FOXO1 orchestrates the intestinal homeostasis via neuronal signaling in group 3 innate lymphoid cells

Innate lymphoid cells (ILCs) are abundant in mucosal surfaces to maintain mucosal homeostasis (Spits et al., 2013; Wang et al., 2017). ILC3s are characterized by retinoic acid–related orphan receptor γ t (RORγt) expression and interleukin (IL)-22 and IL-17 production, contributing to host defense against pathogens and barrier integrity in the intestine (Vivier et al., 2018). However, ILC3s play a double-edged-sword role in regulating intestinal inflammation. Excessive production of IL-17 and IL-22 leads to the over-reaction of neutrophils, exacerbating barrier damage and resulting in colitis (Zhou and Sonnenberg, 2020). Hyperactivation of ILC3s has been reported in patients with inflammatory bowel disease (IBD; Geremia et al., 2011). Increased expression of IL17A and IL17F was observed in the intestinal lamina propria ILC3s from patients with IBD compared with healthy donors. Human intestinal ILC3s produced increased levels of IL-22 in both patients with ulcerative colitis (UC) and those with Crohn’s disease (CD) compared with the non-IBD individuals, indicating the dysfunction of ILC3s in IBD patients (Longman et al., 2014). On the other hand, ILC3s may also maintain intestine homeostasis. ILC3-derived heparin-binding epidermal growth factor–like growth factor is essential to protect the gut against TNF-mediated inflammation (Zhou et al., 2022). However, the balance of ILC3 function in intestine homeostasis is not well studied.

The intestine is innervated by millions of neurons, which modulate immune responses via secreting neurotransmitters and neuropeptides (Veiga-Fernandes and Mucida, 2016). The neuro-immune interactions play central roles in intestinal health and disease, including barrier integrity, host defense, and inflammatory conditions (Jacobson et al., 2021). The autonomic nervous system consists of the sympathetic nervous system, parasympathetic nervous system, and enteric nervous system, all of which are associated with gut immunity (Jakob et al., 2021; Udit et al., 2022). As the primary neurotransmitter of the sympathetic nervous system, noradrenaline plays an essential role in neuro-immune crosstalk (Lorton and Bellinger, 2015). Adrenergic receptors or adrenoceptors (Marino and Cosentino, 2013) mediate the physiological responses to noradrenaline and adrenalin, which also participate in inflammation and infection (Ahrends et al., 2021; Moriyama et al., 2018). For example, β2-AR (ADRB2) protects muscularis macrophages from neuronal loss during challenges with pathogens (Ahrends et al., 2021). ADRB2 prevents T helper 1 (Th1) cell development and promotes the differentiation of Th2, Th17, and regulatory T cells (Godinho-Silva et al., 2019). Recent studies showed that ILC2s expressed ADRB2, which was adjacent to adrenergic neurons. ADRB2 signaling inhibits the type 2 responses of ILC2s (Moriyama et al., 2018). In addition, enteric neurons are located within the myenteric plexus or the submucosal plexus close to immune cells such as goblet cells, macrophages, and ILCs (Jarret et al., 2020; Muller et al., 2014; Talbot et al., 2020; Wallrapp et al., 2017). The neuropeptide neuromedin U released from cholinergic neurons promotes the proliferation and activation of NMUR1⁺ ILC2s (Wallrapp et al., 2017). Vasoactive intestinal peptide (VIP) produced by enteric neurons suppresses the production of inflammatory cytokines by ILC3s (Talbot et al., 2020). Diet is a way to regulate intestinal nerve system. Diets with glucosinolates promote IL-22 production by ILC3s and γδ T cells, and regulate the homeostasis of intestinal epithelial stem cells (Gronke et al., 2019). IL-22 is also critical to protect intestinal barrier against pathogenic bacteria infection (Lee et al., 2011). Stress is another way that contributes to neuronal regulation in the intestine. Depression and anxiety contribute to the onset or exacerbation of IBD (Bisgaard et al., 2022). Chronic stress is a potent risk factor for depressive disorder, and chronic restraint stress mouse models could produce depressive-like behavior and contribute to intestinal barrier dysfunction and inflammatory cell infiltration (Lee et al., 2021). Chronic stress contributes to gut microbiota disturbance, triggers immune responses, and exacerbates colitis (Gao et al., 2018). However, the regulation of neuropeptide receptors and neuronal signaling in ILC3s during colitis is rarely known.

The transcription factor FOXO1, a member of the forkhead box O (FOXO) family, is central to many physiological processes and modulated by the factors from tissue microenvironment (Hedrick et al., 2012). FOXO1 participates in regulating immune responses. Reports showed that FOXO1 maintains the quiescence state of naïve T cells and the function of regulatory T cells, and inhibits the proliferation of Th17 (Ichiyama et al., 2016), T follicular helper cells (Stone et al., 2015), and CD8⁺ T cells (Delpoux et al., 2021). FOXO1 regulates the differentiation of B cells (Dominguez-Sola et al., 2015), activates M2 macrophages (Chung et al., 2019), and induces autophagy of natural killer (NK) cells during their maturation (Wang et al., 2016). Whether FOXO1 regulates the function of ILCs or ILC3s remains elusive. Cyclic adenosine monophosphate (cAMP) is an important second messenger participating in numerous physiological processes (Anton et al., 2022). cAMP also regulates the transcriptional activation of FOXO1 (Shen et al., 2014; Silveira et al., 2020). A previous report showed that cAMP inhibits ILC2 activation and proliferation (Wallrapp et al., 2019). However, other studies revealed that cAMP signaling can also promote the differentiation and activation of Th subsets and play a proinflammatory role (Hernandez et al., 2015; Li et al., 2012). The roles of cAMP and FOXO1 on ILC3s remain poorly understood.

