Integrating single-cell and bulk RNA data to explore the prognostic significance of LIM domain family genes in colorectal cancer and unveil their immune landscape

Figures

• 

  Fig. 1. scRNA-seq Reveals Cellular Composition and Heterogeneity in the CRC Tumor Microenvironment A: UMAP visualization of cell clusters. B: Annotation of clusters with UMAP distribution of major cell types. C: Expression of marker genes across clusters. D: Cellular distribution by cell type. E: Bar plot depicting cell type proportions per CRC sample.

• 

  Fig. 2. Screening of LIM-related Genes Based on scRNA-seq Data A: Venn diagram of overlap between scRNA-seq genes and 47 LIM domain family genes. B: AUCell scoring distribution of all cells based on 42 genes. C: UMAP of LIM scores of all cells. D: Violin plots comparing LIM scores across cell types. E: Distribution of high/low-LIM-score cells per cell type.

• 

  Fig. 3. Screening of LIM Domain Family-related Co-expression Modules Using Weighted Gene Co-expression Network Analysis. A: Analysis of the scale-free topology fit index under different soft-thresholding powers. B: Analysis of mean connectivity under different soft-thresholding powers. C: Hierarchical clustering dendrogram of genes and module assignment based on dynamic tree cutting and module merging. D: Heatmap showing the correlations between module eigengenes and LIM domain family scores. The MEcyan module exhibited the strongest correlation with the LIM score and was selected for subsequent analysis.

• 

  Fig. 4. Construction of the LIM Domain Family-related Prognostic Signature. A: Venn diagram showing the intersection between differentially expressed genes identified from high- and low-LIM-score cells and genes from the key WGCNA module. B: LASSO coefficient profiles of the candidate prognostic genes. C: Ten-fold cross-validation for selection of the optimal penalty parameter in the LASSO Cox regression model. D: Forest plot showing the hazard ratios and 95% confidence intervals of the nine genes included in the final multivariate Cox model.

• 

  Fig. 5. Construction and Validation of the LIM Domain Family-related Risk Score. A–C: Kaplan–Meier curves comparing overall survival between the high- and low-LIMRS groups in the TCGA-CRC training cohort (A), GSE29261 validation cohort (B), and GSE17536 validation cohort (C). D–F: Time-dependent ROC curves evaluating the 1-, 3-, and 5-year predictive performance of LIMRS in the TCGA-CRC cohort (D), GSE29261 cohort (E), and GSE17536 cohort (F). G–H: Univariate (G) and multivariate (H) Cox regression analyses evaluating the prognostic independence of LIMRS after adjustment for tumor purity and cancer-associated fibroblast scores. I: Kaplan–Meier analyses showing the prognostic stratification ability of LIMRS across different consensus molecular subtypes.

• 

  Fig. 6. Expression Patterns of Model Genes and Pseudotime Trajectory Analysis Based on scRNA-seq. A-B: UMAP plot (A) and violin plot (B) displaying the expression patterns of model genes across different cell types. C: Clustering results of fibroblast subpopulations. d-E: UMAP plot (D) and violin plot (E) visualizing the expression patterns of model genes in fibroblast subclusters. F: Differentiation trajectory analysis, with colors representing pseudotime and distinct cell clusters. G: Relative expression profiles of the nine model genes along the differentiation trajectory.

• 

  Fig. 7. Construction and Validation of the Nomogram. A: Nomogram predicting 1-, 3-, and 5-year OS for CRC patients in the TCGA dataset. B: Calibration plots for the predicted 1-, 3-, and 5-year OS of CRC patients. C: Time-dependent ROC curves comparing the nomogram with other clinical factors. D: DCA of the nomogram.

• 

  Fig. 8. Immune Landscape Analysis Between the High- and Low-LIMRS Groups. A: Expression differences in immune checkpoint-related genes between the two LIMRS groups. B: Expression differences in HLA-related genes between the two LIMRS groups. C: Volcano plot showing differences in immune cell infiltration and immune-related functional scores. Positive effect values indicate relatively higher scores in the high-LIMRS group, whereas negative values indicate relatively higher scores in the low-LIMRS group. Only pDCs and the Type II IFN Response remained statistically significant after FDR correction. D: Comparison of TIDE scores between the high- and low-LIMRS groups. E: Comparison of IPS between the high- and low-LIMRS groups. A higher IPS indicates stronger tumor immunogenicity and a potentially higher likelihood of response to immune checkpoint inhibitor therapy. ns, not significant; *p < 0.05; p < 0.01.

• 

  Fig. 9. Functional Enrichment Analysis. A: Volcano plot of DEGs between high- and low-LIMRS groups. B: GO terms. C: KEGG pathways.

• 

  Fig. 10. Small-molecule Drug Sensitivity Across LIMRS Subgroups. (****: p < 0.0001; **: p < 0.01; *: p < 0.05). The Imputed Sensitivity Score refers to an IC50-associated drug response score predicted from GDSC datasets. Lower scores correspond to higher predicted drug sensitivity, whereas higher scores indicate stronger predicted drug resistance.

• 

  Fig. 11. Effects of RGS16 Knockdown on the Biological Behavior and Erlotinib Response of HCT116 Cells. A: Computational molecular docking prediction showing a potential interaction conformation between RGS16 and Erlotinib. B: RGS16 knockdown efficiency in HCT116 cells evaluated by qRT-PCR. C: RGS16 protein expression and knockdown efficiency evaluated by WB. D: Colony formation assay evaluating the proliferative capacity of HCT116 cells in the indicated treatment groups. E: Annexin V-FITC/PI flow cytometry evaluating apoptosis in the indicated treatment groups. F: WB analysis of p-EGFR, total EGFR, p-ERK1/2, and total ERK1/2 in HCT116 cells following RGS16 knockdown and/or Erlotinib treatment. ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001; **p < 0.0001.