The UALCAN Database (http://ualcan.path.uab.edu) is a web-based tool for analyzing differential expression and survival data from the Cancer Genome Atlas (TCGA) and Genotypic Tissue Expression (GTEx) projects. The TIMER database (http://cistrome.dfci.harvard.edu/TIMER/) is a tool for analyzing differential expression and immune infiltration. GTPBP4 expression variations between BRCA and normal tissues were detected using these two online databases (UALCAN, TIMER).

From the TCGA and GEO databases, we collected BRCA datasets with GTPBP4 sequencing data. First, the TCGA database was used to retrieve gene expression and clinical data from BRCA patients. After that, log2 conversion to RNAseq data in FPKM (Fragments Per Kilobase Per Million) format is conducted. The GEO database’s BRCA datasets were then used for validation. The GEO datasets used in this study must adhere to the following guidelines: 1) The species is Homo sapiens; 2) The data comprised the expression profile of GTPBP4 in BRCA; 3) The total number of samples in both the tumor and non-tumor groups was more than 10. All data from the GEO database has been normalized and log2 transformed. Based on R software, ggplot2 package was used to visualize GTPBP4 expression.

To reduce the impact of random error, the meta-analysis combined GTPBP4 expression levels from the tumor group and the normal group from each independent GEO data set. The mean ± standard deviation of each data set was calculated first. The combined SMD (standardized mean difference) and 95 percent confidence interval (CI) were then calculated. The expression level of GTPBP4 in BRCA was substantially higher than that in the normal group if SMD>0 and p < 0.05 were detected. We created the SROC (Summary Receiver Operator Characteristic) curve and evaluated the area under the curve (AUC) to further examine GTPBP4’s sensitivity in diagnosing BRCA.

Furthermore, the variability of particular research was investigated. If I² is more than 50%, the random-effects model is used. The fixed effects model is utilized if I² is less than 50%. Finally, to assess publication bias, the Begg and Egger tests were used. There was no significant publication bias in this study when p > 0.05. Begg’s test, Egger’s test, and SROC analysis were done with Stata program, while other analyses were done with R software’s “meta” and “forestplot” packages.

The TCGA cohort was then used to investigate the predictive usefulness of GTPBP4 in breast cancer. TCGA patients were separated into two groups based on the median value of GTPBP4 expression: high expression and low expression, with 535 patients in each group. The “survival” package in R was then used to investigate survival differences between the two groups. To see if GTPBP4 was an independent predictive marker for breast cancer, researchers used univariate and multivariate COX regression. Statistical significance was set at p < 0.05.

Then, using the R software package ClusterProfiler (version 3.0.4), differential genes in high- and low-expression groups were analyzed for gene ontology (GO) enrichment, and the pathway and ontology of substantial enrichment and linked genes were characterized. The gene set enrichment analysis (GSEA) was the next step in the enrichment analysis process. GSEA starts with a set of biologically significant genes (such as those in a pathway), calculates the genes in the set (all of which have the same meaning/function), and then summarizes them into a single enrichment score. This method of analysis adds interpretability and was used in this study to evaluate changes in pathway/functional activity of the gene set and select the gene set with a p < 0.05.

We investigated differences in immune cell infiltration levels between the high GTPBP4 expression group and the low GTPBP4 expression group using the “CIBERSORT” code data and 22 immune cell comparison files on the CIBERSORT website. Then, we tested the correlation between these 22 immune cells and GTPBP4. Finally, through the intersection of different immune cell types and significantly correlated immune cell types, we obtained the most significant GTPBP4-related immune cell type in breast cancer.

We built a nomogram model incorporating GTPBP4 expression and clinical features in the TCGA cohort to further evaluate breast cancer patient survival. To integrate GTPBP4 expression data with clinical characteristics, the “Regplot” software was used to plot patient “TCGA-AR-A255.” To assess the correctness of this Nomogram, 3- and 5-year ROC curves as well as calibration curves were created.

The Chinese Cell Repository provided MDA-MB-231, BT-549, SUM1315MO2, and ZR-75–1 breast cancer cell lines as well as human breast epithelial cell lines (HBL-100) (Shanghai, China).

All cells were grown at 37°C with 5% CO₂ in DMEM (Gibco) with 10% fetal bovine serum (Gibco) and 1% penicillin-streptomycin solution (Gibco). These cells were transfected using the Lipofectamine3000 (Thermo Fisher Scientific, Waltham, MA, United States) according to the manufacturer’s procedure with previously generated short interfering RNAs (Hippobiotec, Huzhou, China) targeting gene GTPBP4. Supplementary Table S1 lists the siRNA sequences for the gene GTPBP4. All data were presented as the means ± SD of three independent experiments.

Total cellular RNAs were extracted according to the manufacturer’s instructions using Trizol Reagent (Invitrogen, Carlsbad, CA, United States). PrimeScript TM RT reagentKit was used to make cDNAs (Takara, Dalian, China). On an ABI Stepone plus PCR equipment, qRT-PCR was performed using AceQ Universal SYBR qPCR Master Mix (Vazyme, Nanjing, China) (Applied Biosystems, FosterCity, CA, United States). The 2^−ΔΔCt technique was used to determine relative quantification. The level of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA was used to standardize the relative expression of messenger RNA (mRNA) for each gene. In Supplementary Table S2, the primer sequences are listed. All data were presented as the means ± SD of three independent experiments.

Cell proliferation was measured using the Cell Counting Kit-8 (CCK8) technique. A 96-well cell culture plate was used to seed the cells. Each well of the plate received 10 μL of CCK-8 solution (Biosharp, Hefei, China). The plate was then placed in the cell culture incubator for 3 h, away from light. Finally, the absorbance of each well was measured at 450 nm using a microplate reader. All data were presented as the means ± SD of three independent experiments.

