Gastric stem cells promote inflammation and gland remodeling in response to Helicobacter pylori via Rspo3-Lgr4 axis.

Helicobacter pylori is a pathogen that colonizes the stomach and causes chronic gastritis. Helicobacter pylori can colonize deep inside gastric glands, triggering increased R-spondin 3 (Rspo3) signaling. This causes an expansion of the "gland base module," which consists of self-renewing stem cells and antimicrobial secretory cells and results in gland hyperplasia. The contribution of Rspo3 receptors Lgr4 and Lgr5 is not well explored. Here, we identified that Lgr4 regulates Lgr5 expression and is required for H. pylori-induced hyperplasia and inflammation, while Lgr5 alone is not. Using conditional knockout mice, we reveal that R-spondin signaling via Lgr4 drives proliferation of stem cells and also induces NF-κB activity in the proliferative stem cells. Upon exposure to H. pylori, the Lgr4-driven NF-κB activation is responsible for the expansion of the gland base module and simultaneously enables chemokine expression in stem cells, resulting in gland hyperplasia and neutrophil recruitment. This demonstrates a connection between R-spondin-Lgr and NF-κB signaling that links epithelial stem cell behavior and inflammatory responses to gland-invading H. pylori.

Introduction

Helicobacter pylori is a gram‐negative bacterium that has evolved to colonize and persist in the stomach for decades. It is found in approx. 50% of the world's population and infection is the most important risk factor for gastric cancer. It is able to establish free‐swimming colonies that survive within the protective mucus layer in direct contact with the surface epithelium (Amieva & El‐Omar, 2008). In addition, a subpopulation of H. pylori can form colonies deep inside the gastric glands, where stem and progenitor cells reside (Sigal et al, 2015; Fung et al, 2019). It has been shown that gland‐colonizing bacteria, not the bacteria colonizing the surface, induce the typical pathology associated with H. pylori infection, that is, stem cell expansion, gastritis, hyperplasia, and metaplasia—precursor lesions for gastric cancer (Howitt et al, 2011; Sigal et al, 2015; Fischer & Sigal, 2019; Liabeuf et al, 2022).

Upon attachment, highly pathogenic strains of H. pylori use the bacterial type IV secretion system (T4SS) to inject virulence factors into the host cell cytoplasm, including the effector molecule CagA (Backert et al, 2000; Amieva et al, 2003), as well as the LPS biosynthesis metabolite ADP‐glycero‐β‐D‐manno‐heptose (ADP heptose) (Pfannkuch et al, 2019). ADP heptose, as well as its derivative D‐glycero‐β‐D‐manno‐heptose 1,7‐bisphosphate (HBP), has been shown to act as a central mediator of H. pylori‐driven inflammation by inducing activation of the transcription factor NF‐κB via alpha kinase 1 (ALPK1) and TRAF‐interacting protein with forkhead‐associated domain (TIFA) (Zimmermann et al, 2017; Pfannkuch et al, 2019). However, it remains unclear why these inflammatory responses are triggered specifically by bacteria that have invaded the glands and not by those that remain on the surface.

We have previously shown gland responses to H. pylori infection to be driven by increased expression of Rspo3 in the stroma beneath the glands, which drives stem cell expansion and hyperplasia (Sigal et al, 2017). During normal gland homeostasis, a molecular signaling network established by close interplay between epithelial and stromal cells controls epithelial turnover and differentiation (Sigal et al, 2017; Fischer & Sigal, 2019; Kapalczynska et al, 2022). The antral gland base and lower isthmus contains gastric stem cells that are characterized by high Wnt signaling and expression of stem cell markers such as Axin2 and Lgr5 (Barker et al, 2010; Sigal et al, 2017). The epithelium can be divided into the “gland base module” consisting of stem cells as well as gland base secretory cells (Tan et al, 2020), and the “pit module” consisting of surface mucous cells. The base module relies on stromal‐derived Rspo3, which binds to Lgr receptors that associate with and prevent turnover of the Wnt/ Lrp/ Frizzled receptors complex, thereby stabilizing Wnt signaling (de Lau et al, 2011). Lgr4 is expressed broadly throughout the gland, while Lgr5 is itself a target gene of Rspo3 signaling and is found exclusively in the base (Sigal et al, 2017). Loss of Rspo3 results in a loss of the gland base module, while the increased Rspo3 expression upon infection with H. pylori drives stem cell expansion and accumulation of gland base secretory cells. We have previously shown that Rspo3 counterbalances colonization and gland infiltration of H. pylori and mechanistically drives expansion of rapidly cycling Axin2+/Lgr4+/Lgr5‐ cells in the isthmus, as well as slowly cycling Axin2+/Lgr4+/Lgr5+ basal cells that differentiate into secretory cells leading to hyperplasia in the antrum (Sigal et al, 2015, 2017, 2019). While this suggests that both Lgr4 and Lgr5 are likely to play a role in mediating the effects of Rspo, their distinct roles in driving antral pathobiology remain elusive.

Here, we applied conditional knockout mice that lack Lgr4 or Lgr5 to explore their involvement in antral mucosal responses to H. pylori infection. We reveal that Lgr4 expression is required to maintain expression of Lgr5 in the gland base. Depletion of Lgr4, but not Lgr5 alone, results in a diminished hyperplasia as well reduced inflammatory response, and we demonstrate that the Rspo‐Lgr axis does not only promote hyperplasia but is also important for the ability of gland base cells to induce epithelial pro‐inflammatory NF‐κB signaling upon infection. Our data reveal an explanation for the unique ability of the stem cell compartment to respond to H. pylori infection and demonstrate a novel link between epithelial stem cells and innate immunity in the stomach.

Results

Lgr4/5, but not Lgr5 alone, is required for Helicobacter pylori‐driven gland base cell expansion and hyperplasia

Lgr5 and Lgr4 are the Rspo receptors expressed in gastric epithelial cells. While Lgr5 itself is a Wnt target gene, Lgr4 is regulated differently (Barker et al, 2007). Expression of Lgr5 is thus limited to base cells, where Wnt/ Rspo concentrations are highest, while Lgr4 is broadly expressed throughout the antral gland. In situ hybridization and analysis of publicly available single‐cell RNA‐seq data confirmed the broader expression of Lgr4 in the stomach, and we noticed the highest concentration of Lgr4 in the isthmus (Fig EV1A–C). Rspo3 triggers expansion of slower‐cycling Lgr5+ base cells and rapid‐cycling Lgr4+ isthmus cells, leading to hyperplasia in the antrum upon infection with H. pylori. Both cell types are considered progenitor cells that co‐express the classic Wnt target gene Axin2 and are in direct contact with H. pylori micro‐colonies (Sigal et al, 2015, 2017, 2019). To examine the specific roles of Lgr4 and Lgr5 under homeostatic conditions and upon infection with H. pylori, we generated conditional knockout (KO) mice, achieved by floxed sites, which enable a depletion of Lgr5 or Lgr4 (Planas‐Paz et al, 2016) in Axin2+ stem cells upon injection of tamoxifen (B6‐Axin2CreErt2/Lgr5 fl/fl and B6‐Axin2CreErt2/Lgr4 fl/fl, hereafter referred to as Lgr5 −/− and Lgr4 −/−), while B6‐Axin2CreErt2/Lgr5 +/+ and B6‐Axin2CreErt2/Lgr4 +/+ littermates served as controls. For experiments without infection, mice were treated with tamoxifen 14 days before sacrifice, allowing repopulation of glands by mutant cells. For infection experiments, some mice were infected 14 days after tamoxifen treatment with H. pylori and sacrificed 2 months later. All infected mice showed deep gland colonization (examples are shown in Fig EV2A). We first examined the effects of Lgr5 depletion. Infection led to hyperplasia in both the control mice and in the Lgr5 −/− mice (Fig 1A and B). We earlier described that infection also leads to upregulation of Lgr5 (Sigal et al, 2019 ). To visualize mucous gland base cells in the antrum, we used immunofluorescence labeling for lectin GSII (GSII) and gastric intrinsic factor (GIF). Only a subpopulation of GSII+ cells were GIF positive (Figs 1A and EV2B). Proliferating isthmus cells were identified by anti‐Ki67 staining and pit cells by anti‐MUC5AC. In Lgr5 −/− mice, we did not detect changes in GSII+, GIF+, or Ki67+ cells, nor change in the morphology of the gland base. Upon infection, immunofluorescence showed an expansion of the GSII+GIF+ compartment in both the control and the Lgr5 −/− mice. Hyperplastic glands in the Lgr5 −/− mice as well as in the littermate controls were populated by an increased number of Ki67+ cells (Fig 1A and C). We did not observe any significant alterations in H. pylori colony‐forming units (CFUs) in Lgr5 −/− mice compared with control mice (Fig 1D). Gene expression analysis under homeostasis and upon infection similarly revealed no significant changes in base cell markers (Axin2, Gif, Pgc, Tff2, and Muc6) or pit cell marker Muc5ac in control versus in Lgr5 −/− mice (Fig 1E and F). Expression of Itln1, which we previously found to be expressed in gland base secretory cells and which binds H. pylori to counterbalance infection (Sigal et al, 2019), was also not significantly reduced. Although in an earlier study we showed that selective deletion of Lgr5‐expressing cells leads to increased colonization with H. pylori (Sigal et al, 2019 ), the current data suggest that this phenomenon as well as the epithelial response to infection does not require Lgr5 expression.

