TIGIT mediates activation-induced cell death of ILC2s during chronic airway allergy

Overactivated ILC2s express TIGIT that induces activation-induced cell death via cell–cell contact with alveolar macrophages, which suppresses overwhelmed inflammation caused by chronic allergy.

We previously reported that highly activated ILC2s express TIGIT together with KLRG1 and PD-1. We tentatively named these TIGIT + ILC2s "exhausted-like" ILC2s because these cells are characterized by a high expression of inhibitory receptors and IL-10 and a low mRNA expression of Il5 and Il13 Miyamoto et al., 2019). When ILC2s lack runtrelated transcription factor (RUNX), the number of exhaustedlike ILC2s increases, and allergic airway inflammation is improved. However, the physiological importance of TIGIT + ILC2s in chronic allergy remains to be elucidated in RUNXcompetent mice.
The interaction of TIGIT with CD155 on dendritic cells and tumor cells results in suppression of the cytotoxic activity of natural killer cells and CD8 + T cells (Harjunpää and Guillerey, 2020). TIGIT can also bind to CD112 and CD113, though with less affinity (Bottino et al., 2003;Yu et al., 2009). Meanwhile, in regulatory T (Treg) cells, TIGIT-mediated signaling induces Foxp3 expression by inhibiting the T cell receptor-AKT-mTORC1 pathway (Sato et al., 2021). These TIGIT + Treg cells selectively inhibit T H 1/T H 17 differentiation via the secretion of fibrinogen-like protein 2, thereby promoting T H 2 skewing (Joller et al., 2014). TIGIT is also expressed on T H 2 cells and enhances memory immune responses to ovalbumin (Kourepini et al., 2016). However, whether TIGIT plays a regulatory or stimulatory role in chronic allergy has not yet been determined.
Despite their immunosuppressive interactions with soluble factors and ligands, ILC2s tend to be increased in chronic allergy. They also have high resistance to cell death and can proliferate in IL-33-containing medium for several months in vitro (Moro et al., 2016). The fate of overactivated ILC2s, however, has not been well studied. Indeed, they are expected to undergo cell death in chronic allergy due to the limited space in vivo, yet no such phenomena have been identified.
In this study, we evaluated the fate of TIGIT + ILC2s as well as the regulatory effect of TIGIT in ILC2s during chronic allergy by using a fate-mapping mouse model. Chronic airway allergy was induced by repeated papain treatment, and TIGIT + ILC2s were marked and traced by tdTomato expression. During chronic allergy, tdTomato + ILC2s were stably induced and were apoptotic with short life spans. Data from phenotypical studies, transcriptomic analysis, and assay for transposase-accessible chromatin sequencing (ATAC-seq) revealed that tdTomato + ILC2s expressed high levels of IL-5 and IL-10 but exhibited reduced transcription of ILC2 signature genes due to a reduction in chromatin accessibility. Cell death in tdTomato + ILC2s was augmented following interactions with CD155-expressing alveolar macrophages. Both the genetic deletion of Tigit and blockade of TIGIT enhanced the survival of activated ILC2s, resulting in the deterioration of chronic allergy. These data suggest that TIGIT + ILC2s undergo activation-induced cell death (AICD) in chronic airway allergy, with TIGIT accelerating this process; these also suggest a new regulatory mechanism involved in chronic allergy.
To induce TIGIT and tdTomato expression in ILC2s, we established a chronic airway allergy mouse model via the repeated nasal administration of papain twice a week until day 17. Tamoxifen was orally administered on every day of papain treatment and every subsequent day. On day 19, CD4 + T cells were the major population of tdTomato + cells, which accorded with previous reports (Fig. S1, Joller et al., 2014;Kourepini et al., 2016). In lungs that had not been treated with papain, ILC2s did not express tdTomato (Fig. 1 A). Following treatment, a small population of tdTomato + ILC2s (1-2%) was detected in the lungs and a considerably smaller population was detected in the bronchoalveolar space ( Fig. 1 A and Fig. S1 C). Although subtle, TIGIT expression correlated with TIGIT-GFP and tdTomato expression ( Fig. 1 A and Fig. S2 A).
