Using TIMER (http://cistrome.dfci.harvard.edu/TIMER/) and UCSC Xena (https://xena.ucsc.edu/), we analyzed pan-cancer GNPNAT1 expression with the Wilcoxon test [23, 24], complemented by CPTAC-derived proteomics data (https://pdc.cancer.gov/pdc/browse) [25]. For CC, Gene Expression Omnibus (GEO) datasets GSE63514 (24 normal, 28 tumor) and GSE6791 (eight normal, 20 tumor) were utilized, with the Wilcoxon test applied for statistical comparison.

GNPNAT1’s prognostic relevance in CC was assessed via Gene Expression Profiling Interactive Analysis (GEPIA, http://gepia2.cancer-pku.cn/) and the Wilcoxon test [26]. Pan-cancer survival data from UCSC Xena, analyzed with “survminer” (version 0.4.9) and “survival” (version 3.7-0) packages, determined its prognostic and diagnostic value. Receiver operating characteristic (ROC) curves defined diagnostic thresholds: area under the curve (AUC) $>$0.7 for moderate or higher, AUC $>$0.8 for high diagnostic value. The log-rank test (p 0.05) confirmed significance, with HR $>$1 indicating poor prognosis and HR 1 indicating favorable prognosis upon high GNPNAT1 expression.

The University of Alabama at Birmingham CANcer Data Analysis Portal (UALCAN; https://ualcan.path.uab.edu) was used to analyze the relationship between GNPNAT1 expression and clinicopathological staging. Statistical significance was assessed via the Wilcoxon test. The Wilcoxon test was also applied to investigate the correlation between GNPNAT1 expression and the Ki-67 index [27].

We analyzed the biological functions of tumor cells using CancerSEA (http://biocc.hrbmu.edu.cn/CancerSEA/) [28]. The Sangerbox platform (http://vip.sangerbox.com) was used to analyze the pan-cancer correlations of GNPNAT1 with tumor stemness, RNA modifications, and its mutation landscape [29]. The stemness metrics encompassed: DNA methylation-based stemness score (DNAss), the differentially methylated probe-based stemness score (DMPss), epigenetically regulated gene expression-based stemness score (EREG-EXPss), enhancer element/DNA methylation-based stemness score (ENHss), RNA expression-based stemness score (RNAss), and epigenetically regulated DNA methylation-based stemness score (EREG-METHss) [30]. We presented pan-cancer mutation profiles in this study, with a specific focus on mutation distribution in CC and breast cancer. The chi-square test was used to evaluate differences in gene mutation frequency between each group of samples. Additionally, we analyzed the relationship between RNA methylation modifications (m1A, m5C, and m6A) and GNPNAT1 expression.

We used R software (v4.3.2, Posit PBC, Boston, MA, USA) to investigate the relationships between GNPNAT1 and various immune-related genes, including immune checkpoint, immune stimulatory, major histocompatibility complex (MHC), chemokine, immune inhibitory, and chemokine receptor genes. Tumor, stromal, and immune scores were compared via the “ESTIMATE” package (version 1.0.13). Immune cell infiltration differences were assessed using seven TIMER-based algorithms (XCell, TIMER, EPIC, CIBERSORT-abs, CIBERSORT, QUANTISEQ, and MCPcounter). We used the Tumor Immune Single-cell Hub 2 (TISCH2; http://tisch.comp-genomics.org) to analyze GNPNAT1 expression patterns across different tumor cell subpopulations [31]. We used the TISMO website (http://tismo.cistrome.org/) to compare the predictive capabilities of immunotherapy responses across different tumor models [32]. We used GSCALite to evaluate GNPNAT1 sensitivity across various cancers (Supplementary Figs. 1,2) [33].

CC transcriptomes from The Cancer Genome Atlas (TCGA) were grouped by the optimal GNPNAT1 cutoff. Differential expression analysis applied the criteria $|$log₂FC$|$ $\ge$1 and adjusted p 0.05, after which gene set enrichment analysis (GSEA) was performed via “clusterProfiler” packages (version 4.10.1) (p 0.05).

