Integration analysis of single-cell and spatial transcriptomics identifies prognostic genes associated with neddylation in colorectal cancer.

[BACKGROUND] Neddylation modifications in immune and tumor cells are linked to poor tumor prognosis. This study identifies prognostic genes associated with neddylation-related genes (NRGs) in colorectal cancer (CRC) using single-cell and spatial transcriptome (ST) sequencing, aiming to advance CRC treatment strategies.

[METHODS] Datasets included TCGA-CRC (training/internal validation, 7:3 split), GSE28722 (external validation), GSE132257 (scRNA-seq), and GSE226997 (ST). Single-cell analysis annotated seven cell types, with epithelial cells identified as key. Differentially expressed genes (DEGs) from key cells [DEGs(sc)] and bulk analysis of TCGA-CRC [DEGs(bulk)] were intersected with 247 NRGs to yield candidate genes. Regression analyses screened prognostic genes for risk model construction, validated internally and externally. Pseudotime trajectory and ST mapping visualized gene expression, while molecular networks and drug predictions were generated.

[RESULTS] In scRNA-seq dataset, seven cell types were annotated, and epithelial cells were the key cells. A sum of 32 candidate genes were obtained by intersecting 5,131 DEGs(sc)(key cells), 9,089 DEGs(bulk), and 247 NRGs to produce PSMD12, PSMB2, and FBXL5 as prognostic genes. Both prognostic risk model and nomogram model were predictive of CRC. At the ST samples, PSMD12 was expressed at a low level in all sections, whereas PSMB2 and FBXL5 were expressed at a slightly higher level in the sections. In addition, a lncRNA-miRNA-mRNA network and a drug-prognostic gene network were created, getting some potential drugs like bortezomib.

[CONCLUSION] A novel three-gene prognostic model for CRC was developed and validated, offering therapeutic insights through molecular networks and drug predictions.