Scholarlink

Background <p>Pancreatic cancer (PC) remains one of the deadliest malignancies worldwide, characterized by late diagnosis, limited therapeutic options, and dismal prognosis. Post-translational modifications (PTMs) are increasingly recognized as key regulators of tumor biology, yet the role of S-nitrosylation—a redox-based cysteine modification—has not been systematically investigated in PC.</p> Methods <p>We integrated bulk and single-cell transcriptomic data from TCGA, ICGC, and GEO cohorts to comprehensively profile S-nitrosylation-related gene expression. A large-scale machine-learning framework incorporating 101 algorithmic combinations was used to construct an optimized prognostic model. Kaplan–Meier, ROC, and PCA analyses evaluated predictive power and stability across nine independent cohorts. The key hub gene LAMC2 was identified and subjected to single-gene bioinformatic analysis and functional validation through in-vitro experiments using MIA-PaCa-2 cells.</p> Results <p>S-nitrosylation exhibited the highest enrichment among all PTM types in pancreatic cancer and was mainly localized to malignant ductal and fibroblast cells. The StepCox(backward) + GBM model showed excellent prognostic performance (TCGA C-index = 0.827) and successfully stratified patients into high- and low-risk groups across multiple external datasets. High-risk patients displayed distinct transcriptional profiles and significantly worse survival outcomes (all <i>p</i> &lt; 0.01). Among S-nitrosylation-related genes, LAMC2 showed the strongest correlation with risk score (<i>r</i> = 0.63) and was consistently upregulated in tumors. Cox and GSEA analyses linked LAMC2 to epithelial–mesenchymal transition, TGF-β, and hypoxia pathways. Functionally, LAMC2 knockdown suppressed migration and colony formation in pancreatic cancer cells.</p> Conclusions <p>This study establishes S-nitrosylation as a crucial PTM in pancreatic cancer and introduces a robust machine-learning-based prognostic model. LAMC2 is identified as a key oncogenic effector connecting S-nitrosylation dysregulation to tumor aggressiveness, providing novel prognostic and therapeutic insights for pancreatic cancer management.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Single-cell transcriptomic profiling and machine learning identify S-nitrosylation-associated LAMC2 as a key prognostic driver in pancreatic cancer

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月21日

查看全文
  • Chenchen Yang,
  • Hengfa Ge,
  • Yunfan Yin,
  • Jialing Zhang,
  • Aifeng Gong

摘要

Background

Pancreatic cancer (PC) remains one of the deadliest malignancies worldwide, characterized by late diagnosis, limited therapeutic options, and dismal prognosis. Post-translational modifications (PTMs) are increasingly recognized as key regulators of tumor biology, yet the role of S-nitrosylation—a redox-based cysteine modification—has not been systematically investigated in PC.

Methods

We integrated bulk and single-cell transcriptomic data from TCGA, ICGC, and GEO cohorts to comprehensively profile S-nitrosylation-related gene expression. A large-scale machine-learning framework incorporating 101 algorithmic combinations was used to construct an optimized prognostic model. Kaplan–Meier, ROC, and PCA analyses evaluated predictive power and stability across nine independent cohorts. The key hub gene LAMC2 was identified and subjected to single-gene bioinformatic analysis and functional validation through in-vitro experiments using MIA-PaCa-2 cells.

Results

S-nitrosylation exhibited the highest enrichment among all PTM types in pancreatic cancer and was mainly localized to malignant ductal and fibroblast cells. The StepCox(backward) + GBM model showed excellent prognostic performance (TCGA C-index = 0.827) and successfully stratified patients into high- and low-risk groups across multiple external datasets. High-risk patients displayed distinct transcriptional profiles and significantly worse survival outcomes (all p < 0.01). Among S-nitrosylation-related genes, LAMC2 showed the strongest correlation with risk score (r = 0.63) and was consistently upregulated in tumors. Cox and GSEA analyses linked LAMC2 to epithelial–mesenchymal transition, TGF-β, and hypoxia pathways. Functionally, LAMC2 knockdown suppressed migration and colony formation in pancreatic cancer cells.

Conclusions

This study establishes S-nitrosylation as a crucial PTM in pancreatic cancer and introduces a robust machine-learning-based prognostic model. LAMC2 is identified as a key oncogenic effector connecting S-nitrosylation dysregulation to tumor aggressiveness, providing novel prognostic and therapeutic insights for pancreatic cancer management.