Figure EV1

Lgr4 and Lgr5 are expressed in antrum epithelial cells

Single‐molecule ISH for the Lgr homologs Lgr4 and Lgr5 in the antrum tissue of Lgr4 −/−, Lgr5 −/− and control mice. Insets show magnified images of glands. Scale bars: 50 μm.

Quantification of Lgr4 ISH and Lgr5 ISH (n = 2–3 biological replicates per group, n = 20 glands per mouse).

t‐distributed stochastic neighbor embedding (t‐SNE) map displaying scRNA‐seq data from publicly available data of the mouse antrum (Busslinger et al, 2021b) and selected t‐SNE maps for the genes Axin2, Lgr4 and Lgr5.

Data information: For (B) data are mean + SD. *P < 0.05; ns, not significant (t‐test).

Source data are available online for this figure.

Figure EV2

Nuclear beta‐catenin is expressed in Ki67+ and GSII+ cells upon infection and is regulated by Lgr4

Immunofluorescence labeling for Helicobacter pylori in 2‐month‐infected Lgr4 −/−, Lgr5 −/− and control mice confirming gland colonization. Scale bars: 50 μm.

Immunofluorescence labeling for GSII and GIF in uninfected WT antrum mucosa. Scale bars: 50 μm.

Immunofluorescence labeling for Ctnnb1, Ki67 and GSII in uninfected and 2‐month‐infected Lgr4 −/− and control mice. Arrows indicate nuclear beta‐catenin in Ki67+ and GSII+ cells upon infection. In Lgr4 −/− mice the nuclei are not positive for Ctnnb1. Insets show magnified images of glands. Scale bars: 50 μm.

Data information: Images represent findings that were reproduced at least twice in the laboratory in independent biological replicates.

Figure 1

Lgr5 is dispensable for gastric gland hyperplasia, gland base expansion, and antimicrobial defense upon infection with Helicobacter pylori

Lgr5 fl/fl /Axin2 CreErt2 (Lgr5 −/−) and control mice were infected for 2 months with H. pylori.

Representative H&E staining from antral glands (upper row). The dotted line marks the gland base cell area, which expands upon infection in Lgr5 −/− and control mice. Immunofluorescence labeling for GSII and Ki67 (middle row) and GIF and Muc5ac (lower row) of antral epithelium from Lgr5 −/− and control mice uninfected and after 2 months of infection. Scale bars: 50 μm.

Quantification of antral gland height in Lgr5 −/− compared with control mice. Uninfected control (n = 5) versus uninfected Lgr5 −/− (n = 4) versus infected control (n = 4) versus infected Lgr5 −/− (n = 4).

Quantifications of GSII+ area (n = 4–5 biological replicates per group, n = 10 glands per mouse), Ki67 as percentage of total DAPI‐stained cells (n = 3–4 biological replicates per group, n = 20 glands per mouse), GIF+ area (n = 4–5 biological replicates per group, n = 10 glands per mouse), Muc5ac + area (n = 4–5 biological replicates per group, n = 10 glands per mouse).

Helicobacter pylori CFUs in Lgr5 −/− (n = 5) and control mice (n = 5).

qPCR expression levels of Lgr4, Lgr5, and Axin2, gland base cell markers (Gif, Pgc, Tff2, and Muc6), pit cell marker Muc5ac, and Itln1 in uninfected Lgr5 −/− (n = 6–10) and control mice (n = 4–10).

qPCR expression levels of Lgr4, Lgr5, gland base cell markers (Gif, Pgc, and Muc6), and Itln1 in Lgr5 −/− (n = 5) and control mice (n = 5) after 2 months of infection.

Data information: For (B, C, D, E, and F), data are mean + SD. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant (one‐way ANOVA + Tukey's multiple comparison test for B, C, t‐test for D, E and F).

Source data are available online for this figure.

We next explored the responses of Lgr4 −/− mice to H. pylori and found that they failed to develop hyperplasia (Fig 2A and B). Immunofluorescence showed an expansion of both the Ki67+ and the GSII+GIF+ compartment in control, but not in Lgr4 −/− mice (Fig 2A and C). H. pylori CFUs were significantly higher in Lgr4−/− compared with control mice (Fig 2D), mimicking the effects of Rspo3 depletion that we described previously (Sigal et al, 2019). Notably, even in the absence of infection, gland bases of Lgr4 −/− mice contained fewer GSII+ cells. A significant reduction in expression of Lgr5, Axin2, and Gif was detected in uninfected Lgr4 −/− mice compared with uninfected control littermates (Fig 2E), which indicated that Lgr4 is required for Lgr5 expression and Lgr5+ cell differentiation. Our finding that Lgr4 −/− mice harbor fewer stem and secretory cells upon infection was further corroborated by gene expression analysis where we observed a significant decrease in expression of Lgr5, Gif, Pgc, and Muc6 (Fig 2F). Both in homeostasis and in infection conditions, Lgr4 −/− mice also expressed lower levels of Itln1, which is in accordance with the elevated H. pylori colonization levels observed. Thus, we conclude that Lgr4 expression represents a prerequisite for expression of Lgr5. While this makes it impossible to explore the role of Lgr4 alone, our data indicate that expression of Lgr4 is essential for gastric gland hyperplasia upon infection, gland base expansion, and antimicrobial defense upon H. pylori infection, while these parameters are unaffected upon loss of Lgr5 alone.

Figure 2

Lgr4 regulates hyperplasia and expansion of gland base cells upon infection with Helicobacter pylori

Lgr4 fl/fl /Axin2 CreErt2 (Lgr4 −/−) and control mice were infected for 2 months with H. pylori.

Representative H&E staining from antral glands (upper row). The dotted line marks the gland base cell area, which does not expand in infected Lgr4 −/− but does expand in the corresponding control mice. Immunofluorescence labeling for GSII and Ki67 (middle row) and GIF and Muc5ac (bottom row) of antral epithelium from Lgr4 −/− and control mice uninfected and after 2 months of infection. Scale bars: 50 μm.

Quantification of antral gland height in Lgr4 −/− compared with control mice. Uninfected control (n = 2) versus uninfected Lgr4 −/− (n = 4) versus infected control (n = 4) versus infected Lgr4 −/− (n = 4).

Quantifications of GSII+ area (n = 2–4 biological replicates per group, n = 10 glands per mouse), Ki67 as percentage of total DAPI‐stained cells (n = 2–4 biological replicates per group, n = 20 glands per mouse), GIF+ area (n = 2–4 biological replicates per group, n = 10 glands per mouse), Muc5ac + area (n = 4–5 biological replicates per group, n = 10 glands per mouse).

Helicobacter pylori CFUs show higher colonization in Lgr4 −/− (n = 5) compared with control mice (n = 4).

qPCR expression levels of Lgr4, Lgr5, Axin2, gland base cell markers (Gif, Pgc, Tff2, and Muc6), pit cell marker Muc5ac, and Itln1 in uninfected Lgr4 −/− (n = 4–10) and control mice (n = 4–6).

qPCR expression levels of Lgr4, Lgr5, gland base cell markers (Gif, Pgc, and Muc6), and Itln1 in Lgr4 −/− (n = 5) and control mice (n = 5) after 2 months of infection.

Data information: For (B, C, D, E, and F), data are mean + SD. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns, not significant (one‐way ANOVA + Tukey's multiple comparison test for B, C, t‐test for D, E, and F).

Source data are available online for this figure.