HeLa and C33A cells were sourced from Shanghai Gaining Biological Technology Co., Ltd., and SiHa cells from Harbin Medical University’s Laboratory of Medical Genetics. All lines were short tandem repeat-validated and mycoplasma-free. Cultures were grown in DMEM (Gibco, 11885084, Waltham, MA, USA) with 10% fetal bovine serum (FBS; Gibco, Waltham, MA, USA) and 1% penicillin–streptomycin (Sevenbio, Beijing, China) at 37 °C, 5% CO₂.

The SiHa and C33A cells were transfected with GNPNAT1 small interfering RNA (siRNA). The siRNA target sequences are listed in Supplementary Table 1. The primer sequences are provided in Supplementary Table 2. Western blotting was performed as previously described [34]. The following antibodies were used: anti-GNPNAT1 (Rabbit, 16282-1-AP, Proteintech, diluted 1:1000) and anti-$\beta$-actin (Mouse, 66009-1-Ig, Proteintech, diluted 1:10,000).

Cell Counting Kit-8 (CCK-8, Sevenbio, SC119, Beijing, China) assessed cell viability by measuring 450 nm absorbance every 24 hours for 4 days post-seeding in 96-well plates. Colony formation was evaluated by seeding cells in 6-well plates, culturing for 10–14 days, staining with 1% crystal violet, and counting visible colonies.

Migration was measured by Transwell assay (8.0-µm pores) with serum-free medium in the upper chamber and 10% FBS in the lower chamber. After 24 hours, non-migrated cells were swabbed off, and migrated cells were stained (1% crystal violet) and imaged. In the scratch assay, confluent monolayers (95%) in 6-well plates were wounded with a 10 µL pipette tip.

Statistical work was performed in GraphPad Prism (v9.5, Inc., San Diego, CA, USA) and R software, with results expressed as mean $\pm$ SD. The Benjamini–Hochberg procedure corrected for multiple testing. Unless otherwise specified, group comparisons employed one-way analysis of variance or two-tailed Student’s t-test, with p 0.05 considered significant. Correlation strength was measured by Pearson’s r, with $|$r$|$ $>$0.3 and p 0.05 indicating significance (r $>$0, positive; r 0, negative). For all figures, statistical significance is denoted as follows: *p 0.05; **p 0.01; ***p 0.001; ****p 0.0001.

Pan-cancer analysis of TIMER and TCGA data demonstrated markedly higher GNPNAT1 levels in over ten tumor types, including cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC), stomach adenocarcinoma (STAD), and breast invasive carcinoma (BRCA), compared to normal counterparts (Fig. 1A,B). Similarly, paired samples from the TCGA database showed significantly higher GNPNAT1 expression in multiple cancer types (Supplementary Fig. 3). We analyzed pan-cancer data from the CPTAC database to study variations in GNPNAT1 protein expression across different cancer types. Our results revealed that GNPNAT1 was significantly overexpressed in several types of malignant tumors, including uterine, breast, and lung cancers (Fig. 1C).

Fig. 1.

Glucosamine 6-phosphate N-acetyltransferase 1 (GNPNAT1) expression in pan-cancer across different databases. (A) GNPNAT1 expression in pan-cancer from the TIMER database. (B) GNPNAT1 expression in pan-cancer from The Cancer Genome Atlas (TCGA) database. (C) Changes in GNPNAT1 protein expression levels in various cancers. (* p 0.05; ** p 0.01; *** p 0.001). ACC, adrenocortical carcinoma; BLCA, bladder urothelial carcinoma; BRCA, breast invasive carcinoma; CESC, cervical squamous cell carcinoma and endocervical adenocarcinoma; CHOL, cholangiocarcinoma; COAD, colon adenocarcinoma; DLBC, lymphoid neoplasm diffuse large B-cell lymphoma; ESCA, esophageal carcinoma; GBM, glioblastoma multiforme; HNSC, head and neck squamous cell carcinoma; KICH, kidney chromophobe; KIRC, kidney renal clear cell carcinoma; KIRP, kidney renal papillary cell carcinoma; LAML, acute myeloid leukemia; LGG, brain lower grade glioma; LIHC, liver hepatocellular carcinoma; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; MESO, mesothelioma; OV, ovarian serous cystadenocarcinoma; PAAD, pancreatic adenocarcinoma; PCPG, pheochromocytoma and paraganglioma; PRAD, prostate adenocarcinoma; READ, rectum adenocarcinoma; SARC, sarcoma; SKCM, skin cutaneous melanoma; STAD, stomach adenocarcinoma; TGCT, testicular germ cell tumors; THCA, thyroid carcinoma; THYM, thymoma; UCEC, uterine corpus endometrial carcinoma; UCS, uterine carcinosarcoma; UVM, uveal melanoma.

