The RNA-seq data for the 33 cancers and normal tissue in this study were obtained from The Cancer Genome Atlas (TCGA) database (https://portal.gdc.cancer.gov/) and Genotype-Tissue Expression (GTEx) database (https://gtexportal.org/), in which contained 1152 normal human brain tissue and 706 glioma samples. Clinicopathologic information of glioma patients were acquired from TCGA database (comprising 699 tumor samples). The raw data were normalized with the transcripts per million (TPM) and subsequent analyses were performed using log2 (TPM+1). Expression profiling microarray data of GSE14805 were downloaded from the Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/projects/geo/) (containing 34 tumor samples and 4 normal samples) and statistically analyzed by Xiantao database (https://www.xiantaozi.com/), an online bioinformatic analysis tool based on R software.

The mRNA expression data of TGIF2 in normal tissues and subcellular localization of TGIF2 protein in SH-SY5Y cell line were sourced from Human Protein Atlas (HPA) database (http://www.proteinatlas.org/). Statistical analysis of TGIF2 expression in normal, tumor, and various clinicopathologic characteristic groups was performed and visualized by Xiantao tool (“stats”, “car” and “ggplot2” packages).

Glioma patients were categorized into two groups based on the median expression of TGIF2 mRNA (low-expression group: 0%-50%; high-expression group: 50%-100%). Kaplan-Meier analysis and Cox regression analysis were employed for survival analysis. Overall survival (OS), disease-specific survival (DSS), and progression-free interval (PFI) were compared between the high and low TGIF2 expression groups by Xiantao tool (“Survival” package). Additionally, OS, DSS and PFI were compared in different clinicopathologic characteristic subgroups of glioma patients. Kaplan-Meier plots were generated by Xiantao tool (“Survminer” and “ggplot2” packages).

Receiver Operator Characteristic (ROC) curves were utilized to assess the diagnostic value of TGIF2 and previously reported prognostic genes (IDH1, IDH2, EGFR, TP53, CDC6, CDC14B, and CHD5) in glioma. ROC analysis and visualized with Xiantao tool (“pROC”, “timeROC” and “ggplot2” packages). The closer the area under the ROC curve (AUC) is to “1”, the higher the diagnostic value of the gene.

Univariate logistic regression analysis was performed using Xiantao tool (“stats” package). Glioma patients were divided into two groups according to the median mRNA expression of TGIF2, with the low-expression group serving as the control. Logistic regression models were constructed to assess the predictive effects of TGIF2 expression on different clinicopathologic characteristics. P value < 0.05 was considered statistically significant.

Proportional hazards hypothesis testing was conducted using Xiantao tool (“survival” package), followed by univariate and multivariate Cox regression analyses to ascertain the independent prognostic significance of high TGIF2 expression and different clinicopathologic characteristics in the OS of glioma patients. Clinicopathologic factors with p value < 0.1 in the univariate Cox regression analysis were incorporated into the multivariate Cox regression analysis to further delineate their autonomous prognostic value. P value < 0.05 was considered as of statistical significance. Forest plots, risk score curve and point of survival chart were drawn using Xiantao tool (“ggplot2” package).

Utilizing the independent prognostic factors identified post multivariate Cox regression analysis, Xiantao tool (“rms” package) was employed to construct nomogram model predicting 1-, 3-, and 5-year OS in glioma patients. Calibration analysis and visualization were carried out using Xiantao tool (“rms” package), depicting the differences between the predicted and actual probabilities corresponding to the model at different time points.

For TCGA RNA-seq data, glioma patients were categorized into two groups (low-expression group: 0%-50%; high-expression group: 50%-100%) based on the median expression of TGIF2 mRNA. Differentially Expressed Genes (DEGs) of TCGA RNA-seq data and GSE14805 dataset statistically analyzed using Xiantao tool (“DEseq2” and “limma” packages respectively). Genes with an absolute log2 fold change (FC) > 1 and adjusted p value < 0.05 were considered DEGs and used for subsequent analysis. Volcano plots were generated by Xiantao tool (“ggplot2” package).