To explore what cytokines are crucial to induce tdTomato in ILC2s, Tigit fm mice were intranasally instilled with IL-25 or IL-33 with the tamoxifen treatment every 3 d. On day 8, more tdTomato + ILC2s were induced in the mice treated with IL-25 than with IL-33, suggesting that IL-25 plays an important role in induction of TIGIT + ILC2s ( Fig. 1 B). Additionally, inflammatory ILC2s expressed tdTomato more frequently than natural ILC2s after IL-25 treatment, despite tdTomato + ILC2s being natural ILC2s in our papain-induced chronic airway allergy model ( Fig. 1 B and Fig. S1 C).
To characterize tdTomato + ILC2s, expression of possible activation markers was examined. On day 19, tdTomato − and tdTomato + ILC2s expressed comparable levels of Thy1, KLRG1, ST2, and FAS ( Fig. 1 C). In contrast, PD-1 expression at any time point during papain treatment was statistically higher in tdTomato + than in tdTomato − ILC2s ( Fig. 1 C and Fig. S2 B). While a previous study showed that TIGIT + ILC2s express reduced mRNA levels of Il5 and Il13 (Miyamoto et al., 2019), tdTomato + ILC2s in the present study expressed higher levels of IL-5-venus and IL-10-venus proteins than those in tdTomato − ILC2s (Fig. 1, D and E). Although ex vivo IL-5 detection is technically challenging without reporter, more IL-5 protein was detected ex vivo in tdTomato low ILC2s than in tdTomato − ILC2s and tdTomato high ILC2s (Fig. S2, C and D). These data suggest that tdTomato + ILC2s are more activated than tdTomato − ILC2s and that they lose IL-5 during the induction of TIGIT and tdTomato expression.
We next investigated the time point during papain treatment at which tdTomato + ILC2s appeared ( Fig. 1 F). They emerged in the lungs on day 8, with their population increasing proportionally with the number of papain treatments and at a consistent frequency of ∼1%. To determine the survivorship of TIGIT + ILC2s after treatment cessation, we terminated papain treatment on day 7 and then analyzed the tdTomato + ILC2s populations in the lungs later. Until day 36, the number and (F) Number of total (left) and tdTomato + (middle) lung ILC2s and frequency of tdTomato + cells in the indicated lung ILC2s (right) in Tigit fm mice administered papain and tamoxifen for the indicated days. (G) Tigit fm mice were treated with papain and tamoxifen until day 7. Number of total (left) and tdTomato + (middle) lung ILC2s and frequency of tdTomato + cells in the indicated lung ILC2s (right) were analyzed on subsequent indicated days. gMFI, geometric mean fluorescent intensity; nILC2s, natural ILC2s (CD45 + Lin − CD127 + KLRG1 + ST2 + ); iILC2s, inflammatory ILC2s (CD45 + Lin − CD127 + KLRG1 hi ST2 low ). *P < 0.05, **P < 0.01, and ***P < 0.001, as determined by unpaired two-tailed t test. Data represent at least two independent experiments (mean ± SEM of three mice in B, C, and G; mean ± SEM of four mice in E and F). frequency of tdTomato + ILC2s decreased to undetectable levels ( Fig. 1 G). Additionally, during and after the cessation of papain treatment, tdTomato + ILC2s did not lose TIGIT (Fig. S2 A). These data indicate that chronic inflammation is necessary for the maintenance of tdTomato + ILC2s.
TIGIT + ILC2s are dysfunctional due to global transcriptional arrest To examine the transcriptomic changes in tdTomato + ILC2s, we compared the gene expression profiles of tdTomato − and tdTomato + ILC2s by RNA sequences. The expressions of lineage markers were quite low and did not statistically differ between the samples (Fig. S3 A). Although approximately half of the transcripts accounted for protein-coding genes in tdTomato − ILC2s, non-coding transcripts, such as major satellite repeat transcripts and long non-coding RNAs, dominated the proteincoding transcripts in tdTomato + ILC2s (Fig. 2, A and B). Major satellite repeats are transcribed as non-coding RNA from the pericentromere and contribute to heterochromatin structure (Eymery et al., 2009). Examination of differentially expressed genes revealed that the expression of ILC2 signature genes was reduced in tdTomato + ILC2s (Fig. 2, C and D; and Table S1). Regarding apoptosis-related genes, cathepsin genes were upregulated in tdTomato + ILC2s relative to tdTomato − ILC2s.
To gain further insight into the biological processes in tdTomato + ILC2s, we applied gene ontology analysis to the differentially expressed gene sets (Fig. 2 E). Negative enrichment was observed for chromatin organization, transcription, mRNA processing/splicing, and translation. These data suggest that TIGIT + ILC2s are overactivated ILC2s that, despite possessing residual IL-5 and IL-10 proteins, globally retard their basic cellular functions.

