Integrated transcriptomic and immunoinformatics analysis identifies tpx2, bub1b, and ube2c as diagnostic biomarkers and therapeutic targets in endometrial carcinoma

Abstract

Objective

This study aimed to identify novel diagnostic and prognostic biomarkers in endometrial carcinoma (EC) using integrative bioinformatics and immunohistochemical validation, with a focus on potential therapeutic targets and immune correlations.

Methods

Gene expression profiles from the GEO database (GSE63678 and GSE17025) and EC-related genes from the DisGeNET database were analyzed to identify overlapping differentially expressed genes (DEGs). Protein–protein interaction networks were constructed, and hub genes were identified using topological algorithms in Cytoscape. Functional enrichment was conducted via GO and KEGG analyses. Validation of hub gene expression was performed using GEPIA, Xiantao, and immunohistochemistry. Kaplan–Meier and ROC analyses assessed prognostic and diagnostic value. Immune infiltration correlations were analyzed using TIMER 2.0. Exploratory drug sensitivity analysis was performed using the GSCA platform.

Results

Sixty-three EC-specific DEGs were identified. Ten hub genes were prioritized, with showing consistent overexpression in EC tissues across datasets and strong associations with poor survival. These genes were also linked to reduced immune cell infiltration and increased tumor purity. Preliminary analyses suggested positive correlations between tpx2/ube2c expression and trametinib/masitinib sensitivity.

Conclusion

bub1b, tpx2, and ube2c are promising biomarkers for the diagnosis and prognosis of EC. Their association with immunosuppression suggests a potential role in immune evasion. While pharmacogenomic are exploratory, they provide a basis for future investigation into their predictive value for targeted therapy.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-025-03800-9.

Introduction

Endometrial carcinoma (EC) ranks among the most prevalent gynecological malignancies worldwide, particularly affecting women in the perimenopausal and postmenopausal stages [1]. Epidemiological studies over the past decade have shown that, a continuous rise in both the incidence and mortality of EC, underscoring its increasing burden on public health systems and its profound impact on women’s health and overall quality of life [2, 3]. Although advances in diagnostic imaging, surgical techniques, and adjuvant therapies have improved outcomes for patients with early-stage disease, the prognosis remains unsatisfactory for those with advanced or recurrent EC, thereby necessitating the identification of novel biomarkers and therapeutic targets [4].

The molecular etiology of EC is multifactorial, with contributions from hereditary susceptibility [5], hormonal imbalances—particularly elevated estrogen levels [6], and somatic gene alterations [7]. Despite growing insights into these mechanisms, effective strategies for early detection and individualized treatment are still lacking. Thus, the discovery of robust molecular biomarkers is critical to enable timely diagnosis and to facilitate the development of precision medicine approaches.

In recent years, the development of bioinformatics has made it possible to rapidly study the expression of tumor genes or cells under specific conditions. With the advent of high-throughput technologies and the increasing availability of public genomic datasets, bioinformatics has emerged as a powerful approach for elucidating the molecular landscape of complex diseases [8]. Integrated transcriptomic analyses now allow for the comprehensive interrogation of gene expression profiles, aiding in the identification of candidate genes associated with tumor initiation, progression, and patient prognosis [9, 10]. In this study, we employed a systematic bioinformatics workflow to identify differentially expressed genes (DEGs) relevant to EC by analyzing data from the GEO and DisGeNET databases. Protein–protein interaction networks, functional enrichment, and pathway analyses were conducted to uncover key molecular players. The expression and prognostic relevance of selected hub genes were further validated using clinical databases and experimental methods. These efforts aim to lay a foundational framework for the development of novel diagnostic markers and therapeutic strategies for endometrial carcinoma.

Materials and methods

Data sources

Publicly available gene expression datasets were retrieved from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/) using the search term “endometrial cancer.” The GSE63678 dataset was selected for this study, comprising a total of 35 samples. Total RNA was extracted from normal and cancerous tissues and hybridized on Affymetrix HG133_A_2.0 microarray chips corresponding to more than 12.000 uniquely represented genes. For differential expression analysis, a subset including 7 endometrial carcinoma (EC) tissue samples and 5 normal endometrial tissue samples was used after excluding other disease samples. Additionally, EC-associated genes were obtained from the DisGeNET database (https://www.disgenet.org/) using the same keyword to facilitate integrative analysis.

Methods

Identification and visualization of differentially expressed genes (DEGs)

The GSE63678 dataset was accessed and downloaded using the GEOquery package (v2.64.2) in R(v4.2.1). To ensure data accuracy, probes annotated to multiple genes were excluded, and for genes with multiple corresponding probes, the probe exhibiting the highest expression signal was retained. Differential gene expression analysis was carried out using the limma package (v3.52.2). Genes exhibiting an absolute log fold change (|logFC|) ≥ 1.5 and a P-value < 0.05 were considered statistically significant. Visualization of the differential expression results was performed through volcano plots generated by limma and heatmaps constructed using the ComplexHeatmap package (v2.13.1). To identify EC-specific targets, differentially expressed genes (DEGs) from GEO were intersected with EC-related genes obtained from the DisGeNET database using an online Venn diagram tool (http://bioinformatics.psb.ugent.be/webtools/Venn/). This knowledge-based filtering strategy was applied deliberately to increase disease specificity by retaining only DEGs already supported by curated evidence of association with endometrial carcinoma. To enhance the reliability of DEG identification, an additional endometrial carcinoma dataset, GSE17025, was also analyzed using the same preprocessing and statistical pipeline. This dataset was selected due to its larger sample size and its relevance to EC, allowing cross-validation of gene expression patterns observed in GSE63678.

