PTTG1IP (PBF) is a prognostic marker and correlates with immune infiltrate in ovarian cancer

Abstract

Objective: An oncogenic protein, pituitary tumor transforming gene 1 binding factor (PTTG1IP, also called PBF), has been found to be expressed in various cancers. However, few studies have explored its prognostic significance and biologic function in epithelial ovarian cancer (EOC). Methods: Based on the Cancer Genome Atlas (TCGA) database, this study determined the differential expression of PBF at the mRNA level in EOC and normal tissues, which was then verified using real-time PCR and western blotting. Moreover, the Kaplan-Meier method and the Cox regression method were adopted to assess the clinical value of PBF in EOC. A nomogram model was constructed to evaluate the prognostic performance of PBF in EOC. Gene set enrichment analysis (GSEA) was employed to evaluate the signaling and pathway enrichment of PBF in EOC. The association between PBF expression and tumor-infiltrating immune cells (TIICs) in EOC was examined by single-sample GSEA and TIMER. Results: PBF was significantly higher in EOC than normal tissues as shown through TCGA database, and this result was verified by qRT-PCR and western blotting of EOC tissues and different cell lines. High PBF was associated with tumor size and lymphatic metastasis status. Kaplan-Meier (KM) analysis indicated that high PBF expression correlated with poor prognosis in patients with EOC (P < 0.0001). Moreover, multivariate Cox regression analysis was used to verify that PBF is an independent prognostic factor for EOC. The nomogram model exhibited moderate predictive accuracy and clinical utility in predicting EOC prognosis. The GSEA revealed that the expression of signaling pathways, such DNA damage replication, p53 pathway, Akt phosphorylation pathway, and estrogen-dependent nuclear pathway, were increased in the phenotype with high PBF expression. PBF expression was associated with neutrophil cells, iDC cells, NK cells, and Tem cells. Conclusion: As a prognostic biomarker for EOC, PBF was found to be correlated with immune infiltration, and may therefore be a promising target for immunotherapy for EOC.

Introduction

Epithelial ovarian cancer (EOC) ranks among the global top five common causes of female cancer-related death and is one of the most malignant gynecological cancers [1]. Due to a lack of feasible methods for early detection, most cases are diagnosed during the advanced stage (FIGO stage III/IV disease) [2]. Consequently, the 5-year overall survival (OS) rate of ovarian cancer is below 50% (FIGO stage III/IV disease, 5-year OS rate of 29%) [3,4]. Standard therapy is composed of cytoreductive surgery accompanied by platinum and taxane-based chemotherapy. Though the initial response rates of patients after surgery and first-line chemotherapy can reach over 80%, most of these patients eventually experience recurrence due to rapid progression and chemoresistance [5]. Several targeted treatment methods, such as vascular endothelial growth factor (VEGF) inhibitors, have been incorporated into clinical settings as first-line or second-line therapy methods for recurrent disease [6]. However, due to resistance against PARP inhibitors, a majority of patients are likely to develop recurrence due to cancer progression [7]. Therefore, it is critical to explore novel diagnostic or prognostic biomarkers and therapeutic targets for EOC.

Pituitary tumor transforming gene 1 binding factor (PTTG1IP, also called c21orf3, PTTG1-binding factor, or PBF) is a widely expressed proto-oncogene, which was initially measured to be a 22 kDa protein involved in binding to [8]. Previous research has demonstrated that PBF is extensively expressed in normal tissues, such as the thyroid gland, and placenta [9]. Increasing evidence suggests that PBF expression is elevated in different cancers, such as thyroid, breast, and colorectal cancers [10-12]. In addition, PBF overexpression is significantly related to a poor outcome in many cancers. Compared to normal thyroid tissues, PBF is a novel transformation-related and oncogenic gene that is overexpressed in thyroid cancer, and high levels of PBF expression are correlated with neoplasm recurrence and shorter disease-specific survival [13]. Though PBF has hardly been identified in normal breast tissues, it is highly expressed in epithelial cells of ERα-positive breast tumors. Moreover, PBF upregulation in breast MCF-7 cells significantly increases cell invasion in vitro. These results further support the fact that PBF modulates the estrogen-mediated invasion of breast cancer cells by acting as a proto-oncogene [10,14]. Similarly, PBF is also upregulated in colorectal cancer and serves as a novel adjuster through the inhibition of p53 activity, especially in invasive wild-type p53 and mutant p53 tumors, indicating that PBF may be a novel prognostic biomarker of colorectal cancer. However, PBF is apparently downregulated at the mRNA level in NSCLC tissues compared to adjacent tissues. Moreover, the upregulation of PBF significantly inhibits cell proliferation [15]. In summary, the above findings confirm that PBF expression is correlated with endocrine activity and tumorigenesis. However, the role of PBF in EOC has not been elucidated.

Therefore, we analyzed PBF expression in EOC and its association with clinicopathologic features and prognosis, based on data obtained from TCGA. Additionally, a nomogram combined with PBF expression and clinical clinicopathological features was constructed to forecast the risk of peritoneal metastasis for EOC patients. GSEA was applied to evaluate possible molecular functions of PBF. Subsequently, ssGSEA and TIMER were also performed to verify the relationship between PBF expression and TIICs in EOC. This study proves that high expression of PBF is associated with the unsatisfactory OS of EOC patients and that pathways involved in DNA damage replication, Toll pathway, extracellular matrix interaction, p53 pathway, B cell receptor complexes, Wnt signaling pathway, Akt phosphorylation pathway, estrogen-dependent nuclear pathway, and the FGF pathway are associated with PBF expression. Furthermore, we found a correlation between PBF and levels of TIICs, especially in neutrophil cells, iDC cells, NK cells, and Tem cells. Thus, PBF can function as a novel predictive biomarker and immunotherapy target for EOC patients.

