1 Introduction

Endometrial carcinoma (EC) is highly prevalent gynecologic malignancies, accounting for 97,000 deaths and 417,000 new cases in 2020 [1]. Even with a variety of treatment modalities, the 5-year survival rate for EC is still quite low, especially in advanced stages [2]. Multiple therapeutic methods such as surgery, chemotherapy, and radiation therapy have been used to EC treatment, but the incidence and mortality is still increasing [3, 4]. Significant progress has been made in the molecular and genomic profiling of endometrial cancer, enabling better prediction of prognosis and treatment effectiveness in patients with this type of cancer [5, 6]. Therefore, it is imperative to identify novel molecules that can enhance early diagnosis, prognostic evaluation, and treatment outcomes for EC.

Given the tumor complexity, it is imperative to do a pan-cancer expression analysis of genes in order to evaluate association with prognosis and molecular pathways. To undertake a comprehensive and methodical inquiry in this regard, the functional genomics datasets relating to a broad spectrum of malignancies can be obtained by utilizing the publicly funded TCGA and GEO database.

A type I transmembrane protein called GPNMB is overexpressed in multiple malignancies, including as gliomas, triple-negative breast cancer, and melanoma [7,8,9]. GPNMB is essential for several biological processes involved in cancer, including apoptosis, metastasis, and proliferation [10]. We have previously investigated the potential functional connections between GPNMB and ovarian cancer carcinogenesis [11]. Nevertheless, the function of GPNMB in EC is still unknown. Using the GEO and the TCGA project, pan-cancer analysis of GPNMB was performed in this work. Gene expression, survival status, and genetic modification were among the factors examined in order to determine whether GPNMB may play a role in prognosis of cancers. Furthermore, we used in vitro experiments to confirm GPNMB’s suppressive effect in EC, and we also examined immune infiltration and pertinent cellular pathways.

2 Materials and methods

2.1 Gene expression analysis

The TIMER2web was used to assess expression levels of GPNMB based on TCGA. And the GEPIA2web server was used to analyze the expression levels of GPNMB based on GTEx. The CPTAC dataset of UALCAN was used to analyze protein expression levels of GPNMB.

2.2 Survival analysis

Using GEPIA2, the OS (Overall Survival) and DFS (Disease-free Survival) of GPNMB across all tumors were examined. The expression threshold was set at 50% to separate the high-expression and low-expression groups.

2.3 Genetic alteration analysis

The cBioPortal was used to search for the genetic modification characteristics of GPNMB, obtain the results of alteration frequency, mutation type, and CNA (Copy number alteration) of all TCGA tumors, investigate differences in survival between the TCGA cancer cases with and without genetic modifications, and generate the pertinent Kaplan–Meier graphs together with log-rank P values.

2.4 Cell culture and transfection

HEC‐1B and Ishikawa were acquired from the Shanghai Cell Bank of the Chinese Academy of Sciences (Shanghai, China). Cell lines were cultured and transfected using the procedures of our previous article [11].

2.5 Quantitative real-time PCR (qRT-PCR)

Total RNA were extracted, reverse transcript, qRT-PCR was performed using the procedures of our previous article [11]. Primers for GPNMB and β-actin (Beijing Genomics Institute, Beijing, China) were listed as follows: β-actin forward: 5′-TCCCTGGAGAAGAGCTACGA-3′; β-actin reverse: 5′-AGCACTGTGTTGGCGTACAG-3′; GPNMB forward: 5′-CTTCTGCTTACATGAGGGAGC-3′; GPNMB reverse: 5′-GGCTGGTGAGTCACTGGTC-3′; the 2−ΔΔCt method was utilized to analyse the expression levels.

2.6 Western blotting

Western blotting assays were used to analyze the protein expression levels of GPNMB in endometrial cancer cell lines, and Western blotting assays was performed using the procedures of our previous article, and membranes were cropped before incubated with antibodies [11].

2.7 Cell proliferation assay, cell migration and invasion assays

Cells transfected with plasmids were used to perform proliferation, migration and invasion assays, and above experiments were performed according to the procedures of our previous article [11].