To determine the differences in biological functions and signaling pathways between high and low TGIF2 expression groups, the identified DEGs were analyzed for Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment using Xiantao tool (“clusterProfiler” package). Items with adjusted p value < 0.05 were considered as significant enrichment. The gene set “c2.cp.all.v2022.1.Hs.symbols.gmt [All Canonical Pathways]” in Molecular Signatures Database (MSigDB) was chosen as the reference gene set for Gene Set Enrichment Analysis (GSEA), and the items with false discovery rate (FDR) < 0.25 and adjusted p value < 0.05 were considered significantly enriched. Results of GO, KEGG and GSEA enrichment analyses were used Xiantao tool (“ggplot2” package) for visualization.

The “Estimate” package in R software (v 4.2.1) was applied to assess the stromal score, immune score, and estimate score in the glioma microenvironment grouped by high and low TGIF2 expression. Single sample Gene Set Enrichment Analysis (ssGSEA) was used to analyze the immune infiltration of 24 immune cells in glioma by calculating the enrichment score of each immune cell in the tumor samples according to the marker genes of the immune cells by Xiantao tool (“GSVA” package). Wilcoxon rank-sum test compared immune cell enrichment scores between high and low TGIF2 expression groups. Correlation between TGIF2 and various immune cells and immunoregulatory genes was implemented by Spearman analysis (22, 23). Visualization was performed using Xiantao tool (“ggplot2” package).

Spearman’s correlation test was employed to analyze the correlation of each gene with TGIF2. Genes with adjusted p value < 0.05 were sorted by the correlation coefficient, and the top 50 genes positively or negatively correlated with TGIF2 were selected for subsequent analysis. The coexpression heatmap of the top 15 genes positively or negatively correlated with TGIF2 were visualized using Xiantao tool (“ggplot2” package).

To investigate the interaction relationships among coexpression genes correlated with TGIF2, the STRING database (https://string-db.org/) was used to establish a Protein-Protein Interaction (PPI) network based on default parameters. Subsequently, the network was visualized by the Cytoscape software, and the top 10 hub genes were identified using the CytoHubba plugin. Kaplan-Meier plots and ROC plots of these 10 hub genes were generated as previously described.

U251 and 293T cells were cultured in DMEM (Cellmax, CGM102.05) medium supplemented with 10% FBS (Gibco, 2437694). To knockdown TGIF2, U251 cells were transfected with siRNA of TGIF2 (Sigma, SASI_Hs01_00107440) using Lipofectamine 3000 (Invitrogen, L3000015) at a final concentration of 100 pmol according to the manufacturer’s instructions when the cells reached approximately 50% confluence. For TGIF2 knockout, guide RNAs (gRNAs) targeting exon 2 of TGIF2 were designed using the http://crispor.tefor.net/website. The double-stranded DNA formed after annealing gRNA33 and gRNA140 were inserted into the PX459 V2.0 plasmid. We transfected 293T cells with the cloned vectors, extracted genomic DNA and performed Sanger sequencing to determine the knockout efficiency of the vectors. The screened gRNA was transfected into U251 cells, and the monoclones that successfully knocked out TGIF2 were selected for expansion culture for subsequent experiments. The primer sequences for synthesizing gRNA were as follows: gRNA33, Forward: 5′- CACCGACCTAGATCACTGTCCGACA -3′;Reverse: 5′-AAACTGTCGGACAGTGATCTAGGTC -3′. gRNA140, Forward: 5′- CACCGCGGTGAAGATCCTCCGGGAC -3′;Reverse: 5′- AAACGTCCCGGAGGATCTTCACCGC -3′. The primer sequences for verifying knockout efficiency of gRNA were as follows: Forward: 5′- CATCCCCTGTGTCCCTTGTC -3′;Reverse: 5′- ACTTGCAGCACTGACAGGTT -3′.

