Loss of the Wnt receptor frizzled 7 in the mouse gastric epithelium is deleterious and triggers rapid repopulation in vivo

The Wnt signalling pathway regulates multiple cellular functions, including proliferation, migration, differentiation and stem cell function (Clevers and Nusse, 2012), and is crucial during embryonic development (van Amerongen and Nusse, 2009). The control of important cellular functions by Wnt signalling is maintained into adulthood, during which several organs require Wnt signalling for correct homeostasis, including the intestinal tract, hair follicle, mammary gland and liver (Clevers et al., 2014). Wnt signalling has to be tightly regulated during homeostasis, as deregulated Wnt signalling is often one of the earliest oncogenic events in several cancers (Polakis, 2012).

Wnt signalling is divided into three broad pathways: canonical β-catenin, planar cell polarity (PCP) and calcium-dependent signalling (Wnt/Ca²⁺) (Niehrs, 2012). The cytoplasmic signal transducers that regulate these pathways have been the subject of intense research interest, and in the case of the canonical β-catenin pathway, its role is well characterised in several biological contexts, including development, homeostasis, regeneration and cancer, particularly in the intestine.

The receptors that transmit Wnt signalling are beginning to be understood in more detail, with a complex model emerging in which frizzled (Fzd) Wnt receptors associate with various co-receptors to activate different Wnt pathways (Niehrs, 2012). To date, 19 Wnt ligands and 10 Fzd receptors have been discovered in mammals, and the Wnt pathway is highly conserved from humans through to evolutionarily older organisms, including Hydra (Nichols et al., 2006; Srivastava et al., 2008).

The gastric epithelium is composed of parallel, glandular invaginations known as gastric units. Each gastric unit is composed of a pit, which is continuous with the surface epithelium and a flask-shaped gland, which extends down further into the isthmus, neck and base areas. Distinct areas within individual gastric units are characterised by the residency of specialised cell types that regulate various aspects of digestion: gastric mucus cells that secrete protective mucus; parietal cells responsible for secreting hydrochloric acid; chief cells that release active pepsin; and several types of endocrine cells that secrete an array of hormones that aid and regulate digestion and absorption, including ghrelin and somatostatin (Mills and Shivdasani, 2011). Importantly, the precise architecture, cellular heterogeneity and turnover rate of the gastric units varies markedly between the two major anatomical regions of the stomach, the antrum and corpus (Mills and Shivdasani, 2011).

Several studies have implicated Wnt signalling as being important in the gastric epithelium, although its role is poorly understood in comparison with that in the intestinal epithelium. Gastric organoid cultures require Wnt3a in the culture medium in addition to the Wnt agonist R-spondin (Barker et al., 2010; Flanagan et al., 2016), demonstrating that Wnt is required for the gastric epithelium. The R-spondin receptor Lgr5 is expressed in cells that respond to Wnt signals and is a marker of stem cells in several organs, including the gastric epithelium (Barker et al., 2010; de Lau et al., 2011), demonstrating that Wnt-responsive stem cells reside in the gastric epithelium. The Wnt pathway is more active in the antrum than in the corpus; however, Troy⁺ (Tnfrsf19⁺) cells in the corpus express Wnt target genes and stem cell signature genes (Stange et al., 2013). Experimental deregulation of Wnt signalling in the gastric epithelium can also result in tumourigenesis (Radulescu et al., 2013), similar to the intestinal epithelium (Sansom et al., 2004).

We have recently demonstrated that Fzd7 is the predominant Wnt receptor in regulating homeostasis in the intestinal epithelium, in which deletion of Fzd7 in either the whole epithelium or specifically in the Lgr5⁺ intestinal stem cells, triggers rapid repopulation (Flanagan et al., 2015). Here, we show that Fzd7 is also expressed in the antrum of the gastric epithelium, and is required for the growth of gastric organoid cultures. Deletion of Fzd7 in the gastric epithelium in vivo was deleterious and triggered rapid repopulation of the epithelium – the first time repopulation has been reported for the stomach following a genetic insult. These data identify that Fzd7 is crucial for transmitting Wnt signalling to regulate homeostasis in the gastric epithelium.