2.8 GeneMANIA analysis and functional enrichment analysis

With the use of GeneMANIA, which generates genes with activities equivalent to GPNMB and displays the gene–gene interaction network to make connections between GPNMB, and clear, putative functions of GPNMB were predicted [12].

2.9 Relationship between GPNMB expression and immunity in endometrial cancer

The TIMER2 was used to study the relationship between GPNMB expression and immunological infiltrates in EC. Numerous immune cells were selected for examination. The purity-adjusted Spearman’s rank correlation test was used to calculate P values and partial correlation (cor) values. A scatter plot was then used to visually portray the data.

2.10 Statistical analysis

Student’s t test was used to analyze data. The prognostic values of high- and low-expression groups were evaluated according to the hazard ratio (HR), 95% confidence interval (CI), and log-rank P values. P value <0.05 indicated statistically significant differences. Experimental data are presented as the means ± SE and were analyzed using GraphPad Prism 6.0 (GraphPad Software, Inc.; La Jolla, CA, USA). Statistical differences were calculated by the 2-tailed unpaired Student t test. P < 0.05 was considered statistically significant.

3 Results

3.1 The expression of GPNMB in different cancer types

We first analyzed the expression pattern of GPNMB in pan-cancer through TIMER2. As Fig. 1a depicted, the expression level of GPNMB in the tumor tissues of BRCA (Breast invasive carcinoma), READ (Rectum adenocarcinoma), and UCEC (Uterine corpus endometrial carcinoma) was significantly lower than the corresponding control tissues. Conversely, the expression of GPNMB was higher in the tumor tissues of BLCA (Bladder urothelial carcinoma), CHOL (Cholangiocarcinoma), COAD (Colon adenocarcinoma), HNSC (Head and Neck squamous cell carcinoma), KICH (Kidney Chromophobe), KIRC (Kidney renal clear cell carcinoma), KIRP (Kidney renal papillary cell carcinoma), LIHC (Liver hepatocellular carcinoma), LUAD (Lung adenocarcinoma), LUSC (Lung squamous cell carcinoma), and STAD (Stomach adenocarcinoma) than the respective adjacent normal tissues.

Fig. 1
figure 1

Expression levels of GPNMB in different tumors and pathological stages. a The expression status of GPNMB in different cancers or specific cancer subtypes analyzed via TIMER2 (* P < 0.05; ** P < 0.01; *** P < 0.001); b The expression of DLBC, GBM, LGG, SKCM, TGCT, and THYM tissues 80 in the TCGA project and the corresponding normal tissues of the GTEx database were showed in 81 the box plot (* P < 0.05). c The expression levels of GPNMB total protein in primary tissue of breast 82 cancer, ovarian cancer, colon cancer, clear cell RCC, UCEC, and corresponding normal tissues were 83 analyzed using the CPTAC dataset (* P < 0.05; *** P < 0.001). d The expression levels of GPNMB in 84 the main pathological stages (stage I, stage II, stage III, and stage IV) of ESCA, SKCM, STAD, and 85 THCA were analyzed based on the TCGA datasets. Log2 (TPM + 1) was applied for the log scale

Full size image

We subsequently incorporated the normal tissue samples from the GTEx dataset as a control group and conducted an assessment of GPNMB expression levels, comparing them between the normal tissues and the tumor tissues of various cancer types: DLBC (Diffuse Large B-Cell Lymphoma, a type of Lymphoid Neoplasm), GBM (Glioblastoma Multiforme), LGG (Low-Grade Glioma of the Brain), SKCM (Skin Cutaneous Melanoma), TGCT (Testicular Germ Cell Tumors), and THYM (Thymoma). This analysis is depicted in Fig. 1b, where a statistically significant difference was observed.

