Expression of the IL-18-related gene PTX3 correlates with clinicopathological features and prognosis in glioma patients

Gene expression data processing and normalization

RNA sequencing data for glioblastoma (GBM), lower-grade glioma (LGG), and normal brain tissues were obtained from The Cancer Genome Atlas (TCGA) and GTEx through UCSC XENA (https://xenabrowser.net/datapages/). The dataset included 174 GBM cases and 532 LGG samples. Genes with expression levels below 1 in more than 50% of the samples were excluded from the analysis. The validation cohorts, which included comprehensive expression profile data from GSE50161 and GSE7696, were sourced from the GEO database (https://www.ncbi.nlm.nih.gov/gds). The prognostic validation cohort is derived from the Chinese Glioma Genome Atlas (CGGA) database (http://www.cgga.org.cn). Additionally, 256 genes associated with the IL-18 signaling pathway, collectively termed ISRGs, were curated from the WP_IL18_SIGNALING_PATHWAY database, as outlined in Table S1.

Development of protein-protein interactions and identification of central genes

The protein-protein interaction (PPI) network of differentially expressed genes (DEGs) was constructed using STRING (http://string-db.org). The network was then visualized using Cytoscape software (https://www.cytoscape.org).

Functional enrichment analysis

Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were conducted using the clusterProfiler package from Bioconductor (Yu et al., 2012). Gene expression data normalization and identification of differentially expressed genes (DEGs) were performed using the limma package in R (version 3.18) (Li et al., 2021; Wang et al., 2019). The criteria for functional enrichment analysis of DEGs were set as —logFC— > 2 and an adjusted P-value < 0.05.

Gene set enrichment analysis

The clusterProfiler R package (version 3.14.3) was used to examine functional and pathway differences between groups with varying PTX3 expression levels. Significantly enriched hallmarks were identified based on an FDR q-value < 0.05.

Machine learning algorithm to screen key genes

To identify key biomarkers associated with glioma prognosis, we extracted data from the TCGA database and developed gene expression profiles. Using the LASSO algorithm from the glmnet R package, we conducted a comprehensive screening for significant biomarkers. A total of 256 ISRGs were identified through LASSO Cox regression analysis, aimed at minimizing redundancy and preventing model overfitting. The optimal penalty parameter, lambda, was determined through a 10-fold cross-validation procedure. Nine genes were ultimately selected to construct a prognostic risk score model for predicting overall survival (OS) in glioma patients.

Correlation analysis between PTX3 expression levels and immune cell infiltration

The single-sample Gene Set Enrichment Analysis (ssGSEA) algorithm, implemented in the “GSVA” (version 1.34.0) R package, was used to assess the infiltration levels of 28 distinct immune cell types in the tumors. Subsequently, Spearman’s correlation analysis was performed to explore the associations between PTX3 expression levels and immune cell infiltration status.

Correlation analysis between PTX3 expression levels and clinicopathological characteristics

Clinicopathological data for patients diagnosed with glioblastoma multiforme GBM and LGG, including OS, disease-specific survival (DSS), and progression-free interval (PFI), were obtained from the TCGA-GBM and TCGA-LGG projects. Nomograms that integrate clinical features alongside PTX3 models were constructed utilizing the “rms” R package (version 6.3) to forecast the overall survival (OS) of glioma samples. Logistic regression analysis was performed to assess the relationship between PTX3 expression levels and the clinicopathological characteristics of glioma patients.

Construction of transcription factor-gene interaction networks

The hub genes identified previously were subsequently uploaded to NetworkAnalyst (version 3.0, https://www.networkanalyst.ca/) to develop transcription factor (TF) and gene interaction networks. The construction of the TF-gene interaction network was facilitated through the utilization of the JASPAR database (http://jaspar.genereg.net) via the NetworkAnalyst platform.

Construction of the TF-miRNA co-regulatory network

Data regarding the TF-miRNA coregulatory network was sourced from the RegNetwork database (http://www.regnetworkweb.org/), which amalgamates established regulatory interactions from various databases along with prospective regulations inferred from TF binding sites. The visualization of the TF-miRNA coregulatory network was accomplished utilizing the NetworkAnalyst platform.

Ethics statement

Glioma tissues and adjacent paracancerous tissues were obtained from the Department of Pathology, Second Affiliated Hospital, School of Medicine, Zhejiang University. This study adhered to the ethical principles outlined in the Declaration of Helsinki. Given its retrospective nature, patient consent was waived for retrospective data by the Ethics Committee. We randomly selected glioma and adjacent normal brain tissues from 108 patients who had undergone surgical resection at the Second Affiliated Hospital, Zhejiang University. The research protocols received endorsement from the Ethics Committee of the Second Affiliated Hospital, Zhejiang University School of Medicine, located in Hangzhou, China (approval number: 2024-0649).

