Abstract

   Innate lymphoid cells (ILCs) play critical roles in innate immunity,
   epithelial barrier protection, and tissue homeostasis. However, the
   maintenance machinery of intestinal tissue residency of ILCs remains
   elusive. Here, we show that gut microbiota is necessary for the
   maintenance of intestinal tissue residency of ILCs. Microbiota
   metabolite taurodeoxycholic acid (TDCA) binds to P2Y10 receptor on ILCs
   to initiate downstream Ca^2+ and RhoA signaling pathways. TDCA-P2Y10
   engagement induces Zfp414 transcription to prime expression of CD69 and
   integrin αE on ILCs, leading to intestinal residency of ILCs. Moreover,
   decreased levels of TDCA or P2Y10 deficiency abrogates the intestinal
   residency of ILCs, resulting in severer intestinal inflammation. Of
   note, TDCA administration can enhance intestinal tissue residency of
   ILCs and promote protection against intestinal inflammation. Thus, TDCA
   might be used as a potential drug to treat patients with inflammatory
   bowel disease.
     __________________________________________________________________

   Microbiota metabolite TDCA binds P2Y10 on intestinal ILCs to maintain
   tissue residency through CD69 and integrin alphaE.

INTRODUCTION

   Innate lymphoid cells (ILCs) are important guardians in mucosal
   immunity, contributing to barrier immunity, tissue homeostasis, and
   immune regulation. ILCs are classified into three groups on the basis
   of their development and functional characteristics: group 1 (ILC1s),
   group 2 (ILC2s), and group 3 (ILC3s) ([48]1). We also recently defined
   a subpopulation of regulatory ILCs called ILCreg that exist in the gut
   and suppress activation of ILC1s and ILC3s via interleukin-10 (IL-10)
   secretion, leading to protection against intestinal inflammation
   ([49]2). Based on their secretion of specific cytokines, each group of
   ILCs specializes in unique innate immunity against different types of
   pathogens, such as intracellular pathogens, helminths, and
   extracellular pathogens, respectively ([50]3). Tissue residency is a
   hallmark of intestinal ILCs. Upon seeding intestinal tissues during
   fetal development, ILCs maintain tissue sedentary lifestyle and
   replenish their populations predominantly by local expansion ([51]4,
   [52]5). Although long-term parabiosis experiments revealed that, under
   a steady-state condition, resident ILCs keep tissue dwelling, how the
   permanent resident state of ILC populations in the intestine is
   maintained and where ILCs detaching from intestinal tissues migrate
   remain elusive.

   Gut microbiota plays a critical role in shaping an immune system and
   maintaining intestinal homeostasis. We previously showed that bacterial
   metabolite N-undecanoylglycine induces intestinal Tuft-2 cell expansion
   for antimicrobial immunity ([53]6). Recently, accumulating evidence
   suggests that metabolites produced by microbiota regulate the function
   of ILCs ([54]7). For example, tryptophan metabolites derived from
   microbiota promote IL-22 production of ILC3s via activating aryl
   hydrocarbon receptor ([55]8). Likewise, short-chain fatty acids (SCFAs)
   produced by commensal microbiota from dietary fibers differentially
   affect intestinal ILCs in a subset-specific manner ([56]9). They
   promote proliferation and IL-22 production of ILC3s, while they inhibit
   expansion of ILC2s through G protein–coupled receptor 41 (GPR41)
   ([57]10). Recent studies have reported the role of another microbiota
   metabolite, bile acid metabolite, in regulating ILCs’ function and
   intestinal homeostasis. Increased levels of cholic acid activate
   intestinal ILC2 to produce IL-5, leading to increased infiltration of
   eosinophils into intestinal tissues and inducing protective type II
   inflammatory response against worm infection ([58]11). Glycodeoxycholic
   acid binds to TGR5 receptor on ILC3 and activates downstream cyclic
   adenosine 3′,5′-monophosphate signaling pathways, resulting in
   activation of transcription factor (TF)GATA binding protein 3 (GATA3)
   and promotion of IL-22 secretion ([59]12). However, whether microbiota
   and their metabolites sustain retention of ILCs in the intestine is
   unclear.

   As environmental sensors, G protein–coupled receptors (GPCRs) have been
   recognized to maintain retention of tissue-resident lymphocytes. For
   example, adhesion GPCR family member-E5 (Adgre5 or CD97) ensures
   retention of cDC2 in the spleen through mechanosensing red blood cells
   ([60]13). P2Y receptors belong to the GPCR superfamily and are
   expressed in various lymphocytes, implicating their regulatory effects
   on immune system in many ways, such as degranulation, phagocytic
   clearance, cytokine production, and lymphocytes migration
   ([61]14–[62]16). However, how P2Y receptors regulate ILCs residency
   remains unknown. Tissue-resident lymphocytes display unique phenotypic
   characteristics and rely on chemokine receptors and adhesion molecules
   for their retention within tissues, including CXCR6, CCR6, integrin α1,
   and CD69 ([63]5, [64]17, [65]18). Notably, the lectin receptor CD69, a
   marker for tissue-resident cells, is widely expressed in various
   lymphocytes, including tissue-resident memory T (Trm), B (Brm), and
   natural killer (NK) cells ([66]17, [67]19, [68]20). Here, we showed
   that gut microbiota is necessary for the maintenance of intestinal
   tissue residency of ILCs. Microbiota metabolite taurodeoxycholic acid
   (TDCA) is a ligand to P2Y10 receptor that is highly expressed on
   intestinal ILCs. TDCA-P2Y10 engagement initiates downstream Ca^2+ and
   RhoA signaling pathways to promote Zfp414 transcription, leading to
   elevated expression of CD69 and integrin αE (ITAE; also called CD103)
   on ILCs. TDCA administration can enhance intestinal tissue residency of
   ILCs and promote protection against intestinal inflammation.

RESULTS

Gut microbiota is necessary for the maintenance of ILCs intestinal residency

   To explore the effect of gut microbiota on regulating tissue residency
   of ILCs, we analyzed ILCs in intestinal tissues and determined their
   localization in germ-free (GF) mice. We found that intestinal ILCs
   within intestinal epithelia and lamina propria were much lower in GF
   mice compared with those in conventional specific pathogen–free (Ctrl)
   mice ([69]Fig. 1A and fig. S1A). In addition, intestinal
   immunofluorescence staining showed a notable reduction in numbers of
   three ILC subsets in GF mice (fig. S1, B to D). Consistent with GF
   mice, frequencies and numbers of intestinal ILC subsets were decreased
   in mice treated with broad-spectrum antibiotic (Abx) ([70]Fig. 1B and
   fig. S1, A and E). In addition, antibiotic treatment decreased ILC2s
   and ILC3s in Peyer’s patches (PPs) but had no obvious impact on ILC1s
   (fig. S1F). Furthermore, we found that proliferation and apoptosis
   rates were comparable, suggesting that ILC reduction in intestinal
   tissues is not caused by changes of cell proliferation and apoptosis
   (fig. S1G). Given that intestinal and peripheral lymphoid tissues are
   closely interconnected, intestinal lymphocytes are capable of
   trafficking between them. We then examined whether reduced intestinal
   ILCs migrate to lymphoid tissues such as mesenteric lymph nodes (mLNs)
   and spleen. We observed that percentages and numbers of ILCs from mLN
   and spleen were increased in GF ([71]Fig. 1A) and Abx-treated mice
   ([72]Fig. 1, C and D; and fig. S2, A and B) compared with those in Ctrl
   mice. It has been reported that KLRG1^+ ILC2s represent activated
   ILC2s. To verify whether the reduction of intestinal ILC2s caused by
   Abx was due to their migration rather than activation, we sorted
   intestinal ILC2s for in vitro activation assays. We found that
   antibiotic treatment did not affect the activation of ILC2s (fig. S2, C
   and D). Meanwhile, we also used GATA3 as the marker for total ILC2s.
   Consistent with the findings on KLRG1^+Sca-1^+ ILC2s, antibiotic
   treatment decreased numbers of intestinal GATA3^+ ILC2s, while
   increased in mLN and spleen (fig. S2E).

Fig. 1. Gut microbiota is necessary for the intestinal maintenance of ILCs.

   [73]Fig. 1.
   [74]Open in a new tab

   (A) Flow cytometry analysis of total ILCs (Lin^−Thy1.2^+CD127^+) from
   intraepithelial lymphocytes (IELs), lamina propria lymphocytes (LPLs),
   mesenteric lymph node (mLN), and spleen in GF and Ctrl mice. Numbers in
   flow cytometry plots represent percentage of ILCs in each gate. n = 4.
   (B to D) Flow cytometry analysis of ILC1s
   (Lin^−CD127^+CD45.2^+NK1.1^+NKp46^+), ILC2s
   (Lin^−CD127^+Thy1.2^+KLRG1^+Sca-1^+), and ILC3s
   (Lin^−CD127^+Thy1.2^+RORγt^+) in the SI (B), mLN (C), and spleen (D)
   from Ctrl and Abx-treated mice that were fed water containing a
   combination of antibiotics (Abx) for 2 weeks. n = 6. (E to G)
   Intestinal ILC1s (E), ILC2s (F), and ILC3s (G) located in lymphatics
   from Ctrl and Abx-treated mice were visualized by three-dimensional
   (3D) immunofluorescence staining. Lyve1 and EpCAM denote the lymphatic
   vessels and intestinal epithelia separately. White arrows indicate ILC
   subsets within lymphatic vessels. Migratory ILCs subsets were regarded
   as cells within lymphatic vessels. Scale bars, 70 μm. Frequencies of
   migratory ILC subsets in total ILCs located in a villus were
   calculated. Each symbol in the dot plots indicates one villus. n = 11.
   Data are representative of at least three independent experiments and
   are shown as the means ± SD. Statistical analysis was performed by
   using unpaired two-tailed Student’s t test (*P < 0.05; **P < 0.01;
   ***P < 0.001). SSC, side scatter; W, width.

   As lymphocytes egress from small intestine (SI) and enter into lymphoid
   tissues via lymphatic vessels ([75]21), we conducted three-dimensional
   (3D) immunofluorescence of intestinal villi to visualize ILCs
   localization. Lyve1 staining was used to delineate lymphatic vessels
   ([76]22). In a steady state, ILCs predominantly resided in lamina
   propria, and only a rare fraction entered into lymphatics. After Abx
   treatment, much more ILCs exited intestine and entered into lymphatic
   vessels. All of three ILC subsets displayed the same phenomena
   ([77]Fig. 1, E to G). Collectively, these data indicate that microbiota
   environment is necessary for the maintenance of tissue residency of
   intestinal ILCs.

