ICOS Deficiency Boosts IL-10 in ILC2s

ICOS deficiency induces IL-10 secretion in ILC2s and modulates AHR. This study aimed to determine the effects of ICOS on ILC2 effector function. To this end, we administered IL-33 i.n. to WT and ICOS-KO mice for 3 days, followed by lung digestion and ILC2 isolation via cell sorting on the fourth day (Figure 1A). Lung ILC2s were characterized as CD45+, lineage–, CD127+, and ST2+ cells (Supplemental Figure 1A; supplemental material available online with this article; https://doi.org/10.1172/JCI193134DS1). Isolated ILC2s were then cultured for 24 hours, and the concentrations of various cytokines were measured in the culture supernatants by LEGENDplex (BioLegend). We found that ICOS-KO ILC2s produced significantly less type 2 cytokines IL-5 and IL-13 (Figure 1, B and C, and Supplemental Figure 1, B and C), while the production of IL-10, an anti-inflammatory cytokine, was surprisingly increased compared with WT ILC2s (Figure 1D and Supplemental Figure 1D). Furthermore, similar results were obtained utilizing intracellular staining (Supplemental Figure 1, E–G). To assess the impact of secreted IL-10 on ILC2 function, pure populations of activated lung ILC2s isolated from ICOS-KO mice were cultured with isotype control or anti–IL-10R blocking antibody, the receptor for IL-10, for 24 hours (Figure 1E). We found that IL-10R–blocked ILC2s exhibited a significant upregulation in GATA binding protein 3 (GATA-3) expression, a hallmark activation marker for ILC2s (Figure 1F and Supplemental Figure 2A). This observation was accompanied by an enhancement in the production capacity of IL-5 and IL-13 ex vivo (Figure 1, G and H, and Supplemental Figure 2, B and C). In an ILC2-dependent AHR model induced by Alternaria alternata, treatment with anti–IL-10R antibody (200 μg/mice) resulted in a significant upregulation of GATA-3 expression compared with the isotype control group (Supplemental Figure 2, D and E). This finding was consistent with the results observed in the ex vivo experiments. To investigate the effects of IL-10 produced by ILC2s in vivo on the development of AHR, an ILC2-driven AHR model was utilized. WT and ICOS-KO mice were challenged i.n. with IL-33 for 3 consecutive days, concurrently with i.p. administration of anti–IL-10R antibody (200 μg/mice) or isotype control on the first day of the challenge. On day 4, lung resistance and dynamic compliance were measured, bronchoalveolar lavage (BAL) fluid was collected and analyzed, and lung histology was examined to assess lung inflammation (Figure 2A). A comparison of PBS-treated mice with those challenged with IL-33 revealed worsened lung resistance and dynamic compliance in IL-33–challenged mice in response to increased doses of methacholine, a bronchoconstrictor. We notably found that ICOS-KO mice exhibited markedly diminished pulmonary resistance and higher dynamic compliance in comparison with their WT counterparts, confirming our previous findings (17). Conversely, ICOS-KO mice treated with anti–IL-10R antibody exhibited significantly increased lung resistance (Figure 2B) and lower dynamic compliance (Figure 2C) compared with isotype-treated mice. This finding is corroborated by the observation that the anti–IL-10R antibody–treated group exhibited a substantial increase in lung inflammation, as indicated by elevated CD45+ cells reflecting the total number of immune cells in the BAL fluid (Figure 2D and Supplemental Figure 2F) and eosinophils in the BAL fluid (Figure 2E and Supplemental Figure 2F). Furthermore, IL-5 (Figure 2F) and IL-13 levels (Figure 2G) in the BAL fluid were significantly elevated in the anti–IL-10R antibody–treated group compared with the isotype control group. Histological analysis of lung tissue (Figure 2H) further corroborated these results, with significantly increased epithelial thickness, a common manifestation of remodeling during airway inflammation (Figure 2I), and inflammatory cell counts, a common measurement of the extent of inflammation (Figure 2J), in the anti–IL-10R antibody–treated group compared with the isotype control group. Together, our results suggest that ICOS deficiency promotes IL-10 production in ILC2s and that IL-10 plays a role in regulation of ILC2-induced AHR and lung inflammation in ICOS-KO mice.

ICOS mediates IL-10 production from ILC2s. (A–D) WT and ICOS-KO mice received i.n. doses of rmIL-33 over 3 consecutive days. Lung ILC2s were isolated on day 4 and cultured. (B–D) Levels of IL-5 (B), IL-13 (C), and IL-10 (D) production in the culture supernatant were measured. n = 4. (E–H) Cohorts of ICOS-KO mice were i.n. challenged with rmIL-33 over 3 consecutive days. On day 4, lung ILC2s were isolated and cultured with isotype control (ISO) or anti–IL-10R antibody. (F) Representative plots of GATA-3 expression in ISO and anti–IL-10R groups and corresponding quantitation presented as GATA-3 MFI. n = 4. (G and H) Levels of IL-5 (G) and IL-13 (H) production in the culture supernatant were measured. n = 4. Data are presented as mean ± SEM and are representative of 4 independent experiments. Two-tailed Student’s t test was employed for statistical analysis; *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Schematic images were created in Adobe Illustrator. FMO, fluorescence minus one.