Protein–protein interaction (PPI) analysis and identification of hub genes

To investigate the functional connectivity among proteins encoded by EC-related DEGs, a protein–protein interaction (PPI) network was constructed using the STRING database (https://string-db.org/, version 11.0), specifying “Homo sapiens” as the organism of interest. The interaction network data were subsequently imported into Cytoscape software for graphical visualization and further analysis. Hub genes within the PPI network were identified using the cytoHubba plugin, which applies a range of topological algorithms to assess node importance. These included Degree, Edge Percolated Component (EPC), Maximum Neighborhood Component (MNC), Density of Maximum Neighborhood Component (DMNC), Maximal Clique Centrality (MCC), Bottleneck, Eccentricity, Closeness, Radiality, Betweenness, and Stress centrality measures.

GO functional and KEGG pathway enrichment analyses

Gene Ontology (GO) enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were conducted to explore the biological significance of the identified DEGs. These analyses were performed using the ClueGO plugin (v2.5.7) within the Cytoscape environment, with “Homo sapiens” designated as the reference organism. The enrichment analysis encompassed key GO domains—biological processes, molecular functions, and cellular components—as well as relevant KEGG signaling pathways. A P-value < 0.05 was set as the threshold for statistical significance.

Expression validation of hub genes

Application method 1.2.1 Screening differential targets in GSE17025 chip data and verifying key targets. The differential expression of the ten identified hub genes—cdk1, aurka, rad51, ccnb1, ube2c, cdc20, chek1, tyms, bub1b, and tpx2—was validated using two independent online platforms: GEPIA (http://gepia.cancer-pku.cn/) and the Xiantao platform (https://www.xiantaozi.com/). Both platforms utilize RNA sequencing data from The Cancer Genome Atlas (TCGA) and the Genotype-Tissue Expression (GTEx) project. Gene expression comparisons were performed based on the limma package, with statistical significance defined as P < 0.05.

Survival and diagnostic ROC analyses

Transcriptomic and clinical data for endometrial carcinoma were obtained from the TCGA-UCEC cohort through the Xiantao online platform. RNA-seq data, normalized to transcripts per million (TPM) using the STAR pipeline, were preprocessed to exclude duplicate entries and samples lacking clinical annotations. Survival analysis of the hub genes was performed using the survival package (v3.3.1) in R, with graphical visualization via the survminer (v0.4.9) and ggplot2 (v3.4.4) packages. Receiver operating characteristic (ROC) curve analyses were conducted using the pROC package (v1.18.0), and visual outputs were generated with ggplot2 to assess the diagnostic performance of the candidate genes.

Immunohistochemical (IHC) analysis

Formalin-fixed, paraffin-embedded tissue samples were collected from 10 patients with endometrial carcinoma and 10 healthy individuals at Baoding First Central Hospital. The expression of AURKA (Proteintech, Wuhan, China, A00246-4), BUB1B (Proteintech, Wuhan, China, 83920-6RR), TPX2 (Proteintech, Wuhan, China, 11741-1-AP), and UBE2C (Proteintech, Wuhan, China, 12134-2-AP) was assessed using immunohistochemistry (IHC). For each specimen, five random high-power fields (400× magnification) were evaluated.

Staining was scored based on two criteria: staining intensity (0 = no staining or background, 1 = light yellow, 2 = yellow-brown, 3 = dark brown) and the percentage of positive cells (0 = < 5%, 1 = 5–25%, 2 = 25–50%, 3 = 50–75%, 4 = > 75%). The final immunohistochemical score is calculated by multiplying the intensity score and the percentage score. A total score greater than 2 was considered positive expression, while a score of 2 or below was defined as negative. Incubation of sections without primary antibodies was considered a negative control. All histopathological assessments were independently reviewed by experienced pathologists. Ethical approval was obtained from the Clinical Ethics Committee of Baoding First Central Hospital, and written informed consent was secured from all participants (approval number: Fast [2024] No.99). The baseline clinical characteristics of the patients included in the IHC analysis are summarized in Table 1.

Table 1.

Baseline clinical characteristics of patients included in the immunohistochemistry (IHC) analysis

Clinical characteristics		Total	%
Age at diagnosis		33–79
Histopathology	Endometroid adenocarcinoma	8	80
	Serous carcinoma	1	10
	Clear cell carcinoma	1	10
FIGO cancer staging system	Stage I	2	20
	Stage II	5	50
	Stage III	3	30
	Stage IV	0	0
Myometrial invasion	< 50%	7	70
	≥ 50%	3	30

This table summarizes demographic and clinical parameters of 10 patients with endometrial carcinoma and 10 healthy controls, including age, tumor stage, histological type, tumor grade, and menopausal status. These data provide the clinical context for evaluating protein expression patterns in tissue samples via IHC

Immune infiltration analysis

The relationship between the expression of the hub genes bub1b, tpx2, and ube2c and immune cell infiltration in endometrial carcinoma tissues was analyzed using the TIMER 2.0 database (https://cistrome.shinyapps.io/timer/). The “Gene” module was used to input each gene individually, and “Uterine Corpus Endometrial Carcinoma” was selected as the cancer type. Correlation analyses were conducted to assess the association of gene expression with the infiltration levels of six immune cell types: B cells, CD8⁺ T cells, CD4⁺ T cells, macrophages, neutrophils, and dendritic cells. The Pearson correlation coefficient (r) was used to quantify the strength and direction of the association, where r > 0 indicated a positive correlation and r < 0 indicated a negative correlation. Statistical significance was defined as P < 0.05.