Total RNA was extracted with TsingZol total RNA extraction reagent (TSINGKE, TSP401) and cDNA was synthesized using 5× All-In-One RT Master Mix (Applied Biological Materials, G490) following the manufacturer’s protocol. RT-qPCR was performed using PowerUp™ SYBR™ Green Master Mix (Applied Biosystems™, A25742) and PIKOREAL 96 Real-Time PCR System (ThermoFisher). GAPDH served as internal control for quantifying relative gene expression levels using the 2 –ΔΔCT method. The primer sequences were as follows: TGIF2, Forward: 5′- CTTCAACACGCCACCACCCACACC-3′; Reverse: 5′- CCTCTGTAGCGCCACCTCCACCAG -3′.CDH2, Forward: 5′- CCTCCAGAGTTTACTGCCATGAC -3′; Reverse: 5′- GTAGGATCTCCGCCACTGATTC -3′. TWIST1, Forward: 5′- GCCAGGTACATCGACTTCCTCT -3′; Reverse: 5′- TCCATCCTCCAGACCGAGAAGG -3′. TWIST2, Forward: 5′- GCAAGATCCAGACGCTCAAGCT -3′; Reverse: 5′- ACACGGAGAAGGCGTAGCTGAG -3′. GAPDH, Forward: 5′- GGAGTCCACTGGCGTCTTCA -3′; Reverse: 5′- GTCATGAGTCCTTCCACGATACC -3′.

For transwell assay, 5x10⁴ cells were inoculated in transwell (Millicell, PI8P01250) upper chamber containing serum-free DMEM, and the lower chamber was layered with DMEM containing 10% FBS. After 24 h of incubation at 37°C, non-invasive cells on the upper surface of the membrane were wiped off with a cotton swab. Subsequently, invasive cells into the lower surface of the membrane were fixed with 4% paraformaldehyde and stained with 0.1% crystalline violet (Solarbio, G1063), observed under the microscope and selected random fields were photographed. For scratch wound-healing assay, cells were plated into 12-well plates and cultured until 90% confluence, scratched using a 10 μL pipette tip, and washed with PBS to remove dropped cells. The cells migration distance was observed at 0 h and 24 h of scratching and photographed.

Cells were lysed with RIPA lysis buffer (Beyotime, P0013B) supplemented with 1 mM of the protease inhibitor phenylmethylsulfonyl fluoride (Beyotime, ST506) for 20 min on ice. After centrifugation at 12,000 rpm, the supernatant was collected, loaded onto a 10% polyacrylamide SDS gel, and subsequently transferred to a PVDF membrane (Millipore, IPVH00010). Membranes were blocked with TBST containing 5% skimmed milk (BD Pharmingen™, 232100) for 1 h at room temperature, followed by overnight incubation at 4°C with primary antibodies. After 5 washes with TBST, the membranes were incubated with secondary antibody at room temperature for 2 h. The membranes were then scanned and analyzed using the ChemiDoc™ system (Bio-Rad). The following antibodies were used: TGIF2 (11522-1-AP, Proteintech); GAPDH (EPR16891, Abcam); N-Cadherin (22018-1-AP, Proteintech); secondary antibody (AB0101, Abways).

Bioinformatics analysis was performed with Wilcoxon rank sum test to detect statistical significance between the two groups. The relationship between TGIF2 expression and baseline clinicopathologic characteristics of patients was evaluated using the chi-square test by Xiantao tool (“stats” package). Spearman’s correlation coefficient assessed the correlation of TGIF2 expression with various immune cells and other genes. Statistical analysis of in vitro experiments were performed with GraphPad Prism 7.0 and comparisons between two conditions were applied by Student’s t-test and the data were presented as mean ± standard deviation (SD). P value < 0.05 was considered statistically significant.