In this study, we reveal neuronal signaling of ILC3s via the transcriptional activity of FOXO1. FOXO1 suppresses hyperactivation of ILC3s by balancing VIP and adrenergic signaling to maintain intestinal homeostasis.

cAMP as a second messenger is involved in many cellular processes and immune responses (Anton et al., 2022; Hernandez et al., 2015; Li et al., 2012; Shen et al., 2014; Silveira et al., 2020; Wallrapp et al., 2019). To explore the effect of cAMP on ILC3 functions, we isolated ILC3s from mouse small intestines and stimulated them with dibutyryl cAMP (d-cAMP), which mimics cAMP stimulation (Schwede et al., 2000). Although d-cAMP did not apparently promote the proliferation of ILC3s (Fig. 1 A), it significantly primed ILC3 activation in a dose-dependent manner. ILC3s produced elevated levels of IL-17A and IL-22 upon d-cAMP treatment (Fig. 1 B). In addition, d-cAMP increased the proportion and cell numbers of IL-17A–, IL-17F–, and IL-22–positive ILC3s (Fig. 1 C). Reports showed that cAMP modulated the transcriptional activities of many transcription factors, one of which is the FOXO family (Daitoku et al., 2004). cAMP promoted the phosphorylation of FOXO1 at Thr24, Ser256, and Ser319 residues, leading to nuclear export, inactivation, and degradation of FOXO1 (Liu et al., 2014; Silveira et al., 2020). Therefore, we next investigated the expression and activity of FOXO1 in ILC3s after d-cAMP treatment. Intriguingly, we found that d-cAMP decreased the levels of FOXO1 (Fig. 1 D) and promoted its phosphorylation (Fig. 1 E). Imaging flow cytometry revealed that FOXO1 of ILC3s translocated from the nucleus to cytosol upon d-cAMP treatment, suggesting the suppression of its transcriptional activity (Zaiss and Coffer, 2018; Fig. 1 F). These data indicated that d-cAMP elicited ILC3 activation and impaired the transcriptional activity of FOXO1.

We first employed Foxo1^flx/flx;Rorc-Cre mice to delete FOXO1 in RORγt-positive cells, including Th17 cells and ILC3s. We noticed that Foxo1^flx/flx;Rorc-Cre mice developed severe intestinal inflammation (Fig. S1, A–D). In addition, concentrations of inflammatory cytokines IL-17A and IL-22 in murine serum were significantly elevated in Foxo1^flx/flx;Rorc-Cre mice compared with littermate control mice (Fig. S1 E). Apart from the elevated activation of Th17 cells (Fig. S2, A and B), we found that ILC3s exhibited an inflammatory state and produced large amounts of inflammatory cytokines, including IL-17A, IL-17F, and IL-22 (Fig. S2, C and D). It was reported that Citrobacter rodentium (CR) infection could activate ILC3s and induce colitis before T cell activation (Jarade et al., 2021). We noticed that at early stage of CR infection, ILC3s were remarkably activated in Foxo1^flx/flx;Rorc-Cre mice (Fig. S2 D), indicating that FOXO1 might regulate ILC3s independently of T cells. In addition, reports showed that IL-17 and IL-22 recruited neutrophils via augmenting neutrophil-attractant chemokines (Zhou and Sonnenberg, 2020). Accordingly, we found that the proportion and numbers of neutrophils increased in the intestine of Foxo1^flx/flx;Rorc-Cre mice under physiological conditions or at an early stage of infection (Fig. S2 E).

To further exclude the effect of T cells, we used Foxo1^flx/flx;Rorc-Cre;Rag1^−/− mice to evaluate the role of FOXO1 in ILC3s. Similar to that in T cell–intact mice, the small intestine and colon showed more severe intestinal ulceration, a significant increase in mucosal thickness and leukocyte infiltration, and shortened colon length in Foxo1^flx/flx;Rorc-Cre;Rag1^−/− mice (Fig. 2, A–C). These results manifested the gut inflammation in Foxo1^flx/flx;Rorc-Cre;Rag1^−/− mice (Lelievre et al., 2007). In the absence of T and B cells, CR infection induced ILC3-dependent colitis, and FOXO1 abrogation in ILC3s also aggravated CR-induced colitis (Fig. 2, A–C). In addition, we found that the serum levels of IL-17A and IL-22 were elevated in Foxo1^flx/flx;Rorc-Cre;Rag1^−/− mice under physiological condition or during CR-induced colitis (Fig. S2 F). However, the frequencies and total numbers of ILC3s in small intestine and mesenteric lymph nodes did not significantly change (Fig. 2 D). Consistent with the results from Foxo1^flx/flx;Rorc-Cre mice, the numbers and frequencies of IL-17A⁺, IL-17F⁺, and IL-22⁺ ILC3s increased in the small intestine and mesenteric lymph nodes of Foxo1^flx/flx;Rorc-Cre;Rag1^−/− mice (Fig. 2 E). Similar results were observed in mice infected with CR. On day 8 after infection, ILC3s in Foxo1^flx/flx;Rorc-Cre;Rag1^−/− mice produced extremely high levels of inflammatory cytokines, including IL-17A, IL-17F, and IL-22 (Fig. 2 E). Accordingly, neutrophils were remarkably augmented in the intestine of Foxo1^flx/flx;Rorc-Cre;Rag1^−/− mice (Fig. 2 F). Collectively, these findings suggested that FOXO1 contributed to the suppression of the inflammatory response of ILC3s.

To gain insights into the regulated network by FOXO1 that differs between ILC3s and T cells, we isolated ILC3s and CD3⁺ T cells from intestinal lamina propria of Foxo1^flx/flx;Rorc-Cre mice and littermate control mice and examined the transcriptomic change using single-cell RNA sequencing (scRNA-seq). We selected Rorc-positive cells for clustering. The unsupervised clustering partitioned the cells into four clusters, including Th17, CCR6⁺ ILC3, NCR1⁺ ILC3, and NKT17 (Fig. 3 A and Fig. S3 A). We analyzed the differential gene expression between the control and Foxo1^flx/flx;Rorc-Cre mice under steady state or during CR-induced colitis. Volcano plots showed that in Foxo1-deficient ILC3s, the expression of inflammatory cytokine genes, including Il17a, Il17f, and Il22, significantly increased, while some other genes, including Il10, Ret, and Vipr2, were downregulated (Fig. 3 B). Next, we employed gene set enrichment analysis on the differential expressed genes and found that colitis-related genes were enriched in Foxo1-defecient ILC3s (Fig. 3 C; Kibbe et al., 2015), which was consistent with our previous data.

