The spatial transcriptomic landscape of the healing mouse intestine following damage

Abstract

The intestinal barrier is composed of a complex cell network defining highly compartmentalized and specialized structures. Here, we use spatial transcriptomics to define how the transcriptomic landscape is spatially organized in the steady state and healing murine colon. At steady state conditions, we demonstrate a previously unappreciated molecular regionalization of the colon, which dramatically changes during mucosal healing. Here, we identified spatially-organized transcriptional programs defining compartmentalized mucosal healing, and regions with dominant wired pathways. Furthermore, we showed that decreased p53 activation defined areas with increased presence of proliferating epithelial stem cells. Finally, we mapped transcriptomics modules associated with human diseases demonstrating the translational potential of our dataset. Overall, we provide a publicly available resource defining principles of transcriptomic regionalization of the colon during mucosal healing and a framework to develop and progress further hypotheses.

Fig. 1. Spatial transcriptomics reveals molecular regionalization of the murine colonic tissue in steady state.

a Schematics of the experiment: colonic tissue from a naive WT mouse (d0) was processed as a Swiss roll for spatial transcriptomic (ST) with Visium 10X technology (n = 1 Swiss roll). b Colon Swiss rolls shown in hematoxylin and eosin (H&E) staining (left) and with each ST spot color coded based on non-negative matrix factorization (NNMF) (right). ST spots belonging uniquely to one factor are colored in red, blue and green for NNMF1, 2 and 3 respectively. ST spots shared between different factors are colored with respective intermediate gradation of these 3 colors. c Top: spatial distribution of the 3 factors distinguishing muscle, lamina propria (LP) and intestinal epithelial cells (IEC). Bottom: heatmap showing the top 20 genes defining each factor. d Immunohistochemical staining of CDH17, TAGLN, ADH1 in healthy human colonic tissue (from Human Protein Atlas). e Digitally unrolled colonic tissue, showing the distribution of the 3 factors from Fig. 1c along the serosa-luminal and proximal–distal axis. f Proximal to distal distribution of Hmgcs2, Ang4 and B4galt1 expression in colonic swiss rolls (left) and digitally unrolled colon (right). g qPCR validation of regional expression of Hmgcs2, Ang4 and B4galt1 in proximal, mid and distal colonic biopsies from WT mice (n = 3, each dot represents one mouse). Data are presented as mean values ± SEM. Significance was assessed by one-way ANOVA with Bonferroni post-test. *p < 0.05; **p < 0.01. h Spatial distribution of 9 out of 20 factors in the naive colon displaying transcriptional regionalization along the serosa-luminal and proximal-distal axis. Each ST spot is assigned a color-coded score based on the expression of the genes defining each factor. i Schematic representation of the colon (top) and top genes annotated in Factors. Factors are grouped based on their proximal-distal distribution and color-coded (i.e. gray-pink-purple) based on their serosa-luminal distribution.

Fig. 2. Identification and regional distribution of lymphoid follicle, B cell-associated, and enteric nervous system signatures in the naive murine colon.

a Heatmap of the top genes defining factors 1n, 3n and 9n enriched in B cell signature. b Spatial distribution of B cell-associated factors in the naive colon. c Immunohistochemical staining of CLU (enriched in lymphoid follicles) and JCHAIN (localized in the lamina propria) in healthy human colonic tissue (from Human Protein Atlas). d Functional enrichment analysis (Gene Ontology, GO) of factors 1n, 3n and 9n. e Top: spatial distribution (left) and top genes (right) defining factor 6n (enteric nervous system, ENS). Bottom: immunohistochemical staining of UCHL1 (neuronal marker) in the colonic submucosa of healthy human colonic tissue (from Human Protein Atlas). f Pathway analysis (GO) of factor 6n. g Schematic representation of spatial distribution of B cell factors (i.e. 1n, 3n and 9n from Panels A-D) and ENS-factor 6n (from Panel E-F).

Fig. 3. Changes of the molecular topography during mucosal healing are dominant at the distal colon.

a Schematic representation of the experiment: colitis was induced by dextran sodium sulfate (DSS) administration in drinking water for 7 days followed by 7 days of regular water to promote tissue repair. Colonic tissue from a wild-type naive mouse (d0, from Fig. 1) and from a mouse undergoing colonic regeneration (d14) were processed as Swiss roll for spatial transcriptomic using Visium 10X technology. (n = 1 Swiss roll per time point). b Top: Hematoxylin and eosin staining of colonic tissue from d0 and d14. Bottom: spatial representation of UMAP values in CMYK colors on colon d0 and d14. Spots with the same color in the two time points represent transcriptionally similar regions. c Uniform Manifold Approximation and Projection (UMAP) representation of 16 color-coded clusters defining regional transcriptome diversity in the colonic d0 and d14 datasets combined. d Heatmap showing expression of top genes defining each cluster (color-coding on top) in the ST datasets from the two timepoints (light blue columns: colon d0; pink columns: colon d14). e Schematic representation of cluster 0 and cluster 12 distribution in colon d0 (on the left) and d14 (on the right). f Expression of selected genes in cluster 0 onto ST. g qPCR validation of regional expression of Reg3b in proximal, mid and distal colonic biopsies from wild type mice at steady state conditions (d0) and during mucosal healing (d14)(n = 3, each dot represents one mouse). Data are presented as mean values ± SEM. Significance was assessed by one-way ANOVA with Bonferroni post-test. **p < 0.01; ****p < 0.0001.