R‐spondin 3–Lgr signaling enables inflammatory responses to H. pylori

To obtain deeper insight into the effects of Lgr4 depletion in the stomach, we performed whole‐transcriptome microarray analysis on whole antral tissue from 2‐month H. pylori‐infected Lgr4 −/− mice and littermate controls. Gene set enrichment analysis (GSEA) confirmed that the loss of Lgr4 (which, as shown above also leads to the loss of Lgr5 expression) led to a reduction in the antral stem cell signature (Barker et al, 2010) and genes associated with beta‐catenin signaling (Herbst et al, 2014) (Fig 3A and B). The downregulation of the Wnt target genes Axin2 and Sox9 was validated by qPCR (Fig 3C). In infected wildtype control, but not in infected Lgr4−/− mice, we identified nuclear beta‐catenin (Ctnnb1) in Ki67+ and GSII+ cells in the gland (Fig EV2C). Intriguingly, we found that various inflammation‐associated genes were differentially expressed in control versus Lgr4 −/− mice and GO enrichment analysis further revealed that 63 out of the 157 downregulated genes were linked to innate immune signaling via NF‐κB and to TNF‐α signaling via NF‐κB (Table EV1, Fig 3D).

Figure 3

Lgr4 regulates stem cells and NF‐κB signaling in the antrum

GSEA comparing microarray data from 2‐month‐infected Lgr4 fl/fl /Axin2 CreErt2 (Lgr4 −/−) mice with publicly available data demonstrating downregulation of antral stem cell signature (A) and beta‐catenin targets (B).

qPCR expression levels for beta‐catenin targets Axin2 and Sox9 in uninfected and infected Lgr4 −/− and control mice (n = 2–9).

GSEA comparing microarray data from 2‐month‐infected Lgr4 −/− mice with publicly available data demonstrating downregulation of the TNF‐α signaling via NF‐κB.

Immunofluorescence labeling for p65, Ki67, and GSII in the mouse antrum tissue from uninfected and infected Lgr4 −/− and control mice. Insets show magnified images. Scale bars: 50 μm.

Quantification of p65+ area as percentage of total DAPI‐stained cells of uninfected and infected Lgr4 −/− and control mice (n = 4 biological replicates per group, n = 20 glands per mouse).

Immunofluorescence co‐localization of p65 acetyl K310 (activated form of NF‐κB) in the mouse antrum tissue of infected Lgr4 −/− and in control mice. Insets show magnified images. Scale bars: 50 μm.

qPCR expression levels of Cxcl1, Cxcl2, Cxcl10, and Ccl2 in uninfected and infected Lgr4 −/− mice (n = 4–5) and control mice (n = 2–5).

Immunofluorescence labeling for Cxcl1 in the mouse antrum tissue in infected Lgr4 −/− and in control mice. Scale bars: 50 μm.

Data information: For (G and I) images represent findings that were reproduced at least twice in the laboratory in independent biological replicates. For (C, F, and H), data are mean + SD. *P < 0.05; **P < 0.01; ***P < 0.001 (one‐way ANOVA + Tukey's multiple comparison test).

Source data are available online for this figure.

We analyzed by immunofluorescence the key NF‐κB subunit mediating innate immune response p65 as well as p65 acetyl K310 (Fig 3E–G). Acetylation of p65 at lysine 310 on p65 is required for full transcriptional activation of NF‐κB (Chen et al, 2002). Total p65 localized predominantly in the lower Ki67+ gland isthmus compartment in uninfected mice. In infected control mice, p65 level was increased in the expanded Ki67+ compartment and at less intensity in the GSII+ cells. In contrast, p65 level was not increased in H. pylori‐infected Lgr4 −/− mice (Fig 3E and F). Inhibition of H. pylori‐induced NF‐κB signaling by deletion of Lgr4 became obvious by staining for the activated p65 (p65 at K310), which revealed a strong expression exclusively in the glands of infected control, but not of infected Lgr4 −/− mice (Fig 3G).

To corroborate this revelation, we tested for expression of NF‐κB target genes in uninfected and infected Lgr4 −/− mice and controls by qPCR. Expressions of Cxcl1 and Cxcl2, considered murine homologs of IL‐8, as well as Cxcl10 and Ccl2, both prominently regulated in the microarray data (Table EV1), were upregulated upon infection only in control, but not in Lgr4−/− mice (Fig 3H). Immunofluorescence for Cxcl1 further confirmed activation of NF‐κB signaling in the lower gland isthmus proliferative cell compartment (Fig 3I). Taken together, our in vivo experiments reveal that Lgr4 is required for NF‐κB activation and expression of pro‐inflammatory chemokines in response to H. pylori infection.

To validate that these effects are directly related to the Rspo3‐Lgr regulatory axis, we analyzed NF‐κB expression and target genes in tamoxifen‐treated, 2‐week‐ or 2‐month‐infected B6‐Myh11 CreErt2 /Rspondin3 fl/fl mice (here referred to as: Rspo3 −/−), in which Rspo3 is knocked out specifically in myofibroblasts beneath the gland base, as described previously (Sigal et al, 2019). B6‐Myh11 CreErt2 /Rspondin3 +/+ littermates served as controls. In 2‐week‐infected control mice, we observed a strong expression of p65 specifically in the gland isthmus with a pronounced nuclear translocation of p65, while Rspo3 KO led to reduced expression of p65 (Fig 4A and B). In contrast, cells in the pit and surface region of both Rspo3 −/− and control mice were almost completely devoid of p65 protein, despite the infection. Rspo3 KO also led to reduced expression of Cxcl1 and Cxcl2 at 2 weeks and 2 months of infection, as determined by qPCR (Fig 4C and D). We also generated mice to conditionally overexpress Rspo3 in Myh11+ mesenchymal cells (Myh11 CreERT2/Rosa26Sor 6(CAG‐Rspo3), here referred to as: Rspo3 KI) and found that infection in these mice leads to a strong upregulation of p65, GSII and Ki67 compared with infected control mice (Fig 4E–H). Notably, Rspo3 KI upregulated p65 also in uninfected mice in the antrum as well as in the colon (Figs 4F and EV3A). The massive hyperplastic glands in Rspo3 KI mice revealed throughout the epithelium intense signals for nuclear p65 acetyl K310 (Fig 4I). Collectively, our results reveal that the Rspo3‐Lgr axis not only induces Wnt but also enables an activation of NF‐κB signaling baseline and upon infection, facilitating the induction of epithelial pro‐inflammatory chemokines in response to bacteria in vivo.

Figure 4

Rspo promotes accumulation of nuclear p65

Immunofluorescence labeling for p65 and E‐cadherin in the mouse antrum tissue of 2‐week‐infected B6‐Myh11 CreErt2 /Rspondin3 fl/fl (Rspo3 −/−) and control mice. Arrows in infected control mice indicate nuclear p65 that is not observed in Rspo3 −/− mice. Scale bars: 50 μm.

Quantification of p65+ area as percentage of total DAPI‐stained cells of infected Rspo3 −/− and control mice (n = 3 biological replicates per group, n = 20 glands per mouse).

qPCR expression level of Cxcl1 and Cxcl2 in 2‐week‐infected Rspo3 −/− mice (n = 3) versus infected control mice (n = 4).

qPCR expression level of Cxcl1 and Cxcl2 in 2‐month‐infected Rspo3 −/− mice (n = 5) versus infected control mice (n = 6).

Myh11 CreERT2/Rosa26Sor 6(CAG‐Rspo3) (Rspo3 KI) mice were infected for 6 weeks with H. pylori. Immunofluorescence labeling for p65, GSII and KI67 in uninfected and infected Rspo3 KI and control mice. Scale bars: 50 μm.

Quantification of p65+ area as percentage of total DAPI‐stained cells of uninfected and infected Rspo3 KI and control mice (n = 2–3 biological replicates per group, n = 20 glands per mouse).

Quantification of GSII+ area of uninfected and infected Rspo3 KI and control mice (n = 2–3 biological replicates per group, n = 20 glands per mouse).

Quantification of Ki67 as percentage of total DAPI‐stained cells of uninfected and infected Rspo3 KI and control mice (n = 2–3 biological replicates per group, n = 20 glands per mouse).

Immunofluorescence labeling for p65 acetyl K310 in infected Rspo3 KI and control mice. Insets show magnified images. Scale bars: 50 μm.

Data information: For (B, C, D, F, G, and H) data are mean + SD. *P < 0.05; **P < 0.01; ns, not significant (t‐test for B, C, and D, one‐way ANOVA + Tukey's multiple comparison test for F, G and H).

Source data are available online for this figure.

Figure EV3

Rspo–Lgr4 regulates pro‐inflammatory processes

Immunofluorescence labeling for p65 and Ki67 in the mouse colon tissue of Myh11 CreERT2/Rosa26Sor 6(CAG‐Rspo3) mice (Rspo3 KI) and control mice. Tamoxifen was injected 2 months before sacrifice. Insets show magnified images. Scale bars: 50 μm.