We conducted ROC curve analysis by calculating the AUC to assess the diagnostic value of GNPNAT1 expression in pan-cancer. GNPNAT1 expression exhibited exceptional diagnostic efficacy in CESC (AUC = 0.832), kidney renal papillary cell carcinoma (KIRC; AUC = 0.882), and LUAD (AUC = 0.930) (Fig. 2).

Fig. 2.

Accuracy of GNPNAT1 in differentiating tumor from normal tissues in pan-cancer.

We further evaluated the prognostic value of GNPNAT1 expression in human cancers. First, we analyzed the relationship between GNPNAT1 expression levels and the clinical and pathological characteristics of patients. Our results revealed that high GNPNAT1 expression was significantly positively correlated with advanced clinical stages in multiple tumors, including bladder urothelial carcinoma (BLCA), BRCA, and CESC (Fig. 3A). In addition, high GNPNAT1 expression was closely associated with elevated levels of the tumor proliferation marker Ki-67 (Fig. 3B). This suggests that GNPNAT1 may promote tumor progression. Univariate Cox analysis determined GNPNAT1’s prognostic value across cancers (Fig. 4A–D). Kaplan-Meier analyses linked GNPNAT1 expression to overall survival (OS), disease-free interval (DFI), disease-specific survival (DSS), and progression-free interval (PFI) (Supplementary Figs. 4–7). High GNPNAT1 was identified as an OS risk factor in adrenocortical carcinoma (ACC), CESC, and LUAD, and notably served as a dual risk factor for OS and PFI in CESC.

Fig. 3.

Association of tumoral GNPNAT1 expression with clinical features. (A) Correlation with pathological stage across cancer types. (B) Pan-cancer relationship with the Ki-67 proliferation index. (* p 0.05; ** p 0.01; *** p 0.001; ns p $>$ 0.05).

Fig. 4.

GNPNAT1 prognostic potential in pan-cancer. (A–D) Forest plots of univariate Cox regression analyses for overall survival (OS) (A), disease-specific survival (DSS) (B), disease-free interval (DFI) (C), and progression-free interval (PFI) (D).

Further investigation revealed that GNPNAT1 expression exhibits a significant positive correlation with tumor stemness in the majority of malignancies (Fig. 5A–F). This suggests that high GNPNAT1 expression may contribute to maintaining tumor stemness. Additionally, single-cell functional state analysis revealed heterogeneous functional associations of GNPNAT1 across various cancer types. In acute myeloid leukemia, GNPNAT1 expression was positively correlated with cellular hypoxia, apoptosis, proliferation, and invasion, but negatively correlated with DNA repair. In BRCA, GNPNAT1 expression correlated positively with cell cycle, DNA damage, apoptosis, and invasion. Similarly, positive associations were observed in LUAD for invasion and hypoxia, and in gliomas for DNA repair, cell cycle, inflammation, and proliferation. Conversely, melanoma displayed an inverse pattern, with GNPNAT1 exhibiting negative correlations with DNA repair, apoptosis, and invasion (Fig. 5G).

Fig. 5.

Pan-cancer Spearman correlation analysis of GNPNAT1 expression with stemness indices and cellular functional states. (A–F) Correlations with stemness scores: differentially methylated probe-based stemness score (DMPss) (A), DNA methylation-based stemness score (DNAss) (B), enhancer element/DNA methylation-based stemness score (ENHss) (C), epigenetically regulated gene expression-based stemness score (EREG-EXPss) (D), epigenetically regulated DNA methylation-based stemness score (EREG-METHss) (E), and RNA expression-based stemness score (RNAss) (F). (G) Correlation analysis between functional states and GNPNAT1 expression in AML, breast invasive carcinoma (BRCA), LUAD, Glioma, and UVM. (* p 0.05; ** p 0.01; *** p 0.001).

