3.1 Expression landscape of the manganese metabolism family across pan-cancer

To investigate the role of the manganese metabolism family in regulating cancer progression, we examined its differential expression across various cancers, including BLCA, BRCA, CHOL, COAD, ESCA, HNSC, KICH, KIRC, LIHC, PRAD, STAD, THCA, and UCEC. Significant expression differences were identified for the manganese metabolism family across these cancer types. To further explore these variations, we conducted differential expression analyses, presented in a heatmap (Fig. 1A). Additionally, a correlation plot effectively illustrates the expression patterns of manganese metabolism across 14 cancer types (Fig. 1B).

Fig. 1

Differential Expression of the Manganese Metabolism Family A Heatmap depicting the expression landscape of the manganese metabolism family across pan-cancer, with red indicating high expression and blue indicating low expression. B Correlation analysis showing differential expression of the manganese metabolism family across 14 cancers. *P < 0.05, **P < 0.01, ***P < 0.001

3.2 Mutations in the manganese metabolism family accelerate pan-cancer progression

We further analyzed the mutational landscape of the manganese metabolism family. Mutation enrichment analysis (MEs) and TCW mutation count analysis revealed varying degrees of mutational enrichment within the manganese metabolism family genes, with NUDT3 exhibiting particularly significant changes (Fig. 2A). We then categorized NUDT3 expression into low and high score groups, and observed that the low-score group exhibited the highest level of enrichment across 14 cancer types (Fig. 2B). Subsequently, we analyzed the expression levels of TMB, MEs, and TCW mutation rates. The results demonstrated differential expression of manganese metabolism across these genes, indicating that mutations in the manganese metabolism family, especially in NUDT3, play a critical role in exacerbating cancer progression, particularly in lung cancer (Fig. 2C).

Fig. 2

Mutational Landscape of the Manganese Metabolism Family Genes. A MEs enrichment analysis of mutation rates in the manganese metabolism family, and TCW mutation analysis showing the number of mutations in the manganese metabolism family genes. B Correlation analysis illustrating the relationship between low, medium, and high NUDT3 expression scores and their enrichment across 14 cancer types. C Expression analysis of TMB, MEs, and TCW mutation rates in the manganese metabolism family genes across 14 cancer types

3.3 NUDT3 mutations in the manganese metabolism family aggravate lung cancer progression

Next, we focused on analyzing the role of the manganese metabolism gene NUDT3 in lung cancer. Initially, we performed a comprehensive analysis of the manganese metabolism family across 14 cancers, confirming high mutation enrichment in MEs, consistent with our previous findings (Fig. 3A). Further mutation expression analysis revealed an increased mutational frequency of NUDT3 in the MEs (Fig. 3B).We also observed that as the MES levels increased, both co-occurring and mutually exclusive mutations became more frequent, indicating that higher MES levels are associated with elevated somatic mutation activity in lung cancer. Additionally, we categorized MEs into low, medium, and high mutation rate groups, identifying a specific correlation with ZFHX4 and TIN, while no significant correlation was found in the low-score group (Fig. 3C). Upon further investigation of key genes inducing NUDT3 mutations, we discovered that mutations in the ZFHX4 locus could also induce NUDT3 mutations. In conclusion, NUDT3 mutations are likely influenced by ZFHX4 mutations (Fig. 3D).

Fig. 3

Genomic Analysis of NUDT3 in the Manganese Metabolism Family. A MEs enrichment analysis of mutation rates in the manganese metabolism family. B Mutation expression analysis of NUDT3. C Co-occurrence and mutual exclusivity analysis of NUDT3 based on low, medium, and high mutation scores across gene loci. D Analysis of NUDT3 mutations. *P < 0.05, **P < 0.01

3.4 Tumor heterogeneity in lung adenocarcinoma: expression landscape

To further validate our transcriptomic findings, we conducted a single-cell analysis using data from lung adenocarcinoma samples in the GEO database (GSE149655). This analysis included four lung adenocarcinoma samples and two normal samples, which were processed using the Seurat package. During this process, we selected a minimum of 200 genes and a maximum of 4,000 genes for inclusion in the analysis. The "LogNormalize" function was applied to normalize the data, and visualization of expression patterns was performed using the R package. We identified 25 distinct cell clusters. Subsequently, using the SingleR package, we successfully annotated 10 major cell clusters, including epithelial cells (Fig. 4A-B).