Wnt signalling is crucial for homeostasis of the small intestine (Clevers and Nusse, 2012; Flanagan et al., 2015); however, it is less well understood in the gastric epithelium. To examine the requirement for Wnt signalling in the gastric epithelium, we established organoid cultures from the mouse antral gastric epithelium and exposed them to various Wnt pathway inhibitors and activators, which we validated via TOPFLASH assays (Molenaar et al., 1996) and western blots for active β-catenin (van Noort et al., 2002) in HEK293 cells (Fig. S1A,B). Organoids treated with either the porcupine inhibitor IWP-2, which prevents secretion of Wnt ligands (Chen et al., 2009), or the tankyrase inhibitor XAV939, which stabilises the β-catenin degradation complex and consequently inhibits Wnt signalling (Huang et al., 2009), underwent rapid atrophy and organoid death. This was not observed in vehicle-treated organoids, which continued to thrive (Fig. 1A). Conversely, gastric organoids treated with the selective Gsk3β inhibitor CHIR-99021 (CHIR), thereby activating Wnt signalling (Bennett et al., 2002), showed increased organoid size and viability (Fig. 1A-C). These observations were supported by a thiazolyl blue tetrazolium bromide (MTT) assay showing marked reduction in cell viability in gastric organoids treated with either XAV939 or IWP-2, and conversely increased metabolism in organoids treated with Wnt agonist CHIR (Fig. 1B). Quantitative reverse transcriptase PCR (qRT-PCR) was then performed on total RNA extracted from the treated gastric organoids, identifying that expression of Wnt target genes Sox9, Cd44 and Myc was significantly reduced following XAV939 or IWP-2 treatment, and was conversely upregulated following CHIR treatment (Fig. 1D). Interestingly, expression of Fzd genes is increased in organoids treated with IWP-2 or XAZ939, presumably as a mechanism to increase Wnt signalling in response to these compounds inhibiting the pathway, but as Wnt target genes are still reduced, this response is insufficient to activate Wnt signalling and thus the organoids die (Fig. S1C). Collectively, these data demonstrate that Wnt signalling is crucial for gastric organoid growth and maintenance, and they identify that Wnt ligands secreted from the epithelial cells of the gastric organoids are required cell-autonomously for their growth and survival. This strongly implicates an integral role for Fzd receptors in transmitting these essential Wnt signals in the gastric epithelial cells.

Little is known regarding the extent of Wnt signalling and the expression of Fzd receptors in the gastric epithelium. To investigate the expression of Fzd receptors in the gastric epithelium, we extracted RNA from the antrum and corpus epithelium of wild-type mice and performed qRT-PCR. The expression of Fzd2 and Fzd7 was markedly higher in the antrum than in the corpus, both comparatively (Fig. 2A) and in the raw data (Fig. S2A). Fzd7 is of particular interest as it is required for embryonic stem cell activity (Melchior et al., 2008) and we have recently shown it to be the predominant Wnt receptor regulating homeostasis in the intestinal epithelium (Flanagan et al., 2015). The increase in Fzd7 expression, a Wnt target gene (Vincan et al., 2010), in the antrum is also associated with an increase in the expression of other Wnt target genes, including Myc, cyclin D1 (Ccnd1), Cd44 and Lgr5 (Fig. 2B and Fig. S2B), suggesting Fzd7 is transmitting Wnt signalling in the isthmus and base of the antral glands. Although the expression of Lgr5 is approximately sevenfold higher in the antrum compared with the corpus, its comparative expression with other Wnt targets in the antrum is relatively much lower, reflecting its function as a stem cell marker (Fig. S2B). To visualise the expression of Fzd7 in the gastric epithelium, we performed X-gal stains on stomachs isolated from Fzd7^nLacZ/+ mice, which express the LacZ gene under the control of the endogenous Fzd7 regulatory region (Yu et al., 2012). Staining was observed from the base of the antral gastric glands to the isthmus, illustrating that Fzd7 is expressed in these cells (Fig. 2C).

A powerful tool to understand gene function in a particular tissue is to be able to conditionally delete it specifically in the tissue of interest. To drive genetic recombination in the gastric epithelium, we used the tamoxifen-inducible Tff1Cre^ERT2 mouse (Thiem et al., 2016). These mice provide robust recombination in the antral glands, with recombined glands still observed 30 days after tamoxifen induction, demonstrating that recombination must occur in a stem cell population (Fig. S3). As previously reported, some cells are also recombined in the pit region of the corpus but these cells do not give rise to entire glands, suggesting that recombination does not occur in a stem cell with the capacity to populate this tissue with all the differentiated lineages observed (Thiem et al., 2016) (Fig. S3). However, a few recombined cells are still observed long after the continuous renewal of the corpus has replenished the gland cell population, suggesting either a long-lived population of non-stem cells is recombined here or, alternatively, recombination occurs in a population of stem cells here that only gives rise to a small restricted population of cells in the corpus.