Materials and methods

Data collection

The RNA-seq data from TCGA and Genotype-Tissue Expression (GTEx) databases, including data on normal and 33 types of cancer tissue samples were collected from sets of the University of California Santa Cruz (UCSC) Xena browser platform (https://xenabrowser.net/). The gene expression and clinicopathologic data were downloaded for further bioinformatics analysis. According to the intermediate value of PBF expression used as the cutoff, all EOC patients were categorized into low-expression and high-expression groups. The GSE40595 dataset of EOC was collected from the Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/) database using the GEO query R package and was used for validation. Since data were collected from TCGA and GEO databases, approval from the ethics committee was not required for this study.

Human protein atlas (HPA)

HPA (https://www.proteinatlas.org/), a network database, provides immunohistochemistry (IHC) data generated from normal and cancer tissue profiles. The protein expression of PBF in normal tissues and in EOC tissues was measured using IHC data from the HPA database.

Tissue specimens, and cell lines, and culture

We collected 30 ovarian cancer tissues from patients who had undergone surgery at Peking University People’s Hospital between January 2019 and January 2020. As a control, we also acquired 5 normal ovarian samples from bilateral salpingo-oophenrectomy cases. The diagnosis of ovarian cancer grade was performed by two experienced pathologists. All patients involved provided written informed consent, and this study was approved by the Ethics Commitee of the Peking University People’s Hospital. Clinical tissues from EOC patients were used to detect expression at the mRNA and protein level.

Five ovarian cancer cell lines (SKOV3.ip, SKOV3, CAOV3, OVCAR3, and ES2) were also collected for this study. CAOV3 was sustained in DMEM supplemented with 10% FBS; the SKOV3.ip cells, OVCAR3 cells, and ES2 cells were sustained in RPMI-1640 supplemented with 10% FBS; while SKOV3 cells were grown in McCoy’s 5A supplemented with 10% FBS. The cells were all cultured at 37°C in a humid incubator with 5% CO₂. Total RNA and protein were acquired from the ovarian cancer cell line specimens.

RNA isolation and qRT-PCR

The RNA was extracted from the EOC samples and EOC cell lines by TRIzol reagent (Invitrogen, USA). RNA reverse transcription was achieved through the PrimeScript^TM RT reagent Kit (Takara), which was measured by quantitative PCR using the SYBR Green PCR Kit (Applied Biosystems, USA) based on the manufacturer’s instructions. The forward primer of the reference gene was AAGGTGAAGGTCGGAGTCAAC, and the reverse primer of the reference gene was GGGGTCATTGATGGCAACAATA. The forward primer of the target gene was CCTGTGAAGAGTGCCTGAAGAACG, and the reverse primer of the target gene was GAAGCGTCGGGACTGATGTGC. The 2^-ΔΔCT approach was adopted to analyze the results, and the gene expression was normalized relative to GAPDH.

Western blotting analysis

The total proteins extracted from the EOC tissues and EOC cell lines were lysed using the RIPA cell lysate (CST, USA), and were further separated using SDS-PAGE and then transferred to a polyvinylidene fluoride (PVDF) membrane. After blocking using TBS-T with 5% milk for one hour at room temperature, the membranes were cultured with the primary antibodies against PTTG1IP (1:1000, Abcam, USA) at 4°C overnight and with the secondary antibody (1:3000, Abcam, USA). The relative levels of the target proteins were consistent with the protein band intensity of the grey value of the internal reference band (GAPDH). The whole experimental process was performed in triplicate.

Analysis of the differentially expressed genes (DEGs) between the low-PBF and high-PBF expression groups of EOC patients

Using the RNA-seq data based on low-PBF and high-PBF expression groups in the EOC samples, differential mRNA expression was analyzed via the “DESeq2” R package. DEGs with |log₂FC| > 1.5 and false discovery rate (FDR) < 0.05 were incorporated into the following analyses, while volcano plots and heat maps were created using the ggplot2 package in R.

Analysis of gene ontology (GO) and pathway enrichment

GO and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis of the PBF-related DEGs were conducted using the cluster Profiler package. GO was utilized to identify the corresponding biologic processes (BP), cell components (CC), and molecular functions (MF). A P-value of < 0.05 was set as the cut-off criterion for the Benjamini and Hochberg (BH) method.

GSEA and protein-protein interaction (PPI) network

GSEA was employed to verify a significant increase in gene sets between low-PBF and high-PBF expression groups. Pathways with a FDR < 0.25 and a nominal P < 0.05 were regarded as significantly enriched pathways. The PPI network of PBF was evaluated using the Search Tool for the Retrieval of Interacting Genes (STRING) database (https://string-db.org/) with a minimum interaction score of 0.7.

Immune infiltration analysis using single-sample GSEA and tumor immune cells to assess resource (TIMER)

The immune-cell infiltration levels of EOC in the TCGA cohorts were estimated using the single-sample GSEA method in the GSVA R package. In total, 24 different types of immune cells, such as dendritic cells (DCs), T helper 17 (Th17) cells, eosinophils, neutrophils, regulatory T cells, activated DCs, B cells, CD8⁺ T cells, T helper cells, and effector memory T cells, could be discriminated in this study. The immune responses of the 24 TIICs were measured to assess their correlation with PBF expression in the EOC. The correlation analysis between immune infiltration and PBF expression in EOC was done using the Spearman’s rank-correlation coefficient. The Wilcoxon rank sum test was used to compare the infiltration of immune cells between the low-PBF and high-PBF expression groups. P < 0.05 was used to indicate statistical significance.

