression and the immune microenvironment during disease progression or under treatment, and future longitudinal studies will be required to test the stability and predictive value of CDCA3 based signatures. Even with these constraints, our findings point to CDCA3 as a promising prognostic and immuno-biological marker in LUAD with potential translational value. Recent work has shown that artificial intelligence can integrate imaging, pathology and laboratory data to support lung cancer screening and treatment decision making in routine practice, as summarized by Pei et al.⁴⁵, and that immune related gene signatures combined with machine learning can refine risk stratification and predict response to immunotherapy, as reported by Bai et al⁴⁶. These studies suggest that CDCA3 anchored transcriptional and microenvironmental signatures described here could in future be incorporated into similar AI assisted decision support systems, with the aim of improving LUAD risk stratification and individualized therapeutic selection.

Conclusion

CDCA3 demonstrates potential as a prognostic biomarker for LUAD due to its strong association with lung cancer outcomes and immune cell infiltration. Furthermore, it may significantly guide immunotherapy strategies and inform drug selection.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

YanLi Chen and Xinyao Xu: Conceptualization and Writing—original draft; Jipeng Zhang and Chenghui Jia: Data curation; Liang-Guan, Qirui Zhao and Pengyu Jing: Software and Validation; Qiang Lu: Supervision and Writing –review & editing.

Funding

This study was supported by National Natural Fund Youth Project (82002422).

Data availability

Data for this study were obtained from the Genomic Data Commons Data Portal (mRNA-FPKM expression and clinical data, https://portal.gdc.cancer.gov/), Gene Expression Omnibus (GEO)—KM Plotter (mRNA expression and clinical characteristics, https://www.kmplot.com), and The Cancer Immunome Atlas (TCIA) (immunophenoscore, https://tcia.at/). All data are publicly available through these databases.

Declarations

Competing interests

The authors declare no competing interests.