To investigate the requirement for Fzd7 in the gastric epithelium, we grew gastric organoid cultures from the antrum of Tff1Cre^ERT2/+; Fzd7^flox/flox mice (Tff1Cre⁺; Fzd7^fl/fl) to enable tamoxifen-induced deletion of Fzd7. Five days after administration of 4-OHT (the metabolically processed version of tamoxifen) to the medium, we observed widespread atrophy of gastric organoids derived from Tff1Cre⁺; Fzd7^fl/fl mice, whereas gastric organoids derived from Tff1Cre⁺; Fzd7^+/+ mice continued to thrive (Fig. 3A). These observations were supported by a MTT assay, which showed a significant reduction in organoid viability following Fzd7 deletion (Fig. 3B) and no changes in gastric organoids treated with vehicle only (Fig. S4). To confirm robust deletion of Fzd7 we performed PCR on genomic DNA (gDNA) isolated from organoids 3 days after 4-OHT, when organoids were still alive (Fig. 3C). The recombined Fzd7 allele (Fzd7^Δ) showed a very strong amplified product in the organoids derived from Tff1Cre⁺; Fzd7^fl/fl mice, and was undetectable in organoids derived from Tff1Cre⁺; Fzd7^+/+ mice (Fig. 3C). These data demonstrate robust deletion of Fzd7, which was confirmed by performing qRT-PCR at the same 3-day time point, showing ∼80% reduction of Fzd7 and downregulation of Wnt target genes Myc and Cd44 (Fig. 3D). Previously, we have observed upregulation of Fzd1 and Fzd2 expression to partially compensate for the experimental deletion of Fzd7 during intestinal regeneration (Flanagan et al., 2015). Expression analysis of Fzd genes revealed that Fzd1, Fzd2 and Fzd3 levels were elevated in gastric organoids following Fzd7 deletion, although only Fzd3 was significantly different in expression (Fig. 3E). As the organoids still undergo atrophy and die 3 days post-deletion of Fzd7, this suggests that these elevated Fzd receptor levels are unable to compensate for the loss of Fzd7 in this system. Together, these data demonstrate that Wnt signalling is required for the maintenance of gastric epithelial cells ex vivo via the Wnt receptor Fzd7.

Deletion of Fzd7 in organoids derived from the intestinal epithelium resulted in widespread crypt atrophy and organoid death (Flanagan et al., 2015). Interestingly, when Fzd7 was deleted in the intestinal epithelium in vivo, it did not result in denuding of the epithelium, which might have been expected, but instead triggered repopulation of the entire epithelium with non-recombined Fzd7-proficient cells (Flanagan et al., 2015). Although this has been observed previously with deletion of other important genes in the intestinal epithelium, including Myc (Muncan et al., 2006) and Chek1 (Greenow et al., 2009), repopulation has never been observed in the gastric epithelium. To determine the requirement for Fzd7 in the gastric epithelium in vivo, we conditionally deleted Fzd7 using Tff1Cre⁺; Fzd7^fl/fl; Rosa26LacZ^LSL mice in which we could track the fate of recombined cells in the stomach over time. At 3 days after tamoxifen induction Tff1Cre⁺; Fzd7^+/+; Rosa26LacZ^LSL mice displayed lineage tracing throughout the glands of the antrum; however, Tff1Cre⁺; Fzd7^fl/fl; Rosa26LacZ^LSL mice had slightly less recombination at this time point in the base of the glands (Fig. 4A). At 5 days after tamoxifen induction, Fzd7-deleted Tff1Cre⁺; Fzd7^fl/fl; Rosa26LacZ^LSL mice had markedly fewer recombined cells in the lower halves of the glands, whereas Tff1Cre⁺; Fzd7^+/+; Rosa26LacZ^LSL mice continued to contain recombined cells throughout the glands (Fig. 4A). The replacement of Fzd7 recombined cells with non-recombined cells continued over time until the entire epithelium was repopulated by non-recombined (pink) cells (Fig. 4A). Enumeration of this event shows a rapid loss of recombined (blue) glands from 5 days after tamoxifen induction, leading to total repopulation of the epithelium at 30 days (Fig. 4B).

