Gremlin 1 depletion in vivo causes severe enteropathy and bone marrow failure

Abstract

The intestinal epithelium is perpetually renewed from a stem cell niche in the base of crypts to maintain a healthy bowel mucosa. Exit from this niche and maturation of epithelial cells requires tightly controlled gradients in BMP signalling, progressing from low BMP signalling at the crypt base to high signalling at the luminal surface. The BMP antagonist gremlin 1 (Grem1) is highly expressed by subepithelial myofibroblasts adjacent to the intestinal crypts but its role in regulating the stem cell niche and epithelial renewal in vivo has not been explored. To explore the effects of Grem1 loss in adulthood following normal growth and development, we bred mice (ROSA26CreER-Grem1 ^flx/flx ) in which Grem1 could be deleted by tamoxifen administration. While Grem1 remained intact, these mice were healthy, grew normally, and reproduced successfully. Following Grem1 depletion, the mice became unwell and were euthanised (at 7-13 days). Post-mortem examination revealed extensive mucosal abnormalities throughout the small and large intestines with failure of epithelial cell replication and maturation, villous atrophy, and features of malabsorption. Bone marrow hypoplasia was also observed with associated early haematopoietic failure. These results demonstrate an essential homeostatic role for gremlin 1 in maintaining normal bowel epithelial function in adulthood, suggesting that abnormalities in gremlin 1 expression can contribute to enteropathies. We also identified a previously unsuspected requirement for gremlin 1 in normal haematopoiesis. © 2020 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of Pathological Society of Great Britain and Ireland.

Figure 1

Ubiquitous Grem1 deletion in adulthood is lethal and the effects manifest acutely in the bowel. (A) Representative image of the normal intestinal tract of a gremlin 1 (Grem1) intact (Grem1 ^intact) mouse and an image of an abnormal intestine from a Grem1 depleted mouse (Grem1 ^depl) following exposure to tamoxifen. It shows thin transparent intestinal walls distended by clear pale yellow gelatinous material and a loss of mesenteric fat. (B) Kaplan–Meier survival curves of ROSA26CreER‐Grem1 ^fl/fl mice on a normal diet (Grem1 ^intact, n = 20) or a tamoxifen diet (Grem1 ^depl, n = 20, p < 0.001, log‐rank test). Numbers on the graph indicate the number of mice remaining at the end of each day of the tamoxifen diet. (C) Grem1 mRNA assayed by RT‐qPCR in small intestines of Grem1 intact (Grem1 ^intact, n = 8) and Grem1 depleted (Grem1 ^depl, n = 8) mice. (D) Gremlin 2 (Grem2) mRNA assayed by RT‐qPCR in small intestines of Grem1 intact (Grem1 ^intact, n = 8) and Grem1 depleted (Grem1 ^depl, n = 8) mice. (E) Noggin (Nog) mRNA assayed by RT‐qPCR in small intestines of Grem1 intact (Grem1 ^intact, n = 8) and Grem1 depleted (Grem1 ^depl, n = 8) mice.

Figure 2

Attenuated Grem1 expression was associated with epithelial cell abnormalities, villus stunting, and loss of crypts in the small intestines of Grem1 ^depl mice. (A–P) Representative images of the stomach (first row) and small intestine (rows 2–4) of gremlin 1 intact (Grem1 ^intact ) (first and second columns) and gremlin 1 depleted (Grem1 ^depl) mice (third and fourth columns). (A–D) The stomachs of both Grem1 ^intact and Grem1 ^depl mice were normal. (E–P) The intestine of Grem1 ^depl mice showed marked villus stunting and multifocal collapse and loss of crypts in the duodenum (E–H), jejunum (I–L), and ileum (M–P) in comparison to Grem1 ^intact controls. The mucosal epithelial lining of the intestine consisted of disordered, large polygonal cells. (B, F, J, N) In situ hybridisation (ISH) revealed Grem1 mRNA expression as punctate brown labelling outside the base of the crypts and along the muscularis mucosa of the stomach (B) and intestine (F, J, N) of Grem1 ^intact mice. Similar but less intense labelling was shown in the submucosa and the tunica muscularis. (D, H, L, P) In contrast, no Grem1 mRNA labelling was found in the stomach (D) or intestine (H, L, P) of Grem1 ^depl mice. First and third columns, H&E staining; second and fourth columns, ISH for Grem1 mRNA, counterstained with Mayer’s haematoxylin. Scale bar = 50 μm (20× objective, numerical aperture 0.75).

Figure 3

Grem1 depletion induced widespread changes in the large intestinal epithelium, with superficial, abnormal crypts that were reduced in number. (A–H) Representative images of the caecum (first row) and colon (second row) of gremlin 1 intact (Grem1 ^intact) mice (first and second columns) and gremlin 1 depleted (Grem1 ^depl ) mice (third and fourth columns) illustrate the simple columnar epithelial cells and goblet cells in Grem1 ^intact mice which were replaced by large polygonal cells with abundant eosinophilic cytoplasm and a centrally located large round‐to‐oval nucleus in Grem1 ^depl mice. (A, C, E, G) Crypts were pushed to the surface or lost in the caecum and colon of Grem1 ^depl mice in comparison to the caecum and colon of Grem1 ^intact mice. (B, F) In situ hybridisation demonstrated Grem1 mRNA expression as punctate brown labelling at the base of the crypts, along the muscularis mucosa, with occasional brown dots in the submucosa and the tunica muscularis of the caecum and colon of Grem1 ^intact mice. (D, H) In contrast, no Grem1 mRNA labelling was found in the caecum or colon of Grem1 ^depl mice. First and third columns, H&E staining, scale bar = 25 μm (40× objective, numerical aperture 0.95); second and fourth columns, ISH for Grem1 mRNA, counterstained with Mayer's haematoxylin, scale bar = 50 μm (20× objective, numerical aperture 0.75).

Figure 4

Grem1 is essential for normal bone marrow function. (A) Image of the normal bone marrow of a Grem1 intact (Grem1 ^intact) mouse (left) showing marked cellular reduction, with replacement of the marrow by large blood‐filled sinuses in a Grem1 depleted (Grem1 ^depl) mouse (right). H&E, scale bar = 25 μm (40× objective, numerical aperture 0.95). (B) Total number of cells obtained from the long bones of the legs of Grem1 ^intact (n = 6) and Grem1 ^depl (n = 8) mice. (C) Grem1 mRNA expression in bone marrow‐derived macrophages of Grem1 ^intact (n = 6) and Grem1 ^depl (n = 6) mice. (D) Peripheral blood reticulocyte count in Grem1 ^intact (n = 6) and Grem1 ^depl (n = 6) mice. Median (IQR) relative to mean value in the Grem1 ^intact group. Statistical comparisons were made using a Mann–Whitney U‐test.