Microarray gene set of significantly (P < 0.05) downregulated genes upon Lgr4 KO in infected mice was analyzed by DAVID 6.7 (BP direct). From these, NF‐κB target genes (according to Gilmore database of NF‐κB targets) were analyzed as above. Obtained GO terms were processed by REVIGO (2021) to consolidate similar terms, and results mapped on semantic space where proximity between bubbles represents similarity in GO terms and size represents frequency of the GO term in the GOA database. Blue bubbles represent GO terms from all downregulated genes. Yellow bubbles represent GO terms obtained from NF‐κB target genes. GO terms associated with yellow bubbles are shown.

Data information: Images represent findings that were reproduced at least twice in the laboratory in independent biological replicates.

Source data are available online for this figure.

R‐spondin 3–Lgr4–NF‐κB signaling is responsible for neutrophil infiltration in H. pylori gastritis

In H. pylori‐infected mice, we observed a manifestation of pro‐inflammatory processes, including neutrophil chemotaxis and immune response in dependence of Lgr4, by separately examining genes that are experimentally validated as targets of NF‐κB (see https://www.bu.edu/nf-kb/gene-resources/target-genes/) and whose expression was decreased in Lgr4 −/− mice (Fig EV3B). We, therefore, asked if infiltration of immune cells upon infection is dependent on Lgr4 or Lgr5. Using immunofluorescence, we stained for MPO, IBA1, and CD3 to analyze the presence of neutrophils, macrophages, and T cells, respectively, in Lgr4 −/− mice in uninfected and infected conditions as well as in control mice. As expected, infection led to high mucosal influx of all analyzed immune cells, predominantly in the proximal antrum. Interestingly in infected Lgr4 −/− mice, neutrophil infiltration was significantly diminished, indicating that a lack of proper epithelial Lgr signaling has a functional consequence with regard to the recruitment of inflammatory cells (Fig 5A and B). Recruitment of macrophages and T cells was not significantly altered in these mice, which is in line with our similar observations in Rspo3 KO mice (Sigal et al, 2019) and may point toward a distinct immune cell infiltration pattern with various influence factors in gastritis that warrants further investigation (Fig EV4A and B). Equivalent analysis of infected Lgr5 −/− mice failed to detect altered immune cell infiltration (Figs 5A and B, and EV4A and B).

Figure 5

Lgr4 controls mucosal neutrophil infiltration

Immunofluorescence labeling for MPO in uninfected and 2‐month‐infected Lgr4 −/− and control mice (top panel) and 2‐month‐infected Lgr5 −/− and control mice (bottom panel). Asterisks indicate nonspecific staining. Scale bars: 50 μm.

Quantification of MPO+ cells in uninfected and 2‐month‐infected Lgr4 −/− and control mice (n = 2 biological replicates per group, n = 5–10 FOVs per mouse) (left panel) and 2‐month‐infected Lgr5 −/− and control mice (n = 2 biological replicates per group, n = 6–8 FOVs per mouse) (right panel).

Data information: For (B) data are mean + SD. **P < 0.01; ***P < 0.001; ns, not significant (one‐way ANOVA + Tukey's multiple comparison test).

Source data are available online for this figure.

Figure EV4

Lgr is dispensable for macrophage‐ and T cell infiltration

Immunofluorescence labeling for IBA1 and CD3 in uninfected and 2‐month‐infected Lgr4 −/− and control mice (top panel) and 2‐month infected Lgr5 −/− and control mice (bottom panel). Asterisks indicate nonspecific staining. Scale bars: 50 μm.

Quantification of IBA1+ and CD3+ cells in uninfected and 2‐month‐infected Lgr4 −/− and control mice (n = 2 biological replicates per group, n = 5–12 FOVs per mouse) (left panel) and 2‐month‐infected Lgr5 −/− and control mice (n = 2 biological replicates per group, n = 5–10 FOVs per mouse) (right panel).

Data information: For (B) data are mean + SD. *P < 0.05; **P < 0.01; ****P < 0.0001; ns, not significant (one‐way ANOVA + Tukey's multiple comparison test).

Source data are available online for this figure.

Stem cells are the primary epithelial sensors of ADP heptose and promote pro‐inflammatory signaling

To discriminate between effects by immune cells and direct epithelial responses, we took advantage of the epithelial gastric organoid cultures. Gastric organoids were grown in the standard full medium (FM) containing Rspo and Wnt, which resembles the gland base environment and drives stem cell turnover. To test whether these growth factors affect the epithelial ability to mediate pro‐inflammatory responses, some organoids were cultured in medium without Rspo, without Wnt and Rspo, or with the addition of Bmp2, which rapidly induces pit cell differentiation. Organoids grown in FM showed robust growth, while removal of growth factors or addition of Bmp2 led to reduced growth and a more spherical appearance (Fig 6A). Immunofluorescence analysis revealed that organoids grown in FM formed small invaginations, which corresponded to GSII+ gland base cells. As expected, upon removal of growth factors, invaginations and GSII+ cells were lost, while more MUC5AC+ pit cells were detected (Fig 6B). Expression of the stem cell marker Lgr5, as well as the gland base mucous cell marker Gif, was strongly downregulated in all other conditions compared with stem cell‐enriched organoids (FM), as analyzed by qPCR (Fig 6C and D).

Figure 6

Stem cells are the primary epithelial sensors of ADP heptose and promote pro‐inflammatory signaling

Antrum organoids from wildtype mice were grown in full medium (FM) for 2–3 days, followed by respective medium conditions for 3–5 days (FM, −Rspo, −Rspo/‐Wnt, +BMP2). 0.5 μM ADP heptose or 10 ng/ml TNF‐α was added for 2 h to all organoid conditions prior to analysis.

Bright‐field images of organoids in respective conditions before treatment with ADP heptose or TNF‐α. Scale bars: 200 μm.

Immunofluorescence labeling for GSII and Muc5ac of organoids in respective conditions before treatment with ADP heptose or TNF‐α. Scale bars: 50 μm.

qPCR expression level of Lgr5 in untreated organoids in respective medium conditions (n = 4–5 biological replicates).

qPCR expression level of Gif in untreated organoids in respective medium conditions (n = 4–5 biological replicates).

qPCR expression level of Cxcl1, Cxcl2, Cxcl10, and Ccl2 in organoids after treatment with ADP heptose compared with untreated controls within the respective medium conditions (n = 6 biological replicates per condition).

qPCR expression level of Cxcl1, Cxcl2, Cxcl10, and Ccl2 in organoids after treatment with TNF‐α compared with untreated controls within the respective medium conditions (n = 2–3 biological replicates per condition).

Data information: For (C, D, E, and F) data are mean + SD. *P < 0.05; **P < 0.01; ****P < 0.0001 (one‐way ANOVA + Tukey's multiple comparison test for C and D; t‐test for E and F). Genes without asterisks are not significantly altered.

Source data are available online for this figure.

To investigate the mechanism by which NF‐κB could be induced upon H. pylori infection, we treated gastric organoids with ADP heptose, a bacterial sugar, which was previously shown to induce NF‐κB in the context of H. pylori infection (Pfannkuch et al, 2019). We thus exposed organoids grown in the different medium conditions to ADP heptose for 2 h, to examine their ability to activate NF‐κB. The inflammatory response was measured via qPCR of Cxcl1 and Cxcl2, which are classic murine ADP heptose readouts (Brooks et al, 2020) as well as Ccl2 and Cxcl10, which we found to be upregulated in an Rspo3/Lgr4‐dependent manner after H. pylori infection in vivo. ADP heptose triggered NF‐κB target gene expression in organoids grown in FM, while organoids grown in the absence of Rspo and Wnt signaling remained unresponsive (Fig 6E). Notably, the absence of Rspo alone already prevented full activation of NF‐κB signaling, demonstrating the critical role of Rspo signaling for maintaining the ability of the cells to induce epithelial pro‐inflammatory signaling. To address whether these responses are limited to ADP heptose or also affect other pro‐inflammatory pathways, we treated organoids with TNF‐α. We found similar effects of growth factor withdrawal, indicating that NF‐κB activation in gland base cells is not limited to ADP heptose‐induced signaling and that activation via the classical TNF‐α pathway also depends on Rspo signaling (Fig 6F).