Tumor immune to assess resource (TIMER, https://cistrome. shinyapps.io/timer/) is a trustworthy database that can be used to comprehensively evaluate the TIICs identified among the many different cancer types included in TCGA. The Kaplan-Meier method was applied to measure the prognostic value of PTTG1IP in 6 different types of immune cells: B cells, CD8⁺ T cells, dendritic cells (DC), CD4⁺ T cells, neutrophils, and macrophages. TIMER was adopted to explore the association between PBF and immune cell markers to assess the effect of PBF on tumor immunity. The immune cells included monocytes, CD8⁺ T cells, neutrophils, B cells, M1 macrophages, M2 macrophages, T helper 1 (Th1) cells, T helper 2 (Th2) cells, natural killer (NK) cells, DCs, and exhausted T cells.

Statistical analysis and nomogram

The comparison of PBF expression between different clinicopathologic groups was performed using the Kruskal-Wallis test, Wilcoxon signed test, and the Wilcoxon rank sum test. Additionally, the receiver operating characteristic (ROC) curve was used to distinguish EOC from normal tissue. The associations between PBF expression and clinicopathologic features were measured through logistic regression. Survival analyses for disease-specific survival (DSS) and OS were conducted using both the Kaplan-Meier method and Cox regression analysis. The independent prognostic value of PBF was assessed using univariate and multivariate Cox regression models by R (Version v.3.6.2). A two-tailed P-value of < 0.05 was considered significant.

A nomogram was built based on the factors that were of significance to multivariate prognostic parameters and clinicopathologic characteristics using the rms R package (version 5.1.2, https://cran.r-project.org/web/packages/rms/). The nomogram could assess the DSS and 1, 3, and 5-year OS in EOC patients. Discriminations between observations and predictions were quantitatively assessed using the concordance index (C-index). Calibration plots were generated by comparing the relationships between the nomogram prediction probability and the observations. A P-value of < 0.05 was considered significant.

Results

PBF was upregulated in EOC

We first evaluated PBF expression in the TCGA and GTEx pan-cancer database, and observed higher PBF expression in a majority of tumors, including breast invasive carcinoma (BRCA), cholangiocarcinoma (CHOL), thyroid carcinoma (THCA), skin melanoma (SKCM), uterine sarcoma (UCS), and endometrial cancer (UCEC) (Figure 1A). In particular, higher PBF expression was observed in EOC from the TCGA cohort and GSE40595 cohort than in normal tissues (P < 0.001, Figure 1B and 1C). Moreover, the diagnostic value of PBF in EOC patients was assessed by ROC curve analysis. The area under the curve (AUC) of the ROC curve of PBF was 0.606 (95% confidence interval [CI] = 0.556-0.657) (Figure 1D), indicating that PBF can be used to effectively discriminate EOC tissues from normal tissues.

Figure 1.

PBF expression in patients with EOC. A. PBF expression levels in various types of tumor obtained from TCGA database; B. Expression levels of PBF in EOC and normal tissues in TCGA database; C. The PBF expression in EOC and normal tissues in the GEO database; D. ROC analysis of PBF in EOC. PBF, PTTG1-binding factor; EOC, epithelial ovarian cancer.

The immunohistochemistry staining results of HPA suggested that PBF was strongly expressed in EOC tissues but rarely expressed in normal tissues (Figure 2A). Additionally, the mRNA and protein expression patterns of PBF in EOC tissues obtained from Peking University People’s Hospital (PKUPH) and the EOC cell lines were explored. We observed that the mRNA and protein expression of PBF were higher in EOC tissues than in normal tissues (P = 0.004 and P = 0.007) (Figure 2B). Among all the EOC cell lines, a relatively higher level of PBF expression at mRNA and protein levels was observed in SKOV3.ip, SKOV3 and ES2 cells, and a lower level was found in CAOV3 cells and OVCAR3 cells (Figure 2C). In summary, these results suggest that PBF may affect the pathogenesis of EOC.

Figure 2.

PBF expression in EOC tissues and cell lines. A. Representative IHC staining patterns of PBF in EOC tissues obtained from the HPA database; B. PBF protein and mRNA expression in normal tissues (n = 5) and EOC tissues (n = 30); C. PBF protein and mRNA expression in five EOC cell lines (SKOV3.ip, OVCAR3, SKOV3, CAOV3, and ES2). PBF, PTTG1-binding factor; EOC, epithelial ovarian cancer.

Connection between PBF expression and the clinicopathologic characteristics of EOC

To examine the connection between PBF expression and the clinicopathologic features of EOC patients, PBF expression data and clinical data on EOC samples were extracted from the TCGA database (Supplementary Table 1) and analyzed. As Table 1 shows, PBF expression was highly associated with primary treatment outcome (P = 0.028), residual tumors (P = 0.021), and age (P = 0.012). Nevertheless, PBF expression was not strongly associated with other measures, including International Federation Organization of Gynecology and Obstetrics (FIGO) stage, histological grade, anatomical neoplasm subdivision, venous invasion, lymphatic invasion, and TP53 status (all P > 0.05). Furthermore, the univariate logistic regression results confirmed that PBF expression was also highly associated with clinical characteristics, such as primary treatment outcome (CR vs. PD&SD&PR) (OR = 0.60 (0.36-0.98), P = 0.043) and residual tumors (RD vs. NRD) (1.99 (1.15-3.51), P = 0.015), but not FIGO stage, histologic grade, anatomic neoplasm subdivision, venous invasion, lymphatic invasion, and TP53 status (Supplementary Table 2). The analysis results verified that EOC patients with high PBD expression had more progression of EOC.