Global decreases in chromatin accessibility occur in TIGIT + ILC2s
Sustained antigen stimulation induces cell death in exhausted CD8 + T cells accompanied by epigenetic changes (Kurachi, 2019). To assess whether tdTomato + ILC2s acquire a distinct epigenetic state, we performed ATAC-seq using sorted tdTomato − or tdTomato + ILC2s. The results showed that 87% of annotated genes with reduced ATAC peaks in tdTomato + ILC2s were protein-coding genes ( Fig. 2 F). Most genes that were downregulated in tdTomato + ILC2s, determined by RNA sequences, are associated with reduced chromatin accessibility (Fig. 2 G and  Table S2). Motif analysis of the reduced peaks in tdTomato + ILC2s revealed that the loss of chromatin accessibility by GATA-3, Fli1, and KLF1 may lead to the dysfunction in tdTomato + ILC2s ( Fig. 2 H). Chromatin accessibility was reduced for ILC2 signature genes such as Gata3, Il5, and Il13, as well as in Tigit and Il10 loci (Fig. 2 I). Although mRNA expression levels of Tigit and Il10 were comparable between tdTomato − and tdTomato + ILC2 cells, their genomic loci lost chromatin accessibility (Fig. 2, D and I), suggesting that chromatin accessibility in the Tigit and Il10 gene loci may be maintained at the last moment of transcriptional arrest. In contrast, significant changes were not observed at the genomic loci for major satellite repeats ( Fig. 2 I, bottom right). These data indicate that a global reduction in chromatin accessibility may suppress the transcription of coding genes.
TIGIT + ILC2s are apoptotic and short-lived during chronic airway allergy We next wondered why highly activated tdTomato + cells were so rare even while allergic inflammation was getting very severe in the lung. If chronic allergy induced a stable tdTomato + ILC2 population, these cells would be expected to accumulate at sites of chronic allergy. Therefore, we speculated that tdTomato + ILC2s in chronic allergy either migrated to other tissues or had died due to overactivation. While ILC2s are tissue-resident cells, activated ILC2s can enter the bloodstream and lymphatic vessels (Huang et al., 2018;Mathä et al., 2021); however, no tdTomato + ILC2s were observed in the peripheral blood, draining lymph nodes, adipose tissues, and small intestine (Fig. 3 A and Fig. S3  B), suggesting that the recruitment of tdTomato + ILC2s to other tissues is unlikely.
To determine the fate of tdTomato + ILC2s in the airway, we isolated tdTomato − and tdTomato + ILC2s from Tigit fm (CD45.2 + ) mice treated with papain and tamoxifen for 19 d and intratracheally transferred the 500 cells to CD45.1 + mice treated with papain in the same manner for 8 d. 1 h after intratracheal transfer, tdTomato + ILC2s were scarcely recovered from the bronchoalveolar lavage fluid (BALF; Fig. 3 B). Furthermore, the total activity of caspases in tdTomato + ILC2s was greater than that in tdTomato − ILC2s (Fig. 3 C), indicating that tdTomato + ILC2s are apoptotic and quickly disappear from the inflamed airway. The prompt disappearance might enable the cells to die with residual IL-5 and IL-10 reporter proteins before transcriptional arrest impacts the protein expression.