Since the hyperactivation of ILC3s was also observed in Rag1^−/− mice, we intended to find out the unique transcriptional network orchestrated by FOXO1 in ILC3s independent of T cells. We next picked out the genes specifically expressed in ILC3s and changed after FOXO1 depletion (Fig. 3 D). The differentially expressed genes (DEGs) in Foxo1-deficient ILC3s belonged to the pathways in IBD, cAMP signaling, and neuro-associated signaling (Fig. 3 E and Fig. S3 B). Among these, we selected the genes that changed the most in Foxo1-deficient ILC3s, including the upregulated adrenergic receptor gene Adra2a and downregulated VIP receptor gene Vipr2 (Fig. 3 F). Notably, the expression level of Adra2a increased in CR-infected mice and was extremely high in Foxo1-deficient ILC3s after infection (Fig. 3 F). On the contrary, the expression of Vipr2 was attenuated after CR infection and was even lower with FOXO1 abrogation (Fig. 3 F). We further confirmed the expression of Adra2a and Vipr2 in ILC3s, mainly in CCR6⁺ ILC3s, but not in Th17 cells (Fig. 3 G). Notably, the expression levels of other adrenergic receptors or VIP receptors were very low or undetectable in ILC3s (Fig. 3 G and Fig. S3 C). The expression of Il17a, Il17f, and Il22 was correlated with that of Adra2a, but opposite to that of Vipr2 (Fig. 3 H). Additionally, we analyzed the scRNA-seq data of IBD patients (Smillie et al., 2019) and observed that the expression levels of FOXO1 and VIPR2 were higher in the intestine of healthy people, and the levels of ADRA2A and IL17A were augmented in the intestine of IBD patients (Fig. S3 D). Collectively, FOXO1 differentially regulated adrenoceptors and neuropeptide receptors on ILC3s, which might contribute to the regulation of intestinal homeostasis.

To determine the molecular network regulated by FOXO1 in ILC3s, we evaluated the transcription and expression of Adra2a and Vipr2 in Foxo1-deficient ILC3s. Consistent with scRNA-seq data, the transcription and expression of Adra2a were elevated in Foxo1^−/− ILC3s, and those of Vipr2 were attenuated with FOXO1 abrogation (Fig. 4, A and B). CR infection decreased the level of FOXO1 accompanied by upregulation of Adra2a and downregulation of Vipr2 (Fig. S3 E and Fig. 4, A and B), indicating a dynamic regulation of FOXO1 activity during CR-induced colitis. In addition, we verified these results in Foxo1^flx/flx;Rorc-Cre;Rag1^−/− mice (Fig. 4 C). Next, we performed CUT&Tag assay followed by high-throughput sequencing to identify the FOXO1 DNA-binding sites on the genome of ILC3s. We found that FOXO1 directly bound to the promoters of Adra2a and Vipr2 with multiple binding sites (Fig. 4, D and E). The binding regions of FOXO1 were confirmed by chromatin immunoprecipitation–quantitative PCR (ChIP-qPCR; Fig. 4 F). During CR-induced colitis, the binding of FOXO1 on the promoters of Adra2a and Vipr2 was impaired suggesting the suppression of its regulatory function by CR infection (Fig. 4 G). Luciferase reporter assay revealed the transcriptional suppression of Adra2a and the transcriptional activation of Vipr2 by FOXO1 in ILC3s (Fig. 4 H). We also mapped the target regions of FOXO1 on these promoters by using various tractions. We found that −1389 to −940 bp region of the Adra2a promoter was occupied by FOXO1, which suppressed its transcription. Additionally, FOXO1 bound to −1539 to −1290 bp region of the Vipr2 promoter and promoted its expression (Fig. 4 H). These data suggested that FOXO1 directly targeted the promoters of Adra2a and Vipr2 and regulated their transcription.

It was reported that food consumption promoted VIP production in the intestine and attenuated the activation of ILC3s (Talbot et al., 2020). Short-term VIP treatment alleviated CR-induced colitis in mice (Conlin et al., 2009). The function of ADRA2A signaling in ILC3s has not been analyzed. First, we used immunofluorescence to analyze the location of neurons and ILC3s and found there was anatomical correlation between the neurons and ILC3s in the small intestine under steady state or during CR-induced colitis (Fig. S3 F). We next explored how ADRA2A and VIPR2 signaling pathways affected the function of ILC3s. After treatment with ADRA2A agonist clonidine, levels of IL-17A and IL-22 of ILC3s apparently increased in vitro (Fig. 5 A). Similar results were obtained after the inhibition of VIP signaling (Fig. 5 B). To further analyze the effect of ADRA2A and VIPR2 signaling in ILC3s in vivo, we treated the mice with ADRA2A agonist clonidine or VIPR2 antagonist PG99-465 intraperitoneally (i.p.). We found that ADRA2A agonist or VIPR2 antagonist apparently increased the frequency and number of IL-17– and IL-22–producing ILC3s (Fig. 5, C and D). Notably, ADRA2A antagonist or VIPR2 agonist apparently attenuated the inflammation response of ILC3s (Fig. S3, G and H). Next, we conditionally knocked out Adra2a gene in ILC3s. Since RORγt-positive T cells did not express Adra2a (Fig. 3 G), Adra2a^flx/flx;Rorc^cre mice could be used to analyze the function of ADRA2A in ILC3s. Importantly, depletion of Adra2a in ILC3s reduced the production of inflammatory cytokines in ILC3s, including IL-17A, IL-17F, and IL-22 (Fig. 5 E). Accordingly, the inflammation of the intestine was attenuated in Adra2a^flx/flx;Rorc^cre mice after CR infection (Fig. 5 F). Notably, treatment with ADRA2A antagonist and VIPR2 agonist successfully attenuated the hyperactivation of ILC3s in Foxo1^flx/flx;Rorc-Cre mice (Fig. 5 G). These data collectively suggested that ADRA2A and VIPR2 neuronal signalings differentially regulated the function of ILC3s.