Table 1.

Clinical characteristics of ovarian cancer patients based on TCGA

Feature	Level	Low expression of PTTG1IP	High expression of PTTG1IP	p	test
n		188	188
FIGO stage (%)	Stage I	1 (0.5%)	0 (0.0%)	0.321	exact
	Stage II	13 (7.0%)	9 (4.8%)
	Stage III	149 (79.7%)	144 (77.4%)
	Stage IV	24 (12.8%)	33 (17.7%)
Histologic grade (%)	G1	0 (0.0%)	1 (0.5%)	0.166	exact
	G2	17 (9.3%)	25 (13.7%)
	G3	166 (90.7%)	156 (85.2%)
	G4	0 (0.0%)	1 (0.5%)
Primary therapy outcome (%)	CR	115 (75.2%)	98 (64.5%)	0.028
	PD	12 (7.8%)	15 (9.9%)
	PR	13 (8.5%)	30 (19.7%)
	SD	13 (8.5%)	9 (5.9%)
Anatomic neoplasm subdivision (%)	Bilateral	119 (67.6%)	134 (75.3%)	0.139
	Unilateral	57 (32.4%)	44 (24.7%)
Venous invasion (%)	No	21 (41.2%)	19 (36.5%)	0.779
	Yes	30 (58.8%)	33 (63.5%)
Tumor residual (%)	NRD	42 (25.1%)	24 (14.5%)	0.021
	RD	125 (74.9%)	142 (85.5%)
Lymphatic invasion (%)	No	24 (32.0%)	24 (32.9%)	1.000
	Yes	51 (68.0%)	49 (67.1%)
TP53 status (%)	Mut	128 (90.8%)	120 (90.2%)	1.000
	WT	13 (9.2%)	13 (9.8%)
Age (%)	≤ 60	95 (50.5%)	112 (59.6%)	0.097
	> 60	93 (49.5%)	76 (40.4%)
Age (median [IQR])		60.00 [52.00, 70.00]	57.00 [49.00, 66.00]	0.012	nonnorm

Abbreviations: FIGO, International Federation of Gynecology and Obstetrics.

PBF is an independent prognostic predictor of EOC

To measure the prognostic value of PBF in EOC patients, we used EOC sample data obtained from TCGA databases. It was observed that high PTTG1IP expression was strongly associated with a worse OS (HR = 1.42 (1.09-1.85), P = 0.009) and DSS (HR = 1.49 (1.13-1.98), P = 0.005) (Figure 3A and 3B). Moreover, univariate analysis showed that primary thermal outcome, residual tumors, and PBF expression were associated with DSS, while primary thermal outcome, residual tumors, age, and PBF expression were associated with OS (Table 2). As shown in Table 2, the multivariate analysis was performed by adjusting variables with P < 0.1 in the univariate analysis. Primary therapy outcome and high PBF expression were independent predictors for OS as given in the multivariate analyses (P < 0.05). Accordingly, primary treatment outcome, residual tumors, and age were independent prognostic values for DSS (P < 0.05). Hence, the results confirmed that high PBF expression is an independent predictor associated with adverse prognosis of EOC patients.

Figure 3.

Prognostic value of PBF expression in EOC. A. Survival curves of OS constructed using TCGA data; B. Survival curves of DSS constructed using TCGA data; C. OS survival curves of stages I and III subgroups between PBF-high and -low expression EOC cases; D. OS survival curves of stages G1 and 3 subgroups between PBF-high and -low expression EOC cases; E. DSS survival curves of I and III subgroups between PBF-high and -low EOC cases; F. DSS survival curves of G1 and 3 subgroups between PBF-high and -low EOC cases. PBF, PTTG1-binding factor; EOC, epithelial ovarian cancer; OS, overall survival; DSS, disease specific survival.

Table 2.

Univariate and multivariate analyses of disease-specific survival and overall survival in patients with EOC

Characteristic	DSS					OS
	Total (N)	HR (95% CI) Univariate analysis	P value	HR (95% CI) Multivariate analysis	P value	Total (N)	HR (95% CI) Univariate analysis	P value	HR (95% CI) Multivariate analysis	P value
FIGO stage (Stage III & Stage IV vs. Stage I & Stage II)	347	2.244 (0.922-5.462)	0.075	1.294 (0.311-5.394)	0.723	371	2.085 (0.925-4.699)	0.076	2.546 (0.621-10.443)	0.194
Histologic grade (G3 & G4 vs. G1 & G2)	339	1.313 (0.833-2.070)	0.240			364	1.194 (0.797-1.789)	0.389
Primary therapy outcome (CR vs. PD & SD & PR)	298	0.227 (0.163-0.317)	< 0.001	0.274 (0.177-0.423)	< 0.001	304	0.234 (0.169-0.324)	< 0.001	0.269 (0.189-0.384)	< 0.001
Anatomic neoplasm subdivision (Bilateral vs. Unilateral)	329	1.034 (0.747-1.431)	0.841			353	1.041 (0.768-1.410)	0.798
Venous invasion (Yes vs. No)	101	0.846 (0.450-1.591)	0.604			103	0.905 (0.487-1.683)	0.753
Tumor residual (RD vs. NRD)	310	2.559 (1.572-4.166)	< 0.001	2.203 (1.127-4.307)	0.021	332	2.302 (1.479-3.583)	< 0.001	1.591 (0.949-2.667)	0.078
Age (> 60 vs. ≤ 60)	349	1.282 (0.969-1.695)	0.082	1.580 (1.068-2.338)	0.022	374	1.373 (1.059-1.780)	0.017	1.343 (0.980-1.842)	0.067
Lymphatic invasion (Yes vs. No)	143	1.407 (0.816-2.425)	0.219			147	1.422 (0.839-2.411)	0.191
TP53 status (Mut vs. WT)	256	0.643 (0.386-1.070)	0.089	0.911 (0.509-1.628)	0.752	273	0.692 (0.423-1.132)	0.143
PTTG1IP (High vs. Low)	349	1.493 (1.125-1.982)	0.005	1.301 (0.880-1.925)	0.187	374	1.419 (1.091-1.845)	0.009	1.421 (1.029-1.964)	0.033