To determine whether this phenotype is indeed due to the activation of NF‐κB, we quantitated DNA binding activity of the transcription factor by electrophoretic mobility shift assay (EMSA) in organoids treated as above (Fig 7A and B). We observed baseline activity of NF‐κB in dependence of Rspo and demonstrated that the presence of Rspo is necessary for the pronounced induced activation of NF‐κB and of its upstream kinase, IKK (IκB kinase). Removal of Rspo or Wnt led to a partial loss of a bona fide target of NF‐κB, IκBα (Nfkbia). Supershift analysis in FM organoids + ADP heptose with antibodies against the five subunits confirmed p65 as the principal mediator (Fig 7C). These findings confirm that induction of the canonical IKK‐NF‐κB pathway in response to ADP heptose in gastric epithelium requires functional Wnt/ Rspo signaling.

Figure 7

Activation of IKK‐NF‐κB pathway by ADP heptose requires Wnt and Rspo

Antrum organoids from wildtype mice were grown in full medium (FM) for 2–3 days, followed by respective medium conditions for 3–5 days (FM, −Rspo, −Rspo/‐Wnt, +BMP2). 0.5 μM ADP heptose was added for 2 h to all organoid conditions prior to analysis.

Electrophoretic mobility shift assay (EMSA) and SDS–PAGE western blot from whole‐cell protein (WCL) lysates from gastric organoids untreated or treated with ADP heptose, and maintained either in FM in medium without Rspo, without Wnt and Rspo, or + BMP2. Representative images from n = 3 biological replicates.

Quantification of EMSA based on n = 3 biological replicates using ImageJ with normalization to loading (ɑ‐Tubulin).

EMSA supershift from FM + ADP heptose organoids with the antibodies indicated. Blue arrow indicates NF‐κB binding. Red arrow points to gel expected locations of supershifts from antibodies against p65, cRel, or RelB. Gray arrow points to supershift location on the gel from antibodies against p52 or p50. IgG lane serves as a negative control.

Data information: For (B) data are mean + SD. Significance is calculated by one‐way ANOVA, F = 9.21.

Source data are available online for this figure.

To determine which cells in ADP heptose‐treated organoids harbor nuclear p65, we stained organoids treated as above with antibodies against total p65 and Ki67 or GSII (Fig EV5A). Nuclear translocation of p65 was detected in nuclei of Ki67+ and GSII+ cells in organoids in FM treated with ADP heptose. We confirmed our findings by generating gastric organoids from a NF‐κB reporter mouse line, κ‐EGFP (Brischetto et al, 2021). In epithelium, activated NF‐κB localized to Ki67+ and partially to GSII+ cells and was nearly absent in organoids without Rspo or Wnt or in organoids treated with Bmp2. In the latter conditions, NF‐κB activity was not observed upon ADP heptose treatment, indicating that Rspo/Wnt are necessary for an activation of NF‐κB (Fig EV5B and C). Taken together, we find that NF‐κB activation and subsequent chemokine expression occurs in epithelial gland base cells in an Rspo‐dependent manner.

Figure EV5

Rspo promotes NF‐κB activity

WT antrum organoids were grown in full medium (FM) for 6–8 days. 0.5 μM ADP heptose was added for 2 h to the culture. Immunofluorescence labeling for p65, Ki67 and GSII in treated and control organoids. Treated organoids express nuclear p65 in Ki67+ and partially GSII+ cells. Insets show magnified images. Arrows in organoids treated with ADP heptose indicate nuclear p65 that is not observed in untreated organoids. Scale bars: 50 μm.

Antrum organoids from NF‐κB reporter mice using EGFP expression as a readout for NF‐κB activity were grown in full medium (FM) for 2–3 days, followed by respective medium conditions for 3–5 days (FM, −Rspo, −Rspo/‐Wnt, +BMP2). Immunofluorescence labeling for EGFP, Ki67 and GSII of organoids in respective conditions treated with ADP heptose and untreated controls. Asterisks indicate diffuse, nonspecific staining. Scale bars: 50 μm.

Quantification of EGFP+ cells as percentage of total DAPI‐stained cells in respective conditions (n = 4–12 organoids, n = 2 biological replicates).

Data information: For (A) images represent findings that were reproduced at least twice in the laboratory in independent biological replicates. For (C) data are mean + SD. *P < 0.05; ***P < 0.001; ****P < 0.0001; ns, not significant (one‐way ANOVA + Tukey's multiple comparison test).

Source data are available online for this figure.

NF‐κB signaling is important for gland base secretory cell differentiation

As we found that hyperplasia is triggered by Rspo3‐Lgr signaling via base cell and isthmus cell expansion, we asked whether activated NF‐κB, which is predominantly present in isthmus cells, contributes to the maintenance of one of these cell types. Therefore, we obtained mice with ubiquitously suppressed NF‐κB activity, by using a transdominant negative mutant of IkBa (IkBaΔN) that was inserted in‐frame into the beta‐catenin locus (cI kappa B alpha Delta N, hereafter referred to as ΔN). Importantly, the beta‐catenin protein expression and activity in these mice is not impaired (Schmidt‐Ullrich et al, 2001). Strikingly, qPCR on antral organoids (Fig 8A) derived from these mice revealed not only reduced expression of NF‐κB target genes, but also a strong reduction in the expression of gland base cell markers (Fig 8B and C), indicating that NF‐κB activity is not only important for promoting inflammation but is in fact an integral factor determining gland base cell identity. Notably, Lgr4 was higher in ΔN organoids, pointing to a compensatory mechanism upon base cell loss, which is not sufficient to maintain their identity (Fig 8B). Analysis of gastric tissue sections from ΔN mice showed that there was also a significant reduction in the number of GSII+ secretory gland base cells in vivo, but not of proliferating cells (Fig 8D and E). We conclude that NF‐κB signaling in the stem cell compartment is important for shifting cell differentiation in the base toward secretory cells, which could be an explanation for their accumulation upon H. pylori infection.

Figure 8

NF‐κB activity defines gland base compartment in the antrum

Bright‐field images of antrum organoids from 129;129P2‐ctnnb1 (ΔN) mice and control mice, passage 0, day 6. Scale bars: 500 μm.

qPCR expression levels of Lgr4, Lgr5, Gif, Pgc, Muc6 (B) and Cxcl1, Cxcl2 (C) of organoids (n = 3 biological replicates).

Representative H&E staining and immunofluorescence labeling for GSII, Ki67 and MUC5AC in uninfected ΔN and control mice. Scale bars: 50 μm.

Quantification of GSII and Ki67 as percentage of total DAPI‐stained cells in ΔN and control mice (n = 5 biological replicates per group, n = 20 glands per mouse).

Data information: For (B, C, and E) data are mean + SD. *P < 0.05; **P < 0.01; ns, not significant (t‐test).

Source data are available online for this figure.

Discussion

Here, we demonstrate that the stem cell compartment in the stomach antrum is uniquely equipped to respond to infection and initiate mucosal inflammation. Our data reveal a link between Rspo‐Lgr signaling and NF‐κB, which enables the gland base compartment to rapidly initiate a pro‐inflammatory response and simultaneously structurally adapt to H. pylori infection.

Most investigations of epithelial homeostasis that explore Rspo‐Wnt signaling axis put a focus on Lgr5, which was originally identified as a marker of intestinal and later of gastric stem cells (Barker et al, 2007, 2010). Lgr5 is a Wnt target gene that requires high levels of Wnt signaling (Segditsas et al, 2008). Such high Wnt activity depends in vivo in the stomach on the Wnt signaling enhancer Rspo3 produced by stromal myofibroblasts (Sigal et al, 2017; Harnack et al, 2019). Rspo molecules bind to Lgr4 and Lgr5 and prevent turnover of the Wnt receptor Frizzled (de Lau et al, 2011). Thus, Rspo signaling itself drives expression of its own receptor Lgr5, implying that a positive autocrine feedback loop that stabilizes Wnt/ Rspo signaling could exist (Yan et al, 2017). In contrast, unlike Lgr5, Lgr4 is not a Wnt target gene and broadly expressed in the epithelium of the gastrointestinal tract, including the stomach (Sigal et al, 2017; Harnack et al, 2019). It has been shown that Lgr5 and Lgr4 act in concert to maintain homeostasis in the intestine (de Lau et al, 2011). While loss of Lgr5+ cells does not impact epithelial homeostasis and injury repair (Tian et al, 2011; Castillo‐Azofeifa et al, 2019), loss of Lgr4, similar to loss of Rspo3, renders the epithelium more susceptible to injury (Liu et al, 2013; Greicius et al, 2018; Harnack et al, 2019). Depletion of Lgr5+ cells is followed by their recovery in the intestine and stomach, and this process has been shown to require Rspo3, which is likely through interaction with Lgr4 (Sigal et al, 2017; Harnack et al, 2019). Our data here demonstrate that Lgr4 represents a critical receptor for mucosal responses to H. pylori in the antrum. Lgr4 also controls Lgr5 expression; however, Lgr5 alone does not impact H. pylori infection, indicating that while Lgr5+ cells are critical in the context of infection, their maintenance and function is not driven by Lgr5 alone, but requires Lgr4 expression. Therefore, the Rspo‐Lgr4 axis is highly relevant for epithelial responses, while Lgr5 expression is a marker of Rspo signaling, but is largely dispensable for Rspo effects.