ADRA2A signaling is known to be associated with stress responses (Laarakker et al., 2010). Accumulating evidence showed that stress is a potential predisposing factor for gut disorders, such as IBD or irritable bowel syndrome (Gao et al., 2018; Sun et al., 2019). The crosstalk between stress-related neuronal signaling and immune responses during colitis is not well known. It was reported that stress can activate the cAMP–protein kinase A signaling pathway (Thaker et al., 2006). We then investigated whether chronic stress exacerbated gut inflammation via cAMP–FOXO1 axis using a chronic restraint stress model (Gao et al., 2018; Li et al., 2020). The mice under stress exhibited decreased body weight and increased immobility times compared with those in the control group, indicating depression-like behavior (Fig. 6, A–C). The stressed mice showed more severe intestinal damage and increased villus height and crypt depth but decreased colon length (Fig. 6, D and E). Consistent with the previous reports (Thaker et al., 2006), chronic stress elevated cAMP levels in serum (Fig. 6 F). Of note, we found that chronic stress decreased the level of FOXO1 in ILC3s accompanied by its phosphorylation (Fig. 6, G and H). In addition, expression levels of ADRA2A were elevated and expression of VIPR2 was repressed in stressed mice (Fig. 6 I). We next examined the effect of chronic stress on the function of ILC3s. ILC3s in the stressed mice produced larger amounts of inflammatory cytokines under steady state or during CR-induced colitis (Fig. 6 J). Moreover, the ADRA2A antagonist Yohimbine treatment could remedy ILC3-dependent inflammatory response induced by stress (Fig. S3 I). Then, we set up a stress mouse model by using Foxo1^flx/flx;Rorc-Cre;Rag1^−/− mice and found that the proportion of IL-17A–, IL-17F–, and IL-22–positive ILC3s were not affected by stress treatment (Fig. S3 J), suggesting that stress-related inflammatory response of ILC3s is FOXO1 dependent. We further investigated whether the neuro-associated pathways of ILC3s were regulated by cAMP–FOXO1 axis in IBD patients. We found that the IBD patients had higher self-rating depression scale scores (Fig. 6 K) and elevated serum cAMP levels (Fig. 6 L). Consistently, expression levels of FOXO1 and VIPR2 were downregulated and expression levels of ADRA2A were increased in IBD patients (Fig. 6 M). The proportion of IL-17A⁺, IL-22⁺ ILC3s, and neutrophils were elevated in the intestine from IBD patients compared with healthy donors (Fig. 6, N and O). Taken together, these data revealed that chronic stress upregulated cAMP levels and suppressed the transcriptional activity of FOXO1 in ILC3s. As a result, activation of ADRA2A signaling and suppression of VIPR2 signaling in ILC3s aggravated intestinal inflammation.

ILC3s are required for the regulation of intestinal homeostasis. It has been reported that hyperactivation of ILC3s is closely associated with intestinal inflammation. The activation of immune cells in the intestine is modulated by the nervous system (Jakob et al., 2021). However, how gut neuronal signaling coordinates the immune response of ILC3 is largely unknown. In this study, we found that stress-elevating cAMP suppressed the expression of FOXO1. FOXO1 deficiency promoted the activation of ILC3s via upregulating expression levels of Adra2a and suppressing the expression of Vipr2. As a result, ILC3s produced large amounts of inflammatory cytokines and aggravated gut inflammation (Fig. S4).

FOXO1 is essential for the homeostasis of T cells in the intestine. Th17 is the adaptive immune counterpart of ILC3s. FOXO1 inhibits the proliferation and activation of Th17 cells (Ichiyama et al., 2016). Consistent with previous findings, we found that FOXO1 deficiency increased the cell population of Th17 cells. However, we did not observe a significant change in the cell number of ILC3s in Foxo1-deficient mice, indicating a different regulatory mechanism mediated by FOXO1 in ILC3s. Luu et al. found that ILC3 numbers were decreased in BM and thymus, and in contrast, NKp46⁻ ILC3s were increased in spleen of Foxo1,3^ΔVav mice compared with WT mice (Luu et al., 2022). By using Vav-Cre, Foxo1 gene was depleted in all hematopoietic cells, including the progenitor cells. It is possible that the development of ILC3s was affected by Foxo1 depletion. In addition, dysfunction of other immune cells by Foxo1 depletion may also contribute to the altering of ILC3 function. Our results found that the frequencies and total numbers of ILC3s in small intestine and mesenteric lymph nodes did not significantly change in Foxo1^flx/flx;Rorc-Cre;Rag1^−/− mice compared with Foxo1^+/+;Rorc-Cre;Rag1^−/− mice. We used a more specific mouse model for the depletion of Foxo1 in ILC3s. Moreover, deletion of FOXO1 in ILC3s of Rag1^−/− mice exacerbated production of inflammatory cytokines and aggravated intestinal inflammation, indicating that FOXO1 was required to maintain ILC3 homeostasis independently of T cells. Furthermore, we found that FOXO1 regulated the function of ILC3s through a different mechanism from T cells. FOXO1 suppressed the level of Adra2a and promoted the expression of Vipr2 in ILC3s. Interestingly, these two neural receptors were not expressed on Th17 cells. FOXO1 balanced the activation of ILC3s via neuroimmune regulation of these two neuronal signaling pathways. Thus, we provided new insights to distinguish the regulation of ILC3s and Th17 cells by FOXO1 in the intestine. Further studies are needed to determine how FOXO1 is differentially regulated during the development and maturation of ILC3s and Th17 cells.