We explored the basal expression levels of TGIF2 across various tissues. Analysis of HPA database revealed higher TGIF2 mRNA expression in breast and female reproductive system, bone marrow, lymphoid, and muscle tissues, while lower expression was observed in brain tissues ( Figure 1A ). Subsequently, TGIF2 expression was assessed in 33 tumors from TCGA and GTEx databases. Notably, TGIF2 expression levels in 21 tumor tissues were significantly higher than in corresponding normal tissues ( Figure 1B ). Specifically, elevated TGIF2 expression was observed in low-grade glioma (LGG) and glioblastoma (GBM) compared to normal brain tissues ( Figure 1B ). Subcellular localization of TGIF2 protein is predominantly in the nucleus in the human glioma cell line SH-SY5Y ( Figure 1C ). Furthermore, in the GSE14805 dataset, we identified 2954 DEGs, including 1369 upregulated and 1585 downregulated genes, with TGIF2 significantly upregulated in the tumor group ( Figure 1D ). The ROC analysis indicated similar diagnostic ability for TGIF2 (AUC = 0.971) with established glioma prognostic genes (IDH1, IDH2, EGFR, TP53, CDC6, CDC14B, CHD5) (24–30) ( Figure 1E ). Kaplan-Meyer analysis demonstrated that glioma patients with high TGIF2 expression exhibited poor OS, DSS and PFI ( Figures 1F–H ). These findings suggest that TGIF2 expression is elevated in glioma and associated with worse prognosis, warranting further investigation.

We investigated the association between TGIF2 expression and various clinicopathologic characteristics of glioma patients in the TCGA database. A total of 699 patients’ clinicopathologic information was included in the statistics, and the chi-square test revealed that TGIF2 expression was significantly correlated with age, WHO grade, IDH status, 1p/19q codeletion, histological type, OS event, DSS event and PFI event in glioma patients ( Table 1 ). Logistic regression analysis further confirmed significant associations between TGIF2 and age, WHO grade, IDH status, 1p/19q codeletion and histological type ( Table 2 ). TGIF2 expression patterns in different clinicopathologic characteristics showed elevation in age > 60 years, IDH-wild type, 1p/19q non-codeletion, progressive disease and stable disease (PD&SD), OS events, DSS events, and PFI events subgroups ( Figures 2A–G ). TGIF2 expression increased with WHO grade classification, peaking in G4 glioma ( Figure 2H ). For histological type, glioblastoma exhibited significantly higher TGIF2 expression compared to other glioma types ( Figure 2I ). However, no statistically significant differences were observed among glioma patients of different genders and races ( Figures 2J, K ). These results suggest that high TGIF2 expression may be associated with malignant phenotypes of glioma and worsening clinical outcomes.

We assessed the prognostic value of TGIF2 expression for glioma patients in the TCGA database. Firstly, analysis of TGIF2 expression distribution, survival status, and risk scores indicated that higher TGIF2 expression in the high-risk group correlated with increased mortality rates compared to the low-risk group ( Figure 3A ). Time-dependent ROC analysis demonstrated favorable predictive efficacy of TGIF2 expression for OS at 1-year (AUC = 0.653), 3-years (AUC = 0.715), and 5-years (AUC = 0.690) ( Figure 3B ). Univariate Cox regression analysis revealed an association between high TGIF2 expression and poorer OS (HR, 2.401; 95% confidence interval [CI], 1.867-3.089; p < 0.001) ( Table 3 ). Multivariate Cox regression analysis, after confounding with other clinicopathologic variables, identified TGIF2 as an independent prognostic risk factor for glioma patients (HR, 2.366; 95% CI, 1.620-3.455; p < 0.001). Other clinicopathologic parameters, including age (HR, 5.203; 95% CI, 3.291-8.224; p < 0.001), 1p/19q codeletion (HR, 0.490; 95% CI, 0.298-0.807; p = 0.005), histological type (HR, 4.885; 95% CI, 1.709-13.959; p = 0.003), and primary therapy outcome (HR, 0.215; 95% CI, 1.620 -3.455; p < 0.001), also emerged as independent prognostic factors ( Table 3 ; Figure 3C ). A prognostic nomogram model, incorporating TGIF2 expression and other independent prognostic factors analyzed by Cox regression, was constructed for predicting OS at 1, 3, and 5 years ( Figure 3D ). Calibration curves assessed the predictive efficiency of the nomogram ( Figure 3E ).