Abbreviations: FIGO, International Federation of Gynecology and Obstetrics; CR, Complete resolution; PD, Progressive disease; PR, Partial response; SD, Stable disease; NRD, No residual disease; RD, Residual disease; Mut, Mutation; WT, Wild type; PBF, PTTG1-binding factor; EOC, epithelial ovarian cancer; DSS, disease-specific survival.

A subgroup analysis was conducted to measure the effect of PBF expression on OS and DSS and the analysis confirmed that the predictive effect of PBF expression in EOC was significant for OS in Stage III and G3 patients (P < 0.05) (Figure 3C and 3D). This effect was also significant for DSS in Stage III and G3 patients (P < 0.05) (Figure 3E and 3F). The multivariate Cox regression analysis also demonstrated that PBF was the only prognostic indicator of both DSS and OS at Stage III and G3 of EOC patients (Supplementary Tables 3 and 4). The above data indicated that a high PBF level was associated with a poor prognosis.

Construction and validation of a PBF based nomogram

A nomogram that integrated PBF expression with other clinical elements (primary treatment outcome, tumor residual, and age) was constructed to predict the 1, 3, and 5 year DSS and OS of patients with EOC. The total number of points on the nomogram manifested a poorer prognosis. The C-index for OS and DSS prediction were 0.685 (0.663-0.708) and 0.694 (0.670-0.717) (Figure 4A and 4B), respectively. Furthermore, the calibration curves suggested that the predicted rates matched the actual survival rates at 1-, 3-, and 5 years (Figure 4C and 4D). These results confirmed that the nomogram showed a moderate degree of accuracy in predicting the OS of EOC patients.

Figure 4.

A nomogram combining PBF and other prognostic factors for EOC obtained using TCGA data. The nomograms were constructed as PBF expression-based risk scoring models for (A) 1-, 3-, and 5-year overall survival and (B) disease-specific survival. Calibration plots validating the efficiency of the nomograms for (C) OS and (D) disease-specific survival. PBF, PTTG1-binding factor; EOC, epithelial ovarian cancer; OS, overall survival; OS, overall survival.

Identifying the DEGs

The DESeq2 package was adopted to detect DEGs between the high-PBF and low-PBF groups. The criteria of the DEGs were modified to a p value < 0.05 and |log₂ (fold change)| > 1. A total of 278 DEGs (11 upregulated and 267 downregulated genes) were identified (Figure 5A). The expression values of the top ten DEGs between the high-PBF and low-PBF groups are shown in Figure 5B. The adjusted P value and log₂FC of each DEG are shown in Supplementary Table 5.

Figure 5.

A. Volcano Plots of the DEGs; B. Heatmap of the DEGs. C-E. Gene Ontology (GO) analysis showing the top eight genes each for biological processes (BP), cellular components (CC), and molecular functions (MF). F. KEGG pathway analysis and the top eight pathways mapped based on genes co-expressed with NLRP12.

Potential mechanism by which PBF regulates BRC progression

To elucidate the biological functions of PBF in EOC and illustrate key signaling mechanisms regulated by PBF, we performed GO functional enrichment analysis and KEGG pathway analysis. Subsequently, the GO analysis results of the DEGs were classified into three main functional groups: cellular components (GO-CC), biologic processes (GO-BP), and molecular functions (GO-MF) (Supplementary Table 6). The BP enrichment indicated that genes were predominantly involved in mRNA 5’-splice site recognition, mRNA splice site selection, antimicrobial humoral response, nucleosome positioning, and spliceosomal complex assembly. Also, genes from the CC terms were significantly associated with spliceosomal snRNP complex, collagen-containing extracellular matrix, and protein-DNA complex. MF analysis also revealed a positive correlated with hormone activity, endopeptidase inhibitor activity, and receptor ligand activity (Figure 5C-E). The KEGG analysis results showed that the genes were highly upregulated in protein digestion and absorption, neuroactive ligand-receptor interaction, RNA transport, and spliceosomes (Figure 5F). The findings of both analyses confirmed that high PBF expression was associated with multiple biological signaling pathways in EOC.

GSEA identified a PBF-related signaling pathway

To further identify the effects of PBF on EOC, we conducted a GSEA analysis on RNA-seq data on EOC patients from TCGA database. Based on the PBF expression levels in TCGA dataset, EOC samples were sorted into low-PBF and high-PBF expression groups. The significantly enriched biological pathways are shown in Figure 6, and include DNA damage replication, Toll pathway, extracellular matrix interaction, P53 pathway, B cell receptor complexes, Wnt signaling pathway, Akt phosphorylates pathway, estrogen-dependent nuclear pathway and FGF pathway, which reveals possible regulatory mechanisms of PBF in EOC.

