Epithelial GREMLIN1 disrupts intestinal epithelial-mesenchymal crosstalk to induce a wnt-dependent ectopic stem cell niche through stromal remodelling

Abstract

In homeostasis, counterbalanced morphogen signalling gradients along the vertical axis of the intestinal mucosa regulate the fate and function of epithelial and stromal cell compartments. Here, we use a disease-positioned mouse and human tissue to explore the consequences of pathological BMP signalling dysregulation on epithelial-mesenchymal interaction. Aberrant pan-epithelial expression of the secreted BMP antagonist Grem1 results in ectopic crypt formation, with lineage tracing demonstrating the presence of Lgr5(-) stem/progenitor cells. Isolated epithelial cell Grem1 expression has no effect on individual cell fate, indicating an intercompartmental impact of mucosal-wide BMP antagonism. Treatment with an anti-Grem1 antibody abrogates the polyposis phenotype, and triangulation of specific pathway inhibitors defines a pathological sequence of events, with Wnt-ligand-dependent ectopic stem cell niches forming through stromal remodelling following BMP disruption. These data support an emerging co-evolutionary model of intestinal cell compartmentalisation based on bidirectional regulation of epithelial-mesenchymal cell fate and function.

Fig. 1. Ectopic crypt lineage tracing and secretory cell fate.

A Fluorescent co-ISH staining of Sox9 and Fgfbp1 in 120 day Vil1-Grem1 polyp shows co-localised staining both in normal crypt base (green box) and ectopic crypt (red box) (n = 5 mice). B Schematic shows recombination and harvesting timepoint for Sox9-CreER^T2; Rosa26^YFP;Vil1-Grem1 mouse model. C Tissue serial sectioning and lineage tracing ribbon identification with anti-YFP immunohistochemistry (brown), followed by tissue alignment and polyp 3D reconstruction of lineage tracing ribbons (brown) (n = 1 whole mouse SI). D H&E, Grem1 and Sox9 staining of Villin-CreER^T2; Rosa26^Grem1 early and advanced polyps (n = 5). E Secretory cell staining in small intestine and colon 4 days post-recombination in Atoh1-CreER^T2,Rosa26^tdTom or Atoh1-CreER^T2; Rosa26^Grem1 mice stained with mCherry IHC or Grem1 ISH respectively. F Number of mCherry IHC (blue lines) or Grem1 ISH (red lines) stained small intestinal or colonic epithelial cells over time following recombination (n = 5 mice per group). Data were ±s.e.m. G Quantification of number of fully lineage traced colonic crypts across gut roll sections at day 30 post recombination in Atoh1-CreER^T2,Rosa26^tdTom or Atoh1-CreER^T2; Rosa26^Grem1 mice stained with mCherry IHC or Grem1 ISH respectively (n = 5 mice per genotype, two-tailed t test, p value = 0.0013). Data were ±s.e.m. Scale bar 200 µm throughout. Source data are provided as a Source Data file.

Fig. 2. Phenotype reversal through UCB Ab7326 treatment.

A Schematic shows treatment schedule and harvesting timepoints of Vil1-Grem1 mice treated with UCB Ab7326. B Kaplan–Meier Survival Curve showing impact on survival of UCB Ab7326 initiated at weaning, (green line, n = 11, P = 1.5E-06) or from 120 days (yellow line, n = 7, p = 0.00421) in comparison to untreated animals (n = 13, red line). Significance calculated using log-rank tests, with the Benjamini–Hochberg correction for multiple testing. C Schematic shows treatment schedule and harvesting timepoints of Vil1-Grem1 mice treated with UCB Ab7326 from the age of 120 days. D Representative macroscopic and histological images of proximal small intestinal sections harvested from untreated and UCB Ab7326-treated Vil1-Grem1 mice over a range of timepoints (n = 6, scale bars 0.25 cm). E Quantification of Vil1-Grem1 small bowel and colonic surface area, polyp number and polyp area following variable time treatment with UCB Ab7326 antibody (n = 6 mice per group, one way ANOVA with Dunnett post-hoc corrections, p values as stated). Data were ±s.e.m. F Immunohistochemical analysis to show reversion of Vil1-Grem1 ectopic crypt phenotype and restoration of normal crypt-villus staining patterns of Ki67, Sox9, EphB2, Ck20 and Lysozyme following 10 weeks UCB Ab7326 therapy (n = 5 mice). Scale bar 200 µm. Source data are provided as a Source Data file.

Fig. 3. Single-cell RNA-sequencing of proximal small bowel crypts and villi reveal an expansion of an Lgr5(−) intestinal stem/progenitor population that is reversed after UCB Ab7326 treatment.

A T-SNE plot of Epcam-expressing epithelial cells in the proximal small bowel of wild-type, Vil1-Grem1, and UCB Ab7326-treated Vil1-Grem1 mice, classified by cell type, with accompanying subplots separated by mouse genotype/treatment and tissue compartment B The number of cells of each epithelial cell type, as well as the corresponding percentage of total epithelial cells of each cell type for each sample. The number of stem progenitor cells in either the crypt or villi/polyps show a substantial expansion of the stem progenitor cell population in both the crypt and villi of the Vil1-Grem1 mice, which is reversed after subjecting the Vil1-Grem1 mice to 10 weeks of treatment with UCB Ab7326. C The expanded stem progenitor cell population in the Vil1;Grem1 mice is characterised by increased expression of Mki67, Sox9, Olfm4 and Fgfbp1.