Microbiota disruption causes migration of resident ILCs from SI to a
circulation system

   Previous studies have reported that S1PR1 receptor regulates migration
   of central memory T cells (Tcm) from peripheral nonlymphoid tissues to
   lymphoid tissues, and the function of S1PR1 can be antagonized by CD69
   ([78]23). To determine whether CD69 makes similar contribution to
   tissue residency of intestinal ILCs in a microbiota-dependent manner,
   we first examined expression of CD69 and S1PR1 on ILCs in Abx-treated
   versus Ctrl mice. We noticed that CD69^+ ILCs were markedly decreased
   in Abx-treated mice, whereas S1PR1^+ ILCs were increased ([79]Fig. 2A
   and fig. S2, F and G). To determine that the reduction of intestinal
   ILCs caused by Abx treatment is due to their reentry into circulation
   and migration out of SI, we simultaneously treated mice with Abx and
   S1PR1 inhibitor FTY720 to prevent the entrance into lymphatic vessels.
   Flow cytometry analysis confirmed that intestinal ILC1s, ILC2s, and
   ILC3s were increased in Abx and FTY720 cotreated mice compared to those
   in mice treated with Abx alone ([80]Fig. 2B). In contrast, three ILC
   subsets from mLN ([81]Fig. 2C) and spleen ([82]Fig. 2D) were decreased
   in these cotreated mice. In addition, after Abx treatment, numbers of
   S1PR1^+ ILCs in the mLN and spleen increased substantially, while
   blockade of S1PR1 with FTY720 reduced accumulation of migratory ILCs
   (fig. S3A). Moreover, we sorted intestinal ILCs from tdTomato-reporter
   mice and performed adoptive transfer experiments. We found that, with
   increasing duration of antibiotic treatment, numbers of tdTomato^+ ILCs
   decreased in the intestine but increased in the spleen (fig. S3, B to
   D). These results indicate that microbiota disruption abrogates the
   tissue residency of intestinal ILCs and enhances their S1PR1-dependent
   egression.

Fig. 2. Microbiota disruption causes migration of intestinal ILCs to a
circulation system.

   [83]Fig. 2.
   [84]Open in a new tab

   (A) 3D Immunofluorescence staining of CD69^+ and S1PR1^+ ILCs in SI
   from Ctrl and Abx-treated mice. White arrows denote CD69^+ and S1PR1^+
   ILCs. Scale bars, 100 μm. (B to D) Mice were intraperitoneally injected
   FTY720 every other day and treated with Abx for 2 weeks. ILC1s, ILC2s,
   and ILC3s in SI (B), mLN (C), and spleen (D) derived from Ctrl,
   Abx-treated, and Abx + FTY720–treated mice were analyzed by flow
   cytometry. n = 5. (E to G) PE-labeled anti-CD45 antibody (2 μg) was
   intravenously injected into mice. Ten minutes later, PE-labeled
   intestinal ILC1s (E), ILC2s (F), and ILC3s (G) from Ctrl and
   Abx-treated mice were detected by flow cytometry. Frequencies of
   PE-labeled ILC1s, ILC2s, and ILC3s were calculated. n = 4. (H) Lin^−
   intestinal lymphocytes (1 × 10^7) from CD45.2^+ donor mice were sorted
   and intravenously injected into lethally irradiated CD45.1^+ Ctrl or
   Abx-treated recipients. Eighteen hours later, donor-derived ILC subsets
   homing to intestine were analyzed by flow cytometry. Frequencies and
   numbers of intestinal CD45.2^+ ILC subsets in indicated recipients are
   shown in the right panel. n = 4. (I) Bone marrow (BM) cells (5 × 10^6)
   from CD45.2^+ donor mice were intravenously injected into lethally
   irradiated CD45.1^+ Ctrl or Abx-treated recipients. Eight weeks later,
   homing and stable localization of donor-derived ILC subsets in the
   intestine were analyzed by flow cytometry. Frequencies and numbers of
   intestinal CD45.2^+ ILC subsets in indicated recipients are shown in
   the right panel. n = 4. Data are representative of at least three
   independent experiments and are shown as the means ± SD. Statistical
   analysis was performed by using unpaired two-tailed Student’s t test
   (*P < 0.05; **P < 0.01; ***P < 0.001; n.s., not significant). SSC, side
   scatter; W, width.

   To directly detect the circulatory ILCs, we performed intravascular
   antibody staining to label cells in vascular systems. Consistent with
   our above observations, much more ILC1s ([85]Fig. 2E), ILC2s ([86]Fig.
   2F), and ILC3s ([87]Fig. 2G) from Abx-treated mice were labeled by
   intravascular antibody. Next, we determined whether a stable microbiota
   environment is necessary for homing and positioning of ILCs in SI.
   Intestinal Lin^− lymphocytes isolated from CD45.2^+ donor mice were
   injected intravenously into CD45.1^+ recipients for a short-term homing
   assay. In Abx recipients, ILC subsets were lower efficient in migrating
   to SI compared to Abx-untreated (Ctrl) recipients ([88]Fig. 2H).
   Furthermore, we transferred CD45.2^+ bone marrow (BM) cells into same
   CD45.1^+ Abx recipients for a long-term homing assay to determine the
   importance of gut microbiota on retention of intestinal ILCs. Eight
   weeks later, Abx recipients showed a lower proportion and fewer numbers
   of intestinal ILC subsets ([89]Fig. 2I), whereas higher percentages and
   greater numbers of ILC subsets were observed in mLN and spleen (fig.
   S3, E and F), suggesting that the intestinal environment with damaged
   microbiota is not conducive to homing and residency of ILCs. Together,
   our results reveal that disruption of gut microbiota promotes
   intestinal resident ILCs to exit their dwelling tissues and enter a
   circulation system.

P2Y10 is highly expressed on ILCs and mediates intestinal tissue residency

   Intestinal ILCs sense different kinds of contents derived from gut
   microbiota, which plays a critical role in maintaining tissue
   homeostasis ([90]10, [91]24). However, it remains unclear whether gut
   microbiota metabolites contribute to regulate the long-term stability
   of intestinal ILCs in position. To test the effect of gut microbiota on
   regulation of ILCs transcriptome, we conducted RNA sequencing (RNA-seq)
   analysis of ILCs from Ctrl versus Abx-treated mice. We noticed that
   ILCs from Ctrl mice markedly up-regulated Trm signature genes, whereas
   ILCs from Abx-treated mice were more similar to Tcm (fig. S4A). We also
   conducted same transcriptome analysis of ILC1, ILC2, and ILC3 from
   single-cell RNA sequencing (scRNA-seq) data ([92]25) and found that Trm
   signature genes were enriched in all three ILC subsets from Ctrl mice
   (fig. S4B). Additionally, we found that genes up-regulated in Ctrl ILCs
   were associated with tissue residency, mainly encoding integrins and
   some adhesive receptors. By contrast, genes enriched in Abx ILCs were
   related to cell migration and mobilization (fig. S4, C and D).
   Expectedly, gene ontology (GO) pathway enrichment analysis showed that
   pathways associated with innate immune response and cell-cell adhesion
   were enriched in Ctrl ILCs. GPCR activity and downstream signaling
   pathways displayed prominent enrichment ([93]Fig. 3A). In recent years,
   numerous studies have shown that GPCRs bind to various microbial
   metabolites and participate in immune and metabolic regulation in the
   intestine. Gut microbiota typically exerts its functions by recognizing
   GPCRs and modulating downstream signaling ([94]24, [95]26, [96]27).
   Then, we analyzed expression of GPCRs and chose top 10 GPCR genes most
   highly expressed in Ctrl ILCs for further quantitative polymerase chain
   reaction (qPCR) validation ([97]Fig. 3B). We found that several
   chemokine receptors were down-regulated in ILCs from Abx-treated mice,
   such as Cxcr6, Ccr5, and Cxcr4, suggesting that gut microbiota
   modulates chemotaxis and migration of ILCs ([98]Fig. 3C). We further
   noticed that another GPCR gene, P2ry10, was highly expressed in Ctrl
   ILCs, while it was reduced after Abx treatment ([99]Fig. 3C and fig.
   S4, E and F). Then, we sorted various immune cells from SI and detect
   the expression of P2ry10, and we found that P2ry10 is most abundantly
   expressed in ILCs, suggesting that P2ry10 plays an important role in
   intestinal ILCs (fig. S4G). Moreover, mature ILC1s, ILC2s, and ILC3s
   exhibited higher P2ry10 expression compared to their precursors (fig.
   S4H). Compared with ILCs from lymphoid tissues such as spleen and mLN,
   SI-derived ILCs showed higher expression of P2ry10 (fig. S4I).
   Consistently, more ILCs expressing P2Y10 were observed in Ctrl mice
   compared with those in microbiota-deficient mice ([100]Fig. 3D and fig.
   S4, J to L). These data indicate that Abx treatment suppresses resident
   signatures of ILCs and down-regulates P2ry10 expression on intestinal
   ILCs.

Fig. 3. P2Y10 is highly expressed on ILCs and mediates intestinal residency
of ILCs.

   [101]Fig. 3.
   [102]Open in a new tab

   (A) Transcriptome analysis of sorted intestinal ILCs
   (Lin^−CD127^+Thy1.2^+) in Ctrl and Abx-treated mice. Highly expressed
   genes in Ctrl ILCs were performed GO enrichment analysis. (B)
   Expression of GPCRs was shown, and top 10 of the most specifically and
   highly expressed GPCRs in Ctrl ILCs were marked. (C) Relative mRNA
   levels of the candidate GPCRs in sorted Ctrl and Abx ILCs. Fold changes
   were normalized to Actb. (D) Immunofluorescence of P2Y10-expressing
   intestinal ILCs (white arrows). Scale bars, 30 μm. (E to G) Flow
   cytometry analysis of ILC1s, ILC2s, and ILC3s in the SI (E), mLN (F),
   and spleen (G) from P2ry10^+/+ and P2ry10^−/− mice. n = 5. (H)
   Immunofluorescence of the SI from P2ry10^+/+ and P2ry10^−/− mice for
   analysis of gut-associated lymphoid tissues. Scale bars, 2000 μm
   (left), 70 μm (right). n = 5. (I) Flow cytometry analysis of CD69^+ and
   S1PR1^+ ILCs derived from SI of P2ry10^+/+ and P2ry10^−/− mice. n = 5.
   (J) P2ry10^+/+ or P2ry10^−/− CD45.2^+ donor mice underwent parabiosis
   surgery with CD45.1^+ recipient mice. Flow cytometry analysis of
   donor-derived intestinal ILC1s, ILC2s, and ILC3s from the CD45.1^+
   parabionts. n = 4. (K) Relative mRNA levels of P2RY10 in ILCs from
   noninflamed and inflamed intestinal samples of patients with CD. Fold
   changes were normalized to ACTB. n = 4 independent experiments. (L)
   Flow cytometry analysis of the expression of P2Y10 in intestinal ILCs
   from patients with CD. n = 6. Data are representative of at least three
   independent experiments and are shown as the means ± SD. Statistical
   analysis was performed by using unpaired two-tailed Student’s t test
   (*P < 0.05; **P < 0.01; ***P < 0.001; n.s., not significant). SSC, side
   scatter; W, width.

