Vip-deficient mice display an enhanced type 2 immune response
We performed immunophenotyping of Vip mice in different organs, including the mesenteric lymph nodes (mLNs), liver, lung, colon and spleen. We found a consistent increase of eosinophils in Vip mice compared to littermate heterozygous control mice in all organs investigated, whereas comparable frequencies of other myeloid cell subsets were detected (Fig. 1a, Extended Data Fig. 1a-h and Supplementary Table 1). Because eosinophils require IL-5 (refs. ), we examined the source using Il5 reporter mice on a Vip background. Indeed, Vip Il5 mice showed increased IL-5 expression compared to littermate controls (Fig. 1b). Gating on IL-5-tomato cells revealed that ILC2s, and to a lesser extent T cells, are the main producers of IL-5 in Vip mice (Fig. 1b,c). Furthermore, we detected increased mRNA expression of Il5 and Il13 in the small intestine (Fig. 1d) and increased serum levels of IL-5 protein (Extended Data Fig. 1i) in Vip mice. However, the activation of ILC2 and type 2 inflammation was an unexpected finding since previous reports show that VIP stimulates ILC2s via its receptor VIPR2 and we could recapitulate some of these findings (Extended Data Fig. 1j-m).
To clarify this puzzling finding, we performed detailed immunophenotyping of ILC2 in Vip mice. Consistent with the IL-5 reporter data, ILC2s, but not other lymphocytes, were increased in all organs investigated (Fig. 1e and Extended Data Fig. 1n). In contrast, ILC3s, which are known to sense VIP via VIPR2 (refs. ), were decreased along with diminished IL-22 production but without alterations in the CCR6 and NKp46 ILC3s subsets (Extended Data Fig. 2a-c). Moreover, ILC2s adopted an inflammatory phenotype characterized by the upregulation of programmed cell death 1 (PD-1; refs. ), downregulation of ST2 (ref. ) and CD25 (ref. ; Fig. 1f-h and Extended Data Fig. 2d). Interestingly, the increase in Il5- and Il13-producing ILC2s were also found in the colon but not the lung (Extended Data Fig. 2e,f). Taken together, our data show a paradoxical type 2 immune response triggered in the intestines of Vip mice.
We examined the mechanism behind the type 2 inflammation in Vip mice. Previous studies showed that mice with activated ILC2s have a prolonged intestine and expansion of secretory epithelial cells. Indeed, Vip knockout mice had a longer small intestine and a smaller cecum than control mice (Fig. 1i and Extended Data Fig. 2g). We detected an increase in periodic acid-Schiff (PAS) goblet cells, Ulex europaeus agglutinin 1 (UEA-1) cells and Doublecortin-like kinase 1 (DCLK1) tuft cells in all parts of the intestine, including the duodenum, jejunum and ileum (Fig. 1j-m and Extended Data Fig. 2h-j). Thus, Vip mice display an expansion of modules involved in the tuft cell-ILC2 circuit. Consistent with an increase in tuft cells, we detected upregulation of Il25 but not Il33 or Tslp transcripts in the intestine and elevated Il25 expression in sort-purified tuft cells (Fig. 1n,o and Extended Data Fig. 2k). Although we observed reduced ILC3s and IL-22 production in Vip mice, Rorc(γt) and Il22 mice did not show the epithelial disturbances seen in Vip mice, suggesting that ILC3s don’t cause the expansion of secretory epithelial cells and activation of type 2 immunity (Extended Data Fig. 2l-n). Taken together, our data reflect an expansion of secretory epithelial cells and an upregulation of the alarmin Il25, which could explain the activation of ILC2s.
