Machine learning and multi-omics analyses reveal CLDN1 as a candidate marker associated with colorectal cancer invasion and nutlin-3a (-) response.
8 affiliations
- 1The Fifth Clinical Medical College of Henan University of Chinese Medicine (Zhengzhou People's Hospital), Zhengzhou, Henan, China.
- 2Department of Oncology, The Fifth Clinical Medical College of Henan University of Chinese Medicine (Zhengzhou People's Hospital), Zhengzhou, Henan, China.
- 3Translational Medicine Research Center, The Fifth Clinical Medical College of Henan University of Chinese Medicine (Zhengzhou People's Hospital), Zhengzhou, Henan, China.
- 4Scientific Research Management Office, The Fifth Clinical Medical College of Henan University of Chinese Medicine (Zhengzhou People's Hospital), Zhengzhou, Henan, China.
- 5Department of Orthopaedics, The First Affiliated Hospital of Zhengzhou University, Zhengzhou, Henan, China.
- 6Department of Microbiology and Immunology, School of Basic Medical Sciences, Zhengzhou University, Zhengzhou, Henan, China. Electronic address: [email protected].
- 7Translational Medicine Research Center, The Fifth Clinical Medical College of Henan University of Chinese Medicine (Zhengzhou People's Hospital), Zhengzhou, Henan, China. Electronic address: [email protected].
- 8Translational Medicine Research Center, The Fifth Clinical Medical College of Henan University of Chinese Medicine (Zhengzhou People's Hospital), Zhengzhou, Henan, China. Electronic address: [email protected].
Abstract
Objective
Colorectal cancer (CRC) is molecularly, cellularly and spatially heterogeneous, and current staging systems and biomarkers incompletely explain prognosis and treatment response. This study integrated transcriptomics, machine learning, single-cell RNA sequencing, spatial transcriptomics and in vitro experiments to prioritize candidate CRC-associated markers and characterize their potential biological and clinical relevance.Methods
GSE44861 and GSE44076 from the Gene Expression Omnibus (GEO) and The Cancer Genome Atlas Colon Adenocarcinoma (TCGA-COAD) were used for differential expression, consensus differentially expressed gene (DEG) identification and functional enrichment analyses. Four machine-learning algorithms were applied for feature selection. Candidate genes were prioritized by differential expression, model importance and receiver operating characteristic (ROC) performance. GSE231559 single-cell data and spatial transcriptomics characterized the cellular source, pseudotime dynamics and spatial distribution of the core gene. TCGA-COAD assessed prognosis, pathways and computationally predicted drug sensitivity. Western blotting, Transwell and MTT assays were used for experimental validation, and public immunohistochemistry images from the Human Protein Atlas were used as supplementary protein-expression evidence.Results
We identified 544 consensus DEGs enriched in leukocyte migration, chemotaxis, extracellular matrix organization, cytokine activity and metabolic pathways. Six genes were jointly selected: CLDN1, CDH3, KRT24, MMP7, NEBL and SCGN. CLDN1 was markedly upregulated and showed the highest tumour-versus-normal discrimination among the six candidates in the TCGA-COAD cohort. Single-cell and spatial analyses showed that CLDN1 was mainly expressed in CRC epithelial cells, displayed increasing expression across the inferred pseudotime ordering, and was enriched in malignant and mixed regions. In the TCGA-COAD cohort, high CLDN1 expression was associated with poorer overall survival and pathways involving the cell cycle, proteasome, autophagy, Hippo, ErbB and CRC. CLDN1 manipulation altered CRC-cell invasion and in vitro response to Nutlin-3a (-).Conclusion
These findings support CLDN1 as a candidate CRC-associated marker linked to malignant epithelial-cell states, tumour-associated spatial heterogeneity, invasive phenotype and Nutlin-3a (-) response. Further independent and prospective validation is required to determine its clinical diagnostic, prognostic and treatment-response relevance.
Related
- Single-cell transcriptomic and immune microenvironment analyses reveal a TIMP1-associated OLFM4⁺ tumor epithelial program linked to colorectal cancer progression and immune evasion.
- Multi-Omics Analyses Identify Candidate Genes Associated With Stage-Specific Quality Formation in Mulching-Treated Moso Bamboo Shoots.