To genetically monitor the repopulation event, we isolated gDNA from the antral epithelium at 5 and 30 days post-tamoxifen induction and performed PCR for the genetic product of the deleted recombined Fzd7 allele (Fzd7^Δ) and also the non-recombined Fzd7 flox allele (Fzd7^flox). At 5 days after tamoxifen injection there was a large amplification of the Fzd7^Δ product in the Tff1Cre⁺; Fzd7^fl/fl mice that was undetectable in the Cre-negative Tff1Cre⁻; Fzd7^fl/fl mice at the same time point. This indicates that the Fzd7 flox allele had undergone robust recombination only in tamoxifen-treated Tff1Cre⁺; Fzd7^fl/fl mice (Fig. 4C). This was confirmed by a weak non-recombined product for the Fzd7^flox allele in the Tff1Cre⁺; Fzd7^fl/fl mice [some non-recombined product is still present due to the recombination not occurring in 100% of the cells of the gastric antrum (Fig. 4B)], compared with a strong non-recombined Fzd7^flox product in the Tff1Cre⁻; Fzd7^fl/fl mice 5 days after tamoxifen (Fig. 4C). At 30 days after tamoxifen, the Fzd7^Δ product is almost undetectable in Tff1Cre⁺; Fzd7^fl/fl mice, which coincides with a marked increase of the non-recombined Fzd7^flox product, demonstrating the repopulation of the gastric epithelium with non-recombined Fzd7-proficient cells (Fig. 4C). Conversely, the non-recombined Fzd7^flox product remains strong and unchanged 5 days and 30 days after tamoxifen treatment in the Tff1Cre⁻; Fzd7^fl/fl mice, indicating no recombination of this allele and thus no deletion of Fzd7 (Fig. 4C). These data molecularly demonstrate that deletion of Fzd7 in the antrum of the gastric epithelium is a deleterious event and triggers rapid repopulation with Fzd7-proficient cells.

To investigate whether repopulation could be triggered by deletion of a different Fzd receptor, we deleted Fzd5 in the gastric epithelium (Fig. S5A). In contrast to deletion of Fzd7, we did not observe any repopulation after tamoxifen induction in Tff1Cre⁺; Fzd5^fl/fl; Rosa26LacZ^LSL mice, with recombined cells still present 30 days after deletion (Fig. S5A). Deletion of Fzd5 was confirmed by qRT-PCR on cDNA isolated from the antral gastric epithelium, which showed a significant reduction in Fzd5 expression, but no change in the expression of the canonical Wnt target genes Ccnd1, Myc, Cd44 and Lgr5 (Fig. S5B). These data demonstrate that Fzd5 is not required for gastric homeostasis and suggests that the phenotype observed when we delete Fzd7 is not a generic event triggered by deletion of any Wnt receptor from the stomach.

Three days after deletion of Fzd7 in the gastric epithelium, we observe downregulation of Fzd7 expression 2 days before repopulation is apparent at day 5 (Figs 5A and 4A). We therefore analysed the gastric epithelium to determine the phenotype of deleting Fzd7 in this tissue. Caspase 3 immunohistochemistry revealed a marked increase in the number of apoptotic cells after Fzd7 deletion (Fig. 5B,C), which is consistent with our observations in gastric organoid cultures, demonstrating that deletion of Fzd7 triggers apoptosis in gastric epithelial cells.

Immunohistochemical (IHC) staining for Muc5a revealed that deletion of Fzd7 resulted in the dramatic reduction of mucus-secreting cells (Fig. 5B), which was associated with a significant reduction in the expression of Wnt target gene Muc5a (Mucenski et al., 2005) (Fig. 5D), suggesting that Fzd7 regulates Muc5a expression and thus the differentiation of mucus-secreting cells. Immunohistochemistry for gastrin showed that G cells were mislocalised along the length of antral glands following Fzd7 deletion compared with their usual position towards the base of antral glands in control mice (Fig. 5B). Scoring the number of gastrin-positive cells revealed no significant difference in their numbers in each antral gland, which is consistent with the unchanged expression levels of gastrin transcript between Fzd7-deleted and control mice (Fig. 5D,E). To further investigate whether deletion of Fzd7 was altering the differentiation and function of G cells, we performed immunohistochemistry to visualise expression of the gastric hormones ghrelin and somatostatin. The expression of both hormones was consistent between Fzd7-deleted and Fzd7-proficient mice in the antrum, strongly suggesting that the mislocalised G cells are still functional (Fig. S6). These data suggest that Fzd7 regulates differentiation of mucus-secreting cells and also the localisation of G cells along the gastric gland.