The nervous and immune systems play an important role in regulating gastrointestinal diseases. ADRA2A, belonging to class A G protein–coupled receptors, plays a critical role in modulating the inflammatory processes (Leong et al., 2010). ADRA2A is reported to be associated with the development of irritable bowel syndrome and colitis (Sikander et al., 2010). Moreover, VIP, mainly expressed by enteric neurons, participates in maintaining intestinal homeostasis (Lelievre et al., 2007; Talbot et al., 2020; Zhou and Sonnenberg, 2020). Food consumption elevated the production of VIP in the intestine and inhibited the excessive production of IL-22 derived from VIPR2-expressed ILC3s, which increased lipid absorption and played a protective role in the intestinal barrier (Talbot et al., 2020). In contrast, some studies showed that VIP plays a proinflammatory role in IBD (Abad et al., 2003). Our results showed that FOXO1 suppressed the expression of ADRA2A and promoted the expression of VIPR2 in ILC3s to maintain intestinal homeostasis. The stress-associated ADRA2A signaling and diet-associated VIPR2 signaling worked in concert to regulate the function of ILC3s, indicating that stress and diet signaling balanced intestinal homeostasis. Activation of ADRA2A signaling by stress and suppression of VIPR2 signaling by fasting might be risk factors for colitis.

Chronic restraint stress is an early trigger for anxiety and depression disorder (de Kloet et al., 2005). Several studies showed that stress impaired the function of the digestive system and caused IBD bidirectionally (Gao et al., 2018; Sun et al., 2019). Chronic stress induces intestinal inflammation by disrupting gut barrier and hence promoting inflammatory cell infiltration and microbiota disturbance (Zheng et al., 2009). We found that the chronic restraint stress promoted hyperactivation of ILC3s and induced colitis via the cAMP–FOXO1 axis. Elevation of cAMP downregulated FOXO1 level, accompanied by an increased level of ADRA2A and downregulation of VIPR2 expression in ILC3s. The connection between neuronal signaling pathways and immune regulation mediated by FOXO1 might be exploitable in the therapies for stress-related intestinal disorders.

In summary, we revealed that cAMP–FOXO1 signaling balanced intestinal homeostasis via neuronal signaling in ILC3s. Our study showed a regulatory circuit between the immune system and the nervous system in the maintenance of gut homeostasis, which might provide potential targets for treating intestinal inflammation.

Anti-mouse CD3 (17A2), anti-mouse CD19 (eBio1D3), anti-mouse NK1.1 (PK136), anti-mouse KLRG1 (2F1), anti-mouse CD45.2 (104), and anti-mouse CD127 (SB/199) were purchased from eBioscience; anti-mouse TCRβ (H57-597), anti-mouse TCRγδ (GL3), anti-mouse TER-119 (TER-119), anti-mouse Gr-1 (RB6-8C5), anti-mouse CD11b (M1/70), anti-mouse CD11c (N418), anti-mouse CD90.2 (30-H12), anti-mouse IL-17A (TC11-18H10.1), anti-mouse IL-17F (9D3.1C8), anti-mouse IL-22 (Poly5164), anti-human Lineage (CD3/14/19/20/56; UCHT1/HCD14/HIB19/2H7/HCD56), anti-human CD45 (2D1), anti-human CD117 (104D2), anti-human CD127 (IL-7Rα), and anti-GFP (FM264G) were purchased from BioLegend; anti-mouse RORγt (Q31-378) was purchased from BD Bioscience; anti-ADRA2A (NB100-2819) was purchased from Novus Biologicals; anti-VIPR2/VPAC2 (SP235) and anti-VIP (EPR23288-43) were purchased from Abcam; anti-FOXO1 (2880) and anti-phospho-FOXO1 (9464) were purchased from Cell Signaling Technology. Goat anti-rat IgG (H+L) and goat anti-rabbit IgG (H+L) were purchased from Invitrogen. Recombinant murine IL-7 was purchased from PeproTech; recombinant murine IL-23 was purchased from R&D. Brefeldin A was purchased from eBioscience. PMA, ionomycin, N6,2′-O-dibutyryladenosine 3′,5′-cyclic monophosphate sodium salt (cAMP; D0260), and clonidine hydrochloride (C7897) were purchased from Sigma. Yohimbine HCl (V1127) was purchased from InvivoChem. BAY 55-9837 (2711) was purchased from Tocris. PG99-465 (Myristoyl-(Lys^12.27.28)-VIP-Gly-Gly-Thr [free acid]) was purchased from Bachem.

C57BL/6 mice (male and female) were obtained from Beijing Vital River Laboratory Animal Technology. Foxo1^flx/flx mice were kind gifts from Dr. Ming Li (Memorial Sloan-Kettering Cancer Center, New York, NY, USA; Ouyang et al., 2012). Adra2a^flx/flx mice were generated by Cyagen Bioscience. Rorc^cre mice were purchased from the Jackson Laboratory. Rag1^−/− mice were from GemPharmatech Co., Ltd. RORγt-GFP mice were kind gifts from Dr. Xiaohuan Guo. All transgenic mice were bred and maintained in the animal facility of the Institute of Microbiology, Chinese Academy of Sciences (IMCAS) in specific pathogen–free conditions. Age- and sex-matched littermates between 8 and 16 wk of age were used in all experiments. Both male and female mice were used in experiments. Mice were assigned randomly to experimental groups. All animal protocols were approved by Animal Ethics Committee in IMCAS and conducted in compliance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the IMCAS Ethics Committee.

All experiments involving human samples were approved by the Medical Ethics Committee of the IMCAS and Seventh Medical Center of Chinese PLA General Hospital. All human participants provided informed consent. Healthy volunteers and patients with CD or UC were recruited from Seventh Medical Center of Chinese PLA General Hospital. The diagnosis of CD and UC was based on clinical symptoms, endoscopic examination, and histological scores, and the disease severity was tested based on CD activity index or Mayo scores for UC. All patients were invited to report their stress symptoms by means of an adapted version of the self-rating anxiety scale for stress. The intestinal biopsies were taken from participants undergoing routine endoscopy and peripheral blood was collected with procoagulant tubes for further analysis.