Expression of hub genes in different clinical stages

To evaluate the expression patterns of bub1b, tpx2, and ube2c across different clinical stages of endometrial carcinoma, analyses were conducted using the UALCAN database (https://ualcan.path.uab.edu/). Each gene was individually queried under the “TCGA Gene” module, with “Endometrial Cancer” selected as the tumor type. The “Tumor Analysis” option within the expression profile section was used to compare gene expression levels across stages I to IV. Statistical significance was defined as P < 0.05. Using the R package TCGAbiolinks (v2.31.2) and Summarized Experiment (v1.30.2) to download the tpm expression level data and clinical grade staging data of TCGA-UCEC, extract the expression levels of bub1b, tpx2, and ube2c, and use the R package ggplot2(v3.4.4) to draw box plots of the expression levels of each gene in each stage.

Identification of candidate drugs

The potential drug sensitivities associated with the hub genes ube2c, bub1b, and tpx2 were evaluated using the Genomics of Drug Sensitivity in Cancer (GSCA) platform (https://guolab.wchscu.cn/GSCA/#/drug). This analysis aimed to identify candidate therapeutic agents by correlating gene expression levels with drug response profiles across various cancer cell lines. Analyses reflect aggregated multi–cell line correlations from the GDSC pharmacogenomic resource; we did not compile per–cell line IC₅₀/AUC values in the present study.

Statistical analysis

All statistical analyses were conducted using SPSS software (version 27.0). Prior to hypothesis testing, data were assessed for normality to determine the appropriate statistical approach. Comparisons between two groups were performed using either independent-samples t-tests for continuous variables and X² or Fisher’s exact tests (when expected counts were < 5) for categorical variables. Given the small IHC sample size (10 EC vs. 10 controls), categorical comparisons may be underpowered; therefore, results should be interpreted with caution. A P-value less than 0.05 was considered indicative of statistical significance.

Results

Identification of differential targets

To explore genes potentially involved in the molecular pathogenesis of endometrial carcinoma (EC), differential expression analysis was conducted using the GSE63678 microarray dataset. A total of 366 differentially expressed genes (DEGs) were identified based on the defined statistical thresholds (|logFC| ≥ 1.5, P < 0.05), comprising 219 upregulated and 147 downregulated genes. A volcano plot was generated to illustrate the global expression distribution and statistical significance of these DEGs (Fig. 1A), while a corresponding heatmap highlighted the most significantly dysregulated genes across EC and normal samples (Fig. 1B), with red and blue representing upregulation and downregulation, respectively. Using DisGeNET database for EC differential targets, the top 100 scores are shown in Fig. 1C.

Fig. 1.

Identification of endometrial carcinoma (EC)-related differentially expressed genes (DEGs). A Volcano plot of DEGs from the GSE63678 dataset, showing significantly upregulated (red) and downregulated (blue) genes (|log₂FC| ≥ 1.5, P-adj < 0.05). B Heatmap of top 50 DEGs differentiating EC and normal endometrial tissues. C Disease–gene network of EC-related genes from the DisGeNET database, visualized using Cytoscape. For clarity, hub nodes with the highest connectivity are emphasized, while peripheral nodes were minimized. The complete unedited DisGeNET network is provided as Supplementary Figure S1. D Venn diagram showing 63 overlapping genes between GSE63678-derived DEGs and DisGeNET EC-related genes. These overlapping genes were considered EC-specific and used for subsequent analyses

To enhance disease specificity, these DEGs were cross-referenced with EC-related gene sets curated in the DisGeNET database. This integrative approach yielded 63 overlapping genes with established associations to endometrial carcinoma, representing a refined subset of EC-specific DEGs (Fig. 1D). Because DisGeNET aggregates heterogeneous evidence from literature, GWAS, and multiple annotation resources, which do not necessarily align with the microarray platforms analyzed here, a moderate overlap (~ 17%) was expected. This conservative intersection favors specificity over sensitivity, and to mitigate potential exclusion of true positives, we validated the expression of hub genes in an independent dataset (GSE17025) and across TCGA-based platforms (GEPIA, Xiantao). These genes served as the foundation for subsequent functional enrichment and network-based analyses. The complete list of all 366 DEGs, including gene symbol, log₂FC, Benjamini–Hochberg adjusted P-value, and regulation direction (up/down), is provided in Supplementary Table S1.

Construction and analysis of the protein–protein interaction (PPI) network

To investigate the functional interconnectivity of the 63 EC-related differentially expressed genes (DEGs), a protein–protein interaction (PPI) network was constructed using data retrieved from the STRING database. Interaction data, including both experimentally validated and computationally predicted associations, were imported into Cytoscape for visualization and network analysis. In the generated network (Fig. 2), each node represents a protein product of a DEG, while edges denote known or predicted interactions. Node coloring—from yellow to red—reflects increasing degree centrality, with red indicating proteins of higher connectivity and potential biological significance.

Fig. 2.