To monitor the activation of the Wnt pathway during the gastric repopulation event triggered by Fzd7 deletion, we examined the expression of several Wnt target genes in the antrum at different time points. All Wnt target genes examined were downregulated 3 days after Fzd7 deletion, concordant with a reduction in Fzd7 expression (Fig. 6A) and suggesting that Fzd7 is required to transmit Wnt signalling in this tissue. Surprisingly, 5 days after Fzd7 deletion, the expression of Fzd7 and the other Wnt target genes is significantly upregulated, and expression continues to remain high until repopulation has resolved at 14 days (Fig. 6A).

Wnt signalling has been demonstrated to regulate cell proliferation in many tissues and we therefore performed immunohistochemistry for the cell proliferation marker PCNA. Scoring for PCNA-positive cells per gland identified a transient increase in proliferation 5 days and 7 days after Fzd7 deletion, and a return to wild-type levels of proliferation as repopulation was resolved (Fig. 6B,C). This is consistent with the general gene expression profiles of the Wnt target genes analysed, which shows that Wnt signalling is elevated during this period of repopulation and is restored to wild-type levels by 30 days post Fzd7 deletion.

These data strongly suggest that Wnt signalling, via Fzd7, plays an important role during the process of gastric repopulation following Fzd7 deletion. To investigate the functional requirement for Wnt signalling downstream of Fzd7, we deleted Fzd7 from gastric organoids and treated them with the Wnt agonist CHIR. The atrophy and cell death observed when Fzd7 was deleted in gastric organoids was completely rescued when these organoids were additionally treated with CHIR (Fig. 6D). A MTT assay confirmed that the loss of cell viability observed when Fzd7 was deleted was completely rescued if we then activated the Wnt pathway downstream of the receptor using CHIR treatment (Fig. 6E). Analysis of Wnt target genes Fzd7, Sox9 and Cd44 by qRT-PCR showed that they were dramatically upregulated in response to CHIR treatment (Fig. 6F). Conversely, these genes were downregulated when Fzd7 was deleted. However, in Fzd7-deleted organoids treated with CHIR, which do not display the atrophy of Fzd7 deletion alone, the level of these Wnt target genes remained not significantly different from those of the untreated organoids, demonstrating that loss of Wnt signalling is the mechanism responsible for the atrophy and organoid death when Fzd7 is deleted.

Here, we show for the first time a functional requirement for Wnt signalling in the gastric epithelium via the Wnt receptor Fzd7, and that deletion of Fzd7 specifically in the gastric epithelium can trigger repopulation of this tissue, which until now has not been reported. Embryonic development of the corpus (fundus in humans) requires active Wnt signalling, whereas inhibition of Wnt results in antral development (McCracken et al., 2017). However, in the adult gastric epithelium, Wnt signalling is active in different areas. The Axin2-LacZ mouse shows that Wnt signalling is highest in the base and isthmus of antral glands (Barker et al., 2010; Stange et al., 2013), which is consistent with our data here where we also observe expression of Fzd7 in this area of the antrum. In addition, Wnt3a is required in the culture medium for gastric organoids, suggesting an important role for Wnt receptors during gastric homeostasis. Although genetic aberrant activation of Wnt signalling can lead to tumourigenesis (Radulescu et al., 2013), there have been no functional experiments to inhibit the Wnt pathway and examine the consequences to gastric homeostasis. To investigate the requirement for Wnt signalling in gastric epithelial cells, we first treated gastric organoids with Wnt inhibitors, IWP-2 or XAV939, which resulted in reduced organoid cell viability. Interestingly, we have previously shown that removal of Wnt3a from the culture media results in organoid death (Barker et al., 2010). However, as IWP-2 is a porcupine inhibitor and thus prevents cells from secreting all Wnt ligands, this strongly suggests that gastric epithelial cells require cell-autonomously secreted Wnt in addition to supplemental Wnt3a provided by the culture media. Thus, our new data implicate a role for both epithelial and underlying stromal cells as a source of Wnt ligands regulating gastric homeostasis in vivo. Indeed, deletion of Fzd7, which is expressed in the gastric antrum, resulted in cell atrophy and organoid death, similar to organoids treated with Wnt pathway inhibitors. Fzd7 deletion also resulted in the death of intestinal organoids (Flanagan et al., 2015), suggesting a common role for this receptor in both of these tissues to regulate homeostasis. Interestingly, treatment of gastric organoids with IWP-2 or XAV939 resulted in earlier death of organoids than deletion of Fzd7. This could be due to the significant upregulation of Fzd3 following deletion of Fzd7, which is then able to partially and transiently compensate for the loss of Fzd7, which would be ineffective in organoids treated with IWP-2 or XAV939 as they block the Wnt pathway at the level of Wnt secretion or β-catenin, respectively (Fig. S1C). This is similar to the situation we previously reported in the small intestine in which deletion of Fzd7 is partially compensated for by upregulation of Fzd1 and Fzd2 during intestinal regeneration (Flanagan et al., 2015).

