Cdx2 determines the fate of postnatal intestinal endoderm

Abstract

Knock out of intestinal Cdx2 produces different effects depending upon the developmental stage at which this occurs. Early in development it produces histologically ordered stomach mucosa in the midgut. Conditional inactivation of Cdx2 in adult intestinal epithelium, as well as specifically in the Lgr5-positive stem cells, of adult mice allows long-term survival of the animals but fails to produce this phenotype. Instead, the endodermal cells exhibit cell-autonomous expression of gastric genes in an intestinal setting that is not accompanied by mesodermal expression of Barx1, which is necessary for gastric morphogenesis. Cdx2-negative endodermal cells also fail to express Sox2, a marker of gastric morphogenesis. Maturation of the stem cell niche thus appears to be associated with loss of ability to express positional information cues that are required for normal stomach development. Cdx2-negative intestinal crypts produce subsurface cystic vesicles, whereas untargeted crypts hypertrophy to later replace the surface epithelium. These observations are supported by studies involving inactivation of Cdx2 in intestinal crypts cultured in vitro. This abolishes their ability to form long-term growing intestinal organoids that differentiate into intestinal phenotypes. We conclude that expression of Cdx2 is essential for differentiation of gut stem cells into any of the intestinal cell types, but they maintain a degree of cell-autonomous plasticity that allows them to switch on a variety of gastric genes.

Fig. 1.

Validation of the Cdx2 conditional knockout allele. (A) Homologous recombination was used to generate a floxed allele in which exon 2 (E2) was flanked by loxP sites (large arrowheads). Upon addition of Cre recombinase, exon 2 is removed and the Cdx2 gene rendered inactive. (B,C) PCR techniques were used to identify the Cdx2^flox allele (B) and the Cdx2^ko allele (C). (D) (a) Normal E10.5 embryo grown on tetraploid trophoblast; (b) Cdx2^–/– embryo grown on tetraploid trophoblast exhibiting gross posterior truncation; and (c) a Cdx2^ko/– embryo following Sox2Cre-mediated ablation showing identical phenotype to that in b. (E) Section through a heterotopic region from a Cdx2^+/ko mouse generated using the floxed allele shown in A. (a) Transitional region between stratified forestomach type epithelium (SSE) and mucous-secreting region characteristic of cardia (Ca). (b) Heterotopic region showing gastric units typical of stomach corpus containing oxyntic cells (arrows). Scale bars: 200 μm.

Fig. 2.

AH and LGR series intestine 1 week post knock out. (A) Cdx2/AB staining reveals high levels of knock out in the small intestine in both the crypts and villi of AH (Cdx2^ko/ko//AhCreER^+/–) series but principally in the crypts only in LGR (Cdx2^ko/ko//Lgr5Cre^+/–) series. Similar knock out levels in the caecum. Lower recombination levels in the AH series following lower dose still produced villi bearing Cdx2-negative cells. Scale bars: 100 μm. (B) In adjacent sections of proximal small intestine, pepsinogen C expression occurs exclusively in epithelial cells that are negative for Cdx2 (^*); claudin 18 expression can be detected in regions that do not express the intestinal brush border marker alkaline phosphatase (inset). Scale bars: 100 μm.

Fig. 3.

Phenotype 4 weeks post Cdx2 knock out. (A) BrdU pulse labelling following Cdx2 knock out 4 weeks earlier; Cdx2-positive cells (^) are BrdU labelled, indicating that they migrated normally up the villus. Cdx2-negative cells (^*) are unlabelled, indicating that they were present in the villi before labelling and have not been replaced from below. BrdU-labelled Cdx2-negative cells remain in the crypts compared with controls. (B) AB-PAS staining of AH series jejunum. Magenta mucous-secreting cells are present at villous bases in the AH series compared with controls. Scale bars: 100 μm. (C) Electron micrograph of dying Paneth cell from AH series showing swelling and vesiculation of the endoplasmic reticulum and a large autophagosome with degenerating granules, compared with a control. Scale bars: 5 μm. (D) Cell counts of Ki67-labelled duodenal crypts from the AH series comparing Ki67-expressing cells in untargeted (Cdx2 +ve) and targeted (Cdx2–ve) crypts at 1, 4 and 8 weeks (P<<0.001 at all stages) alongside control intestinal crypts. Data are mean±s.e.m. (E) Proximal small intestine showing expression of pepsinogen C in Cdx2-negative crypts but not in crypts that express Cdx2 (black brackets). Scale bars: 100 μm. (F) RNA was extracted from short regions (1-6) along the length of intestines from experimental (AH) and control (CONT) mice. Tissues labelled a-f were embedded for histological analysis as ‘swiss-rolls’. (G) RT-PCR showing expression of stomach-specific Muc6, pepsinogen C and H⁺/K⁺ATPase in the experimental (AH) tissue only. Significantly, Barx1, which is normally found in the gastric mesoderm, was not expressed in experimental tissue. Gapdh is a loading control.

Fig. 4.