Helicobacter pylori colonizes the stomach of about 50% of the world's population and is the main risk factor for gastric cancer (Parsonnet et al, 1997). Infection causes a chronic active gastritis, characterized by mucosal infiltration of immune cells such as neutrophils (Eck et al, 2000), which can progress to atrophic or hyperplastic gastritis, metaplasia and, in about 1% of patients, gastric cancer (Correa, 1995). Neutrophils are associated with mucosal damage in H. pylori gastritis (Ernst et al, 1997; Shimoyama & Crabtree, 1998). It is well established that the inflammatory response to infection is linked to epithelial pathology and mouse experiments have demonstrated that the degree of immune responses against H. pylori correlates with the degree of epithelial premalignant pathology, while the degree of inflammation correlates negatively with H. pylori colonization (Wunder et al, 2006; Arnold et al, 2011). The link between infection, inflammation, and epithelial pathology is not fully resolved. We have previously observed that colonization of gastric glands (in contrast to colonization of the surface pits) triggers inflammation, gland hyperplasia as well as metaplasia (Sigal et al, 2015). These responses result in an accumulation of antimicrobial gland base secretory cells, which limit gland colonization but fail to clear the infection (Sigal et al, 2019). Here, we demonstrate and mechanistically resolve the unique ability of the epithelial gland base module to initiate a pro‐inflammatory NF‐κB response that promotes an accumulation of gland base cells and simultaneously induces an upregulation of epithelial chemokines leading to mucosal neutrophil infiltration that can initiate and maintain a chronic active gastritis. These data provide an explanation for selective pro‐inflammatory responses to gland‐associated bacteria and link epithelial and inflammatory responses observed upon infection.

It is likely that this unique ability of gland and crypt base cells is not restricted to H. pylori infection in the stomach. Glands and crypts show a similar organization throughout the gastrointestinal tract, with stem cells located in the base. It will be interesting to investigate mechanisms of Rspo‐NF‐κB interactions in response to intestinal pathogens. While crypt colonization may be the most beneficial niche for chronic bacterial colonization as it may provide nutritional and physical benefits for colonization (Keilberg et al, 2016), bacterial penetration into this protected compartment that harbors stem cells is likely highly detrimental for the host. Indeed, several pathogens have evolved to target this compartment. In the colon Shigella flexneri reach the bottom of crypts to target stem cells leading to long‐term repercussions for tissue regeneration (Arena et al, 2015). Moreover, Vibrio cholerae in the small intestine and Campylobacter jejuni in the cecum establish crypt colonization by directed and efficient swimming. Vibrio cholerae infect intestinal progenitor cells with a type VI secretion system blocking growth and repair pathways in addition to extensive tissue damage upon infection (Fast et al, 2020). Sensing of bacteria and the ability to initiate inflammation upon gland base cell colonization is, therefore, likely to be a conserved property of gland and crypt base cells, enabling a rapid response to these undesired events.

Progenitor cells in the gland are, therefore, not only responsible for gland repopulation but also play a key role in mucosal immunity. Modulation of gastrointestinal stem cell behavior by immune‐derived signals has been shown previously. Stem cells in the small intestine express NOD2 receptors, and microbiota‐derived peptidoglycans that activate NOD2 support stem cell survival (Nigro et al, 2014). T cell‐derived cytokines have also been shown to modulate crypt differentiation, enabling epithelial adaptation to infectious stimuli (Biton et al, 2018). Here, we demonstrate that antral base cells not only respond to immune signals but act as critical components of the innate immune system by initiating responses to pathogenic bacteria via NF‐κB signaling.

Molecular links between NF‐κB and Wnt signaling in carcinogenesis have been established using genetic cancer models (Greten & Grivennikov, 2019), for example, in Wnt‐induced intestinal cancer, enhanced NF‐κB activity promotes carcinogenesis, while loss of NF‐κB results in prolonged survival and diminished tumor growth (Schwitalla et al, 2013). Constitutive Wnt activation following APC loss leads to hyperproliferation of stem cells and tumor formation via RAC1 (Ras‐related C3 botulinum toxin substrate 1)‐ROS‐NF‐κB pathway activation (Myant et al, 2013). In the small intestine, NF‐κB is required for proper secretory Paneth cell differentiation and epithelial homeostasis by regulating Wnt signaling and Sox9 expression downstream of NF‐κB (Brischetto et al, 2021). Recently, activation of NF‐κB via ALPK1 and TIFA in proliferative intestinal cells has been linked to DNA damage and abrogated replication stress, which predisposes to mutagenesis (Bauer et al, 2020). Our data now reveal that an interplay between these two signaling pathways is in fact an integral part of antral gland physiology. Under healthy conditions, this signaling unit is restricted to the stem cell compartment. This restriction is the basis of selective immune responses if the gland becomes infected. In contrast, mutations in the Wnt signaling pathway may result in an expansion of cells with high Wnt signaling, and thus NF‐κB competence, to the gland surface, where cells are more likely to be exposed to various signals from luminal bacteria. This would in turn lead to a loss of biogeographic selectivity and initiation of inflammation even by surface epithelial cells, which in turn would perpetuate the regenerative process and may contribute to tumor cell proliferation and progression.

Materials and Methods

Mouse experiments

All procedures that involved animals were approved by the institutional, local, and national legal authorities at the Max Planck Institute for Infection Biology (LAGeSo, Berlin, Reg. G0084/17) and were compliant with all of the relevant ethical regulations regarding animal research. Axin2CreErt2 mice were described before (van Amerongen et al, 2012) and were obtained from the Jackson Laboratories. Lgr4 fl/fl mice and Lgr5 fl/fl mice were kindly provided by Dr. Jan Tchorz, Novartis. Mouse models were designed for both genes by flanking exon1 with loxP elements facilitating Cre‐mediated recombination and excision of exon1 including part of the 5′UTR. For Lgr4, Cre‐mediated excision results in a deletion of 1.9 kb including Lgr4 exon1 plus 900 bp of 5′UTR. For Lgr5, Cre‐mediated excision results in a deletion of 1.6 kb including Lgr5 exon1 plus 600 bp of the 5′UTR (Planas‐Paz et al, 2016). Rspo3 fl/fl mice (Neufeld et al, 2012) were a kind gift from John Cobb. Myh11 CreErt2 mice (Herring et al, 2014) were a kind gift from Stefan Offermanns. Mice that conditionally overexpress Rspo3 under the Rosa26 promoter (Rosa26Sor 6(Cag‐Rspo3)) were described previously (Hilkens et al, 2017) and were a gift from John Hilkens and Elvira Bakker. These animals were bred with Myh11 CreERT2 mice to generate double‐heterozygous Myh11 CreERT2/Rosa26Sor 6(CAG‐Rspo3) mice. 129;129P2‐ctnnb1 mice (ΔN) (Schmidt‐Ullrich et al, 2001) and B6‐Tg(k‐EGFP)3Pt/Rsu (k‐EGFP) (Tomann et al, 2016) were a kind gift from Ruth Schmidt‐Ulrich.

All animals were maintained in autoclaved microisolator cages and provided with sterile drinking water and chow ad libitum. Six‐to‐eight‐week‐old male mice were used for this study. Sample size was estimated by own experience in this field (Sigal et al, 2017, 2019). Tamoxifen (Sigma) was injected intraperitoneally into mice on two to three consecutive days (4 mg 25 g−1 bodyweight; diluted in 200 μl corn oil) at the indicated time points.

At the time of collection, the forestomach was removed, and the glandular stomach was opened along the lesser curvature and laid flat. The stomach contents were removed, and the tissue was divided into two halves along the greater curvature. For RNA analysis, the tissue was divided between antrum and corpus, and pieces were snap‐frozen. For microscopy analysis, a similar longitudinal section at the midline along the greater curvature was used for all animals to minimize sampling error. For all mouse experiments, mice were randomized to experimental groups.