Protein–protein interaction (PPI) network of endometrial carcinoma-associated differentially expressed genes (DEGs). The network was constructed using the STRING database and visualized in Cytoscape to highlight key interactions among the 63 EC-specific DEGs identified through integration of GSE63678/GSE17025 datasets and the DisGeNET database. Nodes represent proteins encoded by DEGs, and edges represent predicted functional associations. Hub genes with high connectivity are centrally located, supporting their potential regulatory importance in EC pathogenesis

Topological analysis of the network was conducted using the cytoHubba plugin in Cytoscape, which ranked nodes based on various centrality algorithms. From this analysis, ten hub genes with the highest degrees of connectivity and presumed functional importance were identified: Cyclin-dependent kinase 1 (cdk1 / CDK1), Aurora kinase A (aurka / AURKA), RAD51 recombinase (rad51 / RAD51), Cyclin B1 (ccnb1 / CCNB1), Ubiquitin-conjugating enzyme E2 C (ube2c / UBE2C), Cell division cycle 20 (cdc20 / CDC20), Checkpoint kinase 1 (chek1 / CHEK1), Thymidylate synthase (tyms / TYMS), BUB1 mitotic checkpoint serine/threonine kinase B (bub1b / BUB1B), and Targeting protein for Xklp2 (tpx2 / TPX2). These hub genes were prioritized for further expression validation and functional characterization due to their potential roles in EC pathogenesis.

Enrichment analysis of GO biological processes and KEGG pathways

To gain insight into the functional roles of the identified EC-specific DEGs, Gene Ontology (GO) and Kyoto Encyclopedia of genes and Genomes (KEGG) pathway enrichment analyses were performed using the ClueGO plugin in Cytoscape. GO enrichment revealed that the DEGs were significantly associated with key biological processes, particularly those governing cell division and mitotic regulation. Notably, terms such as negative regulation of the cell cycle and regulation of mitotic cell cycle phase transition were highly enriched, suggesting a central role of cell cycle dysregulation in EC progression (Fig. 3A).

Fig. 3.

GO and KEGG enrichment analysis of EC-specific differentially expressed genes (DEGs). A Biological process enrichment showing key terms such as regulation of mitotic cell cycle phase transition and cell cycle checkpoints. B Cellular component analysis highlighting enrichment in chromosomal regions and anaphase-promoting complex. C Molecular function terms include histone kinase activity and DNA-binding functions. D KEGG pathway analysis reveals significant involvement in p53 signaling, cell cycle, HIF-1 signaling, and other cancer-related pathways. Dot size reflects the number of genes involved; color scale indicates statistical significance (adjusted P-value). These enrichments underline the roles of the DEGs in mitotic regulation, genomic integrity, and oncogenic signaling in endometrial carcinoma

KEGG pathway analysis further highlighted the involvement of these genes in well-established oncogenic pathways (Fig. 3B). The most significantly enriched pathways included the p53 signaling pathway, which is critical for DNA damage response and tumor suppression, and the HIF-1 signaling pathway (Fig. 3C), known for its role in cellular adaptation to hypoxia and angiogenesis—both highly relevant in the tumor microenvironment of EC (Fig. 3D). Bubble plots were generated to visualize these enrichments, where the Y-axis represents the biological processes or pathways, the X-axis indicates the gene ratio, the size of each bubble corresponds to the number of genes involved, and color intensity reflects statistical significance based on adjusted P-values.

2.4 Analysis of cdk1, aurka, rad51, ccnb1, ube2c, cdc20, chek1, tyms, bub1b, and tpx2 mRNA Expression in EC Tissues.

To strengthen the reliability of our DEG selection, we first validated the expression of the ten hub genes in an independent dataset, GSE17025, which includes a larger sample size of EC and normal tissues. The expression profiles from this dataset confirmed that these genes are consistently upregulated in endometrial carcinoma (Fig. 4A). To validate the ctranscriptional relevance of the ten identified hub genes in endometrial carcinoma, mRNA expression levels were analyzed using two independent bioinformatics platforms: GEPIA and Xiantao. In the unpaired analysis performed via the GEPIA database, all ten genes—cdk1, aurka, rad51, ccnb1, ube2c, cdc20, chek1, tyms, bub1b, and tpx2—were found to be significantly upregulated in EC tumor tissues (n = 147) compared to normal endometrial samples (n = 91) (Fig. 4B).

Fig. 4.

Comparative mRNA expression analysis of hub genes in endometrial carcinoma (EC) tissues. A Relative expression levels of ten key genes (aurka, bub1b, ccnb1, cdc20, cdk1, chek1, rad51, tpx2, tyms, and ube2c) in EC versus normal tissues based on the GSE63678 (n = 12; 7 EC and 5 normal) and GSE17025 (n = 103) datasets. B Unpaired analysis of mRNA expression using the GEPIA platform, showing significant upregulation of all ten genes in EC tumor samples (n = 147) compared to normal controls (n = 91). C Paired analysis of tumor versus adjacent normal tissues (n = 23 pairs) using the Xiantao platform, confirming consistent overexpression of these genes. These findings support the transcriptional upregulation of the selected hub genes in EC and reinforce their potential as diagnostic and therapeutic targets

Consistent results were observed in the paired dataset obtained from the Xiantao platform, where mRNA levels of these genes were uniformly elevated in tumor tissues (n = 23) relative to their matched adjacent normal tissues (n = 23) (Fig. 4C). In addition, immunohistochemistry (IHC) was performed on paraffin-embedded tissue specimens from 10 EC patients and 10 controls, further confirming the overexpression of AURKA, BUB1B, TPX2, and UBE2C in patient-derived samples (Fig. 6). These findings demonstrate robust expression trends across multiple datasets and analytical tools, reinforcing the biological relevance of these hub genes in endometrial cancer and supporting their potential as diagnostic or therapeutic targets.