The assessment of the CPTAC dataset indicated that in the primary tissues of breast cancer, ovarian cancer, colon cancer, UCEC, and LUAD, the protein levels of GPNMB were notably lower compared to those in normal tissues (Fig. 1c). Conversely, in clear cell RCC (Renal Cell Carcinoma), the protein level of GPNMB surpassed that of normal tissues (Fig. 1c). To delve deeper into the relationship between GPNMB expression and cancer pathological stages, we employed GEPIA2. Our findings unveiled a significant correlation in ESCA (Esophageal Carcinoma), SKCM, STAD, and THCA (Thyroid Carcinoma) (Fig. 1d).

3.2 Prognostic value of GPNMB in cancers

Utilizing the expression levels of GPNMB, we stratified cancers into high- and low-expression groups. The outcomes presented in Fig. 2a illuminated a significant association between heightened GPNMB expression and adverse OS outcomes in COAD (P = 0.021), LGG (P = 1.6e−05), MESO (Mesothelioma) (P = 0.022), STAD (P = 0.0011), and UVM (Uveal Melanoma) (P = 0.0014) within the TCGA cohort. Additionally, an analysis of DFS, depicted in Fig. 2b, disclosed a notable link between elevated GPNMB expression and poor prognosis in LGG (P = 0.018), PRAD (Prostate Adenocarcinoma) (P = 0.04), and UVM (P = 0.00025). Intriguingly, low GPNMB expression was correlated with unfavorable OS in KIRC (P = 0.044) and DFS in CESC (Cervical Squamous Cell Carcinoma and Endocervical Adenocarcinoma) (P = 0.029).

Fig. 2
figure 2

Correlation between GPNMB expression and prognosis of cancers in TCGA. GEPIA2 was used to analyze the OS (a) and DFS (b) of different tumors in TCGA by GPNMB expression levels

Full size image

3.3 Genetic alteration analysis data

Based on the TCGA cohorts, we examined the genetic alteration status of GPNMB in various tumors. Figure 3a revealed that the highest alteration frequency of GPNMB (>5%) was observed in bladder cancer patients, followed by those with endometrial cancer. Notably, endometrial cancer exhibited the highest frequency of “mutation,” suggesting a potential involvement of GPNMB in EC. Moreover, our investigation extended to exploring the relationship between the genetic alteration of GPNMB and the clinical survival prognosis across different cancer types. Figure 3b indicated that SKCM cases with altered GPNMB exhibited a poorer OS (P = 6.996e−5), DFS (P = 0.0211), and disease-specific survival (P = 5.361e−4) compared to cases without GPNMB alteration. Furthermore, the alteration of GPNMB was found to be inversely correlated with the disease-free survival (P = 8.990e−3) among STAD patients.

Fig. 3
figure 3

Mutation characteristics of GPNMB in different tumors of TCGA using cBioPortal. a Alteration frequency with mutation types of GPNMB in different tumors. b The potential correlation between mutation status and disease-free survival of STAD, and overall, disease-specific and progression-free survival of SKCM

Full size image

3.4 Identification of the GPNMB effect in endometrial cancer via in vitro validation

After reviewing previous studies, it is evident that GPNMB exhibits oncogenic or suppressive functions in cancers including bladder cancer [13], breast cancer [14] and so on. As for endometrial cancer, both mRNA and protein expression data showed that GPNMB was downregulated in EC tissues compared with the corresponding normal tissues. Hence, the focus of this study was to delve into the specific function of GPNMB in endometrial cancer. We confirmed the overexpression of GPNMB in EC cell lines (Fig. 4a, b). And CCK-8 assay showed (Fig. 4c) that overexpression of GPNMB significantly weakened endometrial cancer cell proliferation. The transwell assay (Fig. 4d) indicated that GPNMB overexpressed statistically inhibited the migration and invasion of endometrial cancer cells. In conclusion, the biological functions of GPNMB in endometrial cancer were elucidated through in vitro validation.