Figure 6.

Enrichment plots obtained using GSEA data. PBF was differentially enriched in the (A) P53 pathway, (B) DNA damage replication, (C) B cell receptor complexes, (D) Toll pathway, (E) Wnt signaling pathway, (F) Akt phosphorylation pathway, (G) estrogen dependent nuclear pathway, (H) extracellular matrix interaction, and (I) FGF pathway.

PPI

To detect the role of PBF in EOC, we established a PPI network using the STRING database and evaluated the relationships between the related genes (Figure 7).

Figure 7.

The protein-protein interaction (PPI) network was construction based on PBF co-expression genes. PBF, PTTG1-binding factor.

The connection between PBF expression and immune infiltration

Immune infiltration in the tumor microenvironment (TME) is correlated with the clinical prognosis of survival in cancers. Therefore, ssGSEA was applied to evaluate the involvement of 24 types of TIICs in the EOC, and Spearman’s correlation was used to determine the association between PBF expression and the 24 types of TIICs (Figure 8A). The results revealed that PBF expression levels were positively associated with the infiltration levels of the neutrophils (r = 0.298, P < 0.001), iDCs (r = 0.303, P < 0.001), NK cells (r = 0.305, P < 0.001), and Tem (r = 0.410, P < 0.001) (Figure 8B). In addition, the infiltration levels of the Tgd cells (P < 0.001), Th1 cells (P = 0.012), NK CD56^dim cells (P = 0.033), neutrophils (P < 0.001), DCs (P = 0.005), eosinophils (P < 0.001), iDCs (P < 0.001), macrophages (P < 0.001), mast cells (P = 0.007), NK cells (P < 0.001), Tcm (P = 0.019), and Tem (P < 0.001) were considerably enriched in the PBF-high group (Figure 8C). Altogether these results suggest that PBF affects immune cell infiltration in EOC.

Figure 8.

ssGSEA analyses of PBF and the association between PBF expression and immune infiltration level in EOC. A. The correlation between the infiltration of immune cells and the expression of PBF; B. PBF expression was prominently positively associated with the infiltration levels of neutrophils, iDC cells, NK cells, and Tem cells; C. The infiltration levels of Tgd cells, neutrophils cells, NK CD56dim cells, eosinophils, iDCs, DCs, eosinophils, neutrophils, iDCs cells, macrophages cells, mast cells, NK cells, Tcm cells, and Tem cells were significantly higher in the PBF-high expression group. PBF, PTTG1-binding factor; EOC, epithelial ovarian cancer.

Prognostic value of PBF expression in EOC based on the TIIC subsets

The role of TIIC for the prognosis of EOC was explored using TIMER, which confirmed that a higher expression of PBF in the DCs cells was associated with the prognosis of EOC (Figure 9). However, higher expression of PBF was not associated with any obvious differences in the survival in B cells, CD8⁺ T cells, CD4⁺ T cells, macrophages, and neutrophils of patients with EOC (P > 0.05), which verified that high PBF expression in EOC affected prognosis partly due to TIIC levels.

Figure 9.

Comparison of the Kaplan-Meier survival curves of the high and low PBF expression groups in EOC based on immune cell subgroups analysis conducted using TIMER. PBF, PTTG1-binding factor; EOC, epithelial ovarian cancer.

PBF expression correlates with immune markers

To confirm a correlation between PBF expression and immune infiltration, we explored the association between PBF and multiple immune marker genes of diversified immune cells, such as B cells, neutrophils, CD8⁺ T cells, T cells (general), TAMs, monocytes, and DCs in EOC, using TIMER. Moreover, multiple functional T cells, including Th1 cells, Th2 cells, Th17 cells, and exhausted T cells were explored. After the correlation adjustment for purity, we noticed that the PBF expression level was significantly linked with most immune marker sets of the diverse immune cells and multiple T cells in EOC (Table 3). Overall, the results showed a close relationship between PBF and immune cell infiltration in EOC.

Table 3.

Correlation analysis between PBF and related gene markers of immune cells in EOC by TIMER