Fig. 6.

Immunohistochemical detection of AURKA, BUB1B, TPX2, and UBE2C expression in EC and normal endometrial tissues (400× magnification). Panels A–D represent AURKA, BUB1B, TPX2, and UBE2C expression in normal endometrial tissues, respectively; Panels E–H represent the corresponding expression in endometrial carcinoma tissues. Scale bars are included in all panels for reference. Negative controls (isotype/secondary antibody–only) were performed during staining optimization to confirm antibody specificity but are not shown here; representative examples are available upon request

Kaplan–Meier survival and diagnostic ROC analyses of hub gene expression in EC tissues

The prognostic relevance of the ten hub genes—aurka, bub1b, ccnb1, cdc20, cdk1, chek1, rad51, tpx2, tyms, and ube2c—was evaluated using clinical and transcriptomic data from the Xiantao platform. Kaplan–Meier survival analysis revealed that higher expression levels of aurka, bub1b, tpx2, and ube2c were significantly associated with poorer overall survival in endometrial carcinoma patients. The corresponding hazard ratios (HR) and P-values were as follows: aurka (HR = 2.22, P < 0.01), bub1b (HR = 1.72, P = 0.011), tpx2 (HR = 1.85, P = 0.004), and ube2c (HR = 1.82, P = 0.005), indicating their potential as adverse prognostic indicators (Fig. 5A). These analyses were conducted as univariable comparisons and did not adjust for potential confounding factors such as age, stage, grade, or treatment. Therefore, results should be interpreted with caution, and validation in multivariable models using larger cohorts will be necessary.

Fig. 5.

Prognostic and diagnostic significance of hub genes in endometrial carcinoma (EC). (A) Kaplan–Meier survival curves showing overall survival differences in EC patients with high vs. low expression of ten hub genes. Genes such as aurka, bub1b, tpx2, and ube2c showed significant associations with poor survival. (B) Receiver operating characteristic (ROC) curves evaluating the diagnostic accuracy of each hub gene for EC. All genes demonstrated strong diagnostic power, with AUC values above 0.88, and ube2c showing the highest AUC (0.971)

In contrast, no statistically significant survival differences were observed for ccnb1, cdc20, cdk1, chek1, rad51, or tyms, with P-values exceeding the conventional threshold for significance. To further evaluate their diagnostic potential, receiver operating characteristic (ROC) curve analyses were conducted. Notably, aurka, bub1b, tpx2, and ube2c demonstrated outstanding discriminatory power in distinguishing EC tissues from normal controls, with areas under the curve (AUCs) of 0.984 (95% CI: 0.972–0.995), 0.975 (95% CI: 0.952–0.999), 0.992 (95% CI: 0.984–1.000), and 0.998 (95% CI: 0.996–1.000), respectively (Fig. 5B). Because these AUC values are unusually high, they may reflect cohort composition or potential overfitting, rather than true diagnostic performance. These diagnostic metrics are therefore exploratory, and independent external validation cohorts will be required to confirm their robustness. These results highlight these four genes as promising candidates for both diagnostic biomarker development and prognostic stratification in endometrial cancer.

Immunohistochemical (IHC) analysis

To validate the protein-level expression of selected hub genes in endometrial carcinoma, immunohistochemical (IHC) analysis was performed on formalin-fixed, paraffin-embedded tissue sections obtained from 15 EC patients and 10 healthy controls (Fig. 6). The examined proteins included AURKA, BUB1B, TPX2, and UBE2C. IHC staining revealed cytoplasmic localization of AURKA, whereas BUB1B, TPX2, and UBE2C were detected in both the cytoplasm and nucleus, consistent with their known cellular functions.

The expression patterns varied notably between cancerous and normal tissues. AURKA showed a positive expression rate of 30% (3/10) in EC tissues versus 10% (1/10) in normal tissues; however, this difference did not reach statistical significance (P > 0.05) (Fig. 6A and E). In contrast, BUB1B and TPX2 were each positively expressed in 80% (8/10) of EC samples, compared to only 10% (1/10) in controls, indicating significant overexpression (P < 0.01) (Fig. 6B, C and F, and G). UBE2C demonstrated the most marked differential expression, with 90% (9/10) positivity in EC tissues versus 20% (2/10) in normal tissues (P < 0.01) (Fig. 6D and H). These results corroborate the transcriptomic findings and provide strong experimental support for the diagnostic and potential therapeutic utility of BUB1B, TPX2, and UBE2C in endometrial cancer.

Expression levels of bub1b, tpx2, and ube2c mRNA in EC tumors across different clinical stages

To determine whether the expression of key hub genes is associated with disease progression, the transcript levels of bub1b, tpx2, and ube2c were analyzed across different clinical stages of endometrial carcinoma using the UALCAN platform, based on TCGA datasets (Fig. 7A–C). All three genes exhibited significantly elevated expression in tumor tissues compared to normal endometrium (P < 0.01), and this upregulation persisted consistently across stages I, II, III, and IV. In addition, we assessed gene expression according to tumor grade using TCGA-UCEC clinical data. Compared with normal tissues, the expression levels of bub1b, tpx2, and ube2c mRNA were significantly increased in EC tissues across grades G1, G2, and G3 (P < 0.01) (Fig. 7D–F). These findings indicate that overexpression of bub1b, tpx2, and ube2c is not only associated with tumor initiation but is also sustained across advancing clinical stages and histological grades, reinforcing their utility as biomarkers of endometrial cancer progression.