CD155 on alveolar macrophages induces cell death in TIGIT + ILC2s
To further assess the fate of tdTomato + ILC2s, we cultured the cells under general allergic conditions containing IL-2, IL-7, and IL-33 or under IL-10-inducing conditions containing these three cytokines plus retinoic acid. In both culture media, a lower proliferation and higher number of dead cells was observed in tdTomato + than in tdTomato − ILC2s (Fig. 3 D). TIGIT inhibits the phosphorylation of AKT and, thereby, mTORC1, which is known to negatively regulate apoptosis (Sato et al., 2021;Weichhart et al., 2015). Accordingly, a lower level of AKT phosphorylation was observed in Tigit −/− ILC2s that retrovirally overexpressed TIGIT and were stimulated with soluble CD155 than in non-transduced cells and ILC2s transduced with control vector (Fig. 3 E). Next, to assess the role of TIGIT in the fate of tdTomato + ILC2s, we stimulated these cells in vitro with the anti-TIGIT agonistic antibody 1G9 (Schorer et al., 2020). TIGIT signaling induced the death of tdTomato + ILC2s, while no such results were observed with the anti-TIGIT antagonistic antibody 1B4 (Fig. 3 F).
CD155 was robustly expressed on alveolar macrophages during allergic inflammation (Fig. 4 A). To examine whether alveolar macrophages interact with TIGIT + ILC2s via CD155, alveolar macrophages from either Cd155 +/+ or Cd155 −/− mice were cultured with tdTomato − lung ILC2s from Tigit fm mice in the presence of tamoxifen (Fig. 4, B and C). Cd155 +/+ alveolar macrophages efficiently killed tdTomato + ILC2s, resulting in a reduction in the number of cultured ILC2s (Fig. 4, B-D). This cytotoxic ability was impaired in Cd155 −/− alveolar macrophages (Fig. 4, B and C), indicating that CD155 is required for the cytotoxic effects of alveolar macrophages against tdTomato + ILC2s. Furthermore, we confirmed that macrophages colocalized with CD3 − tdTomato + cells (mostly ILC2s) in the ILC2-rich area of the adventitial cuffs of the lung (Fig. 4 E). Following deletion of alveolar macrophages by intratracheal administration of clodronate liposomes in Tigit fm mice with chronic allergy, the number of tdTomato + ILC2s increased in the airway (Fig. 4 F). Therefore, alveolar macrophages eliminate highly activated TIGIT + ILC2s. We tentatively refer to this form of cell death as AICD since tdTomato + ILC2s are in their final activation state.
TIGIT regulates a pool of activated ILC2s during chronic allergy To determine the intrinsic function of TIGIT in ILC2s, we adoptively transferred a 1:1 ratio of CD45.1 + Tigit +/+ and CD45.2 + Tigit −/− bone marrow (BM)-derived cells to sublethally irradiated CD45.1 + /CD45.2 + recipient mice and induced chronic airway allergy. Before papain treatment, Tigit +/+ and Tigit −/− ILC2s were evenly distributed in the lungs of the recipient mice (Fig. 5, A and  B). However, chronic airway allergy induced a higher number of Tigit −/− ILC2s in the bronchoalveolar space and lungs than that of Tigit +/+ ILC2s (Fig. 5, A and B). Notably, this trend was prominent in the bronchoalveolar space, where the most severe airway inflammation was expected due to the intranasal administration of papain. Functionally, Tigit −/− ILC2s comparably produced IL-5 and IL-13 relative to Tigit +/+ ILC2s (Fig. 5, C and D). We next assessed the physiological relevance of TIGIT to chronic airway allergy by administering anti-TIGIT antagonistic antibody to Tigit fm mice with chronic allergy. TIGIT blockade increased the number of total and tdTomato + lung ILC2s as well as the ratio of tdTomato + cells among ILC2s and decreased the death rate of tdTomato + ILC2s, resulting in an increased number of lung eosinophils and histological inflammation (Fig. 5, E-H). However, the blockade did not statistically increase IL-5 and IL-13 production by lung ILC2s (Fig. 5 I). These data suggest that TIGIT in ILC2s controls ILC2 number, but not cytokine production during chronic allergy.
In this study, we have first provided evidence that chronic allergy constantly induces cell death in overactivated ILC2s via the TIGIT/CD155 axis and that AICD diminishes eosinophilic allergic inflammation. Mechanistically, macrophages interacted with TIGIT + ILC2s via CD155 and enhance AICD, which was found to be accompanied by an increase in caspase activity, transcriptional arrest, and a global reduction in the chromatin accessibility of coding genes. TIGIT + ILC2s were a rare population due to the high cell death rate but possessed significant regulatory functions by AICD in total. Thus, the AICD of ILC2s is a novel  (left) and frequency of dead cells in tdTomato + ILC2s (right) following culture with IL-2, IL-7, IL-33, and RA in the presence of isotype antibody (Iso), anti-TIGIT antagonistic antibody (1B4), or anti-TIGIT agonistic antibody (1G9). *P < 0.05 and **P < 0.01, as determined by unpaired two-tailed t test. Data represent at least two independent experiments (mean ± SEM of three mice in B and C; mean ± SEM of technical triplicates in D-F). mechanism by which allergic inflammation is suppressed during chronic allergy.
We have clarified the critical role of CD155 on alveolar macrophages in the AICD of TIGIT + ILC2s. Alveolar macrophages comprise a major hematopoietic cell population in the airway, highly express CD155, and colocalize with ILC2s in the lungs (Fig. 4, A and E). M2 macrophage activation is dependent on IL-4, IL-5, and IL-13, which are secreted by ILC2s (Bouchery et al., 2015;Shen et al., 2021). In turn, M2 macrophages promote ILC2 differentiation from ILC progenitors (Pei et al., 2020). Metabolic enzymes in macrophages are critical for macrophage-mediated ILC2 activation (Nieves et al., 2016;Yamaguchi et al., 2018). Thus, macrophages interact with ILC2s to initiate type 2 inflammation. Given that macrophages help promote ILC2 activation through close proximity until the acquisition of TIGIT by ILC2s in chronic allergy, they can quickly remove overactivated TIGIT + ILC2s via direct cell-cell contact.
TIGIT/CD155 binding inhibits natural killer cell function via the recruitment of SHIP1 and the inhibition of phosphoinositide 3-kinase and mitogen-activated protein kinase signaling pathways (Liu et al., 2013). However, in Treg cells, TIGIT augments Treg cell functions by inhibiting AKT-mTORC1 signaling (Sato et al., 2021). The function of mTORC1 is to help detect growth factors and nutrients and to inhibit the apoptotic pathway. ILC2s also depend on mTOR signaling for IL-33-mediated cytokine production and proliferation (Salmond et al., 2012). Thus, mTORC1 may be involved in the AICD of tdTomato + ILC2s.
Regulatory roles of IL-10 + ILC2s have been suggested (Golebski et al., 2021). However, Seehus et al. clearly showed that IL-10 + ILC2s produce more IL-5 and IL-13 than IL-10 − ILC2s in vivo (Seehus et al., 2017). Retinoic acid is known to induce IL-10 in ILC2s, but activates ILC2s for cell proliferation in vitro (Seehus et al., 2017). We showed that IL-10 + ILC2s apparently include highly activated cells that can be removed quickly. Regulatory roles of IL-10 + ILC2s may be mediated in part by the AICD of overactivated ILC2s.
The accumulation of activated ILC2s at sites of allergic inflammation is associated with human allergic disease progression (Doherty et al., 2015;Roediger et al., 2013;Smith et al., 2016). Our data suggest that the number of activated ILC2s in mice is regulated by AICD via TIGIT. In humans, a similar mechanism may facilitate the removal of highly activated ILC2s. Future studies are warranted to determine whether TIGIT is expressed on activated human ILC2s or whether other molecules can mediate the AICD of human ILC2s, as well as to determine the potential of enhancing the AICD of ILC2s as a new strategy to treat chronic allergy.