   To further examine the impact of P2Y10 deficiency on ILCs retention, we
   generated P2ry10-deficient mice (fig. S5A). Consistent with the
   phenotype of Abx-treated mice, P2ry10^−/− mice showed decreased ILC1s,
   ILC2s, and ILC3s in SI ([103]Fig. 3E and fig. S5B) but increased ILC
   subsets in mLN ([104]Fig. 3F and fig. S5C) and spleen ([105]Fig. 3G and
   fig. S5D) compared with their P2ry10^+/+ littermates. Subsequent
   immunofluorescence staining further confirmed the reduction of ILC1s,
   ILC2s, and ILC3s in the intestine of P2ry10^−/− mice (fig. S5, E to G).
   Given the involvement of ILC3s, particularly CCR6^+ ILC3s in the
   formation of gut-associated lymphoid tissues ([106]28), we used
   immunofluorescence staining to visualize intestinal lymphoid structures
   and found reduced intestinal PPs and isolated lymphoid follicles in
   P2ry10^−/− mice ([107]Fig. 3H). Meanwhile, flow cytometry analysis
   revealed that P2ry10 knockout reduced numbers of ILC2s and ILC3s in
   PPs, but no impact on ILC1s (fig. S5H). In addition, Ki67 and annexin V
   analysis showed that P2Y10 deficiency did not disturb proliferation and
   apoptosis of ILCs (fig. S5, I and J). Moreover, P2Y10 deficiency caused
   decreased expression of CD69 on ILCs, while expression of S1PR1 was
   increased ([108]Fig. 3I and fig. S5K). Then, we conducted BM
   transplantation experiments to investigate whether intestinal homing
   and long-term residency of ILCs are dependent on P2Y10 receptor.
   Because P2ry10 knockout did not affect the proportion of ILC precursors
   (fig. S6A), an equal number (5 × 10^6) of BM cells from P2ry10^+/+ or
   P2ry10^−/− were, respectively, transplanted into irradiated recipients.
   P2ry10-deficient chimeric mice showed reduced homing ILC subpopulations
   in SI compared with control chimeric mice (fig. S6B). Collectively, we
   conclude that P2Y10 receptor is required for intestinal homing and
   subsequent formation of long-term resident ILC populations.

   In addition, we generated parabiosis mice by jointing P2ry10^+/+ or
   P2ry10^−/− CD45.2^+ mice with CD45.1^+ recipient mice to confirm the
   impact of P2Y10 deficiency on the residency of intestinal ILCs (fig.
   S6C). Due to a shared circulation system, circulating T and B cells in
   blood and spleen could reach equilibrium between parabionts (fig. S6D).
   Next, we analyzed proportions of ILC1s, ILC2s, and ILC3s derived from
   different donor mice in recipient parabionts. We found that, compared
   to recipients jointed with wild-type (WT) donor mice, recipients paired
   with P2ry10^−/− donor mice exhibited increased proportions of CD45.2^+
   ILC subsets in SI ([109]Fig. 3J) and blood (fig. S6E). These results
   indicate that loss of P2Y10 receptor disrupts residency of ILCs,
   leading to their entry into circulation and subsequent migration into
   paired parabionts via a shared circulation system. To further detect
   P2Y10 expression in human intestinal ILCs, we firstly analyzed RNA-seq
   data in patients with inflammatory bowel disease (IBD) from the Gene
   Expression Omnibus database ([110]GSE57945) and observed
   down-regulation of P2RY10 in patients with Crohn’s disease (CD) (fig.
   S7A). Subsequently, we collected ileac biopsies from endoscopically
   inflamed and noninflamed sites in patients with CD and then analyzed
   P2Y10 expression in ILCs by qPCR assay, flow cytometry, and
   immunofluorescence staining. We observed that most of ILCs from
   noninflamed intestinal samples expressed P2Y10 and residency marker
   CD69. However, P2Y10^+ and CD69^+ ILCs were decreased in inflamed
   intestine, while S1PR1^+ ILCs were increased ([111]Fig. 3, K and L; and
   fig. S7, B to E), suggesting that disturbed tissue residency of ILCs
   may contribute to the pathogenesis of intestinal inflammation.
   Together, P2Y10 is highly expressed on ILCs and mediates their
   intestinal tissue residency.

Microbiota metabolite TDCA is a ligand to P2Y10 receptor to mediate
tissue-resident signaling of intestinal ILCs

   We next intended to identify microbiota metabolites that engaged with
   P2Y10 receptor. Through untargeted metabolomics analysis on gut
   contents from Ctrl and GF mice, we found that disruption of microbiota
   caused significant alterations in the abundance of numerous
   metabolites. We chose metabolites specifically and highly enriched in
   Ctrl mice but markedly decreased in GF mice (fold change > 2) for
   further screening ([112]Fig. 4A). Because P2Y10 receptor belongs to
   GPCRs, its activation can be assessed by detecting Ca^2+ flux with a
   calcium indicator, Fluo-4 AM ([113]29). Among these highly enriched
   metabolites, TDCA stimulation showed maximal activation of P2Y10
   receptor ([114]Fig. 4B). TDCA is a secondary bile acid that requires
   microbiota for its generation. We noticed that TDCA level was much
   lower in GF mice compared to that in Ctrl mice, while the levels of its
   precursors, TCDCA and TCA, were higher in GF mice (fig. S8A),
   suggesting that gut microbiota metabolizes TCDCA and TCA to form TDCA.
   In addition, TDCA was able to effectively activate P2Y10 compared to
   TCDCA or TCA ([115]Fig. 4C and fig. S8B). Apart from calcium
   mobilization, previous studies have reported that P2Y10 receptor can
   also activate downstream RhoA signaling ([116]14). Subsequently, G-LISA
   and luciferase reporter assay were used to verify the activation effect
   of TDCA on RhoA signaling pathway ([117]Fig. 4D and fig. S8, C and D).
   Furthermore, median effective concentration for binding of TDCA to
   P2Y10 was 0.27 μM ([118]Fig. 4D).

Fig. 4. TDCA is a ligand to P2Y10 receptor.

   [119]Fig. 4.
   [120]Open in a new tab

   (A) Metabolome analysis of the intestinal contents from Ctrl and GF
   mice. Significantly differentiated metabolites were shown. (B) Response
   of P2Y10 to the metabolites was analyzed by detecting Ca^2+ flux with
   Fluo-4 AM (2 μM). Related fluorescence intensity (F/F[0]) was
   calculated. n = 3. (C) Response of P2Y10 to TDCA and its metabolite
   precursors were analyzed as above. n = 3. (D) Response of P2Y10 to
   different doses of TDCA was analyzed by detecting active RhoA through
   luciferase reporter assay. Median effective concentration (EC[50]) was
   calculated. FC, fold change. (E) Flow cytometry analysis of RhoA
   activation in intestinal ILCs from P2ry10^+/+ or P2ry10^−/− mice after
   TDCA stimulation. (F) Active RhoA staining in sorted intestinal ILCs
   from P2ry10^+/+ or P2ry10^−/− mice. Scale bars, 5 μm. (G) Mice were
   orally administrated with TDCA (1 mg/20 g) daily for 2 weeks while
   maintained on Abx drinking. Flow cytometry analysis of intestinal
   ILC1s, ILC2s, and ILC3s. n = 4. (H) P2ry10^+/+ and P2ry10^−/− mice kept
   on normal drinking were orally administrated TDCA (1 mg/20 g) daily for
   2 weeks. Flow cytometry analysis of intestinal ILC1s, ILC2s, and ILC3s.
   n = 4. (I and J) Similar mice treatment as in (G) and (H) was
   conducted. Flow cytometry analysis of intestinal CD69^+ and S1PR1^+
   ILCs. n = 4. (K) Intestinal ILCs were cultured with TDCA (10 μM), Ca^2+
   inhibitor (U-73122, 1 μM), or ROCK inhibitor [(S)-H-1152, 1 μM], and,
   then, mRNA expression of Cd69 and S1pr1 was analyzed. Fold changes were
   normalized to Actb. Data are representative of at least three
   independent experiments and are shown as the means ± SD. Statistical
   analysis was performed by using unpaired two-tailed Student’s t test
   (*P < 0.05; **P < 0.01; ***P < 0.001; n.s., not significant). SSC, side
   scatter; W, width.

   In addition, we validated the effect of TDCA on intestinal ILCs. Flow
   cytometry analysis showed that TDCA stimulation enhanced the level of
   active RhoA in P2ry10^+/+ ILCs, but not in P2ry10^−/− ILCs ([121]Fig.
   4E). Consistently, immunofluorescence staining showed that ILCs from
   P2ry10^+/+ mice exhibited a marked increase in intracellular active
   RhoA level after TDCA stimulation ([122]Fig. 4F). Next, we explored
   whether TDCA regulates the retention of intestinal ILCs. TDCA orally
   administration rescued a reduction in the number of intestinal ILC
   subsets resulted from Abx treatment ([123]Fig. 4G), whereas such
   protective effect was lacking in P2ry10^−/− mice ([124]Fig. 4H). Of
   note, compared with those in the untreated group, numbers of ILC1s,
   ILC2s, and ILC3s in intestines of P2ry10^+/+ mice slightly increased
   with TDCA treatment ([125]Fig. 4H).

   To further determine whether TDCA-P2Y10 interaction regulates the
   expression of CD69 and S1PR1, we administered TDCA to Abx-treated mice
   followed by flow cytometry analysis. Compared to those in mice treated
   with Abx alone, CD69^+ ILCs were increased in Abx-treated mice
   supplemented with TDCA administration, while S1PR1^+ ILCs were
   decreased ([126]Fig. 4I and fig. S8E). By contrast, in P2ry10^−/− mice,
   the positive regulatory effect of TDCA on CD69 expression was abolished
   ([127]Fig. 4J and fig. S8F). To determine the role of P2Y10 downstream
   signaling in regulating expression of resident molecules, we sorted
   intestinal ILCs and stimulated them with TDCA. We found that TDCA
   increased Cd69 expression in sorted ILCs, while decreased S1pr1
   expression ([128]Fig. 4K). Given that TDCA engagement with P2Y10
   receptor not only stimulated Ca^2+ flux but also activated RhoA
   signaling, we used a phospholipase (PLC) inhibitor (U-73122) and a ROCK
   inhibitor [(S)-H-1152] to separately inhibit Ca^2+ and RhoA signaling
   pathways. Both U-73122 and (S)-H-1152 eliminated the positive effect of
   TDCA on Cd69 expression, while relieved the inhibition on S1pr1
   expression ([129]Fig. 4K). Collectively, microbiota metabolite TDCA
   mediates tissue residency of intestinal ILCs via engagement with P2Y10
   receptor.

TDCA-P2Y10 engagement initiates Zfp414 transcription

   We intended to further define the mechanism by which
   TDCA-P2Y10–mediated signaling regulates long-term retention of
   intestinal resident ILCs. We performed RNA-seq of intestinal ILCs
   isolated from P2ry10^+/+ and P2ry10^−/− mice. Gene set enrichment
   analysis (GSEA) showed that Tcm signature genes were enriched in
   P2ry10^−/− ILCs, while genes up-regulated in Abx-untreated ILCs were
   enriched in P2ry10^+/+ ILCs (fig. S9A). In line with above results
   ([130]Fig. 3A), several pathways such as positive regulation of cell
   adhesion, lymphocyte-mediated immunity, and lymphocyte activation
   involved in immune response were enriched in P2ry10^+/+ ILCs, but not
   in P2ry10^−/− ILCs ([131]Fig. 5A). Of note, the regulatory effects of
   TFs were attenuated in P2ry10^−/− ILCs ([132]Fig. 5A). To further
   explore the TFs involved in regulation of tissue residency of ILCs, we
   identified 227 TFs that were simultaneously highly expressed in all ILC
   subsets from Abx-untreated mice. Subsequently, we analyzed expression
   of the 227 TFs in intestinal ILCs from P2ry10^+/+ and P2ry10^−/− mice
   and picked out eight TFs with higher expression in P2ry10^+/+ ILCs
   ([133]Fig. 5B). Further qPCR validation revealed that Zfp414 was the
   most highly expressed TF in Ctrl ILCs, while its expression was
   down-regulated after Abx treatment and P2ry10 knockout ([134]Fig. 5C).
   Subsequently, to investigate the role of ZFP414 in regulating tissue
   residency of intestinal ILCs, we constructed BM chimeric mice with
   Zfp414 knockout using CRISPR-based mutagenesis in BM cells. Zfp414
   deficiency caused impaired homing of ILC subsets, and reintroduction of
   Zfp414 rescued reduced numbers of intestinal ILC1s, ILC2s, and ILC3s
   ([135]Fig. 5D and fig. S9, B and C). We further performed in vitro
   differentiation assay to test the effect of Zfp414 deficiency on ILC
   differentiation. We found that Zfp414 depletion did not affect ILC
   differentiation (fig. S9D), indicating that the impaired homing of ILC
   subsets caused by Zfp414 deficiency is not due to the defect of ILC
   differentiation. In addition, we observed that P2ry10 reintroduction or
   Zfp414 overexpression (fig. S9E) in P2ry10^−/− ILCs could increase the
   proportion of long-term resident ILC subsets ([136]Fig. 5E), suggesting
   that ZFP414 is necessary for tissue residency of intestinal ILCs.
   Moreover, inhibition of P2Y10 downstream signaling pathways with
   U-73122 and (S)-H-1152 blocked the beneficial effect of TDCA on Zfp414
   expression ([137]Fig. 5F). Consistent with the expression of P2Y10,
   ZFP414 was expressed in human ILCs derived from noninflamed intestine
   but down-regulated in inflamed ILCs ([138]Fig. 5, G and H). These
   results indicate that TDCA-P2Y10–mediated signaling promotes the
   transcription of Zfp414 for tissue residency of intestinal ILCs.