Description	Gene marker	None		Purity
		Cor	p	Cor	p
B cell	CD19	-0.114	4.76e-02	-0.144	2.34e-02
	CD79A	-0.01	8.68e-01	-0.159	1.21e-02
CD8+ T cell	CD8A	-0.134	1.95e-02	-0.012	8.55e-01
	CD8B	0.1	8.32e-02	-0.002	9.72e-01
Dendritic cell	ITGAX	0.275	1.31e-06	0.123	5.25e-02
	NRP1	0.295	1.62e-07	0.182	4.00e-03
	CD1C	0.185	1.2e-03	0.006	3.03e-04
	HLA-DPA1	0.131	2.23e-02	-0.007	9.07e-01
	HLA-DRA	0.116	4.29e-02	0.008	8.96e-01
	HLA-DQB1	0.057	3.21e-01	-0.082	1.99e-01
	HLA-DPB1	0.171	2.92e-03	0.034	5.89e-01
M1 Macrophage	PTGS2	0.148	9.67e-03	0.037	5.58e-01
	IRF5	0.136	1.76e-02	0.043	5.02e-01
	NOS2	0.095	9.84e-02	0.006	2.99e-01
M2 Macrophage	MS4A4A	0.222	1.04e-04	0.053	4.03e-01
	VSIG4	0.225	8.14e-05	0.054	3.93e-01
	CD163	0.261	4.33e-06	0.1	1.17e-01
Monocyte	CSF1R	0.295	1.67e-07	0.126	4.66e-02
	CD86	0.214	1.83e-04	0.036	5.76e-01
Natural killer cell	KIR2DS4	0.151	8.56e-03	0.074	2.46e-01
	KIR3DL3	0.014	7.95e-01	-0.02	7.59e-01
	KIR3DL2	0.143	1.27e-02	0.069	2.78e-01
	KIR3DL1	0.11	5.55e-02	-0.008	9.04e-01
	KIR2DL4	0.094	1.02e-01	-0.019	7.64e-01
	KIR2DL3	0.195	6.44e-04	0.144	2.29e-02
	KIR2DL1	0.136	1.8e-02	0.088	1.66e-01
Neutrophils	CCR7	0.123	3.29e-02	-0.037	5.64e-01
	ITGAM	0.288	3.66e-07	0.103	1.05e-01
	CEACAM8	0.08	1.64e-01	0.007	2.23e-01
T cell (general)	CD3D	0.098	8.71e-02	-0.086	1.78e-01
	CD3E	0.151	8.37e-03	-0.028	6.65e-01
T cell exhaustion	CTLA4	0.116	4.3e-02	-0.035	5.78e-01
	LAG3	-0.02	7.29e-01	-0.13	4.05e-02
	HAVCR2	0.246	1.64e-05	0.066	2.99e-01
	GZMB	0.057	8.2e-01	-0.092	1.49e-01
	PDCD1	0.141	1.41e-02	0.008	8.97e-01
TAM	CCL2	0.154	7.15e-03	-0.028	6.60e-01
	IL10	0.275	1.28e-06	0.148	1.98e-02
	CD68	0.231	5.3e-05	0.064	3.15e-01
Tfh	BCL6	0.099	8.39e-02	0.117	6.48e-02
	IL21	-0.056	3.34e-01	-0.045	4.82e-01
Th1	TBX21	0.129	2.45e-02	-0.055	3.86e-01
	STAT4	0.184	1.27e-03	0.071	2.65e-01
	STAT1	0.082	1.56e-01	0.084	1.84e-01
	IFNG	0.047	4.11e-01	-0.074	2.44e-01
	IL13	0.08	1.66e-01	0.066	2.98e-01
Th2	GATA3	0.219	1.29e-04	0.101	1.12e-01
	STAT6	0.158	5.94e-03	0.161	1.08e-02
	STAT5A	0.084	1.45e-01	0.041	5.21e-01
Th17	STAT3	0.292	2.52e-07	0.238	1.53e-04
	IL17A	0.065	2.57e-01	-0.003	9.58e-01
Treg	FOXP3	0.197	5.63e-04	0.093	1.45e-01
	CCR8	0.178	1.86e-03	0.071	2.67e-01
	STAT5B	0.121	3.6e-02	0.09	1.59e-01
	TGFB1	0.337	2.12e-09	0.217	5.81e-04

Abbreviations: PBF, PTTG1-binding factor; EOC, epithelial ovarian cancer.

Discussion

As an oncogenic protein, pituitary tumor transforming gene 1 binding factor (PTTG1IP, also called PBF) actively participates in the metaphase-anaphase transition of the cell cycle through the activation of securin (PTTG1). The expression of PBF has been identified in many tumors including thyroid, pituitary, and breast [8,12]. Higher PBF expression is associated with early tumor recurrence in thyroid cancer [12]. Functional studies conducted on breast cancer and colorectal carcinoma cells have demonstrated that PBF overexpression promotes cell invasion [11,14]. Collectively, these observations suggest that PBF may be involved in tumorigenesis, but its precise role in EOC development and progression have not been comprehensively studied.

In our study, clinical information download from the TCGA database was used to evaluated PBF expression in different types of cancers, and found that PBF expression was enhanced in a diverse range of cancers compared with corresponding normal tissues, and in particular in EOC, as its high expression was identified in EOC using the GSE40595 dataset and the HPA. Further clinical validation also suggested that PBF was overexpressed in the EOC cell lines and EOC tissues obtained from the PKUPH database, which is in line with previous studies that showed that PBF is highly expressed in various types of cancer, including thyroid cancer, prostate cancer, and head and neck squamous cell carcinoma. PBF and PTTG greatly promote thyroid cancer, which make them candidate biomarkers for prognosis and therapy in EOC patients [12]. Previous immunohistochemical results have demonstrated that PBF expression was higher in prostate cancer than in benign prostatic hyperplasia or adjacent normal prostate specimens [16]. These results are similar to our findings with EOC obtained from TCGA, indicating that PBF may be a diagnostic marker in multiple cancers.

The high expression of PBF in EOC indicates poor prognosis. This study showed that PBF overexpression was associated with clinical values, such as primary therapy outcome, residual tumor, and age. In addition, the Kaplan-Meier survival analysis confirmed that high PBF expression was associated with a worse DSS and OS in EOC based on TCGA data. These studies have highlighted that PBF expression is associated with the clinical features and the survival of EOC patients, and may serve as a prognostic biomarker. Additionally, the univariate Cox analysis conducted on TCGA data showed that PBF expression is a prognostic factor in EOC. Similarly, head and neck squamous cell carcinoma patients with high PBF expression had the poorest OS [17]. The multivariate analysis demonstrated that PBF expression was an independent survival biomarker of DSS and OS in EOC patients. Overall, the results of the analyses confirmed that PBF expression may be a useful prognostic biomarker in EOC. High PBF expression was associated with an unsatisfactory prognosis of stage III-IV and G3 subgroups of EOC patients, with the highest HR associated with the lowest OS and DSS. It was observed that PBF expression was an effective prognostic biomarker in all subsets, indicating that PBF is an independent clinicopathological value.