Euthanasia was by CO₂ inhalation followed by cervical dislocation. The intestines from euthanized mice were removed from mesenteric fat tissue and washed in ice-cold PBS. Peyer’s patches were identified and removed. Next, intestines were opened longitudinally and subsequently cleaned in cold PBS to remove fecal contents. Then, intestines were cut into pieces and epithelial cells were removed by three times 20-min incubation on a horizontal shaker with digestion buffer I (5 mM EDTA Ca²⁺ and Mg²⁺ free Hank’s medium) at 37°C. After each washing step, tissue was vortexed and the epithelial fraction discarded. Then, remaining tissue was cut into fine pieces and digested by two 40-min incubations on a horizontal shaker with digestion buffer II containing 10% FBS, Collagenase II and III (1 mg ml⁻¹; Worthington), DNase I (200 μg ml⁻¹; Roche), and dispase (4 U ml⁻¹; Sigma-Aldrich) at 37°C. After each digestion, cells were harvested and filtered through a 70-μm cell strainer. Purified cells were washed and resuspended in PBS supplemented with FBS (0.5%) and stained for flow cytometry analysis or sorting. Mesenteric lymph nodes were mechanically disrupted and passed through 70-μm cell strainer for single-cell suspension harvest.

Single-cell suspensions were labeled for flow cytometry analysis using fluorophore-conjugated antibodies for 40 min on ice. For intracellular staining, cells were fixed and permeabilized using a Foxp3/Transcription Factor Staining Buffer Set (eBioscience). Then, cells were incubated with intracellular antibodies for 50 min on ice. FACS data were acquired using BD FACSCanto II (BD Biosciences) or BD FACSAria III (BD Biosciences), and analysis was carried out using FlowJo software (V.10.0.8; Tree Star).

The following antibodies were used to label the ILC3s for sorting of flow cytometry: Lin (CD3, CD19, TCRβ, TCRγδ, TER-119, Gr-1, CD11b, CD11c)⁻NK1.1⁻KLRG1⁻CD45^lowCD90.2^hiCD127⁺. The following antibodies were used for ILC3s analysis of flow cytometry: Lin (CD3, CD19, TCRβ, TCRγδ, TER-119, Gr-1, CD11b, CD11c)⁻NK1.1⁻KLRG1⁻CD45⁺CD127⁺RORγt⁺. For inflammatory cytokine (IL-17A, IL-17F, and IL-22) staining, cells were restimulated in completed RPMI1640 media with PMA (50 ng ml⁻¹; Sigma-Aldrich), ionomycin (1 μM; Sigma-Aldrich), and Brefeldin A (eBioscience) at 37°C, 5% CO₂ for 4 h before staining.

Single-cell suspensions from human intestinal tissues were obtained by incubating tissues on a horizontal shaker with digestion buffer I (5 mM EDTA Ca²⁺ and Mg²⁺ free Hank’s medium) at 37°C for 30 min to remove the epithelial layer. Tissues were then cut into small pieces and incubated in the same way as the murine intestine cells harvest method. For flow cytometry, human ILC3s were stained with antibodies according to the following panel: Lin⁻CD45⁺CD127⁺CD117⁺ (Lin = CD3, CD14, CD19, CD20, CD56).

CR (DBS100, ATCC 51459) was cultured overnight in Luria-Bertani broth medium at 37°C, orbital shaking at 180 rpm. After 16 h, the bacteria were centrifuged, the supernatant was discarded, and the pellet was resuspended in sterile PBS. Mice were infected with 2 × 10⁹ colony-forming units (CFU) of CR in 200 μl PBS via oral gavage. CR quantification was measured by plating bacteria in serial dilutions on MacConkey agar plates overnight at 37°C. Mice were sacrificed on day 8 after infection.

Concentrations of IL-17A and IL-22 in mice serum and cell culture supernatants were determined using ELISA MAX Deluxe Set Mouse IL-17A (432504; BioLegend) and ELISA MAX Deluxe Set Mouse IL-22 (436304; BioLegend) according to the manufacturer’s instructions. Quantification of the cAMP concentration was done using Cyclic AMP XP Assay Kit (4339S; Cell Signaling Technology).

Total RNA was extracted using TRIZOL (Invitrogen) and reverse transcribed using FastKing RT Kit (With gDNase; KR116; Tiangen). RT-qPCR analysis was performed with the SuperReal PreMix Plus (SYBR Green; FP205; Tiangen) on an Applied Biosystems QuantStudio 7 machine according to the manufacturer’s instructions. Expression data were normalized to the expression of housekeeping gene Gapdh (reference gene). The following primer sequences were used: Gapdh: 5′-AGG​TCG​GTG​TGA​ACG​GAT​TTG-3′ (forward) and 5′-TGT​AGA​CCA​TGT​AGT​TGA​GGT​CA-3′ (reverse); Adra2a: 5′-GTG​ACA​CTG​ACG​CTG​GTT​TG-3′ (forward) and 5′-CCA​GTA​ACC​CAT​AAC​CTC​GTT​G-3′ (reverse); Vipr2: 5′-GAC​CTG​CTA​CTG​CTG​GTT​G-3′ (forward) and 5′-CAG​CTC​TGC​ACA​TTT​TGT​CTC​T-3′ (reverse); Foxo1: 5′-CCC​AGG​CCG​GAG​TTT​AAC​C-3′ (forward) and 5′-GTT​GCT​CAT​AAA​GTC​GGT​GCT-3′ (reverse). All primer sequences were determined using PrimerBank, a public resource for PCR primers.