Fig. 5. TDCA-P2Y10 engagement initiates Zfp414 transcription.

   [139]Fig. 5.
   [140]Open in a new tab

   (A) Transcriptome analysis of sorted intestinal ILCs from P2ry10^+/+
   and P2ry10^−/− mice. Down-regulated genes in P2ry10^−/− ILCs were
   performed GO enrichment analysis. (B) Venn diagram of transcription
   factors (TFs) that were simultaneously enriched in Ctrl total ILCs and
   ILC subsets. The expression of TFs in intestinal ILCs from P2ry10^+/+
   and P2ry10^−/− mice was analyzed. FC = P2ry10^−/− ILCs/P2ry10^+/+ ILCs.
   (C) Relative mRNA levels of the marked TFs above in P2ry10^+/+ and
   P2ry10^−/− ILCs. Fold changes were normalized to Actb. (D) Construct BM
   chimeric mice with Zfp414 deletion and reintroduction. Flow cytometry
   analysis of CD45.2^+ intestinal ILC1s, ILC2s, and ILC3s in different BM
   chimeras. n = 4. (E) Construct P2ry10^−/− BM chimeras with P2ry10 or
   Zfp414 overexpression. Flow cytometry analysis of CD45.2^+ intestinal
   ILC1s, ILC2s, and ILC3s in different BM chimeras. n = 4. (F) Sorted
   Intestinal ILCs were cultured as in [141]Fig. 4K, and, then, the mRNA
   expression of Zfp414 was analyzed. Fold changes were normalized to
   Actb. (G) Relative mRNA levels of ZFP414 in ILCs from noninflamed and
   inflamed intestinal samples of patients with CD. Fold changes were
   normalized to ACTB. n = 4 independent experiments. (H)
   Immunofluorescence staining of intestinal ILCs expressing ZFP414 in
   noninflamed and inflamed intestinal samples of patients with CD. White
   arrows indicate intestinal ZFP414^+ ILCs. Scale bars, 20 μm.
   Percentages of ZFP414^+ ILCs in CD127^+ cells per field are shown in
   the right panel. n = 6 fields. Data are representative of at least
   three independent experiments and are shown as the means ± SD.
   Statistical analysis was performed by using unpaired two-tailed
   Student’s t test (*P < 0.05; **P < 0.01; ***P < 0.001; n.s., not
   significant).

ZFP414 activates expression of integrin αE and CD69

   We next aimed to explore how TDCA-P2Y10-ZFP414 axis mediates the
   intestinal residency of ILCs. We observed that, besides TF activity,
   cell adhesion mediated by integrins showed a difference after Abx
   treatment ([142]Fig. 3A). Integrins are a family of cell adhesion
   receptors that play a critical role in cell-cell adhesion. Previous
   reports have indicated that integrins play an important role in
   mediating tissue residency of lymphocytes ([143]30, [144]31). To
   determine the role of integrins in intestinal resident ILCs, we
   analyzed expression of integrins and identified five integrins that
   were down-regulated in Abx-treated and P2Y10-deficient ILCs ([145]Fig.
   6A). Further qPCR validation showed that Itgae (coding Integrin αE) and
   its interactor Itgb7 displayed higher expression in untreated WT ILCs,
   whereas they were markedly decreased by Abx treatment and P2Y10
   deficiency ([146]Fig. 6B). Integrin αE typically forms a heterodimer by
   binding to integrin β7, which has been identified as a key molecule in
   maintaining the tissue-resident property of CD8^+ T cells by
   facilitating adhesion of CD8^+ T cells to epithelial cells ([147]17).
   However, it is unknown whether integrin αE regulates the tissue
   residency of intestinal ILCs. Integrin αE exhibited lower expression in
   intestinal ILCs from Abx and P2ry10^−/− mice compared to those from
   their control littermates ([148]Fig. 6, C and D; and fig. S9, F and G).
   Consistently, reduced interaction of ITAE^+ ILCs and E-cadherin^+
   intestinal epithelia was observed in Abx and P2ry10^−/− mice (fig. S9,
   H and I).

Fig. 6. ZFP414 primes expression of integrin αE on ILCs.

   [149]Fig. 6.
   [150]Open in a new tab

   (A) Venn diagram of integrins that were simultaneously down-regulated
   in Abx and P2ry10^−/− ILCs. (B) Relative mRNA levels of integrins that
   were enriched in Ctrl and P2ry10^+/+ ILCs. Fold changes were normalized
   to Actb. (C and D) Flow cytometry analysis of the expression of ITAE on
   intestinal ILCs derived from Abx-treated (C) and p2ry10^−/− mice (D).
   n = 5. (E) Flow cytometry analysis of the expression of ITAE on
   CD45.2^+ intestinal ILCs from different BM chimeras with Zfp414
   deletion and reintroduction. n = 4. (F) Construct P2ry10^−/− BM
   chimeras with P2ry10 or Zfp414 overexpression. Flow cytometry analysis
   of the expression of ITAE on CD45.2^+ intestinal ILCs from different BM
   chimeras. n = 4. (G and H) Mice were treated as in [151]Fig. 4 (G and
   H). Flow cytometry analysis of the expression of ITAE on intestinal
   ILCs from indicated mice. n = 4. (I) Sorted intestinal ILCs were
   cultured as in [152]Fig. 4K, and, then, Itgae expression was analyzed.
   Fold changes were normalized to Actb. (J) ILCs with Itgae silencing
   were labeled with CFSE (5 μM) and transferred (5 × 10^6) into
   immunodeficient mice. Immunofluorescence of SI was performed. Scale
   bars, 70 μm. n = 5. (K) Relative mRNA levels of ITGAE in ILCs from
   noninflamed and inflamed intestinal samples of patients with CD. Fold
   changes were normalized to ACTB. n = 4. (L) Immunofluorescence of
   intestinal ITAE^+ ILCs in noninflamed and inflamed intestinal samples
   of patients with CD. Scale bars, 20 μm. Percentages of ITAE^+ ILCs in
   CD127^+ cells per field are shown. n = 6 fields. Data are
   representative of at least three independent experiments and are shown
   as the means ± SD. Statistical analysis was performed by using unpaired
   two-tailed Student’s t test (*P < 0.05; **P < 0.01; ***P < 0.001; n.s.,
   not significant). SSC, side scatter; W, width.

   To further explore whether ZFP414 regulated the expression of ITAE, we
   performed a chromatin immunoprecipitation (ChIP) assay with anti-ZFP414
   antibody and found that ZFP414 was enriched in the Itgae promoter
   region (−400 to ~−200), suggesting that ZFP414 enriches onto the Itgae
   locus to regulate its transcription (fig. S9J). As expected, Zfp414
   deficiency resulted in decreased expression of integrin αE and CD69 in
   intestinal ILCs, which was restored by Zfp414 reintroduction ([153]Fig.
   6E and fig. S10, A and B). Additionally, reintroducing P2ry10 or
   overexpressing Zfp414 in P2ry10^−/− ILCs partially rescued integrin αE
   and CD69 expression ([154]Fig. 6F and fig. S10C). Consistent with the
   effect on CD69, TDCA administration restored expression of integrin αE
   in Abx ILCs, but not in P2ry10^−/− ILCs ([155]Fig. 6, G and H; and fig.
   S10, D and E). Moreover, blockade of P2Y10 downstream signals
   eliminated the beneficial influence of TDCA on Itgae expression
   ([156]Fig. 6I). To directly verify the function of integrin αE on ILCs
   positioning in the intestine, we silenced Itgae expression in
   intestinal ILCs and then transferred these ILCs into immunodeficient
   recipient NOD-SCID-Il2rg^−/− (NSG) mice to analyze their homing and
   resident ability. Our results showed that Itgae knockdown decreased
   ILCs homing to intestine and adhesion to epithelial cells ([157]Fig. 6J
   and fig. S10F). Subsequently, we analyzed the expression of integrin αE
   on human ILCs with qPCR assay and immunofluorescence staining. Compared
   to noninflamed intestine, inflamed-intestine derived ILCs displayed
   much lower expression of integrin αE ([158]Fig. 6, K and L). Together,
   the TDCA-P2Y10-ZFP414 axis mediates tissue residency of intestinal ILCs
   through integrin αE and CD69.

Maintenance of intestinal resident ILCs contributes to protection against
intestinal inflammation

   Intestinal resident ILCs play a key role in regulating the function of
   intestinal epithelial barrier, contributing to tissue homeostasis and
   host defense ([159]32). To explore the physiological significance for
   maintaining stable residency of ILCs in the intestine, we generated
   Zfp414, Itgae-deficient mice and triple-deficient mice (shZfp414;
   shItgae; P2ry10^−/−) through short hairpin RNA (shRNA)–based gene
   silencing (fig. S10G). Then, we used a dextran sulfate sodium
   (DSS)–induced intestinal injury model to determine whether the
   P2Y10-ZFP414-ITAE axis in ILCs contributes to regulation of intestinal
   injury and inflammation. Deficiency of P2ry10, Zfp414, or Itgae,
   respectively, showed exacerbated intestinal inflammation with DSS
   treatment, as evidenced by increased epithelial damage and mucosal
   thickness ([160]Fig. 7A), higher pathological scores ([161]Fig. 7B),
   and faster weight loss ([162]Fig. 7C). Furthermore, simultaneous
   deficiencies of these three molecules caused most severe intestinal
   inflammation. Next, we investigated whether maintaining stable
   residency of ILCs by TDCA contributes to intestinal homeostasis. With
   DSS treatment, compared to that in the Abx-treated mice, intestinal
   inflammation in the control mice was slightly exacerbated, despite no
   notable difference in weight loss. TDCA treatment obviously alleviated
   intestinal inflammation in both control and Abx-treated mice ([163]Fig.
   7, D to F). In addition, to evaluate the effect of TDCA on host
   defense, Citrobacter rodentium infection model was used ([164]33).
   P2ry10^−/− mice infected with C. rodentium showed more severe colitis
   ([165]Fig. 7, G and H), more rapid weight loss ([166]Fig. 7I) and more
   bacteria load in colons ([167]Fig. 7J). Of note, TDCA administration
   protected against C. rodentium infection in P2ry10^+/+ mice, but not in
   P2ry10^−/− mice ([168]Fig. 7, G to J). To definitively validate that
   the increased susceptibility to intestinal inflammation in C. rodentium
   models is attributed to the decreased tissue residency of ILCs, rather
   than other cell types expressing P2Y10, we transferred P2ry10^+/+ or
   P2ry10^−/− ILCs into immunodeficient NSG mice and subsequently infected
   mice with C. rodentium. We observed that P2ry10^−/− ILCs were less
   effective than P2ry10^+/+ ILCs on protecting NSG mice and clearing C.
   rodentium, displaying more severe colitis ([169]Fig. 7, K and L),
   increased weight loss ([170]Fig. 7M), and higher bacteria load
   ([171]Fig. 7N). These results indicate that maintaining tissue
   residency of intestinal ILCs is beneficial for intestinal homeostasis
   and host defense against pathogen infection.