- Integrating metabolomic and transcriptomic analyses to identify candidate genes and transcriptional regulators associated with triterpenoid saponin accumulation in developing Momordica charantia fruits
- A robustness-first cross-disease framework supports a candidate shared microbial redox axis and disease-specific metabolic divergence in colorectal cancer, Crohn’s disease, and liver cirrhosis
- Two Transcriptomic Analyses Converge on a Five-Gene Core for Colorectal Cancer Discrimination: External Validation in Independent Tissue Cohorts.
BibTeX
@article{shan2026machine,
title = {{Machine learning and multi-omics analyses reveal CLDN1 as a candidate marker associated with colorectal cancer invasion and nutlin-3a (-) response.}},
author = {Wenqi Shan and Qianyu Li and Wei Yuan and Ruiting Feng and Peiyao Hu and Zongmian Song and Yanfeng Li and Chenghan Luo and Ningjing Lei and Haozhe Zhang and Zhaorui Wang},
journal = {Translational oncology},
year = {2026},
month = oct,
doi = {10.1016/j.tranon.2026.103078},
url = {https://europepmc.org/article/MED/42868715},
note = {CC BY-NC-ND},
}
RIS
TY - JOUR
TI - Machine learning and multi-omics analyses reveal CLDN1 as a candidate marker associated with colorectal cancer invasion and nutlin-3a (-) response.
AU - Wenqi Shan
AU - Qianyu Li
AU - Wei Yuan
AU - Ruiting Feng
AU - Peiyao Hu
AU - Zongmian Song
AU - Yanfeng Li
AU - Chenghan Luo
AU - Ningjing Lei
AU - Haozhe Zhang
AU - Zhaorui Wang
PY - 2026
DA - 2026/10/13
JO - Translational oncology
DO - 10.1016/j.tranon.2026.103078
UR - https://europepmc.org/article/MED/42868715
AB - Objective Colorectal cancer (CRC) is molecularly, cellularly and spatially heterogeneous, and current staging systems and biomarkers incompletely explain prognosis and treatment response. This study integrated transcriptomics, machine learning, single-cell RNA sequencing, spatial transcriptomics and in vitro experiments to prioritize candidate CRC-associated markers and characterize their potential biological and clinical relevance.Methods GSE44861 and GSE44076 from the Gene Expression Omnibus (GEO) and The Cancer Genome Atlas Colon Adenocarcinoma (TCGA-COAD) were used for differential expression, consensus differentially expressed gene (DEG) identification and functional enrichment analyses. Four machine-learning algorithms were applied for feature selection. Candidate genes were prioritized by differential expression, model importance and receiver operating characteristic (ROC) performance. GSE231559 single-cell data and spatial transcriptomics characterized the cellular source, pseudotime dynamics and spatial distribution of the core gene. TCGA-COAD assessed prognosis, pathways and computationally predicted drug sensitivity. Western blotting, Transwell and MTT assays were used for experimental validation, and public immunohistochemistry images from the Human Protein Atlas were used as supplementary protein-expression evidence.Results We identified 544 consensus DEGs enriched in leukocyte migration, chemotaxis, extracellular matrix organization, cytokine activity and metabolic pathways. Six genes were jointly selected: CLDN1, CDH3, KRT24, MMP7, NEBL and SCGN. CLDN1 was markedly upregulated and showed the highest tumour-versus-normal discrimination among the six candidates in the TCGA-COAD cohort. Single-cell and spatial analyses showed that CLDN1 was mainly expressed in CRC epithelial cells, displayed increasing expression across the inferred pseudotime ordering, and was enriched in malignant and mixed regions. In the TCGA-COAD cohort, high CLDN1 expression was associated with poorer overall survival and pathways involving the cell cycle, proteasome, autophagy, Hippo, ErbB and CRC. CLDN1 manipulation altered CRC-cell invasion and in vitro response to Nutlin-3a (-).Conclusion These findings support CLDN1 as a candidate CRC-associated marker linked to malignant epithelial-cell states, tumour-associated spatial heterogeneity, invasive phenotype and Nutlin-3a (-) response. Further independent and prospective validation is required to determine its clinical diagnostic, prognostic and treatment-response relevance.
SN - 1936-5233
ER -