Besides neurons, other cell types were reported to secrete VIP. However, we only detected VIP in neurons, which densely innervate all segments of the intestine and are found in close proximity to epithelial cells along the whole crypt-villous axis (Fig. 2a and Extended Data Fig. 3a-f). Therefore, we hypothesized that the nervous system is the main source of VIP causing the phenotype. To restrict Vip deletion to neurons, we generated Vip mice and crossed them to Snap25 mice, a neuronal-specific Cre-deleter mouse line (Extended Data Fig. 3g,h). While Snap25 Vip mice had undetectable VIP expression in neurons, we could not detect any reduction in all enteric neurons reflected by HuC/HuD staining, or in Nos1, Substance P or NMU neurons, arguing that the ENS structure is largely maintained (Extended Data Fig. 3i-m). To dissect the transcriptional changes in the ENS, we used the INTACT system (Rosa26), which allows for the Cre-induced expression of GFP fused to the nuclear protein Sun1. We activated the Rosa26 allele in Snap25 Vip mice, and performed RNA sequencing (RNA-seq) on sort-purified ENS nuclei from Snap25 Vip Rosa26 and Snap25 Vip Rosa26 mice. Vip deletion resulted in a substantial number of downregulated genes but very few upregulated genes (Fig. 2b). A Gene Ontology (GO) analysis revealed that the top downregulated pathways were related to metabolism, synapse processes, cytoskeleton and mRNA processing (Fig. 2c). We mapped differentially expressed genes in the Human Phenotype Ontology (HP) database and identified alterations in physiologic gastrointestinal functions, such as an impaired muscle tone or muscle physiology, indicating gut motility defects in Vip-deficient mice (Fig. 2c). Indeed, absence of neuronal Vip led to altered measures of gut transit time (Extended Data Fig. 3n).
Snap25 Vip mice phenocopied all findings obtained in Vip mice, including the increase in activated ILC2s, type 2 effector cytokines, eosinophil recruitment, an elongated small intestine, the expansion of tuft and goblet cells and Il25 expression (Fig. 2d-m, Extended Data Fig. 3o-r and Supplementary Table 1). We did not measure differences in eosinophil development (Extended Data Fig. 3s). Because VIP was described to regulate the circadian rhythm in the suprachiasmatic nucleus in the central nervous system (CNS), the observed activation of the tuft cell-ILC2 circuit may be due to disturbances in the central clock; thus, we aimed to restrict the conditional deletion to peripheral neurons. We targeted neurons of the sympathetic and parasympathetic nervous system by generating Th Vip and Chat Vip mice, respectively. However, these mice did not show the differences observed in Snap25 Vip mice, excluding the sympathetic and parasympathetic nervous system as a main regulator of VIP functions (Extended Data Fig. 4a-f and Supplementary Table 1). Based on single-cell datasets obtained from neurons of the central and peripheral nervous system, Vip is mainly coexpressed by Nos1 enteric neurons with no coexpression occurring in the CNS. Therefore, we validated coexpression of VIP in Nos1 reporter mice in the ENS, and the suprachiasmatic nucleus in the CNS, the regulator of circadian rhythms. In the ENS, Nos1 and VIP largely overlap, whereas in the suprachiasmatic nucleus only VIP was detected (Extended Data Fig. 4g,h). Thus, the deletion of Vip using Nos1 mice did not affect the central clock and increased ILC2s and eosinophils were consistently detected across circadian phases, arguing against the central clock as a driver of the phenotype (Extended Data Fig. 4i,j). Nos1 Vip mice resembled key features of Vip and Snap25 Vip mice, including ILC2 and eosinophil activation, increased length of the small intestine and expansion of tuft and goblet cells (Fig. 2n-r, Extended Data Fig. 4k and Supplementary Table 1).
To formally demonstrate the involvement of the ENS, we locally administered adeno-associated virus (AAV) 9 particles, carrying an inhibitory Designer Receptor Exclusively Activated by Designer Drugs (DREADD) under the control of a floxed stop codon in the duodenum of Vip and control mice. Upon administration of the clozapine-N-oxide (CNO), the ligand for the DREADD, we could silence VIP neurons in the duodenum but not the ileum (Extended Data Fig. 4l), which served as a control. Vip mice displayed increased tuft cell numbers and ILC2 activation compared to control mice (Fig. 2s-u). Because the inhibition of enteric neurons provoked a similar phenotype observed in Vip-deficient mice, we concluded that VIP Nos1 enteric neurons control epithelial differentiation and type 2 immune responses in the intestine.
