Novel Transcriptomic Signatures in Fibrostenotic Crohn's Disease: Dysregulated Pathways, Promising Biomarkers, and Putative Therapeutic Targets

Abstract

Background: Fibrosis is a common complication in Crohn's disease (CD), often leading to intestinal strictures. This study aims to explore the transcriptomic signature of fibrostenotic ileal CD for a comprehensive characterization of biological and cellular mechanisms underlying intestinal fibrosis.

Methods: Nine CD patients undergoing surgery for fibrotic ileal strictures were prospectively recruited. RNA was extracted from fresh resected samples for bulk transcriptomics. Differentially expressed genes (DEGs) were identified (adj. P value < .05), and machine learning analyses were employed to compare gene expression patterns between strictures and non-strictured margins. Pathway enrichment analysis pinpointed relevant pathways. Furthermore, a random forest model was constructed to evaluate the significance of targeted genes. Relevant genes were subsequently validated through qPCR and further analyzed using a publicly available bulk RNA-seq dataset (GSE192786). Single-cell RNA sequencing (scRNA-seq) analysis was performed using the 10× Chromium Controller platform.

Results: Bulk transcriptomics revealed unique transcriptomes with 81 DEGs, 64 significantly up-regulated, and 17 down-regulated in strictures compared to non-strictured margins. Up-regulated genes were mainly associated with inflammation, matrix and tissue remodeling, adipogenesis and cellular stress, while down-regulated genes were linked to epithelial barrier integrity. LY96, AKAP11, SRM, GREM1, EHD2, SERPINE1, HDAC1, and FGF2 showed high specificity for strictures. scRNA-seq linked up-regulated GREM1 exclusively to fibroblasts, while EHD2 and FGF2 showed upregulation in both fibroblasts and endothelial cells. LY96 and SRM were expressed by immune cells, whereas HDAC1, AKAP11, and SERPINE1 showed low expression across all cellular subsets.

Conclusions: This study comprehensively characterizes resected CD ileal strictures, elucidating main dysregulated pathways and identifying promising biomarkers and putative therapeutic targets.

Keywords: fibroblast; immune cell; intestinal fibrosis; machine learning; single-cell transcriptomic.

Graphical Abstract

Figure 1.

Differentially expressed genes (DEGs) analysis between stricture, non-strictured proximal margin, and non-strictured distal margin. (A-C) Principal component analysis (PCA) score plots performed on the differentially expressed transcriptome datasets demonstrating clustering and variation of subjects within cohorts, that is, stricture versus proximal margin (A), stricture versus distal margin (B), and proximal versus distal margin (C). The dots represent samples and are colored according to the subject cohort. Ellipses represent 95% confidence. Results are plotted according to the PC1 and PC2 scores, with the percent variation explained by the respective axis. D, Principal component analysis score plot for combined analysis of stricture, proximal and distal samples. E, Partial least squares regression score plot for combined analysis of stricture, proximal, and distal samples. F, Overlap of DEGs in stricture versus proximal margin, stricture versus distal margin, and proximal versus distal margin. G, Up and down regulation of 81 DEGs between stricture versus proximal margin and stricture versus distal margin.

Figure 2.

Pathway enrichment analysis of differentially expressed genes (DEGs). A, Heat-map showing differences of expression in ileal stricture compared to non-strictured proximal margin of the 81 previously identified DEGs in our CD population. B, Pathways enriched in the 81 DEGs and their associated genes.

Figure 3.

Genes associated with strictures. A, Box plots showing up- and down-regulated genes of normalized selected genes between stricture, non-strictured proximal margin, and non-strictured distal margin. B, Heatmap of selected gene expression across stricture [S], non-strictured proximal margin [P], and non-strictured distal margin [D]. C, Individual and combined AUC of selected genes from RNAseq analysis and their corresponding confidence interval. *P < .05; **P < .01; ns, not significant.

Figure 4.

Estimated sample size and effect size of genes selected through bulk RNA sequencing. GREM1, LY96, and SERPINE1 exhibited a power of at least 75% to diagnose strictures within a sample size of 20 patients.

Figure 5.

RT-qPCR analysis of genes differentiating stricture. A, Histograms showing up- and down-regulated genes of normalized selected genes between stricture, non-strictured proximal margin, and non-strictured distal margin. B, Individual and combined area under the curve (AUC) of selected genes from qPCR analysis and their corresponding confidence interval. **P < .01

Figure 6.

Single cell RNA sequencing of fibrostenotic CD. 10× scRNA-seq was performed on total cell populations (31.195 cells) isolated from stricture and non-strictured proximal margin of resected bowel of 3 CD patients. Single cells from all patients and locations were pooled and clustered using the UMAP_1 versus UMAP_2 parameters. A, Clusters were identified based on their expression of lineage-defining marker genes. B, UMAP_1 versus UMAP_2 expression plots for genes discriminating strictures and identified through bulk-RNAseq. R1 region showed clusters closely apposed in the UMAP_1 versus UMAP2 plot and containing cells with markers resembling fibroblasts, epithelial cells, endothelial cells, and dendritic cells. Arrows show cell clusters where the target gene is mainly expressed. C, Sub-clustering of R1 region. D, UMAP_1 versus UMAP_2 expression plots for genes associated with strictures, identified through bulk-RNAseq and expressed within R1 sub-clustered map.