Fig. 7.

Expression patterns of bub1b, tpx2, and ube2c across clinical stages and histological grades in endometrial carcinoma (EC). A–C Box plots showing transcript levels of bub1b, tpx2, and ube2c across clinical stages I–IV based on TCGA-UCEC data from the UALCAN platform. All three genes demonstrate consistently elevated expression across progressive tumor stages. D–F mRNA expression of the same genes analyzed by histological grade (G1–G3) using TCGA data. Gene expression levels increase with higher tumor grade, supporting their role in disease aggressiveness. All comparisons show statistical significance (P < 0.01), highlighting the potential of bub1b, tpx2, and ube2c as markers of both EC onset and progression. Box plots display the median, interquartile range (IQR), and whiskers representing 1.5× IQR; individual dots indicate outliers. Thus, the error bars represent data spread based on IQR rather than standard deviation (SD) or standard error of the mean (SEM)

Correlation between bub1b, tpx2, and ube2c mRNA expression and immune cell infiltration

To explore the immunological context of key hub genes in the tumor microenvironment of endometrial carcinoma, the correlation between bub1b, tpx2, and ube2c mRNA expression and immune cell infiltration was evaluated using the TIMER 2.0 database. The analysis revealed distinct immune interaction patterns for each gene.

bub1b expression was positively correlated with neutrophil infiltration, suggesting a potential role in inflammatory signaling, while showing negative associations with B cells and CD4⁺ T cells (Fig. 8A). Similarly, tpx2 expression demonstrated a positive correlation with neutrophils, but was inversely associated with B cells and macrophages (Fig. 8B), indicating possible involvement in immune evasion mechanisms. ube2c showed the most widespread immune associations: it was positively correlated with tumor purity and neutrophil levels, yet negatively correlated with B cells, CD8⁺ T cells, macrophages, and dendritic cells (Fig. 8C). These findings highlight the potential roles of these genes not only in tumor proliferation but also in modulating the immune microenvironment, which may have implications for immunotherapy responsiveness in EC. However, TIMER primarily relies on public databases such as TCGA. Its sample selection criteria (such as tumor staging and treatment history) may affect the representativeness of the results, and its tumor purity correction algorithm may introduce biases into some correlation analyses.

Fig. 8.

Correlation between hub gene expression and immune cell infiltration in endometrial carcinoma (EC). Scatter plots generated using the TIMER 2.0 platform show the relationship between the expression levels of (A) bub1b, (B) tpx2, and (C) ube2c with tumor purity and various immune cell populations, including B cells, CD8⁺ T cells, CD4⁺ T cells, macrophages, neutrophils, and dendritic cells. Partial correlation coefficients (partial.cor) and p-values indicate significant negative associations between gene expression and immune infiltration, particularly for neutrophils and dendritic cells, suggesting an immunosuppressive tumor microenvironment linked to overexpression of these genes

Drug sensitivity analysis

To investigate the potential therapeutic relevance of the identified hub genes, drug response correlations were assessed using the Genomics of Drug Sensitivity in Cancer (GSCA) platform (Fig. 9). Analysis shows that the top three drugs related to tpx2 are trametinib, Nutlin-3a (-), and selumetinib, while the top three drugs related to ube2c are masitinib, TL-1-85, and BHG712. Exploratory GSCA analysis indicated nominal positive correlations between tpx2/ube2c expression and trametinib response (r = 0.14 and r = 0.19). However, these correlation coefficients are negligible by standard interpretation (r < 0.3) and therefore lack predictive value. Findings are presented descriptively only and require independent experimental validation. The complete GSCA output, including correlation coefficients, FDR-adjusted P-values, and drug response scores, is provided in Supplementary Table S3. These results represent aggregated multi–cell line correlations from the GDSC pharmacogenomic resource; raw IC₅₀/AUC values for individual cell lines are not reproduced here and can be accessed directly from the GDSC repository.

Fig. 9.

Correlation between mRNA expression of tpx2 and ube2c and drug sensitivity across GDSC compounds. Bubble plot illustrating the relationship between expression levels of tpx2 and ube2c and sensitivity to various anticancer agents based on the Genomics of Drug Sensitivity in Cancer (GDSC) database. The size of each bubble indicates the –log₁₀(FDR), while the color scale represents the correlation coefficient. Additionally, representative molecular docking images of trametinib with TPX2 and UBE2C are shown, with interaction types explicitly labeled: hydrogen bonds (green dashed lines), hydrophobic contacts (yellow dashed lines), and π–π interactions (purple lines). Distances are indicated in Å to provide clarity on binding modes. Drug–gene associations represent aggregated correlations across ~ 1,000 human cancer cell lines in the GDSC database; raw per–cell line IC₅₀/AUC values are not included here and can be accessed directly from the GDSC repository

Discussion

Endometrial carcinoma (EC) remains one of the most prevalent gynecological malignancies worldwide, representing approximately 30% of all female reproductive system cancers [11]. Despite ongoing advancements in surgical techniques, molecular diagnostics, and targeted therapies, mortality associated with EC—particularly in advanced or recurrent cases—continues to be a major clinical challenge. A significant proportion of cases are diagnosed at later stages due to the disease’s often subtle or nonspecific early symptoms. Consequently, the discovery and validation of reliable molecular biomarkers are critical to improving early diagnosis, prognostication, and the personalization of treatment strategies.