The immunoregulatory effects of VIP via VIPR2 described in the literature and recapitulated by us are difficult to reconcile with the phenotype observed in Vip mice. Because VIP deficiency leads to activation of the tuft cell-ILC2 circuit, we hypothesized that VIP directly regulates epithelial cells via the alternate receptor VIPR1. Indeed, we detected Vipr1 expression in stem cells and epithelial progenitor cells but not in differentiated secretory cells (Fig. 3a). We validated these data using purified IL-25 tuft cells (Extended Data Fig. 5a). As we did not detect an increase in components of the tuft cell-ILC2 circuit in Vipr2 mice, the VIP-VIPR2 axis cannot explain the phenotype observed in Vip-deficient mice (Extended Data Fig. 5b-f and Supplementary Table 1).
To determine whether VIP regulates epithelial cells, we established intestinal organoids and administered VIP or a specific VIPR1 agonist. Untreated organoids readily differentiated, evidenced by the presence of crypts. However, treatment with VIP or VIPR1 agonist prevented differentiation and resulted in spheroids (Fig. 3b-e). Likewise, we detected decreased DCLK1, UEA-1 and Ki-67 cells (Fig. 3c,e and Extended Data Fig. 5g). Organoids from Vip-deficient mice did not preserve an epithelial phenotype in vitro and responded to VIP (Extended Data Fig. 5h). RNA-seq of organoids exposed to VIP at different time points indicated profound changes in epithelial differentiation and revealed upregulation of genes driving processes, including ‘upregulation of the MAPK signaling pathway’, whereas cell proliferation-associated gene sets were downregulated (Fig. 3f,g). While the sequencing results show that VIP triggers a signaling cascade in the epithelium to influence epithelial differentiation, we aimed to explore the molecular signaling pathways behind VIP-mediated epithelial regulation. We noticed that VIP stimulation directs the induction of nuclear factor (NF)-κB target genes. In contrast, the proto-oncogene Myc and E2F pathways were downregulated, which could explain the VIP-mediated effect on cell cycle and proliferation (Extended Data Fig. 5i, j).
To test the relevance of this finding in vivo, we crossed Vil1 with Vipr1 mice to delete Vipr1 in intestinal epithelial cells. Vil1 Vipr1 mice phenocopied Vip-deficient mice and showed an increase in SI length, tuft and goblet cells expansion along with an enhanced type 2 immune response (Fig. 3h-o, Extended Data Fig. 6a-d and Supplementary Table 1). In contrast, we did not detect alterations of ILC3s and IL-22 in Vil1 Vipr1 mice, suggesting that ILC3s or IL-22 are not causing the phenotype (Extended Data Fig. 6e,f). Finally, organoids generated from Vil1 Vipr1 mice were unresponsive to VIP stimulation (Extended Data Fig. 6g,h). However, because the organoids were derived from Vil1 Vipr1 mice, organoids maintained elevated numbers of tuft cells in culture presumably inherited from the signals received previously in vivo (Extended Data Fig. 6i,j).
To investigate whether the VIP-VIPR1 pathway is conserved in humans, we applied VIP to human colonic and small intestinal organoid cultures. VIP and VIPR1 agonist treatment of colonic organoids resulted in round, cystic organoids resembling murine data (Fig. 3p). Similar effects were observed in small intestinal organoids (Fig. 3q). Further, we detected a reduction in UEA-1 staining in VIP-treated organoids (Fig. 3r). Therefore, we concluded that the VIP-VIPR1 pathway influences epithelial differentiation in mice and humans.