Surprisingly, when we deleted Fzd7 from the gastric epithelium in vivo, we did not observe widespread atrophy and denuding of the epithelium, as might be expected. However, using lineage tracing of recombined Fzd7-deficient cells, we were able to track a repopulation event in the gastric epithelium for the first time, in which non-recombined, Fzd7-proficient cells replaced the Fzd7-deficient cells over the course of 7-10 days. This is consistent with previous lineage tracing showing full glands could be generated from Lgr5⁺ cells in the same way in 7-10 days (Barker et al., 2010). Repopulation does not occur from a denuded epithelium, as in regeneration, and therefore does not preclude that a large apoptotic event is associated with it. We and others have previously reported that the intestinal epithelium is able to repopulate after deletion of crucial genes such as Myc (Muncan et al., 2006), Stat3 (Matthews et al., 2011) or Chek1 (Greenow et al., 2009). Indeed, we also recently observed repopulation when we deleted Fzd7 in the intestinal epithelium, again suggesting a common role for this receptor in gastric and intestinal homeostasis. This mechanism of repopulation is an important adaptation to allow these epithelial layers to rapidly respond to damaging molecular events that could otherwise disrupt the delicate homeostasis of these tissues, resulting in possible pathologies that include colitis/gastritis and neoplasia (Clevers et al., 2014). Repopulation of the gastric epithelium with bone marrow-derived cells (BMDCs) has been previously reported in vivo, but only 30 weeks after experimental infection with Helicobacter, which eventually resulted in the development of gastric tumours, with no repopulating cells observed at earlier time points (Houghton et al., 2004). This then represents a very different kind of repopulation to the rapid event we describe here, which results in the gastric epithelium returning to a normal homeostatic state after the repopulation event, rather than any associated pathology as observed with the slow BMDC repopulation. Epithelial damage and gastric ulceration are common pathologies associated with radiotherapy in humans (Coia et al., 1995; Henriksson et al., 1999). This suggests that manipulation of the Wnt pathway following irradiation could be of therapeutic benefit for individuals receiving radiotherapy, as has been suggested in the intestine (Ashton et al., 2010; Phesse and Sansom, 2013; Zhou et al., 2013).

Gastric repopulation is not observed until 5 days after Fzd7 deletion. Therefore, before this time point we can analyse the requirement for Fzd7 in the gastric epithelium. At 3 days post-deletion we could demonstrate robust deletion of Fzd7 from the gastric epithelium, observed as perturbed differentiation of Muc5a⁺ mucus-secreting cells, which was also previously reported in mice with hyperactive Notch signalling (Demitrack et al., 2015). However, neither inhibition nor hyperactivation of Notch signalling triggered repopulation in the gastric epithelium, suggesting that this mechanism is exquisitely sensitive to loss of Wnt signalling. These data suggest that Notch and Wnt signalling work in parallel to regulate gastric homeostasis, with distinct functions from one another. In support of this, proliferation is also altered in the gastric epithelium in response to modulated Notch signalling (Demitrack et al., 2015; Kim and Shivdasani, 2011), which we did not observe in the Fzd7-deficient gastric epithelium 3 days after Fzd7 deletion. Transient changes in proliferation were observed only during the gastric epithelial repopulation event, which was characterised by the return of Fzd7-proficient cells and increased Wnt activation. A small increase in the number of apoptotic cells per gland was also observed following Fzd7 deletion. These apoptotic events were located in the isthmus of antral glands, which is the location of a population of stem cells marked by either Lrig1, Sox2 or Cckbr (Hayakawa et al., 2016). This suggests that deletion of Fzd7 may be deleterious to these stem cells, and consequently triggers repopulation, which will be important to investigate in future studies. Furthermore, Lgr5⁺ cells are located in the base rather than in the isthmus of the antrum, suggesting that, in contrast to the intestine, Fzd7 may be regulating a population of stem cells that are not expressing Lgr5.

