Single cell sequencing data identify distinct B cell and fibroblast populations in stricturing Crohn's disease

Abstract

Single cell RNA sequencing of human full thickness Crohn's disease (CD) small bowel resection specimens was used to identify potential therapeutic targets for stricturing (S) CD. Using an unbiased approach, 16 cell lineages were assigned within 14,539 sequenced cells from patient-matched SCD and non-stricturing (NSCD) preparations. SCD and NSCD contained identical cell types. Amongst immune cells, B cells and plasma cells were selectively increased in SCD samples. B cell subsets suggested formation of tertiary lymphoid tissue in SCD and compared with NSCD there was an increase in IgG, and a decrease in IgA plasma cells, consistent with their potential role in CD fibrosis. Two Lumican-positive fibroblast subtypes were identified and subclassified based on expression of selectively enriched genes as fibroblast clusters (C) 12 and C9. Cells within these clusters expressed the profibrotic genes Decorin (C12) and JUN (C9). C9 cells expressed ACTA2; ECM genes COL4A1, COL4A2, COL15A1, COL6A3, COL18A1 and ADAMDEC1; LAMB1 and GREM1. GO and KEGG Biological terms showed extracellular matrix and stricture organization associated with C12 and C9, and regulation of WNT pathway genes with C9. Trajectory and differential gene analysis of C12 and C9 identified four sub-clusters. Intra sub-cluster gene analysis detected 13 co-regulated gene modules that aligned along predicted pseudotime trajectories. CXCL14 and ADAMDEC1 were key markers in module 1. Our findings support further investigation of fibroblast heterogeneity and interactions with local and circulating immune cells at earlier time points in fibrosis progression. Breaking these interactions by targeting one or other population may improve therapeutic management for SCD.

Keywords: B cells; Crohn's disease; fibroblasts; fibrosis; stricturing.

FIGURE 1

Cellular profile of stricturing Crohn's disease (SCD). ScRNA‐seq was performed on cells from full‐thickness tissue surgically resected specimens from stricturing (SCD; n = 4) and non‐stricturing (NSCD; n = 4) ileum of CD patients (n = 5). Three patients provided both SCD and NSCD specimens. No prior cell selection was applied (e.g., FACs sorting) and all cells (n = 14,539) were sequenced. (A) Uniform Manifold Approximation (UMAP) identified 24 clusters assigned to 16 definable cell lineages (Figure S1A). Undifferentiated cells (Undiff1/2) contained markers of undifferentiated cells as defined by protein atlas webtool, and expressed TOP2A and MKI67 unlike any other cell type. Unknown (U) had no identifiable gene marker linked to any cell type. (B) Bubble plot shows levels of expression of key markers for different cell types identified from published data sets.

FIGURE 2

(A) Stricturing (SCD) and (B) non‐stricturing (NSCD) small bowel tissue comprise the same cell types but proportions differ markedly for B cells and plasma cells. A UMAP of NSCD and SCD preparations indicate they are comprised of the same cell types. B cells and plasma cells showed the most marked increased in frequency in SCD compared to NSCD (see Table 2).

FIGURE 3

Analysis of B cells and plasma cells in stricturing (SCD) and non‐stricturing (NSCD) small bowel tissue. (A) Proportions of antibody subclasses expressed by plasma cells in SCD and NSCD tissues. (B) Expression of CXCR4 in plasma cells. (C) Sub‐clustering of B cells identified three distinct clusters consistent with naïve (1), memory (2) and germinal centre (3) subsets.

FIGURE 4

Identification of the C12 and C9 fibroblast subtypes in stricturing and non‐stricturing small bowel tissue. UMAP of fibroblasts identifying cluster 12 (all C12, blue symbols) and cluster 9 (all C9, red symbols) from the integrated stricturing (SCD; triangles) and non‐stricturing (NSCD, dots) scRNA data set. Insert shows a comparison of the numbers of LUM+ve C12 and C9 clusters in resected small bowel from both NSCD and SCD tissue. The percentage of each cluster in either the NSCD or SCD data set is given in parenthesis. The fold change in C12 and C9 fibroblasts in NSCD and SCD is shown as SCD/ NSCD.

FIGURE 5

Workflow and stringdb network assembly, trajectory and differential gene analysis for fibroblast populations and clusters. (A) Schematic of scRNA‐seq workflow, incorporating stringdb network assembly for C9‐like fibroblasts and target validation workflow. (B) Stringdb network for C9‐like SCD fibroblasts. (C) Four clusters resolved from differentiation trajectories of the two fibroblast populations were analysed using Monocle3. (D) Identification of the most significant gene marker for each cluster using differential gene analysis. (E) Co‐regulated gene modules identified from predicted pseudotime trajectories using Monocle3. Each module contains multiple genes that have similar expression patterns between cell neighbours (i.e. intracluster gene analysis). Gene members of each gene module are listed in Figure S3.