Because Vipr1 is expressed in progenitor cells but barely detectable in secretory epithelial cells, we hypothesized that this pathway controls differentiation of epithelial cells. To investigate which epithelial cell lineages are affected, we performed a multiplex staining for epithelial cell markers, including DCLK1 (tuft cells), MUC2 (goblet cells), LYZ1 (Paneth cells), CHGA (enteroendocrine cells), OLFM4 (olfactomedin 4; pan stem cell marker) and Ki-67 (transit-amplifying cells; Fig. 4a and Extended Data Fig. 7a). Our data revealed an expansion of tuft cells, goblet cells and Paneth cells but not enteroendocrine cells in Snap25 Vip mice (Fig. 4b and Extended Data Fig. 7a,b). Further, we detected increased OLFM4 and Ki-67 cells at the base of the crypts, suggesting that VIP suppresses stem cell proliferation. The fact that Vipr1 is expressed on RNA and protein levels in stem cells and colocalizes with LGR5 cells in situ prompted us to investigate whether VIPR1 signaling regulates LGR5 stem cells (Extended Data Fig. 7c-e). For this purpose, we generated Lgr5 Vipr1 mice for inducible ablation of Vipr1 in LGR5 stem cells and performed in vitro organoids. VIP-mediated suppression of epithelial differentiation was abolished in organoids with LGR5-specific deletion of Vipr1 (Fig. 4c-e), suggesting that VIP already affected LGR5 stem cells. To investigate if VIPR1 regulates stem cells in vivo, we performed short-term tamoxifen administration to Lgr5 Vipr1 and Lgr5 Vipr1 mice. Tamoxifen administration resulted in increased EpCAMLGR5 stem cells in Lgr5 Vipr1 (Fig. 4f). RNA-seq of sort-purified LGR5 stem cells lacking Vipr1 revealed gene expression changes opposite to those in the organoid gain-of-function experiment, including upregulation of proliferation and cell cycle-associated genes (Myc and E2F pathways) and downregulation of apoptosis (Fig. 4f,g and Extended Data Fig. 7f). We further observed upregulation of antimicrobial peptide genes and modulation of Wnt receptor genes (Extended Data Fig. 7g-i). Therefore, these data document direct effects of VIP on LGR5 stem cells and reveal the molecular signaling pathways.
The test whether the alterations in secretory epithelial cells affect barrier functions, we assessed mucus characteristics in Vip mice. We detected an increase in mucus thickness and reduced penetrability indicating a denser mucus structure (Fig. 4h). Consistent with previous findings, we detected a drastic increase in UEA-1 but not in wheat germ agglutinin (WGA) sugars, suggesting differences in fucosylation recognized by UEA-1 (Fig. 4i). Next, we performed mass spectrometry on small intestinal samples of Snap25 Vip mice and measured downregulation of VIP and upregulation of antimicrobial peptides (Extended Data Fig. 7j,k). To assess luminal secretion, we performed proteomic analyses of the mucus, which revealed a broad array of proteins differentially abundant, including the enrichment of goblet and Paneth cell proteins in the Vip mice, whereas enterocyte-associated proteins were enriched in Vip mice (Fig. 4j-n and Extended Data Fig. 7l,m). In summary, loss of Vip unleashes aberrant LGR5 stem cell proliferation and differentiation into secretory epithelial cell lineages, leading to enhanced mucus thickness and maturation.
Before weaning, pups have limited microbiota. Since we previously observed expansion of secretory precursor Math1 (Atoh1) cells in Vip-deficient mice, we asked if VIP is required early during development. Unlike adult mice, pups at day one after birth did not show expansion of Math1 secretory epithelial cells indicating that the phenotype is not present at birth (Fig. 5a). Therefore, we performed a kinetic experiment to identify when the phenotype starts to materialize. We did not detect expansion of secretory precursors and activation of the type 2 response before postnatal day 21, coinciding with weaning (Fig. 5b-e and Extended Data Fig. 8a-c). To dissect epithelial cell trajectories occurring in the epithelium, we performed single-cell sequencing of sort-purified live EpCAM CD45 cells from the intestinal epithelial fraction of Vip and Vip and detected increased stem cells in Vip-deficient mice (Fig. 5f and Extended Data Fig. 8d-f). KEGG pathway analysis of the stem cell cluster revealed enrichment of genes of the DNA replication and cell cycle pathway (Extended Data Fig. 8g). Furthermore, processes responsible for protein secretion and antimicrobial defense, including ‘positive regulation of secretion’, ‘glycoprotein metabolic process’ and ‘glycosylation’ were upregulated in secretory epithelial cell clusters of the knockout mice (Fig. 5g). Similarly to the RNA-seq of Vipr1-deficient LGR5 stem cells, Myc targets genes were upregulated, whereas NF-κB signaling was downregulated (Extended Data Fig. 8h). Collectively, our sequencing approaches point toward the cell cycle, epithelial differentiation, NF-κB and Myc-E2F signaling pathways.