Fig. 4
figure 4

Effect of GPNMB overexpression on EC cell lines. a Western blotting results showed that GPNMB was successfully overexpressed by plasmid transfection. b qRT-PCR results showed that GPNMB was successfully overexpressed by plasmid transfection. c Transfection of GPNMB overexpression plasmid inhibited the growth of HEC-1B and Ishikawa cells. d Transfection of GPNMB overexpression plasmid significantly weakened the motility and invasion of HEC-1B and Ishikawa cells. All experiments were carried out in triplicate and data were presented as means ± SD. Calculation of statistical significance was performed using a two-tailed t test (* P < 0.05; ** P < 0.01; *** P < 0.001)

Full size image

3.5 Functions enrichment and immunity analysis of GPNMB in endometrial cancer

To unravel the potential mechanisms of GPNMB in endometrial carcinogenesis, we utilized GeneMANIA to establish a protein–protein interaction (PPI) network centred on GPNMB. This network revealed that 19 genes exhibited the strongest associations with GPNMB among the interconnected genes, including SDC4, PTK6, SMAD4, PMEL, SORCS2, ATP1A3, SORCS1, SORCS3, PLA2G4A, KIAA0319L, TMEM130, PKD1L1, APOCA, GPR137B, APOE, KIAA0319, CCL18, LIPA and SLC38A6. Functional analysis (Fig. 5b) revealed that the top 7 pathways related were regulation of plasma lipoprotein particle levels, neuropeptide signaling pathway, regulation of ERK1 and ERK2 cascade, regulation of cell activation, negative regulation of growth, IL-18 signaling pathway and metal ion transport.

Fig. 5
figure 5

Functions enrichment and immunity analysis of GPNMB. a Genes associated with GPNMB using GeneMANIA. b KEGG pathway analysis based on the GPNMB-related genes. c Correlation between GPNMB expression and tumor-infiltrating immune cells in EC

Full size image

Tumor-infiltrating immune cells were important components of the tumor microenvironment, which were closely correlated with tumor initiation, progression [15]. Using the TIMER2 web server, we verified the strong connection between GPNMB expression and immune cells in EC (Fig. 5c). The expression of GPNMB was connected with B cell (P = 2.39e−06), CD8+ T cell (P = 1.02e−05), CD4+ T cell (P = 3.91e−03), Macrophage (P = 1.39e−05), Neutrophil (P = 4.46e−12), and Dendritic cell (P = 1.07e−10).

4 Discussion

Pan-cancer analysis, aimed at integrated analysis across multiple cancer types or subtypes rather than focusing on a single cancer type, plays a crucial role in identifying shared genes, pathways, biomarkers, or other features among various cancer types. This shared molecular landscape may be indicative of common molecular mechanisms or biological processes driving carcinogenesis. By leveraging large amounts of tumor data, pan-cancer analysis can elucidate tumor evolution patterns and metastasis mechanisms across different cancer types. This knowledge contributes to a deeper understanding of cancer origin and development, enables the discovery of molecular markers associated with metastasis, and aids in predicting tumor metastasis risk and formulating treatment strategies. Additionally, pan-cancer analysis can support the reconstruction of gene expression regulatory networks in cancer. By identifying critical nodes and patterns within these networks, researchers gain insights into key genes and signaling pathways implicated in cancer progression and development. Consequently, this analysis offers novel targets and strategies for the development of targeted therapies. Ultimately, the significance of pan-cancer research lies in its capacity to comprehensively unveil the essence of cancer through cross-cancer type analysis, thereby enhancing strategies for cancer diagnosis, treatment, and prevention.

EC is one of the most commonly diagnosed gynecologic malignancies worldwide, recent advancements have significantly improved the management of EC patients, especially in the areas of molecular biology and minimally invasive treatments [16, 17]. In 2013, TCGA Research Network revolutionized the classification of EC by incorporating molecular characterization. Since then, molecular and genomic profiling in endometrial cancer has become increasingly popular [18]. For instance, L1 cell adhesion molecule (L1CAM) is often altered in endometrial cancer. Additional research suggests that L1CAM has a predictive function in endometrial cancer stage I, offering a potentially helpful tool for adjusting when adjuvant therapy is necessary [5, 6]. Hence, it is crucial to discover new molecules that can improve early diagnosis, prognostic assessment, and treatment results for EC.