Fig. 4. Aberrant pan-epithelial Grem1 expression causes intercompartmental remodelling with de-repression of stem cell supporting fibroblast cells.

A p-Smad1, 5 immunohistochemistry shows physiological villus epithelial cell nuclear staining from the crypt isthmus upwards (brown stain). Pan-epithelial expression loss is seen in Vil1-Grem1 mice, with little reversion of epithelial stain seen following 10 weeks UCB Ab7326 therapy. Automated stain quantification using digital pathology platform used to exclude stromal cell staining (QuPath) (n = 3 per group, >15 villi or polyps total). B p-Smad1,5 immunohistochemistry shows physiological stromal cell staining in the villus stomal compartment (brown stain). Loss of expression is seen in untreated Vil1-Grem1 animals as a consequence of intercompartmental BMP antagonism. Treatment with UCB Ab7326 recovers stromal expression levels of p-Smad1,5 in treated Vil1-Grem1 mice. Automated stain quantification using digital pathology platform used to exclude epithelial cell staining (QuPath, n = 3 per group, >15 villi or polyps total). C Hi-plex in situ hybridization (12 markers) was used to identify and topographically map functionally distinct fibroblast subtypes (SEMF, Cd55(+) Wnt2b(+), and Trophocytes) in wildtype and Vil1-Grem1 mouse tissue. Image analysis software was used to identify each subtype and map cell location back onto the tissue based on density distribution (coloured circles, within shaded areas - HALO, Indica Labs) (n = 5 mice per group, ≥10 villi, crypts or polyps total, one way ANOVA with Dunnett post-hoc corrections, p values as stated). Data were ±s.e.m. Scale bars 200 µm. Source data are provided as a Source Data file.

Fig. 5. Formation of the ectopic niche requires ligand-dependent Wnt signalling.

A Representative images of in situ hybridisation of Wnt ligands (Wnt2b, Wnt5a), receptors (Fzd5, Lgr4) and target gene (Axin2) in Vil1-Grem1 versus wild-type small intestine (n = 5 mice). Scale bar 200 µm B Schematic shows treatment and harvesting timepoints of Vil1-Grem1 animals treated with the porcupine inhibitor, LGK-974. C Digital pathology-based quantification of polyp number per section, polyp area, ectopic crypts and villus Ki67+ cell number per gut roll following 2 weeks of porcupine inhibition (n = 5 mice per group, two-tailed t test, p values as stated). Data were ±s.e.m. D Representative H&E images of untreated and LGK-974-treated Vil1-Grem1 mouse small intestine (10 weeks) showing loss of polyps and ectopic crypts but ongoing villus widening/deformity in treated animals. Scale bar 0.25 cm for gut rolls E Quantification of Cd55(+) Wnt2b(+) cell proportion above the crypt isthmus in wildtype and Vil1-Grem1 animals including after UCB Ab7326 and LGK-974 therapy (n = 5 mice, ≥15 villi or polyps total, one way ANOVA with Dunnett post-hoc corrections, p values as stated). Data were ±s.e.m. F Quantification of wnt target, receptor and target stained cell proportion above the crypt isthmus in wildtype and Vil1-Grem1 animals including after UCB Ab7326 and LGK-974 therapy (n = 5 mice per group, one-way ANOVA with Dunnett post-hoc corrections, p values as stated). Data were ±s.e.m. Source data are provided as a Source Data file.

Fig. 6. Investigating ectopic niche and advanced polyp signalling.

A Schematic shows recombination and harvesting timepoint for acute Lgr4 knockout in 120-day-old Rosa-CreER^T2, Lgr4^fl/fl; Vil1-Grem1 mice. B Ki67 staining and C quantification of proliferating crypt base and ectopic crypt cells following acute Lgr4 knockout (n = 15 crypts and n = 35 polyps total per group with n = 5 mice per group, t test, P < 0.001). Data were ±s.e.m. D Change in wnt target and receptor staining in advanced Vil1-Grem1 polyps (n = 11 advanced polyps). E Pie chart shows identified mutations in Ctnnb1, detected by Sanger sequencing from micro-dissected advanced (Lgr5(+)) polyps in Vil1-Grem1 animals (n = 11 advanced polyps). ISH confirms epithelial staining of RSpo3 in a single polyp with an identified Ptprk-Rspo3 fusion mutation. F Representative fluorescent co-ISH staining images, and G quantification of CD55(+) WNT2B(+) cell proportions in the stroma of small sporadic tubulovillous adenoma (TVA) (n = 6), Hereditary Mixed Polyposis Syndrome (HMPS) (n = 5) and sporadic traditional serrated adenoma (TSA) (n = 6) lesions, one way ANOVA with Dunnett post-hoc corrections, (*P  =  0.05). Data were ±s.e.m. Scale bars 200 µm. Source data are provided as a Source Data file.

Fig. 7. Human polyp stromal remodelling and model summary.

Model summary depicting the disruption of BMP signalling gradients leading to a remodelling of the stromal microenvironment and establishment of wnt ligand-dependent ectopic crypt stem cell niches.