Weaning coincides with bacterial colonization. To determine if the microbiota is altered, we performed fecal metagenomics sequencing in different segments of the intestine from littermate, co-housed Vip and Vip, as well as Snap25 Vip and control mice. We observed clustering of either knockout strain and control mice independently of the intestinal segment (Fig. 5h and Extended Data Fig. 8i,j). In particular, loss of Vip decreased microbial diversity and concomitant enrichment in proteobacteria and Firmicutes (Fig. 5i,j). We did not detect parasites in the feces of these mice, excluding other microorganisms triggering tuft cell expansion (Supplementary Tables 2 and 3). Moreover, we did not detect expansion of succinate-producing microorganisms or differences in succinate concentrations in feces comparing Vip and Vip mice (Extended Data Fig. 8k).
Because VIP deficiency resulted in dysbiosis, we hypothesized that microbial disturbances may activate the tuft cell-ILC2 circuit. To assess a causal effect of the commensal microbiota, we re-derived Snap25 Vip mice under germ-free conditions (Extended Data Fig. 8l,m). However, germ-free Snap25 Vip mice and control mice recapitulated the phenotype observed in specific pathogen-free Snap25 Vip animals (Fig. 5k-s). Therefore, we conclude that dysbiosis did not cause the phenotype, but developed as a consequence of VIP deficiency.
Because the differences in Vip-deficient mice were not dictated by the commensal microbiota, we considered additional events occurring at weaning, such as the switch from milk to solid food and immune activation. To investigate whether consumption of solid food independent of the commensal microbiota might be sensed, for instance by the neuroepithelial unit, we fed germ-free Snap25 Vip and control mice either a liquid diet or normal chow and compared the immune and epithelial parameters. The phenotype characterized by ILC2 activation, eosinophilia and expansion of secretory epithelial cells was only minorly detected when the mice were fed a liquid diet (Fig. 6a-g). Therefore, these data indicate that food formulations influence type 2 immunity in Vip-deficient mice (Fig. 6a-g).
To dissect how the unleashed type 2 immune response affects epithelial cells, we examined the top hits from our single-cell sequencing experiment conducted in 3-week-old mice comparing all clusters of comparing Vip and Vip mice (Fig. 6h). Notably, many genes, among the top hits, such as Pla2g4c, Gsdmc4, Gsdmc2, Ang4, Gsdmc3, Sprr2a3 and Car4 were found in epithelial cells during worm infections or following treatment with IL-4 or IL-13 and linked to STAT6 activity downstream of the IL-4rα/IL-13rα1 receptor (Fig. 6h). To test activation of IL-4rα/IL13ra1 downstream signaling, we performed phospho-flow staining for p-ERK in epithelial cells. Given that we detected increased p-ERK epithelial cells in Snap25 Vip mice (Fig. 6i), epithelial cells might have persistently active p-ERK as a result of the type 2 response.
Because ILC2s were activated in Vip-deficient mice and are known producers of IL-13, we hypothesized that ILC2s may promote epithelial cell hyperplasia similarly to worm infections. By crossing Nmur1 Id2 to Vip mice, we genetically ablated ILC2 in Vip mice (Fig. 6j and Supplementary Table 1). The absence of ILC2s resulted in reduced numbers of eosinophils and rescued the type 2 immune response (Fig. 6k and Supplementary Table 1). Further, the length of the small intestine as well as the number of secretory epithelial cells became comparable to the control mice (Fig. 6l-o and Extended Data Fig. 9a,b). Despite the reduction in UEA-1 cells, their localization did not return to normal, indicating that some of the epithelial alterations are ILC2 independent (Fig. 6n). In summary these data identify food and ILC2s as pivotal players, driving the phenotype in Vip-deficient mice and complementing the processes autonomously regulated in the epithelium by VIP.