According to reports, GPNMB plays a variety of functions, such as controlling cell adhesion and migration between cells, encouraging tissue healing, activating kinase signaling, and controlling cell development and differentiation [19,20,21,22]. In addition, GPNMB is expressed by immune cells including dendritic and macrophage cells, and it may prevent T-cell activation, which would reduce the immune system’s ability to fight tumors [23]. GPNMB exerts complicated effects in cancer since it has both oncogenic and anti-tumorigenic qualities. However, there is a paucity of research on a pan-cancer study of GPNMB that concentrates on all tumor types.

In the present study, we analyzed the expression of GPNMB in cancers using GEPIA2, TIMER2, and UALCAN. GPNMB is known to be highly expressed in most cancers and has been linked to increased proliferation, invasion, migration, and decreased tumor cell apoptosis [19, 24,25,26,27,28,29]. Consistent with earlier research findings, overexpression of GPNMB has been found to play oncogenic roles in a variety of cancers including bladder cancer [13], breast cancer [14], GBM [30], HNSC [31], renal cell carcinoma [32], hepatocellular carcinoma [33], lung cancer [34], and STAD [35]. Interestingly, our bioinformatic analysis revealed a discrepancy between the mRNA and protein levels of GPNMB in tumor tissues compared to normal tissues. For example, Ashktorab et al. [36] found that high methylation of GPNMB in colon cancer led to lower GPNMB expression. For SKCM, our analysis indicated that high GPNMB expression was correlated with the pathological stages, as well as associations between alteration status and disease-specific, progress-free, and overall survival rates. While a tissue microarray study confirmed GPNMB overexpression in malignant melanoma, further research is needed to fully understand the underlying mechanisms [37].

Our previous data suggested that GPNMB was important for the proliferation, migration, and invasion of ovarian cancer cells, and was regulated by miR-532-3p [11], but we failed to observe correlation between GPNB expression and the prognosis of ovarian cancer patients. And for CESC, dysregulation of GPNMB was not observed in cervical cancer tissues, a negative correlation was observed between GPNMB and DFS of cervical cancer patients. However, Xu et al. [38] reported the oncogenic role of GPNMB in cervical cancer, by regulating MMP-2/MMP-9 activity.

Based on the pan-cancer analysis, we observed downregulation of GPNMB in EC both at mRNA and protein levels, high alteration frequency of GPNMB, and literature searching didn’t find any studies about the function of GPNMB in EC. Therefore, we conducted the following in vitro experiments to identify the effect of GPNMB in EC. CCK-8 and transwell analysis results indicated the suppressive effect of GPNMB in the proliferation, migration, and invasion of EC. Further function enrichment indicated that GPNMB participated in the regulation of plasma lipoprotein particle levels. Previous research uncovered the association between plasma lipoprotein levels and the risk of EC, which might account for the potential mechanisms of GPNMB in EC [39]. In the context of cancer, the immune system was comprised of innate and adaptive immune cells and collectively functioned to eliminate tumor cells, and evidence of previous studies supported the role of GPNMB in cancers [40]. In this study, we explored the relationship between GPNMB expression and immune cells in EC, revealing potential pro-tumor immune functions of GPNMB, this also suggested that GPNMB may serve as a potential molecular biomarker for the diagnosis and treatment of endometrial cancer, especially for immunotherapy.

Limitations in the current study include the lack of detailed mechanism research and in vivo experiments to validate the role of GPNMB in EC. Further studies are planned to elucidate the specific mechanisms of GPNMB in EC.

5 Conclusions

Taken together, our first pan-cancer analyses of GPNMB indicated statistical correlations of GPNMB expression with survival status, genetic alteration to investigate the potential mechanism of GPNMB in the pathogenesis or prognosis of different cancers, which implied the role of GPNMB in tumorigenesis. Based on that, we identified the suppressive effect of GPNMB in EC and predicted the potential mechanisms and connection between GPNMB expression and immune cells.