Immunohistochemical staining

Antigen retrieval was performed using an autoclave after tissue sections were dewaxed in water. The sections were equilibrated to room temperature and then treated with 3% hydrogen peroxide for 10 min. To block non-specific binding, a 10% serum solution was applied. The sections were incubated overnight at 4 °C with the primary antibody, PTX3 (Proteintech, dilution 1:1000). The following day, the sections were incubated with the secondary antibody at room temperature for one hour. Positive signals were visualized in brown using DAB as the chromogenic substrate. The nuclei were counterstained with hematoxylin for 30 s. Finally, the sections were mounted with a neutral dendrimer, examined microscopically, and photographed for documentation. The staining outcomes were systematically evaluated by a duo of pathologists, who assigned scores according to specified criteria: tumor intensity was categorized as 0 (negative), 1 (weak positive), 2 (moderate positive), or 3 (strong positive); tumor extent was classified as 0 (0–10%), 1 (10%–25%), 2 (26%–50%), 3 (51%–75%), or 4 (76%–100%). The overall score was determined by multiplying the intensity score with the extent score. Scores between six and 12 were interpreted as indicative of elevated PTX3 expression or positive status, while scores from 0 to five denoted low expression or the absence of PTX3, thus indicating a negative outcome.

Cell culture and transfection procedures

The U251 cell line was acquired from the American Type Culture Collection (ATCC) (Manassas, VA, USA). U251 cells were maintained in DMEM/F12 medium (Gibco, Waltham, MA, USA), which was enriched with 10% fetal bovine serum (Lonsera, Australia) and 1% penicillin/streptomycin (Beyotime Biotechnology, Jiangsu, China). Following this, the silencing of PTX3 was achieved through cell transfection. Small interfering RNA (siRNA) that specifically targets PTX3 (designated as si-PTX3) was procured alongside a negative control (si-NC) from He Yuan Biology (Shanghai, China). The sequence for si-PTX3 was identified as GGGAUAGUGUUCUUAGCAATT. The transfection of U251 cells was performed utilizing Lipofectamine 3000 (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer’s protocols, with a transfection duration of 48 h.

Quantitative real-time PCR

Total RNA extraction was conducted employing the TRIzol reagent produced by Invitrogen (Waltham, MA, USA). The complementary DNA (cDNA) synthesis was carried out utilizing the PrimeScript RT kit provided by Takara (Shiga, Japan). The mRNA expression levels were quantified through the application of qRT-PCR. The primers were meticulously designed by Qingke (Hangzhou, China), specifically: PTX3 forward primer: 5′-GCATAATAGGAACACTTGAGAC-3′ and reverse primer: 5′-CTGACAGAGACACAGCATT-3′.

Wound healing assay

In the wound healing assay, cells were cultivated in 24-well plates. A sterile tip was utilized to create scratches in the cells, aligned perpendicularly to the previously marked line. Following the scratching process, the resulting wound area was captured using a light microscope, and cell migration was assessed at designated time intervals of 0 and 24 h. Each experiment was conducted in triplicate to ensure reliability of the results.

Cell proliferation assay

Cells were harvested via centrifugation and subsequently resuspended in the culture medium to achieve a concentration of 10⁴ cells per milliliter. Following this, 200 µL of the culture medium was dispensed into each well of a 96-well plate. The evaluation of cell proliferation was performed utilizing the CCK-8 assay (Beyotime Biotechnology, Shanghai, China), with measurements taken at 0, 24, 48, and 72 h in accordance with the manufacturer’s guidelines. The optical density at 450 nm (OD450) was quantified and documented using GraphPad Prism software to determine the proliferative capacity of the cells. Each experimental procedure was conducted in triplicate.

Colony formation assay

In order to conduct the colony formation assay, transfected U251 cells were seeded into a 6-well plate at a density of 500 cells per well and cultured for a duration of 14 days. Following this incubation period, the resulting colonies were fixed using a 4% paraformaldehyde solution (Beyotime, Shanghai, China) for 30 min and subsequently stained with a 0.1% crystal violet solution (C0121; Beyotime, Shanghai, China). Ultimately, the colonies were counted to facilitate further analysis.

Statistical analysis

Bioinformatics analyses were conducted using R software (version 4.2.0). A one-way ANOVA was performed to assess differences among multiple groups. The Wilcoxon rank-sum test was conducted for comparing two groups.