Tumor mutation is a key driver of malignant transformation, promoting cell proliferation and immune evasion and thereby propelling cancer progression. In this study, we examined GNPNAT1 mutation patterns in different tumors. Mutational landscape comparison between high- and low-GNPNAT1 groups uncovered significantly mutated genes in each subgroup. For example, we discovered mutations in genes such as TTN, SYNE1, and DNAH10 in CESC, and we observed high-frequency mutations in TP53 and GATA3 in breast cancer (Fig. 6A–C). Further correlation analysis demonstrated that GNPNAT1 expression was positively associated with m1A, m5C, and m6A methylation levels (Fig. 7). This suggests that GNPNAT1 may influence tumorigenesis and development by regulating epigenetic transcription processes.

Fig. 6.

Mutation characterization of GNPNAT1. (A) Mutation map across cancer types. (B,C) The top ten most frequently mutated genes in CESC and BRCA patients.

Fig. 7.

Relationship of GNPNAT1 expression with RNA modifications. (* p 0.05).

Further investigation demonstrated that GNPNAT1 expression was positively linked to immune checkpoints (PD-L1, CTLA-4, LAG-3) in cancers such as colorectal adenocarcinoma (COAD) and uterine corpus endometrial carcinoma (UCEC) (Fig. 8A). Conversely, in tumors such as serous ovarian cancer (OV) and rectal adenocarcinoma (READ), GNPNAT1 expression was positively correlated with MHC genes, including HLA-A, HLA-DR, and HLA-DM (Fig. 8B). In STAD and thymoma (THYM), GNPNAT1 expression was positively correlated with T-cell chemokines, including CXCL10 and CCL5. These findings imply that GNPNAT1 may promote immune cell recruitment into the tumor microenvironment. Conversely, GNPNAT1 showed a negative correlation with immunosuppressive chemokines, such as CCL17 (Fig. 8C,D). Additionally, in LUAD and OV tumors, GNPNAT1 was positively correlated with immunostimulatory (CD276 and HHLA2) and immunosuppressive (TGFBR1 and KDR) (Fig. 8E,F).

Fig. 8.

Correlation of GNPNAT1 expression with immune-related genes. (A) Immune checkpoint genes. (B) Major histocompatibility complex (MHC) genes. (C,D) Chemokines and chemokine receptors. (E,F) Immunostimulatory genes and immunosuppressive genes. (* p 0.05; ** p 0.01; *** p 0.001).

We analyzed the cell-type-specific expression of GNPNAT1 in the tumor microenvironment (TME) using single-cell transcriptomic data. We investigated its distribution across different cell types and found that in non-Hodgkin lymphoma (NHL), GNPNAT1 is primarily expressed in keratinocytes and fibroblasts (Fig. 9A). In non-small cell lung cancer (NSCLC), GNPNAT1 is primarily distributed in macrophages and natural killer (NK) cells (Fig. 9B). In pancreatic adenocarcinoma (PAAD), GNPNAT1 is predominantly expressed in endothelial cells, fibroblasts, and malignant cells (Fig. 9C). Immune infiltration analysis demonstrated that GNPNAT1 expression is positively associated with B cell, T cell, and dendritic cell abundance in tumors such as thyroid carcinoma (THCA) and prostate adenocarcinoma (PRAD) (Fig. 9D). We further comprehensively assessed this association using seven independent algorithms (Supplementary Figs. 8,9). The results revealed that, in most tumors, GNPNAT1 expression correlated positively with neutrophil and CD4⁺ T cell infiltration, while exhibiting negative associations with macrophages, CD8⁺ T cells, and NK cells. Such findings imply that GNPNAT1 may function through immune regulatory mechanisms.

Fig. 9.

Correlation of GNPNAT1 with immune infiltration, microsatellite instability (MSI), tumor mutational burden (TMB), and drug sensitivity. (A–C) GNPNAT1 distribution in non-Hodgkin lymphoma (NHL), non-small cell lung cancer (NSCLC), and PAAD. (D) Correlation with infiltrating immune cells. (E,F) Radar plots for MSI (E) and TMB (F). (G) Correlation with immune infiltration scores. (H) Top 30 drug sensitivity correlations (GDSC). (* p 0.05; ** p 0.01; *** p 0.001; **** p 0.0001).