Mice
All mice were bred and maintained at a specific pathogen-free facility at the Akita University Graduate School of Medicine, Akita, Japan, and the animal protocols were approved by the ethics review board of Akita University. C57BL/6 and congenic  (left) and tdTomato + ILC2s (right) in the BALF of Tigit fm mice intratracheally treated with control liposomes or clodronate liposomes. *P < 0.05, **P < 0.01, and ***P < 0.001, as determined by unpaired two-tailed t test. Data represent at least two independent experiments (mean ± SEM of technical triplicates in B-D; mean ± SEM of four mice in F).

Chronic airway allergy and tamoxifen treatment
To establish chronic airway allergy models, mice were intranasally administered 200 µg papain (FUJIFILM Wako) in 50 μl of sterile PBS every 3-4 d for 19 d; alternatively, they were intranasally administered 500 ng of either IL-25 (Peprotech) or IL-33 (Peprotech) in 50 μl of PBS on days 0, 3, and 6. To induce tdTomato expression, 2 mg of tamoxifen in 100 μl corn oil was orally administered on every day of papain treatment and every subsequent day. BALF and lungs were harvested 24 h after the final administration of tamoxifen. To block TIGIT-ligand interaction, mice were intraperitoneally treated with 100 µg of isotype antibody or anti-TIGIT antagonistic antibody (1B4; Bio X Cell) on days 12, 14, and 16. To delete alveolar macrophages, 100 μl of control liposomes (F70101-N; FormuMax) or clodronate liposomes (F70101C-N; FormuMax) was intratracheally administered to mice on days 15 and 16 during papain treatment. BALF was collected to count alveolar macrophages and tdTomato + ILC2s 24 h after the final papain and tamoxifen treatments on day 17.