Fig. 4

Tumor Heterogeneity in Lung Adenocarcinoma A tSNE plot showing the distribution of different clusters prior to annotation. B tSNE plot displaying cell types annotated using the SingleR package. C Significance analysis of various cell subpopulations. D Proportional analysis of the distribution of each cell subpopulation. E Enrichment scores of manganese metabolism family genes across different cell subpopulations. F Expression landscape of the manganese metabolism family genes. G Cell communication analysis depicting interactions among various cell subpopulations,*P < 0.05,**P < 0.01,***P < 0.001, ****P < 0.0001

Following enrichment scoring and significance analysis of different cell subgroups, we observed specific variations within epithelial cells (Fig. 4C, E). A proportional analysis of the 10 cell subgroups revealed a marked decrease in the percentage of epithelial cells in the tumor group compared to the normal group (Fig. 4D). We then excluded normal epithelial cells and extracted tumor-specific epithelial subpopulations, identifying 12 smaller subpopulations. Among these, clusters 0, 1, 4, and 5, which exhibited reduced expression, were marked as key cell subpopulations (Fig. 5A).Subsequent cell communication analysis revealed strong synergistic interactions between these key epithelial subpopulations and other cell groups (Fig. 4G). Moreover, the single-cell expression landscape of manganese metabolism genes, particularly NUDT3, highlighted their critical role in exacerbating lung adenocarcinoma progression (Fig. 4F).

Fig. 5

Identification of NUDT3 in Epithelial Cells of Lung Adenocarcinoma. A Expression of different clusters in epithelial cells. B Pathway scoring for manganese metabolism genes, with red indicating high scores and blue indicating low scores. C Pseudotime analysis of manganese metabolism genes. D-E Key cell communication analysis after NMF clustering. F Spatial expression of NUDT3, PPM1A, B4GALT1, MGAT1, LAP3, and GALNT1 in epithelial cells. G Pathway enrichment analysis in epithelial cells

3.5 Expression and localization of NUDT3 in lung adenocarcinoma epithelial cells

After excluding normal epithelial cells, tumor epithelial cells were analyzed to investigate the spatial and temporal expression patterns of manganese metabolism-related genes. The distribution of epithelial cell subclusters between tumor and normal samples is shown (Fig. 5A), where subclusters such as GALNT1-C0 and LAP3-C1 were predominantly observed in tumor samples, while other subclusters were more evenly distributed. Differential expression analysis across these subclusters highlighted significant variations in the expression of manganese metabolism-related genes (Fig. 5B). For instance, NUDT3 showed increased expression in the GALNT1-C0 subcluster, while MGAT1 expression was enriched in LAP3-C1. NUDT3 was also found to have significantly higher expression in late-stage tumors compared to early-stage tumors (Fig. 5C).Using the NMF clustering method, six key manganese metabolism-related genes were identified: NUDT3, PPM1A, B4GALT1, MGAT1, LAP3, and GALNT1. These genes displayed distinct spatial expression patterns within epithelial cells, as shown in the UMAP plot (Fig. 5F). LAP3 was predominantly expressed in the lower region of the UMAP plot, overlapping with LAP3-C1 subclusters, while B4GALT1 was mainly expressed in the upper region. NUDT3 and PPM1A showed broad expression across epithelial cells, but with higher intensities in specific subclusters, such as GALNT1-C0 and LAP3-C1.Cell communication analysis revealed differences in outgoing and incoming signaling patterns across epithelial cell subclusters (Fig. 5D, E). The GALNT1-C0 subcluster had the highest outgoing signaling strength, while LAP3-C1 demonstrated strong incoming signaling activity. These findings indicate that these subclusters may serve as key hubs in epithelial cell signaling networks.Pathway enrichment analysis using GSVA identified multiple biological pathways associated with manganese metabolism-related genes (Fig. 5G). Enriched pathways included epithelial-mesenchymal transition (EMT) and xenobiotic metabolism, which showed higher activity in tumor epithelial cells compared to normal cells. In addition, through correlation analysis, we found that NUDT3 had a negative correlation with MGAT1, while GALNT1 had a positive correlation, while other genes had no correlation (Supplementary 1A).