To visualize intracellular localization of FOXO1 protein, ILCs in small intestine from WT mice were isolated using MojoSort Mouse Hematopoietic Progenitor Cell Isolation Kit (480003; BioLegend) and cultured in RPMI1640 complete medium supplemented with 10% FBS, 1× penicillin-streptomycin, 2 mM glutamine, 25 ng ml⁻¹ IL-7 (PeproTech), 25 ng ml⁻¹ IL-23 (R&D), and vehicle or cAMP (50 μM) for 16 h. ILCs were stained with PI, FOXO1, Lin (CD3, CD19, CD11b, CD11c, Gr-1, TER-119, TCRβ, TCRγδ, NK1.1), CD45, and RORγt, and analyzed by imaging flow cytometry (Amnis ImageStream MarkII; Merck). The nuclear localization of FOXO1 was analyzed using the IDEAS software v.6.2 (Merck).

ILC3s were isolated from small intestine of WT mice using flow cytometry (ILC3 = Lin⁻NK1.1⁻KLRG1⁻CD45^lowCD90^hiCD127⁺, Lin = CD3, CD19, TCRβ, TCRγδ, TER-119, Gr-1, CD11b, and CD11c). For in vitro experiments of cytokines secretion, 5 × 10⁴ ILC3s were cultured per well in a 96-well round bottom plate in RPMI1640 complete medium supplemented with 10% FBS, 1× penicillin-streptomycin, 2 mM glutamine. ILC3s were stimulated with 25 ng ml⁻¹ IL-7 (PeproTech), 25 ng ml⁻¹ IL-23 (R&D), and vehicle or cAMP with indicated concentrations. The supernatants were collected for ELISA, and the cell pellets were measured by flow cytometry for IL-17A⁺, IL-17F⁺, and IL-22⁺ ILC3s.

Mouse colon and small intestine tissues were prepared in the Swiss roll technique, fixed with 4% paraformaldehyde, and embedded in paraffin. Serial 5-μm sections were stained with hematoxylin and eosin (H&E). The arrowheads indicated the inflammatory lesions with swollen and infiltration of immune cells.

Mouse small intestine tissues were prepared in the Swiss roll technique, fixed in 4% paraformaldehyde overnight, and dehydrated in 30% sucrose overnight at 4°C. Dehydrated small intestine tissues were embedded in optimal cutting temperature compound medium (Tissue-Tek) and stored at −80°C until sectioning at a thickness of 8 μm using a cryotome (Leica). Frozen slides were retrieved at room temperature and excess optimal cutting temperature compound medium was removed in PBS before blocking in PBS with 10% normal goat serum and 0.3% Triton X-100 for 30 min. Tissue sections were then stained with the following primary antibodies diluted in blocking buffer overnight at 4°C: anti-GFP 1:100 and anti-VIP 1:500. Sections were washed three times with PBST (PBS + 0.05% Tween-20) and then incubated with secondary antibodies (anti-rat IgG-594 and anti-rabbit IgG-488) diluted 1:500 in blocking buffer for 1 h at room temperature. Tissue sections were then washed three times in PBST and incubated with anti-mouse CD3 (17A2) diluted 1:500 in blocking buffer for 2 h at room temperature. Tissue sections were then washed three times in PBST, incubated with DAPI (Sigma-Aldrich) for 3 min before a final wash in PBS, and mounted with antifade regent. Stained sections were imaged on a confocal microscope (Leica SP8).

10x Genomics scRNA-seq was performed as previously described (Ma et al., 2023). Briefly, we isolated mouse intestinal CD3⁺ cells (Lin⁻CD45.2⁺CD3⁺, Lin = CD19, CD11b, CD11c, TER-119, Gr-1) and ILC3s (Lin⁻CD3⁻CD45^lowCD127⁺CD90^hi). Sorted cells were encapsulated into droplets, and libraries were prepared using Chromium Single Cell 3′ Regent Kit V3.1 according to the manufacturer’s protocol (10x Genomics). scRNA-seq libraries were sequenced on an Illumina Novaseq 6000. Reads were processed by CellRanger software (10x Genomics) using the mm10 reference genome. According to the flow cytometry analysis, the same mouse strains have similar lymphocyte distributions. Therefore, we pooled ILC3s and CD3⁺ T cells from three mice of the same mouse strain together as one sample for scRNA-seq per experimental condition. One scRNA-seq experiment was run for each sample.

Downstream analysis was performed with Seurat v.4.1.2. We further filtered out low-quality cells and cell doublets with <200 genes detected, >5,000 genes detected, or >10% mitochondrial reads. Normalization was performed with Seurat (v.4.1.2) and integrated by canonical correlation analysis. The graph representing cells with similar expression patterns was generated with the FindNeighbors function using the 20 largest principal components. Cell clusters were generated using the Louvain algorithm implemented by the FindClusters function with the resolution parameter equal to 0.9. Marker genes for each cluster were determined using the Wilcoxon test on the raw counts, implemented by the function FindAllMarkers, and included only positive marker genes with log-fold changes >0.25 and Bonferroni-corrected P values <0.01. Dimensionality reduction by t-distributed stochastic neighbor embedding was performed using the RuntSNE function with the 20 largest principal components. All visualizations of scRNA-seq data were generated using the Seurat package as well as ggplot2 version 4.1.2.

The CRS procedure was performed based on previous studies (Li et al., 2020). In brief, the C57BL/6J mice were placed in 50 ml conical centrifuge tubes with ventilation holes for 4 h (10:00 am–2:00 pm) per day for 29 d. A hole was created in the center of the lid to allow the tail of the mouse to pass through. The stress (CR) group mice were infected 2 × 10⁹ CR via oral gavage on day 21 of CRS treatment and sacrificed on day 8 after CR infection. Mice were not allowed forward and backward movement, were not physically compressed, and did not experience pain.

Mice were placed individually in a 3-liter glass chamber filled with 15 cm of water (temperature 25 ± 1°C) for 6 min. The lengths of periods of immobility or struggling during the 6-min test period were measured. Water was replaced between every test. Following swim sessions, mice were removed from the cylinder, dried with a towel, and returned to their home cages.

Mice were individually suspended by the distal portion of their tails using adhesive tape, keeping their head 15 cm from the table. The test sessions lasted 6 min, and the immobility time was determined by a skilled observer. The mice were regarded as immobile only when they hung passively and fully motionless.