Fig. 4. Non-negative matrix factorization reveals eight distinct molecular patterns during colon mucosal healing.

a UMAP representation of 16 clusters in d0 and d14 colon. b Hematoxylin and eosin images displaying overlaid spots with the highest factor weight. c Top 10 genes defining the indicated NNMFs (factors). d Functional enrichment analysis (Gene Ontology, GO) based on the top genes defining factor 5, 7, 14 and 20. e Schematic representation summarizing the expression pattern between selected factors. Biological processes associated with each factor are indicated in brackets.

Fig. 5. Predictive algorithm reveals pathway-specific spatial patterns during mucosal healing.

a Correlation matrix between non-negative matrix factorization (NNMF) and pathway activity scores determined by PROGENy. b Schematic of the colon area at d14 displaying the distribution of the indicated factors. c Spatial transcriptomic (ST) spot heatmaps of the colon at d0 (upper swiss rolls) and d14 (lower swiss rolls) showing pathways scores predicted by PROGENy. Arrows indicate areas of tissue damaged as defined by factor 14 shown in Supplementary Fig. 7 and Fig. 4a–c. d ST spot heatmaps showing TNFα, NFkB, JAK-STAT pathway activity on d0 (upper Swiss rolls) and d14 (lower Swiss rolls). Selected areas indicated as ILF (isolated lymphoid follicles) or “i” and “ii” and outlined in black are magnified below each Swiss roll. Arrows in “ii” indicate the presence of an ILF. e Spatial distribution of androgen and estrogen pathway activity at d14. Selected areas (indicated as “i” and “ii”) are magnified. Arrows indicate an example of the muscle layer showing opposite expression patterns between the two pathways. f Spatial distribution of p53 pathway activity at d0 and d14. Selected areas indicated as “i” and “ii” on colon d14 are magnified on the right. Hematoxylin and eosin magnifications show the overlaid spots with the lowest p53 activity shown in “i” and “ii”. g Left: UMAP visualization of intestinal epithelial cells (IEC) clusters from scRNAseq on colon d14 (GSE163638). Middle: ST spots from colon d14 are color-coded based on the enrichment of stem cell core signature identified from scRNAseq dataset. Selected areas indicated as “i” and “ii” on colon d14 are magnified on the right. Right: Hematoxylin and eosin magnifications showing the overlaid spots with the highest stem cell signature shown in “i” and “ii”. h Pearson correlation between PROGENy predicted pathways and stem cell signature on the ST dataset.

Fig. 6. Human cell type mapping onto murine spatial transcriptomic datasets.

a Scheme showing the integration of published human single cell RNAseq and our mouse Visium datasets. b Correlation matrix between transcriptomic profiles from human single cell datasets and factors defining transcriptomics patterns in mouse ST. c Integration of human stromal cell transcriptomic profiles (S4.CCL21+ and S4.CXCL13 + ) onto visium datasets at day 0. d Integration of human intestinal cell transcriptomic profiles onto visium datasets at day 14. e Integration of human immune cell transcriptomic profiles onto visium datasets at day 14 and magnification of the isolated lymphoid follicle area.

Fig. 7. Spatial transcriptomic (ST) allows mapping transcriptomic signatures with clinical relevance.

a Scheme illustrating the analysis designed and dataset sources. b Correlation matrix between transcriptomic modules distinguishing the processes of inflammation and mucosal healing during DSS-induced colitis and factors defining transcriptomics patterns in ST. c (i) Venn diagram and spatial representation of overlapping genes between module 1 and factor 9, (ii) relative mean expression and Gene Ontology (GO) of genes belonging to module 1. d (i) Venn diagram and spatial representation of overlapping genes between module 6 and factor 15, (ii) relative mean expression and Gene Ontology (GO) of genes belonging to module 6. ENS: enteric nervous system. e Spatial distribution of genes defining UC1 and UC2 patients on mouse ST colon d0 and d14. UC: Ulcerative Colitis. f Spatial distribution of Inflammatory Bowel Disease (IBD) risk genes from Cluster 3 (Supplementary Fig. 10b) on colon d0 (left) and d14 (right). g Gene Set Enrichment Analysis for IBD risk genes in Cluster 3 (Supplementary Fig. 10b) and NNMF factors (Supplementary Figs. 6 and 7).