Fig. 7. TDCA administration promotes intestinal residency of ILCs and
protection against intestinal inflammation.

   [172]Fig. 7.
   [173]Open in a new tab

   (A) Mice were treated with 2% (w/v) DSS in drinking water for 5 days
   and followed by regular drinking water for 5 days. At day 10, colons
   were collected and fixed with 4% PFA for paraffin section and H&E
   staining. Scale bars, 50 μm. (B and C) Pathological scores (B) and
   weight change (C) of indicated mice. n = 6. (D) Mice were treated as in
   [174]Fig. 4G, and, then, intestinal injury was induced as above with 2%
   (w/v) DSS. At day 10, colons were collected and fixed with 4% PFA for
   paraffin section and H&E staining. Scale bars, 50 μm. (E and F)
   Pathological scores (E) and weight change (F) of indicated mice. n = 5.
   (G) Mice were pretreated as in [175]Fig. 4H and then were orally
   infected with C. rodentium (5 × 10^9 CFU). At day 10, colons were
   collected for paraffin section and H&E staining. Scale bars, 50 μm. (H
   and I) Pathological scores (H) and body weight change (I) of indicated
   mice. n = 5. (J) Bacteria loads in feces were calculated. n = 5. (K)
   ILCs were isolated from P2ry10^+/+ or P2ry10^−/− mice by flow cytometry
   and transferred (1 × 10^6) into NSG mice on day −2 and 2 postinfection
   with C. rodentium (1 × 10^9 CFU). Ten days later, colons were collected
   for paraffin section and H&E staining. Scale bars, 50 μm. (L and M)
   Pathological scores (L) and body weight change (M) of indicated mice.
   n = 5. (N) Bacteria loads in feces were calculated. n = 5. Data are
   representative of at least three independent experiments and are shown
   as the means ± SD. Statistical analysis was performed by using unpaired
   two-tailed Student’s t test (*P < 0.05; ***P < 0.001; n.s., not
   significant).

DISCUSSION

   Tissue residency is a remarkable characteristic of ILCs. Upon reaching
   their destination, tissue-resident ILCs develop unique molecular
   characteristics influenced by local microenvironment, irrespective of
   their origin. However, little is known about how the tissue
   environment, such as the abundant microbiota in the gut, affects the
   stable residency of ILCs. In this study, we showed that gut microbiota
   is necessary for the maintenance of intestinal tissue residency of
   ILCs. Microbiota metabolite TDCA binds to P2Y10 receptor on ILCs to
   initiate downstream Ca^2+ and RhoA signaling pathways. TDCA-P2Y10
   engagement induces Zfp414 transcription to prime expression of resident
   molecules CD69 and integrin αE on ILCs. These two molecules mediate the
   maintenance of intestinal tissue residency of ILCs by tethering ILCs to
   intestinal epithelial cells and restrict them out of tissue.

   Microbiota and their metabolites represent an additional layer of
   regulation for the intestinal immune system. Microbiota-mediated
   signals have been defined to modulate tissue-resident lymphocytes, such
   as Trm and liver-resident Kupffer cells ([176]34). Intestinal
   microbiota-derived SCFAs like acetate can directly regulate glycolytic
   activity of CD8^+ T cells by GPCRs to prolong Trm survival ([177]35).
   In addition, intestinal commensal bacteria elicit tissue-resident
   intestinal T helper 17 (T[H]17) cells, which are homeostatic and
   beneficial for barrier maintenance ([178]36). Dysregulation of
   microbiota causes less butyrate in the liver and reduces IL-18
   production in hepatocytes, leading to defective maturation of
   tissue-resident NK in the liver ([179]37). The involvement of
   microbiota in regulating the function and development of intestinal
   ILCs were observed ([180]38, [181]39), but how the microbiota modulates
   intestinal tissue residency of ILCs has not yet been investigated.
   Herein, we found that all three ILC subsets were decreased in the SI of
   GF and Abx-treated mice, while ILC subsets were increased in lymphoid
   organs such as mLNs and spleen. Although total numbers of ILCs were
   reduced in the intestinal tracts, the proportion of ILCs in the
   intestinal lymphatic vessels was increased, suggesting that ILCs are no
   longer in a resident state and have acquired characteristics of
   circulation. Although conventional NK cells belong to ILC1s, we
   excluded them from our study and distinguished ILC1s from NK cells with
   CD127 expression. Now, ILC1s are known to be enriched in tissues and
   are generally considered to be tissue-resident cells, whereas NK cells
   are often regarded as circulatory cells ([182]40). Furthermore, our
   results indicate that the expression levels of P2ry10 in intestinal NK
   cells are lower than those in ILCs (fig. S3G), suggesting that P2Y10
   may not play a major regulatory role in intestinal NK cells.

   P2Y10 belongs to a purinergic receptor family and is classified as a
   GPCR due to its sevenfold transmembrane structure. Previous studies
   have reported that P2Y10 plays a role in the regulation of T cell
   migration and is expressed by other immune cells ([183]14, [184]15,
   [185]41). We showed that, among intestinal immune cell populations,
   P2ry10 exhibited the highest expression levels in ILCs, suggesting its
   important regulatory function in ILCs. In addition, we used an ILC
   adoptive transfer model combined with a C. rodentium–induced intestinal
   inflammation model in immunodeficient mice to exclude the role of T
   cells in the susceptibility to inflammation caused by P2ry10
   deficiency. We identified the gut microbiota metabolite TDCA, a
   secondary bile acid whose metabolism and abundance are modulated by gut
   microbiota ([186]42), as a ligand for the P2Y10 receptor. Emerging
   evidence suggests that microbiota-derived secondary bile acids play a
   critical role in regulation of intestinal immunity ([187]43, [188]44).
   As derivatives of chlorogenic acid (LCA) and deoxycholic acid (DCA),
   3-oxoLCA can directly bind to retinoid-related orphan receptor γt
   (RORγt) and inhibit differentiation of T[H]17 cells, while isoallo-LCA
   enhances the differentiation of regulatory T cells with
   anti-inflammatory function by promotion of Foxp3 expression ([189]45).
   Here, we demonstrated that TDCA acts as a ligand for P2Y10 receptor.
   TDCA-P2Y10 engagement activates its downstream Ca^2+ and RhoA signaling
   pathways to mediate tissue residency of ILCs.

   Activation of GPCRs downstream Ca^2+ and RhoA signaling pathways is
   involved in various biological processes. For instance, GPR55 on
   macrophage responds to l-α-lysophosphatidylinositol, leading to release
   of Ca^2+ and activation of nuclear factor of activated T cells
   ([190]46). The RhoA signaling pathway is also involved in the
   regulation of TFs transcription. RhoA activates protein kinase N to
   regulate the activity of myocardin-related TF A ([191]47).
   Additionally, hyperactivate RhoA signaling is both necessary and
   sufficient to drive oncogenic TEAD/YAP transcription in tumor
   ([192]48). In our study, we showed that TDCA-P2Y10–mediated Ca^2+ and
   RhoA signaling pathways initiate Zfp414 transcription to promote
   expression of resident molecules CD69 and integrin αE on ILCs.

   In SI, CD8^+ Trm cells inhibit SIPR1 by expressing unique molecules
   such as CD69 to prevent cell egression. Integrin αE binds E-cadherin on
   epithelial cells to tether Trm cells in the intestinal tissue.
   Similarly, resident T cells in the lamina propria also express CD69 and
   CD49a for their retention in a transforming growth factor–β–dependent
   manner ([193]49). In this study, we showed that disruption of
   TDCA-P2Y10 axis down-regulates CD69 and integrin αE on intestinal ILCs,
   leading to their tissue egress. A recent study showed that human
   intestinal ILCs also exhibit resident characteristics, with low
   expression of S1PR1 and SELL but with high expression of CD69
   ([194]50). Reduced TDCA has been defined in the gut of CD ([195]51). We
   validated that patients with CD lowly expressed P2Y10 and resident
   molecules, suggesting that the TDCA-P2Y10 pathway could be involved in
   the pathogenesis of CD. Of note, TDCA administration can enhance
   intestinal tissue residency of ILCs and promote protection against
   intestinal inflammation. Thus, TDCA could be used as a potential drug
   to treat patients with IBD.