Next, we explored the relationship between TGIF2 and prognosis (OS, DSS, and PFI) across different clinicopathologic subgroups of glioma. Kaplan-Meyer analysis revealed that higher TGIF2 expression associated with worse OS in subgroups including age >60, age ≤60, male, female, Race (White), WHO grade (G3), 1p/19q codeletion status (non-codel), histological type (astrocytoma), and primary therapy outcome (PD) ( Figures 4A–I ). For DSS, patients with higher TGIF2 expression had worse DSS in subgroups such as age >60, age ≤60, male, female, race (White), WHO grade (G3), 1p/19q codeletion status (non-codel), histological type (astrocytoma), primary therapy outcome (PD) and primary therapy outcome (SD) ( Supplementary Figures S1A–J ). For PFI, higher TGIF2 expression was associated with worse PFI in subgroups including age >60, age ≤60, male, female, race (White), WHO grade (G3), 1p/19q codeletion status (non-codel) ( Supplementary Figures S2A–G ).

We analyzed DEGs between TGIF2 high and low expression groups based on the median TGIF2 expression level. A total of 2696 DEGs were identified, comprising 1353 upregulated and 1343 downregulated genes ( Figure 5A ). In GO enrichment analysis, the biological process (BP) mainly contained embryonic organ development, regulation of membrane potential, extracellular matrix organization, calcium-mediated signaling and chemokine-mediated signaling pathway. The cellular component (CC) was enriched in ion channel complex, collagen-containing extracellular matrix, glutamatergic synapse, T cell receptor complex and voltage-gated potassium channel complex. The molecular function (MF) was mainly involved in DNA-binding transcription activator activity, extracellular matrix structural constituent, voltage-gated potassium channel activity, cytokine activity and chemokine activity. KEGG pathway enrichment analysis indicated that TGIF2 potential participation in the regulation of neuroactive ligand-receptor interaction, calcium signaling pathway, cAMP signaling pathway, glutamatergic synapse and ECM-receptor interaction ( Figure 5B ).

GSEA was applied to further investigate the biological functions of TGIF2. Upregulation of TGIF2 expression correlated with cell cycle and DNA replication pathways, including DNA replication, cell cycle, synthesis of DNA, DNA repair, and cell cycle checkpoints ( Figure 5C ). ECM-related functions, such as extracellular matrix organization, ECM receptor interaction, and collagens and laminin interactions, were enriched in the TGIF2 high-expression phenotype, suggesting involvement in ECM formation in glioma ( Figure 5D ). TGIF2 also impacted pathways related to tumor immunity, including cytokine cytokine receptor interaction, signaling by interleukins, inflam pathway, chemokine signaling pathway, toll-like receptor signaling pathway, and Pd-1 signaling ( Figure 5E ). Additionally, pathways related to tumor progression, such as Jak Stat signaling pathway, signaling by Notch, and Pi3kakt signaling pathway, were enriched ( Figure 5F ). Ion channel and glutamate signaling are facilitators of glioma cell invasion and migration (31, 32), yet they were negatively correlated with high TGIF2 expression ( Figures 5G, H ), which may be related to glioma grading (33, 34). These findings suggest that TGIF2 may play a role in glioma cell proliferation, invasion and migration, and tumor immunity, making it a promising target for glioma treatment.

To unravel the intricate relationship between TGIF2 expression and the tumor immune response in glioma, we meticulously analyzed the disparities in the glioma immune microenvironment between high and low TGIF2 expression groups. The outcomes unveiled a significant elevation in stromal scores ( Figure 6A ), immunity scores ( Figure 6B ), and estimated scores ( Figure 6C ) in glioma patients with high TGIF2 expression. Furthermore, multiple immune cell subtypes were significantly enriched in the TGIF2 high-expression group, including aDC, cytotoxic cells, eosinophils, iDC, macrophages, neutrophils, NK CD56dim cells, NK cells, T cells, T helper cells, Th17 cells, and Th2 cells ( Figure 6D ). In contrast, the TGIF2 low-expression group exhibited higher enrichment of DC, mast cells, NK CD56bright cells, pDC, TFH, TReg, and Tcm ( Figure 6E ). Subsequently, we further delved into the association between TGIF2 expression levels and the infiltration levels of different immune cells in gliomas. The results demonstrated a positive correlation between TGIF2 expression and the infiltration levels of Th2 cells, macrophages, eosinophils and neutrophils, etc. Nevertheless, there was a negative correlation with infiltration levels of mast cells, NK CD56bright cells, pDC and TFH, etc. ( Figures 6F–N ). In addition, we investigated the correlation between TGIF2 and various immunoregulatory molecules to further understand its immune modulating function. These immunoregulatory genes include antigen presentation, cell adhesion, co-inhibitory, co-stimulatory, ligand, receptor, and other according to the classification of Thorsson et al. (23). The findings indicated a positive correlation between TGIF2 expression and most immunoregulatory genes, implying a potential role for TGIF2 in regulating glioma immune infiltration ( Supplementary Figure S3 ).