Cell preparation and flow cytometry
Cell-containing BALF was obtained via intratracheal infusion of PBS through a catheter. Lungs were dissected and minced using scissors and then incubated in 8 ml of digestion buffer containing RPMI medium supplemented with 2% FBS, 4 mg collagenase IV (C5138; Sigma-Aldrich), and 400 μg of DNase (043-26773; FUJIFILM) at 200 rpm and 37°C for 45 min. The digested cells were forced through a 70-µm strainer and subjected to red blood cell lysis before being used for flow cytometry and cell culture. For flow cytometry, cells were stained with Fixable Viability Dye eFluor 506 (eBioscience) to detect dead cells. The antibodies used for flow cytometry are listed in Table S3. Data were acquired on a FACSAria system (BD Biosciences) and analyzed using FlowJo software (TreeStar). For cell culture, RNA sequencing (RNA-seq), and ATAC-seq, tdTomato − and tdTomato + ILC2s were sorted and separated from lung lymphocytes using FACSAria (BD Biosciences). To detect intracellular cytokines, cells were cultured with 50 ng/ml of phorbol myristate acetate, 0.5 µg/ml of ionomycin, and Gol-giPlug for 3 h followed by antibody staining using the Cytofix/ Cytoperm Buffer Set (BD Biosciences). Foxp3/Transcription Factor Staining Buffer Set (eBioscience) was used to stain the transcription factors. To detect active caspase enzymes via flow cytometry, the sorted cells were stained with FAM-VAD-FMK FLICA (ImmunoChemistry Technologies) as per the manufacturer's protocol.

Histological analysis
To prepare paraffin blocks, mouse lungs were intratracheally infused with 1 ml of 10% formalin, removed, and incubated in 10% formalin at 4°C overnight. Hematoxylin and eosin staining or PAS staining was performed as previously described (Miyamoto et al., 2019). Airway allergy was assessed in two ways: first, by counting the number of cell layers around six bronchi per mouse (using ×400 magnification); second, by determining the frequency of PAS + cells among eight views of bronchial epithelial cells per mouse (using ×400 magnification), which was performed by a researcher blinded to the experimental grouping.
To make frozen sections, lungs were intratracheally infused with 1 ml of 10% formalin, incubated for 10 min, and infused with 50% OCT compound (Sakura Finetech) in PBS after removal of the formalin. The frozen sections were incubated with anti-mouse CD68 rabbit antibody (E3O7V; Cell Signaling Technology) and anti-mouse CD3e rat antibody (17A2; Biolegend) overnight at 4°C. After washing with PBS, the sections were stained with goat anti-rabbit IgG (H + L) secondary antibody conjugated to Alexa Fluor 405 (Thermo Fisher Scientific) and goat anti-rat IgG (H + L) secondary antibody conjugated to Alexa Fluor 647 (Thermo Fisher Scientific) at room temperature for 1 h. All images were analyzed using a BZ-X800 (Keyence).

Transcriptomic analysis
The tdTomato − and tdTomato + ILC2 cells (1.0 × 10 2 cells) were sorted from male Tigit fm mouse cells using FACSAria (BD Biosciences) and then flash-frozen in liquid nitrogen. RNA-seq libraries were generated using SMART-Seq Stranded Kit (Takara Bio) according to the "low-input protocol" provided by the manufacturer. Sequencing was performed using a HiSeqX (Illumina). After quality checking with FASTP, paired-end reads were aligned to the mouse reference genome (Ensembl GRCm39) using STAR with standard input parameters. Transcript counts were determined using featureCounts (subread package) of the Ensembl annotation (Release 104) and processed to identify differentially expressed genes and generate volcano plots and heatmaps using EdgeR. Gene ontology analysis was performed using the Database for Annotation, Visualization, and Integrated Discovery.