Given that PBF is a significant prognostic factor, we created a nomogram, in which PBF is a prognostic marker for EOC. The results of the multivariate Cox analysis were used to construct a nomogram with PBF being an independent factor of clinical risk (primary therapy outcome, tumor residual and age). In TCGA cohorts, the C-indexes and calibration plots demonstrated that the nomogram performed well in predicting the 1-, 3-, 5-year OS and DSS in patients with EOC, and provided a personalized score that could be used to identify high-risk EOC patients and provide them with more aggressive therapy regimens.

To verify the molecular mechanism of PBF in EOC, we implemented the GSEA to analyze the pathways that were upregulated under PBF expression. The results showed that PBF may be involved in multiple signaling pathways, including DNA damage replication, P53 pathway, B cell receptor complexes, Wnt signaling pathway, Akt phosphorylates pathway, and estrogen dependent nuclear pathway. Certain studies have revealed that PBF may affect several biologic functions, including cell transformation, migration, and invasion [11,18,19]. For instance, head and neck squamous cell carcinoma patients with high PBF suffered a worse outcome partially due to greater aberration of the p53-dependent signaling pathway [17]. Previous research has demonstrated that PBF may be a prognostic indicator in invasive tumors through its regulation of p53 activity in colorectal tumorigenesis [11]. It was also reported that as a proto-oncogene, PBF may serve as a negative regulator of p53 function in thyroid tumorigenesis [20]. Evidence has also indicated that PBF and PTTG play a crucial role in regulating genes associated with DNA damage response, which is related to poor clinical outcome [12]. Functional regulatory variants of PBF have been associated with the risk of developing ER-positive breast cancer, which was then confirmed through its function as a proto-oncogene in breast cancer [10]. PBF over-expression activates PI3K/Akt signaling and may contribute to enhance the vulnerability of females towards thyroid disease [13]. In general, these results verified that PBF is strongly associated with tumorigenesis. Therefore, PBF may affect the carcinogenesis of EOC by regulating these signaling pathways, leading to a more unsatisfactory prognosis in EOC patients.

In recent years, many studies have revealed that TIICs can regulate the development and progression of tumors, and immune cell infiltration exerts an influence on the survival of EOC patients [21]. In our study, the ssGSEA analysis was used to examine the association between PBF expression and immune cell infiltration in EOC. By analyzing an estimated fraction of the 24 TIICs in EOC, we found that PBF expression was positively associated with the infiltration levels of the iDCs, neutrophils, NK cells, and Tem. Infiltrating NK cells exert a strong immunosuppressive effect on the tumor microenvironment, reducing the secretion of IFN-γ and inducing T cell dysfunction [22]. Neutrophils play an essential role in innate immunity, which initiates an adaptive immune system response towards antigen stimuli. Simultaneously, neutrophils are also possible immunotherapy targets [23]. The analysis results proved that PBF may play a role in the recruitment and activation of CTLs by neutrophils. In addition, PBF is associated with most immune markers in EOC, such as CD8⁺ T cells and T cells, indicating that PBF may affect the regulation of T cell responses. Furthermore, PBF was highly associated with other T-cell markers, such as the different subtypes of T-helper cells in EOC, which indicated that PBF may regulate T lymphocyte immunity in EOC. PBF was also negatively associated with B cell markers (CD19 and CD79a) and TAM markers (CCL2), which suggests that it may affect immunosuppression and the regulation of macrophage polarization in EOC. The weak relationship between PBF and M1/M2 macrophage markers, such as PTGS2, IRF5, CD163, VSIG4, CD163, and MS4A4A, suggests that PBF affects the regulation of TAM polarization. Furthermore, PBF was associated with T cell exhaustion markers (PDCD1, CTLA4, LAG3, HAVCR2 and GZMB) and Treg markers (FOXP3, CCR8, STAT5B and TGFB1), which verified that PBF may influence immune escape in EOC. Together, these analysis results prove that PBF is associated with immune cell infiltration, which may serve as a novel immunotherapy target during EOC development, that affects the prognosis of EOC patients. Further studies are required to expound the relevant mechanisms between immune cell infiltration and PBF.

Our study, which was based on preliminary data obtained from TCGA has certain limitations. First, the sample size in the cohort study of EOC was relatively small, and more precise data can be obtained using a larger sample size and sufficient clinical information. Second, the possible regulatory mechanisms of PBF in EOC need to be further validated. Furthermore, more efforts are required to further illustrate the mechanisms involved in PBF-related immune infiltration for the development of immunotherapies for EOC.

In brief, this study verified that PBF may act as a promising biomarker and a marker of poor prognosis in EOC. Additionally, a strong correlation between PBF and pathways in EOC, such as the P53 pathway, B cell receptor pathway, toll pathway, Wnt pathway, FGF pathway, and Akt phosphorylation, was observed. Furthermore, PBF may also affect the microenvironment of EOC by regulating the tumor-infiltration of immune cells, indicating that PBF is a therapeutic target that regulates anti-tumor immune response. High PBF expression may be an independent prognostic factor for EOC patients and might be developed as a novel therapeutic target for EOC patients.

Disclosure of conflict of interest

None.