In this study, we leveraged integrative bioinformatics to identify 366 differentially expressed genes (DEGs) in EC, 63 of which overlapped with curated EC-related genes from DisGeNET. These DEGs were implicated in key oncogenic signaling cascades, particularly the p53 and HIF-1 pathways, both of which are central to cell cycle regulation, apoptosis, and tumor adaptation to hypoxia. Gene ontology (GO) enrichment further confirmed their roles in critical biological functions such as mitotic regulation and cell cycle progression. Notably, hub gene analysis identified aurka, bub1b, tpx2, and ube2c as key players in the EC transcriptome.

To strengthen the reliability of DEG identification, we validated the expression of these hub genes using an independent cohort (GSE17025) in addition to the original GSE63678 dataset. The concordant expression patterns across both datasets demonstrate that these genes are consistently dysregulated in EC, reinforcing their potential as reproducible biomarkers across different populations.

Among these, BUB1B emerged as a compelling candidate due to its central role in the spindle assembly checkpoint (SAC), where it safeguards against chromosomal missegregation during mitosis. Dysregulation of BUB1B disrupts mitotic fidelity, contributing to aneuploidy and tumorigenesis [12]. Its overexpression has been documented across multiple tumor types, including prostate [13], breast [14], lung adenocarcinoma [15], and ovarian cancers [16]. In colorectal carcinoma, BUB1B suppression has been shown to inhibit the JNK/c-Jun pathway and alter the apoptotic balance, suppressing malignancy [17]. In the present study, bub1b was significantly overexpressed in EC tissues and correlated with poorer survival, reinforcing its prognostic and potentially therapeutic relevance.

TPX2, a microtubule-associated protein involved in mitotic spindle assembly, is another gene with established oncogenic functions. Elevated TPX2 expression disrupts chromosomal segregation and genomic stability, facilitating malignant transformation. Its interaction with KIF4A, which inhibits TPX2 ubiquitination, has been shown to promote EC progression by maintaining mitotic integrity [18]. Furthermore, TPX2 regulates chemokine signaling via the CX3CR1/CXCL10–PI3K/Akt axis, contributing to immune evasion and tumor progression [19]. In our study, TPX2 overexpression was associated with neutrophil infiltration and poor prognosis, indicating its dual role in tumor proliferation and immune modulation.

UBE2C, a core member of the E2 ubiquitin-conjugating enzyme family, regulates the ubiquitin–proteasome degradation of mitotic regulators via the anaphase-promoting complex/cyclosome (APC/C). Overexpression of UBE2C has been implicated in the progression of various cancers and is frequently associated with higher tumor grade and poor prognosis [15]. In EC, UBE2C expression is driven by estrogen receptor-mediated transcription, linking hormonal signaling to genomic instability [20]. Previous studies have shown that UBE2C promotes epithelial-to-mesenchymal transition (EMT) through modulation of the p53 pathway, facilitating invasiveness and therapeutic resistance [15]. Our immune infiltration analysis further demonstrated that ube2c expression negatively correlated with macrophage and dendritic cell infiltration, suggesting an immunosuppressive tumor phenotype consistent with immune escape.

The immunological analysis revealed a complex interplay between gene expression and tumor-infiltrating immune cells. High expression of tpx2 and ube2c was associated with neutrophil enrichment and a concomitant reduction in B cells, CD8⁺ T cells, and antigen-presenting cells, such as dendritic cells and macrophages. These patterns suggest that these genes may contribute to immune exclusion or an immunosuppressive microenvironment—an increasingly recognized hallmark of aggressive EC subtypes. Given the growing interest in immunotherapy for EC, particularly immune checkpoint inhibitors, these genes could also serve as indicators of immune phenotype and responsiveness to immunomodulatory therapies.

While the findings of this study provide valuable mechanistic and translational insights, several limitations must be acknowledged. The reliance on publicly available datasets and in silico predictions, although robust, calls for further validation in large-scale prospective cohorts and functional assays. Moreover, immune infiltration data derived from bulk transcriptomic profiles may lack resolution compared to single-cell or spatial transcriptomics approaches. Although GSCA analysis suggested possible drug associations such as trametinib, the observed correlations with tpx2/ube2c expression were negligible (r < 0.2). Thus, any potential therapeutic relevance should be regarded as exploratory only, requiring experimental validation before clinical inference. Given the weak drug–gene correlations and the small size of the IHC validation cohort, these results should be interpreted as hypothesis-generating rather than confirmatory. Nonetheless, the combined use of differential expression, survival analysis, IHC validation and immune profiling offers a comprehensive multi-level assessment of potential biomarkers.

In conclusion, this study identifies bub1b, tpx2, and ube2c as key genes involved in the pathogenesis and prognosis of endometrial carcinoma. Their overexpression is linked to poor clinical outcomes, immune modulation. The drug sensitivity findings are preliminary and descriptive, and the overall conclusions should be considered exploratory due to the limited cohort size. These findings provide preliminary evidence supporting further investigation of these genes as potential biomarkers, with confirmatory validation in larger, independent cohorts and functional studies required before clinical translation.

Conclusion

This study employed an integrative bioinformatics and experimental framework to identify and validate key molecular targets associated with endometrial carcinoma (EC). Among the differentially expressed genes, bub1b, tpx2, and ube2c emerged as critical players implicated in mitotic regulation, genomic instability, and tumor progression. Their overexpression in EC tissues, association with adverse clinical outcomes, and correlation with immune cell infiltration underscore their dual role in tumor biology and immune modulation. Incorporating two independent datasets (GSE63678 and GSE17025), we demonstrated the reproducibility of expression trends across multiple EC cohorts.