Helicobacter pylori infection and CFU analysis

Animals were infected with a single oral dose of 1 × 108 H. pylori strain PMSS1 and euthanized at the indicated time points as previously described (Sigal et al, 2015). Longitudinal sections of stomach tissue (antrum and corpus) were weighed and then mechanically homogenized in brain‐heart‐infusion medium. Serial dilutions of the homogenates were plated for H. pylori CFU enumeration, and bacterial counts were expressed as CFU g−1 of stomach. Lgr4−/− and Lgr5−/− mice plus littermate controls were infected with separate H. pylori cultures leading to variability in CFUs between two different experiments. Experiments were performed in at least three biological replicates per condition. The investigator was blinded for CFU analysis.

Tissue processing and microscopy

Tissue samples were processed for microscopy as previously described, with minor modifications (Sigal et al, 2015). For paraffin embedding, tissues were fixed in 2% paraformaldehyde for 1 day. Paraffin embedding and sectioning, as well as H&E staining, were performed by the Charité Core Unit Immunopathology for Experimental Models. Paraffin‐embedded sections were rehydrated, followed by an antigen retrieval step, then incubated with primary antibodies, followed by a washing step, and incubated with secondary antibody for 1 h. Samples were imaged using a Leica Sp8 confocal microscope (LAS X acquisition software) or a Zeiss Observer 7 microscope (ZEN acquisition software). The investigator performing staining was blinded for treatment and genotype.

The following antibodies were used for this study: rabbit anti‐GIF (custom‐made by David H. Alpers); mouse anti‐Muc5ac (Invitrogen, MA5‐12178, lot RL2313146B); rabbit anti‐Ki67 (Cell Signaling, 9129S, clone D3B5, lot 3); rabbit anti‐NF‐κB p65 (Cell Signaling, D14E12, #8242, lot 16); mouse anti‐E‐Cadherin (BD Biosciences, 610,181); DAPI (Roche, Cat. No. 10236276001); goat anti‐GFP (Abcam, ab6673); rabbit anti‐CD3 (Abcam, ab16669); rabbit anti‐IBA1 (VWR, 100369–764); rabbit anti‐MPO (Abcam, ab208670); mouse anti‐beta‐catenin (Becton Dickenson, 610,153); rabbit anti‐H. pylori (custom‐made and described previously (Sigal et al, 2015)); Alexa Fluor 647 Phalloidin (Invitrogen, A22287); rabbit anti‐ NF‐κB p65 acetyl K310 (Cell Signaling, #3045); rabbit anti‐Cxcl1 (Ptglab, 12335). GSII+ cells were stained with Lectin GSII from Griffonia simplicifolia, Alexa Fluor 647 conjugate (Life Technologies, L32451, lot 1774229), nuclei were counterstained with DAPI (Roche, Cat. No. 10236276001).

Organoid cultures

Antrum tissue was sterilized in 0.04% (w/v) sodium hypochlorite (Roth) for 15 min at room temperature, followed by 90‐min incubation in buffered saline solution containing 0.5 mM DTT and 3 mM EDTA to dissociate gastric glands. Isolated glands were washed with PBS and mixed with 50 μl Matrigel (BD) and plated in 24‐well plates. After polymerization of Matrigel, murine gastric culture medium (Advanced Dulbecco's Modified Eagle Medium/F12 supplemented with B‐27 (Invitrogen), N‐2 (Invitrogen), N‐acetyl cysteine (Sigma), 100 U/ml penicillin–streptomycin (Invitrogen) and containing 50 ng ml−1 epidermal growth factor (Invitrogen), 100 ng ml−1 Noggin (Peprotech), 100 ng ml−1 fibroblast growth factor 10 (Peptrotech), 10 μM gastrin (Sigma), 7.5 μM Y‐27632 (Sigma), Rspo1‐conditioned medium and Wnt3A‐conditioned medium) was overlaid. The medium supplemented with growth factors was replaced every 2–3 days. After 2–3 days in culture, for some experiments Wnt3a and/or Rpso1 were removed or 50 ng ml−1 BMP2 (R&D Systems) was added for a further 3–5 days. Subsequently, organoids were treated with 0.5 μM ADP heptose (Invivogen) or 10 ng/ml TNF‐α (Peprotech) for 2 h. For qPCR analysis, RNA from the organoids was then extracted as described below. For immunofluorescence, organoids were fixed with 3.7% formaldehyde for 1–2 h at room temperature and embedded in paraffin as described before (Grinat et al, 2022).

Microarray analysis

For microarray analysis, RNA of infected Lgr4 −/− mice and the corresponding controls treated with tamoxifen 2 weeks before euthanasia were isolated from the antrum. Total RNA from bulk tissue was isolated using the RNEasy mini kit (Qiagen) following the manufacturer's protocol.

Microarray experiments were performed as independent dual‐color dye hybridizations using two biological replicates. Quality control and quantification of total RNA were carried out using an Agilent 2100 Bioanalyzer (Agilent Technologies) and a NanoDrop 1,000 UV–Vis spectrophotometer (Kisker). RNA labeling was performed using the dual‐color Quick‐Amp Labeling Kit (Agilent Technologies). In brief, mRNA was reverse transcribed and amplified using an oligo‐dT‐T7 promoter primer, and the resulting complementary RNA (cRNA) was labeled with cyanine 3‐CTP or cyanine 5‐CTP. After precipitation, purification, and quantification, 1.25 μg of each labeled cRNA was fragmented and hybridized to whole mouse genome 4 × 44 K multipack microarrays (Agilent‐014868, whole mouse genome 4 × 44 K microarray kit) according to the supplier's protocol (Agilent Technologies). Scanning of microarrays was performed with 5 μm resolution using a G2565CA high‐resolution laser microarray scanner (Agilent Technologies) with extended dynamic range. Microarray image data were analyzed using the Image Analysis/Feature Extraction software G2567AA v.A.11.5.1.1 (Agilent Technologies) using the default settings and the GE2_1105_Oct12 extraction protocol. The extracted MAGE‐ML files were analyzed further using the Rosetta Resolver Biosoftware, build 7.2.2 SP1.31 (Rosetta Biosoftware). Ratio profiles comprising single hybridizations were combined in an error‐weighted manner to create ratio experiments. A 1.5‐fold‐change expression cut‐off for ratio experiments was applied together with anti‐correlation of dye‐swapped ratio profiles, yielding microarray analysis results that were highly significant (P < 0.01), robust, and reproducible (Churchill, 2002).

GSEA analysis

We used antral stem cell signature genes (Barker et al, 2010), beta‐catenin target genes (Herbst et al, 2014), and NF‐κB signature genes from the MsigDB (Liberzon et al, 2015) on genes pre‐ranked by gene‐expression‐based t‐score between the antrum tissue of Lgr4 fl/fl /Axin2 CreErt2 and Lgr4 +/+ /Axin2 CreErt2 animals using the fgsea R package (preprint: Sergushichev, 2016) with 5,000 permutations; R v. 4.0.2 was used (https://cran.r-project.org/). For each experiment, P‐values from all gene sets were jointly adjusted for multiple testing using FDR—the method of Benjamini–Hochberg (Benjamini & Hochberg, 1995). For further information regarding GSEA and data interpretation, please refer to the original publication (Subramanian et al, 2005).

The computational code for GSEA analysis of microarray in this manuscript data can be accessed under https://github.com/Sigal-Lab/Wizenty_et_al_Lgr4_stomach_gland.

DAVID and Revigo analysis

Genes showing significant negative changes in expression (P < 0.05) were analyzed by DAVID 6.7 BP direct (https://david.ncifcrf.gov). For NF‐κB target genes, Gilmore database (https://www.bu.edu/nf-kb/gene-resources/target-genes/) was used to select genes showing a significant decrease in expression in Lgr4 KO (P < 0.05). Separate GO terms were obtained for these from DAVID 6.7 BP direct. GO terms were refined by Revigo (Uniprot mus musculus, Resnik Large; Supek et al, 2011).

RT–PCR

For qPCR from mouse tissue, RNA was extracted from the snap‐frozen antrum tissue using the RNeasy RNA Purification Kit (Qiagen). For the analysis of organoid RNA, organoids were released from Matrigel with cold Dulbecco's phosphate‐buffered saline (DPBS) and pelleted by centrifugation (7 min at 4°C, 300 g), and RNA was then isolated using the TRIzol (Invitrogen) and the RNeasy RNA Purification Kit according to the manufacturer's protocol. Quality and quantity controls were performed for each extract using the NanoDrop 1,000 Spectrophotometer (Thermo Scientific). qPCR was performed using the Power SYBR Green RNA‐to‐CT 1‐Step Kit (Applied Biosystems). Reactions were performed in 25 μl containing 50 ng RNA, 12.5 μl SYBR Green mix, 0.2 μl RT mix, and 0.2 μM per primer.