Regulation of differentiation by Wnt signalling is also observed in the intestinal epithelium, where activation or inhibition of the pathway can result in perturbed differentiation and mislocalisation of Paneth cells (Phesse et al., 2008; Sansom et al., 2004). Indeed, G cells are mislocalised throughout the gastric antral glands after Fzd7 deletion, rather than located at their usual position at the base of the glands. These data strongly suggest a conserved function for Wnt signalling in regulating the location of differentiated cells within the gastric and intestinal epithelium.

Deletion of Fzd7 perturbs gastric organoid viability. In contrast, treatment of gastric organoids with the Wnt pathway activator CHIR, which inhibits Gsk3β, increases Wnt target gene expression and cell viability. These data identify Wnt as an important regulator of gastric epithelial cell function. Intriguingly, the deletion of Fzd7 in these gastric organoids prevents CHIR treatment from activating Wnt target genes to the levels observed in Fzd7-proficient organoids. These results demonstrate that modulation of Wnt/Fzd receptor interactions can still influence the outcome of cells in which the cytoplasmic downstream signal transducers of the pathway have been mutated. This is consistent with previous findings in which we and others have shown that Wnt pathway activity can still be modulated in colon cancer cells with mutant APC (Caldwell et al., 2004; Suzuki et al., 2004; Vincan et al., 2007, 2005). As the Wnt pathway is also deregulated in gastric cancer (Phesse et al., 2016), these data also suggest that Fzd receptors could be a target for therapeutic intervention for this disease.

The BAC transgenic Tg(Tff1Cre^ERT2) (Thiem et al., 2016), Fzd7^fl/fl (Flanagan et al., 2015), Fzd7^nLacZ (Yu et al., 2012), Fzd5^fl/fl (van Es et al., 2005) and Rosa26LacZ^LSL (Soriano, 1999) have all been previously described. Mice were interbred to generate compound mice with appropriate alleles. All mice were co-housed, and with the exception of Fzd7^nLacZ mice, all mice were on an inbred C57Bl/6 genetic background, using males and females and appropriate littermates as controls. The Fzd7^nLacZ mice were on a mixed C57Bl/6×Sv129 background. All animal experiments were approved by the Animal Ethics Committee, Office for Research Ethics and Integrity, University of Melbourne, Australia.

Short-term in vivo Cre induction (<7 days post induction) was performed in 6- to 10-week-old mice with a single intraperitoneal (i.p.) injection of 2 mg of tamoxifen per mouse. Long-term in vivo Cre induction (>14 days post induction) was performed in 6- to 10-week-old mice with a single daily i.p. injection of 2 mg of tamoxifen per mouse per day over four consecutive days.

Freshly isolated stomachs were cut along their greater curvature, washed with PBS and immediately fixed (1% formaldehyde, 0.2% gluteraldehyde, 0.02% NP-40 in PBS) for 2 h at 4°C. The fixative was removed and stomachs were washed in PBS. Stomachs were incubated in β-galactosidase detection substrate [5 mM K₃Fe(CN)₆, 5 mM K₄Fe(CN)₆.3H₂0, 2 mM MgCl₂, 0.02% NP-40, 0.1% sodium deoxycholate, 1 mg/ml X-gal in PBS] in the dark, overnight at room temperature. The detection substrate was removed the following day and stomachs were washed in PBS, followed by an overnight incubation in 4% PFA at 4°C in the dark. The PFA was removed and stomachs were washed in PBS. Stained stomachs were placed into histological cassettes, embedded in paraffin wax, sectioned at 5 µm, mounted onto slides and counterstained with Neutral Red.