3.6 Spatial heterogeneity of NUDT3 in lung adenocarcinoma

To investigate the spatial organization of cell populations in lung adenocarcinoma tissues, spatial transcriptomics analysis was performed. The results revealed a heterogeneous distribution of cell populations, with tumor cells predominantly localized in mixed regions of the tissue (Fig. 6A). These regions were characterized by a high density of immune cells, such as macrophages, CD4 + T cells, and CD8 + T cells, reflecting the complexity of the tumor microenvironment.Using spatial annotation, 11 distinct cell subpopulations were identified and visualized, including epithelial cells, tumor cells, B cells, macrophages, neutrophils, endothelial cells, fibroblasts, dendritic cells, and plasma cells (Fig. 6B). The spatial distribution map highlights the co-localization of tumor cells with various immune and stromal cell types, suggesting dynamic cellular interactions within the tumor microenvironment.The spatial analysis provides a comprehensive overview of cell type heterogeneity within lung adenocarcinoma tissues, highlighting the intricate organization of tumor and immune cells.

Fig. 6

Spatial transcriptomics analysis of lung adenocarcinoma tissues. A Distribution of CD4 + T cells, CD8 + T cells, and tumor cells, with tumor cells predominantly located in mixed regions surrounded by immune cells. B Annotation of 11 distinct cell subpopulations, including epithelial cells, tumor cells, macrophages, dendritic cells, B cells, neutrophils, fibroblasts, endothelial cells, and plasma cells, highlighting the heterogeneity and spatial organization of the tumor microenvironment

3.7 Increased expression of NUDT3 as a key driver of lung adenocarcinoma progression

In our analysis of lung adenocarcinoma samples, we identified NUDT3 as a key factor associated with disease progression. Using data from the TCGA-LUSC and GSE21933 datasets, we evaluated the expression and diagnostic significance of NUDT3. ROC curve analysis revealed a high diagnostic value for NUDT3 in lung adenocarcinoma, with an AUC of 0.912 (95% CI: 0.886–0.936) (Fig. 7A). Paired sample analysis indicated significantly elevated NUDT3 expression in tumor tissues compared to normal tissues (p = 2.411e − 08) (Fig. 7B).Further investigation into tumor staging showed that NUDT3 expression increases progressively with tumor stage, with the highest expression observed in stage IV tumors (Fig. 7C). These findings were consistent across datasets, as demonstrated by the density plot and box plot from GSE21933, confirming upregulated NUDT3 expression in tumor tissues (Fig. 7D).

Fig. 7

NUDT3 expression and its diagnostic and prognostic significance in lung adenocarcinoma. A ROC curve analysis (TCGA-LUSC) showing high diagnostic accuracy for NUDT3 (AUC = 0.912, 95% CI 0.886–0.936). B NUDT3 expression is significantly upregulated in tumor tissues compared to normal tissues (C) Progressive increase in median NUDT3 expression across tumor stages (I–IV), correlating with tumor progression. D Density and box plots (GSE21933) confirm higher NUDT3 expression in tumor tissues

3.8 Elevated NUDT3 expression in A549 lung adenocarcinoma cells

The relative mRNA expression of NUDT3 was significantly higher in the A549 lung adenocarcinoma cell line compared to the BEAS-2B normal bronchial epithelial cell line. As illustrated in Fig. 8, NUDT3 expression in A549 cells is elevated by approximately 40-fold relative to BEAS-2B cells. This upregulation is statistically significant (p < 0.01), highlighting a pronounced increase in NUDT3 levels in lung adenocarcinoma cells. These findings point to a potential role for NUDT3 in promoting tumor progression.

Fig. 8

Relative mRNA expression of NUDT3 in BEAS-2B and A549 cells. Quantitative PCR analysis shows significantly higher mRNA expression of NUDT3 in A549 lung cancer cells compared to BEAS-2B normal lung epithelial cells. Data are presented as mean ± SD, P < 0.01