The cycles used were as follows: 30 min at 48°C; 10 min at 95°C; followed by 40 cycles of 15 s at 95°C; and 60 s at 60°C. For each oligonucleotide pair and RNA sample, the reaction was performed in duplicate. The expression levels of the target genes were normalized to the levels of Gapdh gene expression in each individual sample and expressed as fold‐change compared with the expression in control tissue (WT, uninfected or nontreated), using the 2−ΔΔCt equation (Pfaffl, 2001).

The following murine primers (Sigma) were used: Axin2 forward 5′‐TGACTCTCCTTCCAGATCCCA‐3′, reverse 5′‐TGCCCACACTAGGCTGACA‐3′; Ccl2 forward 5′‐CAGGTCCCTGTCATGCTTCT‐3′, reverse 5′‐GTGGGGCGTTAACTGCATCT‐3′; Cxcl1 forward 5′‐CCCAAACCGAAGTCATAGCCA‐3′, reverse 5′‐CTCCGTTACTTGGGGACACC‐3′; Cxcl2 forward 5′‐CCCAGACAGAAGTCATAGCCAC‐3′, reverse 5′‐CTTCCGTTGAGGGACAGCAG‐3′; Cxcl10 forward 5′‐CAAGCCATGGTCCTGAGACA‐3′, reverse 5′‐TGAGCTAGGGAGGACAAGGA‐3′; Gapdh forward 5′‐TCACCATCTTCCAGGAGCG‐3′, reverse 5′‐AAGCAGTTGGTGGTGCAGG‐3′; Gif forward 5′‐AAGCACAGCGCAAAAACTCC‐3′, reverse 5′‐GCAACCCCTTCATCCAAAGG‐3′; Itln1 forward 5′‐GTTGCTACCAGAGGTTGCAGT‐3′, reverse 5′‐AGACCATCTTGTGCCTTTGTGT‐3′; Lgr4 forward 5′‐TACAACTGGCTGGTAACGACC‐3′, reverse 5′‐GGAGGGTTAGGACTTTGAGTTCT‐3′; Lgr5 forward 5′‐CCTACTCGAAGACTTACCCAGT‐3′, reverse 5′‐GCATTGGGGTGAATGATAGCA‐3′; Muc5ac forward 5′‐ CCTGAGGGTATGGTGCTTGA‐3′, reverse 5´ TGTGTTGGTGCAGTCAGTAGAG3´; Muc6 forward 5′‐CAGCTCAACAAGGTGTGTGC‐3′, reverse 5′‐GGTCTCCTCGTAGTTGCAGG‐3′; Pgc forward 5′‐CCCATGGCCTATATGGATGCTT‐3′, reverse 5′‐CAGGCCGAACTCCTGGTTAG‐3′; Sox9 forward 5´‐GAGCCGGATCTGAAGAGGGA‐3′, reverse 5´GCTTGACGTGTGGCTTGTTC‐3′; Tff2 forward 5′‐CTGGTAGAGGGCGAGAAACC‐3′, reverse 5′‐CCATGACACACTGCTCCGAT‐3′.

Electrophoretic mobility shift assay (EMSA) and sodium dodecyl sulfate‐polyacrylamide gel electrophoreses (SDS–PAGE)

The gel electrophoresis mobility shift assay (EMSA) is used to detect protein complexes with nucleic acids. Here, EMSA was used to determine the affinity of NF‐κB to the specific sequence on DNA (the κB sequence). EMSA provides a readout of the activation state of the NF‐κB pathway. The 32p radiolabeled DNA probe containing the κB sequence was incubated with the protein lysate and then separated on a gel.

EMSA was performed with whole‐cell lysates (WCL). Whole‐cell lysates were prepared using Baeuerle buffer (20 mM HEPES pH 7.9; 350 mM NaCl; 20% glycerol; 1 mM MgCl2; 0.5 mM EDTA; 0.1 mM EGTA; 1% NP‐40), including complete protease inhibitor cocktail (Roche), 10 mM NaF, 8 mM β‐glycerophosphate, 0.2 mM Na3VO4, and 1 mM DTT. H2KκB probe from MHC promoter was used and incubated with 5 μg of whole‐cell lysate for 30 min. The protein and DNA was then separated on a non‐denaturing TBE‐polyacrylamide gel over 90 min. The gel was vacuum dried, and signal was visualized on an X‐ray film.

EMSA protocol and SDS–PAGE protocol was described previously (Mikuda et al, 2018). Supershift protocol was described previously (Kolesnichenko et al, 2021).

The following antibodies were used for this study: IκBα C‐15 (Santa Cruz, sc‐203), IKK‐p (S176‐180) (Cell Signaling, #2697), alpha‐tubulin DM1A (Santa Cruz, sc‐ 32,293), p65 C‐20 (Santa Cruz, sc‐372 X), p50 E‐10 (Santa Cruz, sc‐8,414 X), RelB D‐4 (Santa Cruz, sc‐48,366 X), p52 (Millipore, 06–413), and cRel (Cell Signaling, #4727).

Single‐molecule RNA ISH

Tissue sections cut at 5 μm thickness were processed for RNA in situ detection using the RNAscope Red Detection Kit according to the manufacturer's instructions (Advanced Cell Diagnostics). Positive and negative control probes were used for each experiment according to the manufacturer's instructions. The following probes were used in this study: Lgr4 (Order no.: 318321, Lot: 18053A, Target region: 1973–2895); Lgr5 (Order no.: 312171, Lot: 16351A, Target region: 2165–3082).

ScRNA‐seq analysis

ScRNA‐seq data from epithelial cells of murine gastric antrum were used from a published data set (Busslinger et al, 2021b). GEO accession number for the source data is GSE157694 (Data ref: Busslinger et al, 2021a). Seurat R package (Butler et al, 2018) was used to perform t‐distributed stochastic neighbor embedding (t‐SNE) analysis. t‐SNE analysis was applied to visualize cell clusters. Datasets were filtered (> 50 transcripts, < 5% mitochondrial reads per cell), integrated with SCTransform to eliminate the batch effect of samples, and log‐normalized, and the first 20 principal components were used to reduce dimensionality by t‐SNE with a resolution of 0.1.

Statistics

GraphPad Prism (v.8.4.0) was used for statistical analysis. Data are displayed as mean + SD. P < 0.05 was considered to be statistically significant.

For comparison of two groups, we used an unpaired two‐sided t‐test, for multiple comparisons one‐way ANOVA as indicated. Data are displayed for independent biological samples, unless otherwise indicated. The numbers are indicated in the figure legends. Microarray data displayed in Figs 3 and EV4, and Table EV1 are derived from n = 2 biological replicates.

Author contributions

Jonas Wizenty: Conceptualization; data curation; software; formal analysis; validation; investigation; visualization; methodology; project administration. Stefanie Müllerke: Investigation. Marina Kolesnichenko: Software; formal analysis; investigation. Julian Heuberger: Investigation. Manqiang Lin: Investigation. Anne‐Sophie Fischer: Investigation. Hans‐Joachim Mollenkopf: Software; formal analysis. Hilmar Berger: Software; formal analysis; visualization. Frank Tacke: Resources; supervision. Michael Sigal: Conceptualization; resources; data curation; software; formal analysis; supervision; funding acquisition; validation; investigation; visualization; methodology; project administration.

In addition to the CRediT author contributions listed above, the contributions in detail are:

This study was designed by JW and MS. JW and SM performed experiments and analysis. SM performed animal experiments. MK, JH, ML, and A‐SF performed experiments and provided intellectual support. H‐JM performed microarray studies and HB, H‐JM, and MK analyzed the microarray data. FT gave advice for experimental design and data interpretation. The manuscript was written by JW and MS; all authors provided comments.

Disclosure and competing interests statement

The authors declare that they have no conflict of interest.

Expanded View Figures PDF

Click here for additional data file.

Table EV1

Click here for additional data file.

Source Data for Expanded View

Click here for additional data file.

Source Data for Figure 1

Click here for additional data file.

Source Data for Figure 2

Click here for additional data file.

Source Data for Figure 3

Click here for additional data file.

Source Data for Figure 4

Click here for additional data file.

Source Data for Figure 5

Click here for additional data file.

Source Data for Figure 6

Click here for additional data file.

Source Data for Figure 7

Click here for additional data file.

Source Data for Figure 8

Click here for additional data file.