These results not only enhance our understanding of EC pathogenesis but also provide preliminary evidence suggesting these genes as potential candidates for biomarker development. Given the small validation cohort and weak drug–gene correlations, the findings should be regarded as exploratory and hypothesis-generating rather than confirmatory. Overall, the integration of transcriptomic profiling, immune landscape analysis, and computational predictions provides a comprehensive framework for identifying promising molecular targets. Future studies with larger cohorts and mechanistic experiments are required to confirm their diagnostic, prognostic, and therapeutic relevance before clinical application.

Limitation of study

While this study presents valuable insights into potential diagnostic and therapeutic targets in endometrial carcinoma, several limitations should be acknowledged. First, the bioinformatics analyses were primarily based on publicly available datasets, which, despite their robustness, may introduce bias due to variability in sample collection, processing, and batch effects. Although we used two independent cohorts (GSE63678 and GSE17025) to enhance reproducibility, additional datasets from prospective studies would further strengthen these findings.

Second, the transcriptomic data used were derived from bulk RNA-sequencing, which may mask cell-type-specific gene expression patterns and limit the resolution of tumor heterogeneity and immune infiltration dynamics. In particular, TIMER provides computational estimates of immune cell abundance from bulk transcriptomes, which may not fully reflect actual immune infiltration levels or spatial distribution in EC tissues. The integration of single-cell RNA-seq or spatial transcriptomic approaches in future studies could provide a more granular view of gene expression and cell–cell interactions within the tumor microenvironment.

Third, although immunohistochemical validation confirmed the overexpression of selected hub genes in EC tissues, the relatively small sample size (n = 10 EC and 10 control) limits the generalizability of the findings. Larger, multicenter cohorts are needed to validate the clinical relevance of bub1b, tpx2, and ube2c across diverse patient popula tions and disease subtypes. In addition, the IHC validation cohort was predominantly composed of endometrioid adenocarcinoma cases (80%), which may restrict the generalizability of our findings to non-endometrioid EC subtypes. Furthermore, statistical analyses of categorical variables in this small IHC cohort may be underpowered; Fisher’s exact test was applied when appropriate, but results should still be interpreted cautiously.

In addition, the survival analyses performed in this study were univariable and did not adjust for potential confounders such as age, stage, grade, or treatment, which may have influenced the associations observed. Similarly, the diagnostic ROC metrics were derived from public datasets and yielded unusually high AUC values (> 0.97), which may reflect overfitting or cohort-specific bias rather than true clinical performance. Finally, the drug–gene correlations identified through GSCA were very weak (r < 0.2), underscoring that these findings are descriptive and exploratory only, without predictive value until validated in independent experimental studies. These findings should therefore be considered exploratory, and independent external validation will be required to confirm their robustness.

Future studies

Building on the findings of this study, future research should aim to experimentally validate and functionally characterize the roles of bub1b, tpx2, and ube2c in the pathogenesis and progression of endometrial carcinoma (EC). In particular, loss- and gain-of-function approaches such as CRISPR-Cas9 gene editing, RNA interference, and overexpression models in EC cell lines and xenograft models can be employed to delineate the precise biological mechanisms regulated by these genes, including their impact on cell cycle regulation, DNA damage response, apoptosis, and epithelial–mesenchymal transition (EMT).

Given their association with immune cell infiltration, integrative immunogenomic profiling using single-cell RNA sequencing (scRNA-seq) or spatial transcriptomics could reveal how these genes modulate the tumor–immune interface—particularly with respect to the exclusion or suppression of cytotoxic T cells and antigen-presenting cells. This could open avenues for combinatorial therapeutic strategies, such as alongside immune checkpoint blockade, especially in immunologically “cold” EC subtypes that overexpress tpx2 and ube2c.

Additionally, prospective, multicenter clinical studies involving large, diverse patient cohorts are essential to validate the prognostic and predictive value of bub1b, tpx2, and ube2c. Their expression levels may be integrated into molecular risk stratification models to guide personalized therapy and improve clinical outcomes.

Author contributions

S.Z. and J.S. performed the bioinformatics analyses and data interpretation. J.L. and X.Z. conducted the immunohistochemistry experiments and validation studies. X.S. and M.Y. contributed to statistical analysis and figure preparation. Y.W. assisted with manuscript drafting and data curation. J.Z. conceived and supervised the study, and finalized the manuscript. All authors reviewed and approved the final version of the manuscript.

Funding

This research was supported by the project titled “Clinical study of a new biomarker cervical cytology screening method p16 series ICC” (Grant No. 2441ZF036).

Data availability

The datasets analyzed in this study are publicly available. Gene expression data were obtained from the Gene Expression Omnibus (GEO) under accession number GSE63678 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE63678). Additional gene–disease associations were retrieved from the DisGeNET database (https://www.disgenet.org/). All other data supporting the findings of this study are available from the corresponding author upon reasonable request. Per–cell line pharmacologic response values underlying our summary correlations can be accessed directly from the GDSC repository.

Declarations

Ethics approval and consent to participate

This study was approved by the Clinical Ethics Committee of Baoding First Central Hospital under approval number (Fast [2024] No.99). Formalin-fixed, paraffin-embedded tissue samples were obtained with written informed consent from all participants. All procedures involving human tissues were conducted in accordance with the ethical standards of the institutional research committee and the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.