The mRNA expression profiles of selected ISRGs in glioma

To identify the most suitable candidate genes, we employed a LASSO Cox regression method to develop a gene signature based on ISRGs. This analysis resulted in the creation of a prognostic model consisting of nine genes with non-zero coefficients: BMP2, NCF1, HSPB1, PIGT, PTX3, CCNA2, CCNB2, CCN4, and DES (Fig. 1A). A univariate Cox regression analysis was conducted to assess the correlation between ISRG expression levels and OS in glioma patients, as shown in Fig. 1B. Notably, the upregulation of NCF1, HSPB1, PIGT, PTX3, CCNA2, CCNB2, CCN4, and DES was significantly associated with poor prognostic outcomes. The gene expression heatmap, integrating clinical parameters, shows that higher expression levels of these genes correlate with advanced WHO grades, IDH wild-type status, 1p/19q non-codel status, and poor prognosis (Fig. 1C). Furthermore, BMP2, NCF1, HSPB1, PIGT, PTX3, CCNA2, and CCNB2 exhibited increased expression, while DES displayed decreased expression in glioma tissues (Fig. 1D). These gene expression patterns not only reflect the biological characteristics of the tumors but may also be pivotal for evaluating clinical outcomes. Thus, future research should focus on elucidating the functional roles of these genes and their mechanisms in tumor progression.

Expression correlation and prognosis of ISRGs

Subsequent investigations examined the interactions among these seven key genes in glioma. Using the Spearman correlation method, we identified significant correlations between these genes (Figs. 2A–2C). Additionally, Kaplan–Meier survival analyses confirmed the strong association between increased expression levels of NCF1, HSPB1, PIGT, PTX3, CCNA2, CCNB2, CCN4, and DES and poor prognostic outcomes in glioma patients. Notably, elevated expression of these genes is associated with shorter survival and disease progression, whereas increased BMP2 expression exhibited a contrasting effect (Fig. 2D).

PTX3 expression levels are significantly elevated in multiple cancers including glioma

The analysis of PTX3 expression across various cancer datasets provides strong evidence for its role in tumor biology. In our study, we observed an upregulation of PTX3 in gliomas, while its downregulation in 20 different cancer types suggests a potential tumor-suppressive role in these contexts (Figs. S1A, S1B). The correlation between PTX3 and immune infiltrating cells is noteworthy (Fig. S1C). PTX3 interacts with immune components, influencing tumor progression and the inflammatory microenvironment in cancers.

Functional enrichment analysis of PTX3 in glioma

The PPI network of DEGs was constructed using the STRING database to analyze their interactions, which were visualized with Cytoscape software (Fig. 3A). The interaction frequency of each gene is also shown (Fig. 3B). The results of the GO functional analysis and KEGG enrichment analysis are summarized below. The GO enrichment analysis identified significantly enriched biological processes, cellular components, and molecular functions, as detailed in Fig. 4 and Table 1. The key biological processes include leukocyte migration, leukocyte chemotaxis, myeloid leukocyte migration, and cytokine-mediated signaling pathways. The most prominent molecular functions include receptor–ligand activity, signaling receptor activator activity, cytokine activity, and extracellular matrix structural constituents. The cellular components with the highest enrichment levels are the secretory granule lumen, cytoplasmic vesicle lumen, vesicle lumen, and protein-DNA complex. Among the KEGG pathways, notable entries include cytokine-cytokine receptor interaction, transcriptional misregulation in cancer, the PI3K-Akt signaling pathway, and protein digestion and absorption (Figs. 4A, 4B) (Table 1). Additionally, GSEA revealed that the DEGs associated with PTX3 were significantly enriched in the integrin1, syndecan1, aurora-b, and IL18 signaling pathways (Figs. 5A–5C).

Association between the expression of PTX3 and survival outcomes in glioma

The mRNA expression of PTX3 in the validation datasets (GSE50161, GSE7696) was comparable. Upregulated mRNA levels of PTX3 were observed in tumor tissues across these datasets (Figs. 6A, 6D). The volcano plot clearly illustrates the differentially expressed genes. These findings suggest that PTX3 is differentially expressed in glioma, with elevated expression levels (Figs. 6B, 6E). Receiver operating characteristic (ROC) curve analysis further confirmed the robust performance of our risk model (Figs. 6C, 6F).

To assess the reliability of the LASSO-derived prognostic model developed from the previous TCGA dataset, we utilized the CGGA dataset for external validation purposes. In addition, we analyzed the expression profiles of PTX3 in conjunction with risk scores, survival durations, and the distribution of survival statuses within the CGGA dataset. The analytical results demonstrated that PTX3 expression levels in the high-risk group were significantly higher than those in the low-risk group, correlating with poorer prognoses (Fig. 6G). The Kaplan–Meier (KM) curve analysis revealed a significant association between elevated expression levels of PTX3 and negative clinical outcomes (Fig. 6H). Additionally, the results showed that elevated PTX3 expression was associated with reduced survival rates in patients with G4, IDH wild-type, 1p/19q non-codel in patients diagnosed with glioma (Figs. 6I–6K).