We hypothesize that the ILC2-tuft cell circuit provides the missing piece linking the epithelial disturbances to immune activation in Vip-deficient mice. Tuft cells are the main source of IL-25 (refs. ). Because we also detected increased Il25 in Vip mice, we hypothesized that this cytokine is essential to trigger type 2 inflammation in Vip mice. To this end, we crossed IL-25-receptor-deficient mice (Il17rb) to Vip mice. We compared Il17rb Vip mice with littermates of the co-housed Il17rb Vip and Il17rb Vip mice. Indeed, Il17rb Vip mice had a type 2 immune response at the level of control mice, characterized by lower ILC2s and PD-1 ILC2s as well as eosinophils (Fig. 7a-c). Consequently, deletion of IL-25R in Vip mice rescued the length of the small intestine, accompanied by normalization in the number of secretory epithelial cells (Fig. 7d-f). Comparable to ILC2-deficient mice on the Vip background, the localization of UEA-1 cells was still altered in the absence of the IL-25 receptor, indicating that the phenotype is partially autonomously regulated in the epithelium.
Given that we measured STAT6 target genes and activation of p-ERK, which are downstream of the IL-4rα/IL13ra1 receptor, we hypothesized that IL-4/IL-13 effector cytokines secreted by ILC2s impose the phenotype on the epithelium. We crossed Nmur1 Il4/I13 to Vip mice, and genetically ablated Il4 and Il13 in ILC2s, as both cytokines bind to the IL-4rα/IL13rα1 receptor. Consistent with a pivotal role for IL-4 or IL-13, we measured a rescue of the epithelial phenotype, gut length and components of the type 2 immune response (Fig. 7g-o). We also considered IL-5 as a potential effector cytokine responsible for the phenotype. However, VIP expression in Il5 mice was not different compared to Il5 mice (Extended Data Fig. 9c,d). Therefore, epithelial disturbances did not follow the phenotype observed in Vip-deficient mice (Extended Data Fig. 9e-g). Next, we generated Vip Il5 mice to assess the role of IL-5 in this model. As expected, IL-5 deficiency led to a strong reduction in eosinophils (Extended Data Fig. 9h), but the length of the small intestine was unchanged compared to Vip mice, suggesting that IL-5 regulates eosinophils but is not involved in other components of the phenotype (Extended Data Fig. 9i).
Taken together, a tuft cell-IL-25-ILC2 circuit activated in Vip-deficient mice is responsible for secretion of IL-4/IL-13, driving epithelial differentiation and type 2 immune reaction.
Given that Vip-deficient mice display an enhanced type 2 response in the intestine, we examined if the absence of VIP improves clearance of intestinal worm parasites. For this purpose, we infected Snap25 Vip and control mice with the nematode Nippostrongylus brasiliensis. Snap25 Vip mice largely expelled the parasite on day 7 when littermate control mice still had an uncontrolled worm infection (Fig. 8a). Although N. brasiliensis is a strong inducer of type 2 inflammation, Snap25 Vip mice had an enhanced type 2 response compared to the littermate controls (Fig. 8b-d). Therefore, our data show superior immunity to helminth infection in the absence of VIP.
Snap25 Vip mice have increased ILC2s and eosinophils in several tissues even though the phenotype is originating from the intestine, arguing for a systemic type 2 immune response. Therefore, we explored if Vip deficiency can affect immunohomeostasis at other mucosal barrier sites, such as in the lung. Although the trachea and large bronchi of the lung are innervated by VIP neurons (Extended Data Fig. 10a), we did not find evidence that lung epithelial cells are direct sensors of VIP, because we did not detect Vipr1 expression in the lung epithelium nor altered tuft and goblet cells in the respiratory tract (Extended Data Fig. 10b-e). However, VIP deficiency leads to increased lung eosinophils, and lung ILC2s were slightly more activated although not expressing more type 2 cytokines. Therefore, the increase in lung eosinophils is most likely explained by the elevated serum IL-5 levels detected in Vip and Snap25 Vip (Extended Data Figs. 1i, 2f and 10f-j).
When we challenged Snap25 Vip and control mice intranasally with the protease allergen papain to induce allergic lung inflammation, we detected an increase in eosinophils, ILC2s and total CD45 cells and an elevated lung inflammation score in Snap25 Vip mice, suggesting an enhanced allergic lung inflammation (Fig. 8e-j). However, we did not measure increased lung fibrosis in Vip-deficient mice or papain-challenged Snap25 Vip mice (Extended Data Fig. 10k,l).
Therefore, these data confirm an exaggerated type 2 response triggered in the absence of VIP with protective and detrimental effects in worm infection and allergic lung inflammation.