Freshly isolated mouse stomachs were flushed with PBS and fixed overnight at 4°C in 10% neutral buffered formalin (NBF) and washed twice in 70% ethanol at room temperature. Tissues were placed into histological cassettes, embedded in paraffin wax, sectioned at 5 µm and mounted onto slides as described previously (Flanagan et al., 2015). Paraffin sections were de-waxed, re-hydrated, blocked and incubated in primary antibody overnight at 4°C. Sections were washed and incubated in secondary antibody (polymer horse radish peroxidase-conjugated mouse/rabbit/goat) for 30 min at room temperature. Sections were rinsed in and bound peroxidase was detected and developed by adding diaminobuteric acid substrate (DAB) at room temperature. Slides were washed in MilliQ water and nuclei counterstained with Mayer’s haematoxylin. Antibodies used were mouse anti-Muc5aC (1:400, Thermoscientific, MS-145B0), rabbit anti-PCNA (1:300, Santa Cruz, SC-7907), rabbit anti-caspase 3 (1:1000, R&D systems, AF-835) and goat anti-gastrin-C20 (1:400, Santa Cruz, SC-7783).

The stomachs from mice were dissected out, cut along the greater curvature and flushed in ice-cold PBS, then incubated in a 50 mM EDTA (pH 8.0) in PBS chelating solution for 1 h on a roller at 4°C. Stomachs were then transferred to tubes containing PBS and vigorously shaken to dissociate gastric glands from the underlying stroma (Flanagan et al., 2016). Isolated gastric gland suspension was filtered through a 70 µM cell strainer (BD Biosciences, #352350), which was collected and counted using a haemocytometer. The gastric glands were resuspended in Matrigel (∼100 glands/50 µl of Matrigel) and plated onto a 24-well tissue culture plate. Once the Matrigel had set at 37°C, organoids were covered with 500 µl of gastric culture medium as previously described (Barker et al., 2010; Flanagan et al., 2016). Gastric medium containing growth factors was replenished every other day and cultures were passaged and split once a week. In vitro Cre recombinase was activated by treating gastric organoid cultures with 100 nM 4-hydroxytamoxifen (4-OHT) as previously described (Barker et al., 2010; Flanagan et al., 2016). Organoid cultures were imaged on a Nikon Ti-E microscope using DIC (differential interference contrast) with a 4× PlanApo NA 0.3 objective. A focal stack of images was collected 10 µm apart and processed through the ‘Best Focus’ function of MetaMorph v7.7.7 (Molecular Devices) to generate the final image of individual organoids as previously described (Flanagan et al., 2015; Phesse et al., 2014).

Gastric glands were homogenised in TRizol and total RNA was purified and DNAse treated on Qiagen columns (Promega) and quantified using a DNA/RNA nanodrop spectrophotometer. Four μg of each RNA sample was reverse transcribed using anchored oligodT primers (Promega) and Moloney Murine Leukemia Virus Reverse Transcriptase (M-MLV RT, Promega, #M1705) following the manufacturers' instructions in a final volume of 100 µl, as previously described (Vincan et al., 2007). Real-time RT–PCR was performed using the SYBR green PCRmaster mix and the ABI PRISM 7500 sequence detection system (Applied Biosystems) on cDNA synthesized from DNase-treated total RNA as previously described (Flanagan et al., 2015). Gene expression levels were calculated relative to the housekeeping gene 18S. The 2-^ΔΔCT method (Bustin et al., 2009) was used to calculate the fold change as previously described (Phesse et al., 2008; Vincan et al., 2007). Primer sequences are available in Table S1.

Following treatment, gastric organoids were mechanically dissociated, washed with ADF, counted, resuspended in fresh Matrigel and seeded in a flat-bottomed 96-well tissue culture plate and incubated for 24 h at 37°C in a 5% CO₂ chamber. Organoids were incubated with MTT (thiazolyl blue tetrazolium bromide, Sigma, #M2128) for 4 h at 37°C in 5% CO₂ chamber. Gastric culture medium was removed from organoids and replaced with lysis buffer (50% DMF, SDS, acetic acid+2.5% 1 M HCl) and incubated overnight at 37°C. Solution (100 µl) was transferred to a clean flat-bottomed 96-well plate and optical density determined using BMG lumistar plate reader (Hansen et al., 1989).

Data are expressed as mean±s.e.m., where mean represents number of mice (≥3 per genotype) or number of independent experiments (≥3). Statistical tests used were Mann–Whitney with Prism7 (GraphPad software) where P≤0.05 was considered significant.

We thank Lachlan Whitehead for assistance in the imaging of organoids and Damian Neate and Jean Moselen for technical assistance.