Furthermore, we analyzed OS outcomes in patients categorized by low and high PTX3 expression levels across different phenotypes (Fig. 7). The results showed that elevated PTX3 expression was associated with reduced survival rates in patients with G3, G4, IDH wild-type, 1p/19q alterations, as well as in both female and male patients diagnosed with astrocytoma or glioblastoma. In contrast, no significant correlation was observed in patients classified as G2, IDH-mutated, oligoastrocytoma, or oligodendroglioma.

PTX3 expression levels correlate with multiple clinicopathological characteristics in glioma

Significant differences were observed in WHO grade, IDH status, 1p/19q codeletion, age, histological classification, OS, DSS, and PFI between glioma patients with high and low PTX3 expression levels (Figs. 8A–8H). ROC curve analysis further validated the robust performance of our risk model (Figs. 8I–8L). The correlations between clinicopathological characteristics and PTX3 protein levels in glioma patients are presented in Table 2. Patients with high PTX3 expression exhibited poorer survival outcomes. We subsequently combined the WHO grade, IDH status, 1p/19q co-deletion, and PTX3 expression levels to develop a nomogram aimed at predicting survival outcomes (Fig. 9A). Importantly, the expression of PTX3 has shown promise in improving the precision of survival probability estimations at the 1-, 3-, and 5-year marks. Furthermore, a calibration chart was employed to assess the accuracy of the predictions generated by the model (Fig. 9B). The time-dependent ROC curves revealed AUC values for the 1, 3, and 5-year periods that surpassed 0.75, signifying the robust performance of the model (Fig. 9C). Univariate Cox regression analysis identified WHO grade, IDH status, 1p/19q codeletion, age, and PTX3 expression as independent prognostic markers for glioma patients. Furthermore, multivariate Cox regression analysis confirmed WHO grade, IDH status, and age as independent prognostic determinants (Fig. 9D).

TF-gene interaction and TF-miRNA coregulatory network

The NetworkAnalyst database facilitated the prediction and visualization of interactions between transcription factors (TFs) and hub genes. The resulting TF-gene interaction network comprised 18 nodes and 19 edges, as illustrated in Fig. 10A. Specifically, FOXL1 demonstrated interactions with two hub genes, namely PTX3 and CXCL8, whereas TP53 was found to interact with PTX3 and MMP9. Nonetheless, these observations necessitate additional validation. Following this, the NetworkAnalyst tool was employed to develop the TF-miRNA co-regulatory network, depicted in Fig. 10B.

PTX3 expression levels in tissue samples of glioma patients

To assess the differences in PTX3 protein expression between adjacent paracancerous and glioma tissues, we employed IHC to examine the expression levels of PTX3 in both tissue types. HE staining revealed that tumor cells in LGG displayed a small, rounded, adipose-like morphology, characterized by invasive growth. These cells exhibited a mild increase in density, substantial cytoplasm, and infrequent mitotic figures. In contrast, tumor cells in GBM were marked by invasive growth and a rounded shape, with a moderate to high increase in cell density and reduced cytoplasmic volume. Mitotic activity was notably high, with over 20 mitoses observed per 10 high-power fields (HPF), along with significant pseudopalisading necrosis and endothelial proliferation. IHC results showed that PTX3 expression was significantly higher in glioma tissues compared to adjacent paracancerous tissues (Fig. 11A). Furthermore, we assessed the relationship between PTX3 expression and various clinical parameters in patients. Elevated levels of PTX3 expression were found to be associated with a more advanced WHO grade, non-codel status of 1p/19q, as well as IDH wild type (Figs. 11B–11E).

Biological significances of PTX3 in glioma progression

In order to evaluate the function of PTX3 in glioma, a series of in vitro experiments were conducted. qPCR analysis demonstrated that the suppression of PTX3 resulted in a reduction of its mRNA expression levels in U251 cells (Fig. 12A). Furthermore, a significant increase in the number of clones was observed in the si-NC group compared to the si-PTX3 group (Figs. 12B, 12C). The results from the CCK-8 assay indicated that the proliferative capacity of U251 cells was markedly diminished following the knockdown of PTX3 (Fig. 12D). Additionally, wound healing assays indicated a decline in the metastatic potential of U251 cells subsequent to PTX3 knockdown (Figs. 12E, 12F). The inhibition of glioma cell growth and migration upon PTX3 knockdown suggests that PTX3 plays a role in promoting glioma progression.