The analysis of genes coexpressed with TGIF2 in glioma provided valuable insights into potential regulatory networks and prognostic markers. The coexpression heatmap demonstrated the top 15 genes that were positively and negatively correlated with TGIF2 ( Figure 7A ). We performed PPI network analysis of the top 50 protein-coding genes positively and negatively correlated with TGIF2 using the STRING database and Cytoscape ( Figure 7B ) and identified the top 10 hub genes using the CytoHubba plugin (HDAC1, CASP3, REST, HMG20B, FZD7, GNAI3, FZD1, EPHB4, KCNJ9, SCRT1) ( Table 4 , Figure 7C). The top 10 hub genes highlighted potential central regulators in the context of TGIF2-associated glioma biology. Among these 10 genes HDAC1, CASP3, REST, HMG20B, FZD7,GNAI3, FZD1 and EPHB4 had elevated expression in tumor group compared to the normal group, while KCNJ9 and SCRT1 were decreased ( Figure 7D ). Kaplan-Meyer analysis showed that high expression of HDAC1, CASP3, REST, HMG20B, FZD7,GNAI3, FZD1 and EPHB4 was associated with worse OS, DSS and PFI, whereas high expression of KCNJ9 and SCRT1 suggested better OS, DSS and PFI ( Figures 8A–J ; Supplementary Figures S3A–J , S4A–J ). ROC analysis further validated the good diagnostic capabilities of these 10 hub genes in glioma ( Figures 9A–J ). In summary, this comprehensive analysis not only revealed potential regulatory networks with TGIF2 in glioma but also highlighted a promising gene cluster for prognostic assessment. Integrating these hub genes with TGIF2 may enhance the accuracy of prognostic predictions in glioma patients.

We explored the function of TGIF2 in vitro. Initially, we employed siRNA to knock down TGIF2 expression in the U251 glioma cell line. Both RT-qPCR and Western blot analyses confirmed the efficacy of TGIF2 knockdown, ensuring the reliability of subsequent experiments ( Figures 10A, B ). Subsequent transwell assays and scratch wound-healing assays revealed that TGIF2 inhibition suppressed the invasion and migration of U251 cells ( Figures 10C, D ). Furthermore, our investigations revealed a downregulation of N-Cadherin, a well-established marker of EMT, in the TGIF2-inhibited group ( Figure 10B ). This observation suggests that TGIF2 may play a role in promoting the EMT phenotype, thereby influencing glioma cell invasion and migration through the regulation of N-Cadherin expression. We used CRISPR/Cas9 gene editing to further validate the function of TGIF2. We designed gRNA sequences gRNA33 and gRNA140 targeting exon 2 of TGIF2, cloned them into the CRISPER Cas9 plasmid PX459 and verified the editing efficiency on TGIF2 ( Figure 11A ). Next, we transfected the PX459-gRNA33 plasmid, which was effective in editing TGIF2, into glioma U251 cells, and screened for TGIF2 knockout cell lines using Sanger sequencing ( Figures 11B, C ). Western blot analyses verified the knockout efficiency for TGIF2 of the PX459-gRNA33 plasmid in U251 cells ( Figure 11D ). Subsequently, transwell assays and scratch wound-healing assays showed that knockout of TGIF2 inhibited invasion and migration of U251 cells ( Figures 11E, F ). In addition, the expression of the signature genes of EMT (CDH2, TWIST1 and TWIST2) were also suppressed in the TGIF2-knockout group ( Supplementary Figure S6 ). These results strengthen the evidence that TGIF2 promotes glioma cell invasion, migration and EMT.