MATERIALS AND METHODS

Antibodies and reagents

   Anti-human CD3–fluorescein isothiocyanate (FITC; UCHT1, RRID:
   AB_314059), anti-human CD3-allophycocyanin (APC;HIT3a, RRID:
   AB_314047), anti-Thy1.2-APC/cyanine7 (30-H12, RRID: AB_10613280),
   anti-CD127-APC (SB/199, RRID: AB_1595577), anti-CD127-Brilliant Violet
   711 (A7R34, RRID: AB_2564577), anti-CD127-phycoerythrin/cyanine7
   (PE-cyanine7; A7R34, RRID: AB_1937266), anti-NKp46 FITC (29A1.4, RRID:
   AB_2149150), anti-NKp46 (29A1.4, RRID: AB_10551441),
   anti-CD45.1-PE-cyanine7 (A20, RRID: AB_1134170),
   anti-CD45.2-PE-cyanine7 (104, RRID: AB_313444), anti-CD45.2-peridinin
   chlorophyll protein-cyanine5.5 (PerCP-cyanine5.5; 104, RRID:
   AB_893352), anti–GATA3–Alexa Fluor 647 (16E10A23, RRID: AB_2563216),
   and anti–epithelial cell adhesion molecule (EpCAM)–Alexa Fluor 647
   (G8.8, RRID: AB_1134101) were purchased from BioLegend. Anti–CD3–eFluor
   450 (17A2, RRID: AB_1272193), mouse hematopoietic lineage eFluor 450
   Cocktail (RRID: AB_10426799), mouse hematopoietic lineage FITC cocktail
   (RRID: AB_2644066), human hematopoietic lineage eFluor 450 cocktail
   (RRID: AB_2642588), anti–CD19–eFluor 450 (eBio1D3 (1D3), RRID:
   AB_2734905), anti-Thy1.2-FITC (30-H12, RRID: AB_2735033), anti-human
   CD90–APC–eFluor 780 (5E10, RRID: AB_2848360), anti-B220-PE (RA3-6B2,
   RRID: AB_465671), anti-CCR6-APC (R6H1, RRID: AB_10733388), anti-human
   CD127-PE (eBioRDR5, RRID: AB_10853334), anti-CD127-PE (A7R34, RRID:
   AB_465843), anti-CD127 (A7R34, RRID: AB_467457),
   anti–CD127–PerCP–eFluor 710 (SB/199, RRID: AB_2573710),
   anti–NK1.1–eFluor 450 (PK136, RRID: AB_2043878), anti-NK1.1-APC (PK136,
   RRID: AB_469479), anti-NKp46-PE-cyanine7 (29A1.4, RRID: AB_2573441),
   anti-NKp46-PE (29A1.4, RRID: AB_1210743), anti–Sca-1–PerCP–cyanine5.5
   (D7, RRID: AB_914370), anti–Sca-1–APC (D7, RRID: AB_469486),
   anti-KLRG1-PE (2F1, RRID: AB_10597431), anti-KLRG1-APC (2F1, RRID:
   AB_469469), anti-KLRG1 (2F1, RRID: AB_469131), anti-RORγt-APC (B2D,
   RRID: AB_2573253), anti-RORγt (AFKJS-9, RRID: AB_1311291),
   anti-PLZF–Alexa Fluor 488 (Mags.21F7, RRID: AB_2574444), anti-ID2-PE
   (ILCID2, RRID: AB_2735055), anti–c-Kit–PE (2B8, RRID: AB_465813),
   anti–integrin α4β7–APC (DATK32, RRID: AB_1210577), anti–integrin
   α4β7–biotin (DATK32, RRID: AB_529602), anti-CD25-PE-cyanine7 (PC61.5,
   RRID: AB_469608), anti-Flt3-PE (A2F10, RRID: AB_465859),
   anti-CD45.1-FITC (A20, RRID: AB_465057), anti-CD45.1-PerCP-cyanine5.5
   (A20, RRID: AB_925750), anti–CD45.1–APC–eFluor 780 (A20. RRID:
   AB_1582229), anti-CD45.2-PE (104, RRID: AB_465678),
   anti–CD45.2–APC–eFluor 780 (104, RRID: AB_1272211), anti-CD11C-FITC
   (N418, RRID: AB_464940), anti–CD11b–eFluor 450 (M1/70, RRID:
   AB_1582236), anti-F4/80-PE (BM8, RRID: AB_465923),
   anti-CD69-PE-cyanine7 (H1.2F3, RRID: AB_469636), anti–integrin alpha
   E–FITC (2E7, RRID: AB_465175), anti-Ki67-FITC (SolA15, RRID:
   AB_11151689), anti–IL-13–PE (eBio13A, RRID: AB_763559),
   anti–LYVE1–Alexa Fluor 488 (ALY7, RRID: AB_1633417), ant–LYVE1–eFluor
   450 (ALY7, RRID: AB_2784723), goat anti-rabbit immunoglobulin G (IgG)
   (H+L)–Alexa Fluor 647 (RRID: AB_2576217), donkey anti-rabbit IgG
   (H+L)–Alexa Fluor 488 (RRID: AB_2535792), donkey anti-rabbit IgG
   (H+L)–Alexa Fluor 594 (RRID: AB_141637), goat anti-rat IgG (H+L)–Alexa
   Fluor 594 (RRID: AB_2535795), and goat anti-rat IgG (H+L)–Alexa Fluor
   488 (RRID: AB_2535794) were purchased from eBioscience.
   Anti–RORγt–Brilliant Violet 421 (Q31-378, RRID: AB_2687545) and
   anti–IL-5–PE (TRFK-5, RRID: AB_10894193) were purchased from BD
   Horizon. Anti-CD69 antibody was purchased from Bioss (RRID:
   AB_10855614). Anti-S1PR1-PE (T4-H28, RRID: AB_10994187) was purchased
   from R&D Systems. Purified anti-S1PR1 antibody (RRID: AB_789722) was
   purchased from Novus Biologicals. Anti–E-cadherin (Ag14973, RRID:
   AB_10697811) was purchased from Proteintech. Purified anti-P2Y10
   polyclonal antibody was purchased from Signalway Antibody and Novus.
   Anti-ZNF414 antibody (RRID: AB_10722362) was purchased from GeneTex.
   Anti–β-actin, goat anti-mouse IgG (H+L)–horseradish peroxidase (HRP)
   and goat anti-rabbit IgG (H+L)–HRP were purchased from Sungene Biotech.
   Recombinant murine IL-3, recombinant murine IL-6, recombinant murine
   IL-7, recombinant murine IL-2, recombinant murine IL-25, recombinant
   murine IL-33, and recombinant murine stem cell factor (SCF) were
   purchased from PeproTech. Fluo-4 AM was purchased from Yeasen. Type II
   collagenase and type III collagenase were purchased from Worthington.
   4′,6-Diamidino-2-phenylindole (DAPI), Percoll, and paraformaldehyde
   (PFA) were purchased from Sigma-Aldrich. TRIzol, Lipofectamine 3000
   transfection reagent, and the FoxP3 Transcription Factor Staining Kit
   were purchased from Thermo Fisher Scientific. Hispidol, spermidine, and
   dihydrostreptomycin were purchased from Psaitong.
   l-α-Glycerophosphorylcholine was purchased from Sigma-Aldrich.
   Butobarbital was purchased from Toronto Research Chemicals.
   Propoxyphenyl thiosildenafil and creatinine were purchased from
   Aladdin. Linoleoyl ethanolamide was purchased from TargetMol.
   Taurochenodeoxycholic acid, glutathione reduced, taurocholic acid, and
   fibronectin (FN) were purchased from Solarbio. TDCA was purchased from
   APExBIO. (S)-H-1152 was purchased from Cayman. U-73122 was purchased
   from MedChemExpress. DSS was purchased from MP Biomedicals. An RhoA
   G-LISA activation kit was purchased from Cytoskeleton. ONE-Glo
   Luciferase Assay System was purchased from Promega. The Annexin V
   Apoptosis Detection Kit was purchased from eBioscience.

Mice

   GF C57BL/6 and NSG mice were purchased from GemPharmatech Co. Ltd.
   (Nanjing, China). P2ry10^−/− mice were generated on the basis of the
   methods described before ([196]6). In brief, mixtures of Cas9 mRNA (100
   ng/ml) and single guide RNA (sgRNA; 50 ng/ml) (table S1) were
   microinjected into the cytoplasm of C57BL/6 fertilized eggs, followed
   by transferring to the uterus of pseudo-pregnant ICR females, from
   which viable founder mice were obtained. The following primers were
   used for deleted locus detection: 5′-TGTGCTGTGGGTCCTGTGTCCAAAGT-3′ and
   5′-TATTGAAACTTTTGTGTCATTTTGTC-3′.

   For generation of mice with Itgae and Zfp414 silencing, shItgae
   (5′-GGTCATGAACTGCAAGATT-3′) and shZfp414 (5′-GCTACTTCAAGTGTGAGAA-3′)
   were designed on online with RNAi designer (Merck or Thermo Fisher
   Scientific) and cloned into pLVshRNA-Puro lentivirus vector. Human
   embryonic kidney (HEK) 293T cell line (American Type Culture
   Collection, CRL-3216) was used for lentivirus packaging. When cells
   were 80 to 90% confluent, 8 μg of pLVshRNA-Puro plasmid, 6 μg of
   packaging plasmid psPAX2, and 2 μg of envelope plasmid pMD2.G were
   mixed and transfected into HEK293T cells by using Lipofectamine 3000
   reagent. Cell supernatants were collected on the first, second, and
   third days after transfection and filtered through a 0.45-μm filter.
   Then, supernatants containing virus particles were ultracentrifuged at
   25,000 rpm for 2 hours at 4°C for viral concentration. The virus
   pellets were resuspended with phosphate-buffered saline (PBS) and
   injected intravenously into P2ry10^+/+ or P2ry10^−/− mice. Virus
   injections were performed every other day and continued for 2 weeks.

   Cas9-KI, PLZF^GFPcre, and Id2^+/GFP mice were purchased from the
   Jackson Laboratory. All the mice were C57BL/6 background and 8 to 10
   weeks old. We used littermates with the same age and gender for each
   group. Animal use and protocols were approved by the Institutional
   Animal Care and Use Committees at the Institute of Biophysics, Chinese
   Academy of Sciences.

Mouse treatment

   For antibiotic (Abx) treatment, 8-week mice were fed water containing
   vancomycin (0.5 g/liter), ampicillin (1 g/liter), kanamycin (1
   g/liter), and metronidazole (1 g/liter) for 2 weeks. Fresh Abx was
   replaced every other day. For microbiota metabolite treatment,
   metabolites were dissolved in drinking water. Mice were treated with
   200 μl of indicated concentration of metabolites once a day by gavage
   administration at the beginning of antibiotic treatment. Treatment
   continued for 2 weeks.

IBD sample collection and ethical approval

   All patients are from The Affiliated Wuxi People’s Hospital of Nanjing
   Medical University. The patients were treatment naive and included upon
   diagnostic colonoscopy. IBD colitis diagnosis was endoscopically and
   histologically verified. Biopsies were obtained from two areas in the
   ileum of patients with IBD: the endoscopically most affected area
   (inflamed) and the endoscopically least-affected area (noninflamed).
   All participants provided written informed consent to sample collection
   and data analysis. This study was approved by the Research Ethics
   Committee at The Affiliated Wuxi People’s Hospital of Nanjing Medical
   University.

Parabiosis surgery

   Female 8-week-old congenic CD45.1 and CD45.2 mice were surgically
   connected in parabiosis according to a published protocol ([197]52). In
   brief, the mice were first anesthetized with tribromoethanol (300
   mg/kg). Then, the skin along a continuous line from the elbow to the
   knee on the side to be jointed was shaved and sterilized with iodine
   and alcohol. After corresponding lateral skin incisions were made from
   elbow to knee in each mouse, forelimbs and hindlimbs were tied together
   using nylon suture, and the skin incisions were closed using stainless
   steel wound clips. The mice were placed on a heated pad in a supine
   position and allowed to wake up in ambient air. After the surgery, each
   pair of mice was housed separately, and subcutaneous injections of
   0.25% bupivacaine (2.5 mg/ml, 8 mg/kg) were given daily for 3 days for
   pain management. The mice remained surgically jointed for at least 2
   months. T cell and B cell equilibration in the spleen and blood was
   confirmed by assessing presence of CD45.1/CD45.2 congenic markers.
   Last, the parabionts were euthanized, and tissues from CD45.1^+
   parabionts were processed and analyzed by flow cytometry for the
   presence of CD45.2^+ ILC subsets.

Isolation of mononuclear cells

   BM cells were flushed out from femurs in staining buffer [PBS buffer
   containing 1% fetal bovine serum (FBS)] and filtered through 40-μm cell
   strainers. Spleens and mLN were physically dissociated using ground
   glass slides and filtered through a 40-μm strainer. Red blood cells in
   spleen were lysed using red lysis buffer. The SI was dissected, and the
   fat and PPs were removed. Next, the SI was open up lengthwise, and the
   lumen contents were washed out. Then, SI tissue was cut into 0.5
   cm-long pieces, placed in solution I buffer [10 mM Hepes and 5 mM EDTA
   in Hanks’ balanced salt solution (HBSS), pH 7.2] to incubate three
   times for 15 min each at 37°C. The cell suspension was collected and
   centrifuged after three rounds of digestion. Density gradient step was
   done by resuspending cells in 44% Percoll and carefully laid on 67%
   Percoll solution and centrifuged at 600g and room temperature for 20
   min. Intraepithelial lymphocytes (IELs) were at the 44 to 67%
   interface. For lamina propria lymphocyte (LPL) isolation, remaining
   intestinal fragments were cut into fine pieces and digested three times
   for 20 min each at 37°C with solution II buffer [2% FBS and collagenase
   II and collagenase III (0.5 mg/ml)], and, then, cell suspension was
   collected after three rounds of digestion. Collected IELs and LPLs were
   filtered through 40-μm strainers before being centrifuged and
   resuspended in an appropriate amount of staining buffer. For cell
   isolation from ileac biopsies, biopsies were cut into pieces and
   digested three times for 20 min each at 37°C with digestion buffer [2%
   FBS and collagenase II and collagenase III (0.5 mg/ml)], and, then,
   cell suspension was collected and filtered through 40-μm strainers.
   After centrifugation, the cells were resuspended in staining buffer.