Immunotherapy is a key approach to cancer treatment, and its efficacy is significantly influenced by microsatellite instability (MSI) and tumor mutational burden (TMB). In tumors such as UCEC, STAD, and COAD, GNPNAT1 expression correlates positively with both MSI and TMB. However, in skin cutaneous melanoma (SKCM), PRAD, LUAD, and lymphoid neoplasm diffuse large B-cell lymphoma (DLBC), GNPNAT1 expression is negatively correlated with MSI. In contrast, in THCA, kidney renal papillary cell carcinoma (KIRP), and kidney renal clear cell carcinoma (KIRC), GNPNAT1 expression is negatively correlated with TMB (Fig. 9E,F). GNPNAT1 was negatively correlated with stromal, immune, and ESTIMATE scores in most cancers (Fig. 9G), pointing to its possible involvement in fostering an immunosuppressive microenvironment.

Furthermore, analysis of mouse models based on the TISMO database indicates that GNPNAT1 expression may predict responses to treatment with PD-L1 and CTLA-4 inhibitors in 4T1, T22, and CT26 models. Changes in GNPNAT1 expression before and after cytokine intervention further support the notion that it may exhibit plasticity in immune regulation (Supplementary Figs. 10,11). To explore the potential of GNPNAT1 as a therapeutic target, we used the GSCALite database to analyse its relationship with drug sensitivity. Analysis of CTRP database revealed a positive correlation between GNPNAT1 and sensitivity to BIX-01294, CCT036477, GSK525762A, I-BET151, NVP-BSK805, and a negative correlation with IC₅₀ values for ABT-737 and oseltamivir (Supplementary Fig. 12). Analysis of GDSC database indicated that GNPNAT1 was positively correlated with the IC₅₀ values of AR-42, BX-912, CAY10603, CP466722, methotrexate, and TG101348, and negatively correlated with the IC₅₀ values of 17-AAG (Fig. 9H).

Subsequently, DEG analysis between CESC groups with high versus low GNPNAT1 expression identified the top 15 up- and downregulated genes (Fig. 10A). The upregulated genes in the high-expression group primarily included immunoglobulin-related genes (IGHA1, IGHA2, IGKC, IGLC3, and IGHG3) and genes related to cell structure (KRT5, KRT15, and DSC3). These results suggest that CC with GNPNAT1 overexpression may exhibit enhanced humoral immune responses and epithelial structural characteristics. Enrichment analysis revealed that GNPNAT1 overexpression significantly enriched multiple pathways, including chemokine-mediated signaling pathways, the humoral immune response, the Ig-mediated immune response, cell adhesion, and the epidermal growth factor receptor binding pathway (Fig. 10B,C).

Fig. 10.

Enrichment analysis based on CESC patients. (A) Differential analysis heatmap. (B,C) Enrichment analysis.

Finally, we validated the expression of GNPNAT1 in CC cohorts. Our findings demonstrated that GNPNAT1 expression was markedly elevated in CC tissues compared to normal counterparts (Fig. 11A,B,E,F). Furthermore, elevated GNPNAT1 expression was associated with worse clinical outcomes (Fig. 11C). Time-dependent ROC curve analysis revealed that GNPNAT1 exhibited excellent predictive ability for one-, three-, and five-year survival rates, with an AUC value of 0.944 for 1-year survival (Fig. 11D). Functional experiments confirmed that GNPNAT1 knockdown markedly suppressed the proliferation, colony formation, and migration abilities of CC cells (Fig. 11G–L).

Fig. 11.

The role of GNPNAT1 in CC. (A,B) Differential expression of GNPNAT1 in normal versus CC tissues. (C) DFS for CC in the GEPIA database. (D) The area under the receiver operating characteristic (ROC) showing the prognostic value of GNPNAT1 for CC at 1, 3, and 5 years. (E,F) GNPNAT1 protein and mRNA expression in CC cells. (G,H) GNPNAT1 mRNA (G) and protein (H) levels in SiHa and C33A cells following si-NC or si-GNPNAT1 transfection. (I–L) GNPNAT1-knockdown SiHa and C33A cells were subjected to (I) CCK8 assay, (J) wound healing assay (scale bar: 200 µm, 4$\times$ objective), (K) colony formation assay, and (L) Transwell assay (scale bar: 100 µm, 10$\times$ objective). (* p 0.05; ** p 0.01; *** p 0.001). The experiment was repeated biologically with n=3.