ATAC-seq
The tdTomato − and tdTomato + ILC2 cells (2,000-5,000 cells) were sorted into CELLBANKER (Takara Bio) and stored at −80°C. After thawing, cells were pelleted via centrifugation, and the supernatants removed. Pellets were reconstituted by mixing with 35 μl of transposase mixture (10 mM TAPS-NaOH [pH 8.5], 5 mM MgCl 2 , 10% N,N-dimethylformamide, 0.2 mg/ml digitonin, and 50 µg/ml Tn5 preloaded linker oligo) and incubated with gentle agitation at 30°C for 30 min. The reaction was stopped by adding 0.3% sodium dodecyl sulfate and 15 mM EDTA solution. Tn5-tagged DNA fragments were purified using a Monarch PCR purification column (New England Biolabs) and amplified by KAPA HiFi DNA polymerase using a unique indexing primer pair for each reaction. Indexed libraries were purified using AMPure beads (Beckman Coulter), quantified using quantitative PCR, and then sequenced as described earlier for RNA-seq. Paired-end reads were quality-checked, trimmed using FASTP, and aligned to the mouse reference genome (UCSC mm10) using STAR. Peaks were identified by aggregating the data by each unique cell type, peak summit, and motif using HOMER. To create Venn diagrams, reads were reanalyzed using the UCSC mm10 reference genome and annotations.

Statistical analysis
Data were analyzed using two-tailed Student's t test with or without Welch's correction. P values <0.05 were considered significant. No randomization was used in the animal studies. Other than for genotype, no attempt was made to study deliberately selected mice. Blinding was only used in the histological analysis.
Online supplemental material Fig. S1 relates to Fig. 1 and depicts the genome structure of Tigit fm mice and the gating used for ILC2s in flow cytometric analysis.  Fig. 1 and shows the kinetic analysis of TIGIT, TIGIT-GFP, and PD-1 during papain treatment, ex vivo IL-5 detection in tdTomato + ILC2s, and tdTomato expression in ILC2s from adipose tissues and the small intestine. Fig. S3 relates to Figs. 2 and 3 and shows the lineage marker expression of RNAseq samples and tdTomato + ILC2s in tissues. Table S1 provides a list of differentially expressed genes. Table S2 provides a list of the genes used in Fig. 2 G. Table S3 provides a list of antibodies used for flow cytometry.

Data availability
The presented data sets can be found in the Gene Expression Omnibus under accession no. GSE212180. Figure S1. Expression of tdTomato in Tigit fm mice. (A) Schematic figure depicting the genome structure of Tigit fm mice in which TIGIT + cells can be fatemapped based on tdTomato expression. (B) Frequency of indicated cells among total tdTomato + cells from the lungs of Tigit fm mice treated with papain and tamoxifen for 19 d. Data represent more than three experiments (mean ± SEM of three mice). (C) Representative gating for total ILC2s and tdTomato + ILC2s (CD45 + Lin − CD127 + CD25 + ST2 + ) in flow cytometric analysis. Data represent at least two independent experiments (mean ± SEM of three mice in B).

Yamada et al.
Journal of Experimental Medicine S1 TIGIT mediates AICD of ILC2s during chronic allergy https://doi.org/10.1084/jem.20222005 Figure S2. Phenotypical analysis of tdTomato + ILC2s in the lungs and other tissues. (A-E) Tigit fm mice were treated with papain and tamoxifen on the days indicated in Fig. 1, F and G. (A) Expression of TIGIT-GFP and TIGIT in tdTomato − and tdTomato + lung ILC2s from Tigit fm mice. (B) Representative histograms (left) and quantification (right) of PD-1 expression in tdTomato − and tdTomato + lung ILC2s from Tigit fm mice. (C) Gating for tdTomato − (tdTom. − ), tdTomato low (tdTom. low), and tdTomato high (tdTom. high) ILC2s sorted for ex vivo intracellular IL-5 staining. (D) Representative histograms (left) and quantification (right) of IL-5 production in the indicated lung ILC2s. gMFI, geometric mean fluorescent intensity. *P < 0.05 and **P < 0.01, as determined by unpaired two-tailed t test. Data represent at least two independent experiments (mean ± SEM of three mice in A and D). Provided online are three tables. Table S1 provides a list of differentially expressed genes used in this study. Table S2 provides a list of the genes used in this study. Table S3 provides a list of antibodies used in this study.