Adjacent sections of AH series 8 weeks after Cdx2 knock out. (A) Serial sections through mucosa of upper jejunum. Cdx2- and villin-negative cells line cystic vesicles in the subsurface mucosa (X). Cells lining these vesicles secrete PAS-positive mucous (^*) with pepsinogen C expression also evident basally (black arrow). Sonic hedgehog (Shh) expression is present at low levels in the cyst (V) and more strongly in an adjacent Cdx2-negative crypt (Δ). Scale bars: 100 μm. (B) A separate section demonstrates that cystic regions express claudin 18, whereas villi express alkaline phosphastase. (C) Submucous cyst from LGR series showing GFP staining (a reporter of Lgr5 expression indicating a stem cell marker). On the right is an untargeted crypt showing GFP expression in cells lying between the Paneth cells (arrows). Scale bar: 50 μm. (D) Nuclear β-catenin expression is evident in Paneth cells as expected (arrowheads) but not in the cells lining the cysts. Scale bar: 50 μm. (E) Electron micrograph showing sparse microvilli and electron-dense secretory profiles in cells lining a submucous cyst. Also present (^*) is a pale, apparently undifferentiated, cell deep from the surface containing few mature organelles. Scale bar: 2 μm. (F) A Cdx2-negative ileal region that has lost villous morphology, stained for Cdx2 and Alcian Blue, AB-PAS, pepsinogen C and villin. The Cdx2-negative region is also villin negative but contains cells that secrete pepsinogen C- and PAS-positive mucous (black arrows provide orientation). Scale bar: 250 μm.

Fig. 5.

Phenotype of the AH series 29 weeks after Cdx2 knock out and beyond. (A) (a) Haematoxylin and Eosin montage through the polyp-containing caecal region. Scale bar: 1 mm. (b,c) Adjacent sections showing absence of Cdx2 staining (b) and secretion of pepsinogen C in the same region (c). Scale bars: 500 μm. (d) Staining of a formalin-fixed section for β-catenin shows no evidence of nuclear localisation. Scale bar: 50 μm. (B) (i) Normal adult antral mucosa showing expression of Sox2, (ii) constitutional ablation of Cdx2 (Cdx2^+/ko) showing Sox2 expression in heterotopic polyp next to normal Sox2 negative colon and (iii) conditional ablation of Cdx2 in AH series showing absence of Sox2 expression in caecal polyp. Scale bars: 250 μm. (C) RT-PCR of AH experimental versus control caecum 62 weeks post Cdx2 knock out. The experimental caecum contained a large polyp but did not express Barx1. Embryonic and adult stomach controls. (D) Jejunal region stained with AB-PAS and showing excessive development of Paneth cells within Cdx2-positive regions, copious magenta mucous secretion and PAS-positive subepithelial cysts. Scale bars: 100 μm. A formalin-fixed section stained for lysozyme confirms the excessive numbers of Paneth cells present. (E) Electron micrograph of normal Paneth cells 29 weeks after knock out showing abundance of healthy secretory granules. Scale bar: 5 μm.

Fig. 6.

Cdx1 compensates for loss of Cdx2 but is itself downregulated. (A-D) Cdx1 continues to be expressed both in proximal jejunum (A) and in distal ileum (B) 1 week after Cdx2 knock out but Cdx1 is downregulated proximally (C) and distally to a lesser extent (D) by 4 weeks after Cdx2 knock out. (E) Adjacent sections of jejunum from an AH1//2 mouse 1 week after conditional knock out of Cdx2, stained for Cdx2 and Alcian Blue, pepsinogen C and with AB-PAS. Pepsinogen C secretion is apparent in Cdx2-negative crypts and along Cdx2-negative regions of villi. Small PAS-positive cysts are beginning to develop in the subsurface mucosa (high magnification inset). Scale bars: 100 μm. (F) Schematic diagram illustrating a mechanism to account for the difference between constitutive and conditional Cdx2 ablation based upon the loss of the ability of adult intestinal mesoderm to express Barx1. Genes are either expressed (green text) or repressed (red text), depending on whether pathways are active (solid lines) or inactive (broken lines).

Fig. 7.

Cdx2 inactivation causes cultured small intestinal crypts to become empty cysts instead of growing organoids. (A-C) Overview of the process for testing the growth of Cdx2 knockout small intestine organoids. (A) Immunostaining of proximal small intestine used as starting material for crypt isolation. Magenta, Cdx2; brown, lysozyme. Scale bar: 100 μm. (B) Representative organoid before induction with tamoxifen. (C) Representative cystic organoid after 7 days of culture post-tamoxifen induction. Scale bar: 200 μm. (D,E) View of culture wells containing control (uninduced) and tamoxifen-induced organoids. (F) Genotyping of ‘normal’ and ‘cystic’ organoids after manual picking. The lowest band in the normal and cystic organoids corresponds to the knockout allele, and the highest band, only present in the normal-looking organoids, corresponds to the non-deleted floxed allele; the highest band in the Cdx2Δfl control lane corresponds to the wild-type Cdx2 allele. Cdx2fl is from a Cdx2^flox/flox DNA sample and Cdx2Δfl is from a Cdx2^+/ko DNA sample; both act as controls for the different Cdx2 bands (G-J) Representative organoids cultured for 5 days post tamoxifen induction: (G-I) ‘cystic’ Cdx2 knockout organoids, (J) phenotypically normal intestinal organoid. Scale bar: 200 μm. (K-N) Representative organoids cultured for 7 days post-tamoxifen induction. (K-M) ‘Cystic’ Cdx2 knockout organoids, (N) normal-looking intestinal organoid. Even after 7 days of culture, ‘cystic’ organoids do not make buds and remain empty. Scale bar: 250 μm. (O,P) Sections of organoids and cysts from 7 day post-tamoxifen induction cultures stained for the Paneth cell marker lysozyme (brown). Cysts do not express lysozyme (Δ), whereas the organoids do (^*). Scale bar: 50 μm.