The migration of ZR-75-1 and MDA-MB-231 cells was assessed using a transwell assay. Cells were seeded into the upper well for 36 h after transfection and allowed to invade through the transwell plate. Methanol was used to fix the cells on the inserts, which were then stained with crystal violet and counted under a light microscope. All data were presented as the means ± SD of three independent experiments.

Transfected ZR-75-1 and MDA-MB-231 cells were plated in 6-well plates and wounds were made with a pipette tip. Cell migration was assessed by measuring wound closure at 0 and 24 h. All data were presented as the means ± SD of three independent experiments.

Differences between normal tissue and tumor samples were analyzed using the Wilcoxon rank sum test. For survival analysis, Cox analysis and the K-M curve were utilized. For drug sensitivity correlation analysis, the Pearson correlation test was performed. The signature’s accuracy was assessed using a ROC curve. In the absence of special instructions, the test level was set at p < 0.05. The platform for analysis is R software (4.1.2).

To confirm GTPBP4 expression in breast cancer, we looked at it in multiple independent cohorts, including one TCGA cohort and five GEO cohorts. GTPBP4 expression was elevated in breast cancer in each independent cohort (p < 0.001), as seen in Figures 3A–F.

Following that, we built ROC curves in the 6 cohorts mentioned above, including 1 TCGA cohort and 5 GEO cohorts, to determine the diagnostic utility of GTPBP4 for breast cancer. The area under the curve (AUC) is then calculated. The AUC for the TCGA cohort is 0.762, while the AUCs for the five GEO cohorts were 0.854, 0.874, 0.956, 0.777, and 0.859, respectively (Figures 4A–F). GTPBP4 has a high diagnosis accuracy for breast cancer, according to these findings.

To increase the credibility of the results, we conducted a meta-analysis of GTPBP4 expression in 5 GEO Cohorts. As shown in Figure 5A, the random effect model was used for analysis due to heterogeneity I²>50%. The combined SMD was 1.31, 95%CI was (0.84, 1.78). GTPBP4 expression upregulation in breast cancer is thus plausible. Sensitivity analysis revealed that the findings of our meta-analysis were trustworthy (Figure 5B). The AUC value of sROC curve was 0.80, indicating good diagnostic accuracy of GTPBP4 (Figure 5C). Finally, the p-value of Egger test was 0.599 (Figure 5D), and the p-value of Begg test was 0.462 (Figure 5E), indicating that there was no obvious publication bias.

Then there was a survival analysis. According to the median expression value of GTPBP4, TCGA breast cancer patients were separated into two groups: high expression and low expression. The difference in survival time between the two groups was then examined. The findings revealed that elevated GTPBP4 expression was linked to a poor breast cancer outcome (p = 0.012, Figure 6A). Following that, univariate and multivariate COX regressions were performed, with the results presented in Table 1 GTPBP4 was found to be an independent breast cancer predictive factor using COX regression. Table 2 shows the clinical information for these patients.

Gene ontology enrichment analysis was used to find pathways that differed between the high and low GTPBP4 expression groups. The differently elevated genes in the two groups were mostly enriched in organelle fission, humoral immunological response, and nuclear division, according to the findings (Figure 6B). GSEA showed that glycolysis, DNA repair, E2F targets and G2M checkpoint pathways were significantly enriched in GTPBP4 high expression group (Figures 6C–F). This is significant for revealing the role of GTPBP4 in breast cancer.

The immunological landscape of breast cancer is depicted in Figure 7A. The presence of diverse immune cells is plainly obvious. Following that, we looked into the changes in immune cell infiltration levels between the high and low GTPBP4 expression groups, and discovered eight different immune cells (Figure 7B). Following that, correlation analysis revealed that GTPBP4 was significantly linked with 12 different types of immune cells (Figures 7C–N). Finally, the intersection of eight separate immune cell types and twelve related immune cell types yielded the five most important immune cell types (Figure 7O). T cells CD4 memory, T cells Follicular helper, T cells regulatory (Tregs), Macrophages M2, and Mast cells resting were identified. Supplementary Figure S2 showed the correlation analysis between GTPBP4 and tumor mutation load (TMB). It can be shown that the TMB level of the GTPBP4 high expression group was similarly high, and the TMB and GTPBP4 expression levels had a positive correlation. Supplementary Figure 3 depicted the relationship between GTPBP4 and the ESTIMATE score, Immune score, Stromal score, and tumor purity, highlighting GTPBP4’s function in the immune milieu.

We created a nomogram integrating GTPBP4 expression and clinical features to further evaluate the survival of breast cancer patients. Patient “TCGA-AR-A255” had 1-, 3-, and 5-year death rates of 0.0113, 0.0617, and 0.112, respectively, as shown in Figure 8A. The AUC values of the 3-year and 5-year ROC curves for this nomogram were 0.77 and 0.75, respectively (Figure8B, 8c). The nomogram can accurately predict the survival of breast cancer patients, according to the 3-year and 5-year calibration curves (Figures 8D,E).

The relative expression levels of GTPBP4 in Breast Cancer cell lines were quantified using qRT-PCR. In the cell lines ZR-75-1, SUM1315MO2, BT-549, and MDA-MB-231, we discovered that GTPB4 is overexpressed (Figure 9A). Because GTPBP4 expression was higher in the ZR-75-1 and MDA-MB-231 breast cancer cell lines, gene knockdown was performed in these two cell lines. With three different siRNA sequences, we inhibited the expression of GTPBP4 in ZR-75-1 and MDA-MB-231 breast cancer cell lines using RNAi technology. The knockdown potency of all three sequences was significant, although sequence 2’s knockdown effect was more desirable. As a result, it was employed in subsequent research.