Flow cytometry

   For flow cytometry, cells were isolated and stained with antibodies
   targeting surface antigens for 30 to 40 min on ice, followed by washing
   with staining buffer (PBS containing 1% FBS). For intracellular
   staining, cells were harvested after surface marker staining, then
   fixed, and permeabilized with the FoxP3 Transcription Factor Staining
   Kit (Thermo Fisher Scientific) according to the manufacturer’s
   instructions. Antibodies for intracellular staining were incubated with
   cells on ice for 30 to 40 min in 1× permeabilization buffer. For flow
   cytometry analysis, ILCs (Lin^−Thy1.2^+CD127^+), ILC1s
   (Lin^−CD127^+CD45.2^+NK1.1^+NKp46^+), ILC2s
   (Lin^−CD127^+Thy1.2^+KLRG1^+Sca-1^+), ILC3s
   (Lin^−CD127^+Thy1.2^+RORγt^+), common lymphoid progenitor (CLP,
   Lin^−CD127^+c-Kit^loSca-1^loFlt3^+α4β7^−), common ILC progenitor (CILP,
   Lin^−CD25^−CD127^+Flt3^−α4β7^+), common helper-like ILC progenitor
   (CHILP, Lin^−CD25^−CD127^+Flt3^−α4β7^+Id2^+), and ILC progenitor (ILCP,
   Lin^−CD127^+Flt3^−α4β7^+c-kit^+PLZF^+) were analyzed with a FACSAria
   III instrument (BD Biosciences). Results were analyzed by FlowJo 10.

Quantitative real-time PCR

   For analysis of sorted cell populations, total RNA was extracted
   through the RNA Miniprep Kit (Tiangen, Beijing, China) following the
   manufacturer’s instructions. For analysis of SI tissues and IBD
   samples, tissues or samples were homogenized directly into TRIzol
   (QIAGEN), and RNA was extracted via chloroform extraction. The quantity
   and quality of RNA were determined by using a NanoDrop (Thermo Fisher
   Scientific). For both methods, cDNA was synthesized with 5× All-In-one
   RT MasterMix (Abm, Vancouver, Canada) and analyzed on QuantStudio1 qPCR
   system using specific primer pairs listed in table S2. Relative
   expression was calculated and normalized to endogenous Actb.

Immunofluorescence staining

   SIs and colons were cut open longitudinally and rinsed with precold 1×
   PBS. The tissues were then rolled into a “Swiss roll” shape and fixed
   in 4% PFA for 1 hour at room temperature. After fixation, the samples
   were dehydrated in a 30% sucrose solution for 24 hours and then
   embedded in optimal cutting temperature (OCT) freezing medium for
   sectioning. For immunofluorescence staining, the tissue sections were
   briefly rehydrated in PBS and then permeabilized with 1% Triton X-100
   solution (PBS diluted) for 1 hour, followed with blocking in 10% donkey
   serum (diluted in PBS) for 30 min. The blocked sections were incubated
   with prediluted primary antibodies overnight at 4°C on a shaker. After
   washing with PBST (0.05% Tween 20 in PBS) three times, 5 min each time,
   sections were incubated with secondary antibodies for 1 hour at room
   temperature and then washed with PBST. Last, stained sections were
   mounted with Fluormount G (SouthernBiotech) and examined on a Nikon
   laser scanning confocal microscope. Images were analyzed by Imaris 9
   software.

ILC2s’ in vitro stimulation

   Sort-purified small intestinal ILC2s were incubated in complete RPMI
   1640 medium [containing 10% FBS, 50 μM 2-mercaptoethanol, 1 mM
   l-glutamine, 1 mM sodium pyruvate, penicillin (100 U/ml), and
   streptomycin (100 μg/ml)] for 24 hours at 37°C and 5% CO[2] with IL-2
   (10 ng/ml) and IL-7 (10 ng/ml). IL-25 (50 ng/ml) and IL-33 (50 ng/ml)
   were added if indicated. For intracellular cytokine staining, the
   culture was supplemented with brefeldin A (10 μg/ml) for 4 hours.

In vivo antibody labeling

   PE-conjugated anti-CD45.2 antibody (2 μg) was intravenously injected
   (intravenously) into mice, which had been pretreated with antibiotics
   for 2 weeks. Ten minutes later, mice were euthanized, and SI was
   harvested to isolate mononuclear cells for further flow cytometry
   analysis.

Intestinal homing of ILCs

   To analyze short-term homing of ILCs, intestinal lymphocytes were
   isolated from CD45.2^+ WT mice as described above, and, then, Lin^−
   lymphocytes were sorted. Sorted CD45.2^+ Lin^− lymphocytes (1 × 10^7)
   were transplanted into lethally irradiated (2 × 5 gray with a 4-hour
   interval) recipient mice (CD45.1^+) pretreated with or without
   antibiotics for 2 weeks. The recipients were euthanized 18 hours later,
   and SI was harvested and examined by flow cytometry for frequencies of
   transferred ILCs.

3D fluorescence imaging

   3D fluorescence imaging was performed by using a slightly modified
   protocol previously published ([198]53). In brief, mice were perfused
   with cold PBS, followed by 4% PFA. Then, strips of SI were pinned onto
   agarose plates and fixed overnight with 4% PFA in PBS. After several
   rinses in PBS, samples were passed through a glycerol and sucrose
   gradient, permeabilized, and blocked in permeabilizing solution [0.1%
   Tween 20, 0.5% Triton X-100, 0.1% deoxycholate, 0.1% NP-40, and 10%
   bovine serum albumin (BSA) in PBS] at 4°C for 1 day with gentle
   shaking. An appropriate primary antibody mixture in blocking buffer
   (0.5 Triton X-100 and 3% BSA in PBS) was performed on samples and
   incubated for 1 day on a shaker at 4°C. After washing in washing buffer
   (1% Triton X-100 in PBS) for 5 hours with a change of buffer every 30
   min on a shaker at 4°C, secondary antibodies were diluted in blocking
   buffer and performed on samples for 1 day on a shaker at 4°C, and,
   then, samples were washed in washing buffer as mentioned above. For
   confocal imaging, sample was properly sliced to 1-mm fragments with
   blades and placed on glass slides preadded with mounting medium
   Fluormount G (SouthernBiotech) and then covered by a coverslip. Samples
   were imaged by using confocal scanning (Nikon), and z-stacks
   projections were compiled by using Imaris 9 software.

Ca^2+ mobilization assay

   P2ry10 sequence was cloned into pCDNA4 plasmid for overexpression.
   Gα12, Gα13, Gα15, and Gα16 sequences were cloned into p3 × FLAG-CMV
   plasmid. Promiscuous expression vectors were cotransfected into HEK293T
   cells using Lipofectamine 3000 (Invitrogen) according to the
   manufacturer’s instructions. The transfected cells were cultured for 24
   hours at 37°C and then were collected and seeded in fibronectin
   (Sigma-Aldrich)–coated 96-well plates. After continuing to culture for
   16 hours, cells were washed with washing buffer [1× HBSS with
   Ca^2+/Mg^2+, 10 mM Hepes, and 0.1% BSA (pH 7.4)]. Then, cells were
   loaded with 2 μM Ca^2+ indicator (Fluo-4 AM) diluted with washing
   buffer supplemented with 0.05% pluronic-F127 (Sigma-Aldrich) and 1 mM
   probenecid (MCE) and then cultured in 37°C incubator for 30 min. Later,
   loading buffer with indicator was replaced with washing buffer to
   incubate for another 30 min, and, then, cells were washed again and
   loaded with washing buffer. Intracellular Ca^2+ flux induced by
   stimulation of different metabolites was analyzed by Nikon confocal
   through collecting fluorescence images at one frame per second after
   the metabolite was added into cells. Collection process was lasted for
   180 s, and, then, the fluorescence intensity of every image was
   analyzed by NIS Elements AR software, and fluorescence/background
   (F/F[0]) was quantified over time as an index of P2Y10 receptor
   activation.

Analysis of activated RhoA signaling

   Active RhoA (guanosine 5′-triphosphate bound) was determined by RhoA
   G-LISA activation kits (Cytoskeleton). HEK293T cells overexpressing
   P2Y10 and combined G proteins were obtained as above and then
   stimulated by indicated metabolite. Basal levels and metabolite-induced
   changes in activation were determined.

   For luciferase reporter assay to determine the activation of RhoA
   signaling, HEK293T cells overexpressing P2Y10 and combined G proteins
   were seeded into six-well plates. Subsequently, 2 μg of
   pGL4.34[luc2P/SRF-RE/Hygro] plasmid (ZHUANGMO Bioscience, ZK39238) was
   transfected into the cells, which were then collected after 6 hours and
   seeded into 96-well plates. Following a 12-hour incubation, the culture
   medium was replaced with serum-free Dulbecco’s modified Eagle’s medium
   (DMEM) (100 μl per well) and continued to incubate for 8 hours. To
   stimulate the cells, the metabolites with indicated concentrations were
   diluted in DMEM containing 10% FBS and added to each well after
   removing 50 μl of the original culture medium. The cells were then
   incubated at 37°C for 6 hours. A control group without stimulation was
   set up. The relative light units (RLU) of luminescence signals were
   detected with the ONE-Glo Luciferase Assay System (Promega, E6110)
   according to the manufacturer’s protocol. Fold
   induction = RLU[induced]/RLU[noninduced].

Analysis of activated RhoA signaling in ILCs

   To analyze the active RhoA levels in intestinal ILCs by flow cytometry,
   LPLs were isolated and stained with specific antibodies against
   lineage, Thy1.2, and CD127. After washing, the cells were stimulated
   with 10 μM TDCA for 5 min. Subsequently, the cells were fixed with 4%
   PFA and permeabilized with 0.1% Triton X-100 for 30 min on ice. The
   cells were then incubated with glutathione S-transferase (GST)–tagged
   Rho binding domain (RBD) of rhotekin for 3 hours at room temperature,
   followed by washing with TBST and incubation with rabbit anti-GST
   antibodies (no. ab111947, Abcam; 1:200) for 2 hours at room
   temperature. After washing, the cells were incubated with Alexa Fluor
   488–labeled goat anti-rabbit secondary antibodies for 1 hour at room
   temperature. The flow cytometry analysis was performed to determine the
   active RhoA levels in ILCs.