FOXO1 genome-wide binding regions in ILC3s were detected by CUT&Tag assay using Hyperactive In-Situ ChIP Library Prep Kit for Illumina (TD901; Vazyme Biotech Co., Ltd) according to the manufacturer’s recommendation. Briefly, ConA Beads (10 μl per sample) were washed with ConA Binding Buffer. 1 × 10⁵ ILC3s were isolated and washed with wash buffer. The cells were mixed with ConA beads and incubated with anti-FOXO1 (Abcam) or anti-IgG overnight at 4°C. Goat anti-rabbit IgG antibody was added into the sample and incubated for 1 h at room temperature. After washing away the unbounded secondary antibody, 0.04 μM Transposon was added and incubated with cells for 1 h at room temperature. The sample was washed by Dig-300 buffer and followed with incubation by Tagmentation Buffer for 1 h. Next, DNA was isolated and amplified with P5 and P7 primers and purified for high-throughput sequencing. Paired-end sequencing was carried out with the Illumina NovaSeq 6000 with read length of 150 bp. Paired-end reads were aligned using Bowtie2. The peak visualization in genome was shown by IGV software.

1 × 10⁵ ILC3s from small intestine of WT mice were analyzed by FOXO1 CHIP-qPCR using ChIP-IT High Sensitivity kit (53040; Active Motif) according to the manufacturer’s instruction. Anti-FOXO1 (39670; Abcam) was used for immunoprecipitation. Primers used for ChIP-qPCR were as follows: Adra2a-N: 5′-CGT​GAT​CTT​CCT​GCT​GCT​CGT​C-3′ (forward) and 5′-GCT​TGG​CAG​TAG​ATG​GCA​AGG​T-3′ (reverse); Adra2a-B1: 5′-GGG​ACA​GAC​AGC​CAT​CTT​GGT​T-3′ (forward) and 5′-GCA​CCT​TTA​CCC​TGC​AAC​CCT​T-3′ (reverse); Adra2a-B2: 5′-TGT​CAT​GGC​GCT​CTT​CAT​AAA​G-3′ (forward) and 5′-CTC​TCT​GCT​TGC​CTG​GCT​TT-3′ (reverse); Vipr2-N: 5′-GGC​ACA​ACA​CAA​GGC​TGA​ACT​C-3′ (forward) and 5′-GGT​GAC​TGC​TTG​GTT​CTG​ATG​G-3′ (reverse); Vipr2-B1: 5′-GCT​GCT​GCT​GCT​TTG​GGT​ATG​A-3′ (forward) and 5′-TCC​TGG​CAC​CTT​CCA​CTT​GAG​T-3′ (reverse); Vipr2-B2: 5′-ACC​TTC​GCT​CCT​CCA​GGA​TAC​C-3′ (forward) and 5′-CCA​GGG​TTC​CAA​CCA​CAG​CAA​T-3′ (reverse).

All reporter assays were performed by co-transfection of HEK293T cells with pcDNA4-FOXO1 and pGL3 vectors containing different regions of the mouse Adra2a promoter or Vipr2 promoter. Firefly and Renilla luciferase actives were measured by the Dual-Luciferase Reporter Assay System (E2920; Promega). Data were normalized to the activity of Renilla luciferase. Cell line HEK293T was tested negative for mycoplasma contamination.

Statistical analysis was performed using GraphPad Prism v9. All statistical graphs represented the mean ± SD and were determined by unpaired two-tailed Student’s t test, one-way ANOVA, and two-way ANOVA according to the type of experiments. The statistical significances of differences (*, P < 0.05, **, P < 0.01, ***, P < 0.001, ****, P < 0.0001) are specified throughout the figures and legends. All flow cytometry data were analyzed with FlowJo (Treestar). No statistical methods were used to predetermine sample sizes.

Fig. S1 shows that FOXO1 deficiency in ILC3s and Th17 cells results in intestinal inflammation. Fig. S2 shows that FOXO1 suppresses the expression of inflammatory cytokines in ILC3s. Fig. S3 shows that FOXO1 regulates ADRA2A and VIPR2 signaling in ILC3s. Fig. S4 demonstrates the work model for FOXO1-induced neuronal signaling regulation of ILC3s in the intestine. Table S1 shows the quality control metrics of scRNA-seq data. Table S2 shows the colitis-related gene set.

The scRNA-seq and CUT&Tag data generated in this study have been deposited in the Genome Sequence Archive (GSA) of the National Genomics Data Center, China National Center for Bioinformation database under accession code CRA011143 and CRA011145. All other relevant data are available from the corresponding author upon reasonable request.

We thank Dr. Ming Li (Memorial Sloan-Kettering Cancer Center, USA) for providing Foxo1^flx/flx mice. We thank Dr. Xiaohuan Guo (Tsinghua University, Beijing, China) for providing RORγt-GFP mice. We thank Dr. Mingzhao Zhu (Institute of Biophysics, Chinese Academy of Sciences, Beijing, China) for providing CR. We thank Dr. Pengyan Xia (Peking University, Beijing, China) for technical support. We thank Tong Zhao (Institute of Microbiology, Chinese Academy of Sciences, Beijing, China) for technical support. Cartoons in Fig. 6 A and Fig. S4 and graphical abstract were created with BioRender.com.

This work was supported by the National Key R&D Program of China (2021YFA1300202, 2022YFC2302900, 2020YFA0803501), the National Natural Science Foundation of China (92169113, 31930036, 81921003), Key Research Program of Frontier Sciences of Chinese Academy of Sciences (ZDBS-LY-SM025), CAS Project for Young Scientists in Basic Research (YSBR-010), and Youth Innovation Promotion Association of CAS to S. Wang.

Author contributions: F. Shao performed experiments, analyzed data, and wrote the paper; Z. Liu performed experiments and analyzed data; Q. Wei provided IBD tissue samples and analyzed data; D. Yu, M. Zhao, X. Zhang, and X. Gao performed experiments. Z. Fan provided mouse models. S. Wang initiated and organized the study, and organized, designed, and wrote the paper.