   To analyze the intracellular distribution of active RhoA using
   immunofluorescence, total ILCs were sorted from the SI of P2ry10^+/+
   and P2ry10^−/− mice. The sorted cells were seeded on chambered
   coverslips coated with 10% poly-l-lysine solution. The cells were then
   stimulated with 10 μM TDCA for 5 min. After fixation with 4% PFA and
   permeabilization with 0.1% Triton X-100 for 10 min at 4°C, the cells
   were blocked with 10% donkey serum in PBS at room temperature.
   Subsequently, the cells were incubated with GST-tagged RBD of rhotekin
   for 24 hours at 4°C, then washed, and incubated with rabbit anti-GST
   antibodies overnight at 4°C. After washing, the cells were incubated
   with Alexa Fluor 488–labeled goat anti-rabbit secondary antibodies for
   1 hour at room temperature. Last, the cells were stained with DAPI and
   imaged by using a Nikon confocal microscope.

DSS-induced inflammation model

   P2ry10^+/+ and P2ry10^−/− mice were treated with 2% (w/v) DSS (MP
   Biomedicals) in the drinking water for 5 days and followed by regular
   drinking water for 5 days. For TDCA-treated mice, mice were treated
   with TDCA (1 mg/20 g of body weight, diluted in drinking water) via
   orally gavage daily for 2 weeks while maintained on Abx drinking for 2
   weeks, and, then, mice with different treatment were treated with 2%
   (w/v) DSS as above. Body weight was measured every day and mice were
   euthanized at day 10. Colons were collected for pathological analysis.

BM transductions and transplantation

   For Zfp414 knockout in BM cells, sgZfp414s (table S1) were cloned into
   the adeno-associated virus (AAV) vector with Cre recombinase. sgRNA
   sequences were selected by using Benchling’s CRISPR Guide tool. HEK293T
   cells were used for adenovirus packaging. AAV plasmid (8 8 μg),
   packaging plasmid pHelper (4 μg), and pAnc80L65 (4 μg) were mixed and
   transfected into HEK293T cells by using Lipofectamine 3000 reagent. On
   the second day after transfection, intracellular AAV was collected by
   freeze thawing with liquid nitrogen and then concentrated via
   ultracentrifuged at 25,000 rpm for 2 hours at 4°C. For Zfp414
   overexpression, the CDS sequence of Zfp414 was cloned into
   PLVX-IERS-Puro lentivirus vector. Concentrated lentivirus was obtained
   as above. For BM transduction, Rosa26-cas9-P2A-EGFP mice were injected
   intravenously with 3 mg of 5-fluorouracil (Sigma-Aldrich). After 4 to 5
   days, BM was collected and cultured in complete RPMI 1640 medium
   supplemented with IL-3 (20 ng/ml), IL-6 (50 ng/ml), and SCF (100
   ng/ml). BM cells were spin infected (1000g, 2 hours, 32°C) at days 1
   and 2 to introduce viral vectors. After the second infection, the
   infected BM cells (5 × 10^6) were 1:1 mixed with BM cells from CD45.1^+
   mice. The mixed cells were then transplanted into lethally irradiated
   CD45.1^+ recipients. Two months later, flow cytometry analysis was
   performed.

In vitro differentiation assay and ILCs labeling

   Intestinal ILC in vitro differentiation assay was performed as
   previously described ([199]54). Briefly, cells from femurs were flushed
   out by using PBS containing 1% FBS and filtered through 40-μm
   strainers. Collected cells were treated with red blood cell lysis
   buffer (Tiangen, Beijing) to exclude red cells. CHILPs
   (Lin^−CD25^−CD127^+Flt3^−α4β7^+Id2^GFP) and ILCPs
   (Lin^−CD127^+Flt3^−α4β7^+c-Kit^+PLZF^GFP) were sorted by FACSAria III
   instrument and transduced by lentivirus containing shItgae through spin
   infection (1000g, 2 hours, 32°C) at days 1 and 2. OP9-DL1 cells were
   maintained in complete minimum essential medium alpha (αMEM) medium
   (supplemented with 10% FBS, 1% streptomycin, and 1% penicillin). Before
   seeding of progenitor cells, OP9-DL1 cells were treated with mitomycin
   (4 μg/ml) for 2 hours to inhibit cell division. Then, cells were
   digested and seeded at the density of 1 × 10^6 cells per well in a
   24-well plate. After OP9-DL1 cells were adhered, and indicated
   progenitor cells were inoculated on OP9- DL1 feeder cells in a complete
   RPMI 1640 medium [supplemented with 10% FBS, 1% streptomycin, 1%
   penicillin, IL-7 (20 ng/ml), and SCF (20 ng/ml)]. After 14 days,
   differentiated ILCs were harvested and labeled with carboxyfluorescein
   diacetate succinimidyl ester (CFSE; 5 μM; Thermo Fisher Scientific,
   [200]C34570) for 15 min at 37°C, and, then, CFSE was quenched with PBS
   for 3 min. Labeling ILCs (5 × 10^6) were resuspended with complete RPMI
   1640 medium and injected intravenously into immunodeficient NSG mice.
   The recipients were euthanized 18 hours later for further
   immunofluorescence.

C. rodentium infection

   C. rodentium was a gift from B. Ge (Shanghai Institutes for Biological
   Sciences, Chinese Academy of Sciences). Mice were fasted for 5 hours
   and then orally infected with 5 × 10^9 colony-forming units (CFU) of C.
   rodentium. Mice were weighed daily; on day 10 after infection, mice
   were euthanized, and colons were collected for pathological analysis.
   Feces were collected and homogenized for bacterial load measurement.

Histopathological analysis

   Colons from indicated mice were cut open longitudinally and rinsed with
   PBS before fixation in 4% PFA followed by paraffin sectioning and
   hematoxylin and eosin (H&E) staining. The histopathological score was
   evaluated and analyzed according to a previous study ([201]55).
   Briefly, the sections were evaluated for leukocyte infiltration (0 for
   absent, 1 for mild, 2 for moderate, and 3 for severe infiltration),
   epithelial injury (0 for normal villus structure, 1 for mild epithelium
   loss, 2 for moderate epithelium loss, and 3 for severe epithelium
   loss), submucosal edema (0 for no edema, 1 for <50% submucosal area
   and <100 mm wide, 2 for 50 to 80% submucosal area and <200 mm wide, and
   3 for >80% submucosal area and > 200 mm wide), and the number of goblet
   cells (0 for >40 goblet cells per visual field, 1 for 26 to 40 goblet
   cells per visual field, 2 for 16 to 25 goblet cells per visual field,
   and 3 for 0 to 15 goblet cells per visual field).

Inhibition of RhoA and Ca^2+ signaling cascade

   Sorted intestinal ILCs were rested for 2 hours in RPMI 1640 medium at
   37°C. Then, cells were incubated with 1 μM Ca^2+ inhibitor U-73122
   (HY-13419-5, MCE), 1 μM ROCK inhibitor (S)-H-1152, 1 μM (10007653,
   Cayman), or vehicle (dimethyl sulfoxide) during overnight stimulation
   with TDCA at 37°C. Then, the mRNA expression of Cd69, S1pr1, and Itgae
   was analyzed by qPCR.

ChIP assay

   Intestinal ILCs were isolated and crosslinked with 1% formaldehyde at
   37°C for 10 min. After two PBS washes, cells were lysed with
   SDS-containing buffer and subjected to sonication to generate DNA
   fragments between 200 and 500 base pairs. Lysates underwent
   preclearance with Protein A/G magnetic beads before overnight
   incubation at 4°C with 2 μg of ZFP414 antibodies. Following
   immunoprecipitation and sequential washes, bound DNA complexes were
   eluted and purified for subsequent analysis. Quantitative detection was
   performed using qPCR with primer pairs detailed in table S3.

Transcriptome and bioinformatic analysis

   ILCs (Lin^−Thy1.2^+CD127^+; 1 × 10^6) from Ctrl and Abx-treated or
   P2ry10^+/+ and P2ry10^−/− mice were sorted by using flow cytometry, and
   cells were lysed with TRIzol reagent, and, then, total RNA was
   extracted and subjected to RNA-seq by BGI Tech Company. R packages of
   “clusterProfiler” in Bioconductor were used for generating GO analysis.
   For bioinformatic analysis of RNA-seq of Ctrl and Abx-treated ILC1,
   ILC2s, and ILC3s, the scRNA-seq data were from [202]GSE85154 ([203]25).

Liquid chromatography–mass spectrometry

   For screening metabolites affected by microbiota deficiency, the
   contents of intestinal lumen were obtained and homogenized. Then, 400
   μl of precooled methanol/acetonitrile/water solution (4:4:2, v/v/v) was
   added to 1 ml of sample. The mixture was vortexed, incubated at −20°C
   for 60 min, and then centrifuged at 14,000g for 20 min at 4°C. The
   supernatant was collected and dried in vacuum. For mass spectrometry
   (MS) analysis, 100 μl of acetonitrile-water solution
   (acetonitrile:water = 1:1, v/v) was added to reconstitute the samples.
   The mixture was vortexed and centrifuged at 14,000g for 15 min at 4°C,
   and 2 μl of the supernatant was taken for analysis. The chromatographic
   separation was performed by using an ACQUITY UPLC BEH C18 column (100
   mm by 2.1 mm, 1.7 μm, Waters, USA) at a column temperature of 40°C and
   a flow rate of 0.3 ml/min. The mobile phase consisted of water with
   0.1% formic acid as solvent A and acetonitrile as solvent B. The
   following gradient was used for metabolite elution: 0 to 0.5 min, 5% B;
   0.5 to 1.0 min, 5% B; 1.0 to 9.0 min, 5 to 100% B; 9.0 to 12.0 min,
   100% B; 12.0 to 15.0 min, 5% B. The injection volume for each sample
   was 5 μl. After separation by ultrahigh performance liquid
   chromatography (UHPLC), the samples were subjected to MS analysis using
   a Q-Exactive quadrupole-Orbitrap high-resolution mass spectrometer
   (Thermo Fisher Scientific). The analysis was operated in positive and
   negative electrospray ionization (ESI) mode. ESI settings were as
   follows: Ion Source Gas1, 60; Ion Source Gas2, 60; Curtain gas, 30;
   source temperature, 320°C; and IonSapary Voltage Floating, ±3500 V for
   both positive and negative ion modes. The MS scan mass/charge ratio
   range was set from 80 to 1200 Da, with a product ion scan resolution of
   17,500. The MS scan accumulation time was 0.20 s per spectra, and the
   product ion scan accumulation time was 0.05 s per spectra. The
   second-stage MS was performed by using information-dependent
   acquisition in high-sensitivity mode. The following settings were used:
   exclude isotopes within 4 Da; candidate ions to monitor per cycle, 6.
   The declustering potential was set at ±60 V for both positive and
   negative ion modes, and the collision energy was set at 35 ± 15 eV.
   Data from liquid chromatography–MS/MS was analyzed with Compound
   Discoverer 3.0 (Thermo Fisher Scientific) and searched in BioCyc, HMDB,
   METLIN, HFMDB, and LIPID MAPS databases.

Statistics and reproducibility

   The numbers of mice and experiments, along with the statistical tests,
   are reported in each figure legend. For statistical evaluation, data
   were analyzed by GraphPad Prism 8.0. Two-tailed unpaired Student’s t
   test was used in this study. Results are shown as means ±SD. P values
   of ≤ 0.05 were considered significant (*P < 0.05; **P < 0.01;
   ***P < 0.001), and P > 0.05 was considered nonsignificant (n.s.). All
   flow cytometry data were analyzed by FlowJo 10.

Acknowledgments