IL-10 affects the AHR. (A–J) WT mice received i.p. injections of isotype control or anti–IL-10R antibody (200 μg) on day 1 and were i.n. exposed to 0.5 μg of rmIL-33 or PBS for 3 days. Lung function and inflammation were assessed on the day 4. (B and C) Lung resistance (B) and dynamic compliance (C) in response to elevating doses of methacholine. n = 4. (D and E) The total number of CD45+ cells (D) and CD45+, Gr-1–, CD11c–, and SiglecF+ eosinophils (E) in BAL fluid are demonstrated in bar graphs. n = 4. (F and G) Levels of IL-5 (F) and IL-13 (G) in the BAL fluid are shown in bar graphs. n = 4. (H) Lung histologic sections stained with H&E are presented. Scale bars: 50 μm. (I and J) Quantification of airway epithelium thickness (I) and infiltrating cells (J). n = 4. Data are presented as mean ± SD or SEM and are representative of 3 independent experiments. Two-tailed Student’s t test or 1-way ANOVA followed by Tukey’s post hoc tests was employed for statistical analysis; *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Schematic images were created in Adobe Illustrator.

ICOS signaling limits IL-10 production by regulating MAF and NFIL3 expression in ILC2s. We next isolated activated ILC2s from the lungs of WT and ICOS-KO mice and performed a transcriptomic analysis to investigate how ICOS regulates IL-10 production in these cells. We first isolated pulmonary active ILC2s (aILC2s) from WT and ICOS-KO mice subjected to i.n. IL-33 challenge for 3 days. Subsequently, we performed RNA sequencing (RNA-Seq) on ILC2s following 18-hour culture (Figure 3A). We found 809 differentially expressed genes between WT and ICOS-KO (337 downregulated genes and 472 upregulated genes) (Figure 3B). Furthermore, the expression of Il22 and Il24, which are members of the IL-10 superfamily (26) as well as Maf and Nfil3, known to induce IL-10 (12, 27), were also significantly increased (Figure 3B). In support of our transcriptomic analysis, we confirmed the expression and upregulation of MAF and NFIL3 in ICOS-KO ILC2s compared with WT ILC2s at the protein level (Figure 3, C and D). To further support these findings, analysis using the Alternaria stimulation model revealed a significant increase in the expression of MAF and NFIL3 in ICOS-KO ILC2s (Supplemental Figure 3, A–C). We next investigated the role of ICOS, MAF, and/or NFIL3 in the regulation of IL-10 production. Activated lung ILC2s were isolated from ICOS-KO mice and cultured with Maf siRNA, Nfil3 siRNA, or Scramble siRNA for 48 hours (Figure 3E). As expected, the expression of MAF in ILC2s cultured with Maf siRNA (Supplemental Figure 3D) and NFIL3 in ILC2s cultured with Nfil3 siRNA (Supplemental Figure 3E) was significantly reduced compared with Scramble siRNA. Interestingly, however, ILC2s with reduced MAF and NFIL3 expression exhibited significantly lower IL-10 levels in the culture medium compared with Scramble siRNA (Figure 3F). To confirm the effects of ICOS on MAF, NFIL3, and IL-10 production, we next cultured pure populations of activated WT ILC2s with an anti-ICOS antibody or isotype control (Figure 3G). In confirmation of our findings using ICOS-KO mice, the levels of IL-10 in culture supernatants were significantly increased in WT ILC2s treated with an anti-ICOS antibody (Figure 3H). Similarly, the expression of MAF (Figure 3I) and NFIL3 (Figure 3J) was also increased in ILC2s treated with anti-ICOS antibody compared with controls. Consistent with these findings, analysis of the Alternaria stimulation model demonstrated a significant upregulation of MAF and NFIL3 expression in ILC2s following anti-ICOS antibody treatment (Supplemental Figure 3, F–H). Finally, we validated these findings in vivo using a mouse model of IL33-mediated airway inflammation. Cohorts of WT mice were challenged with IL-33 i.n. on days 1–3, concurrently with anti-ICOS antibody (500 μg/mouse) or an isotype control i.p. on day 1 (Figure 3K). Consistent with our previous findings, the frequency of IL-10+ ILC2s in lung ILC2s was significantly elevated in the anti-ICOS antibody–treated group compared with controls (Figure 3L). Furthermore, the expression of MAF (Figure 3M) and NFIL3 (Figure 3N) in ILC2s was also elevated in anti-ICOS antibody–treated mice compared with controls. To support these ex vivo and in vivo findings and our previous report (17), we employed an ILC2-induced AHR model with anti-ICOS antibody and observed that mice treated with anti-ICOS antibody exhibited significantly improved AHR compared with controls (Supplemental Figure 4, A–K). Together, these results suggest that ICOS controls IL-10 production by regulating MAF and NFIL3 expression.

NFIL3 and MAF regulate IL-10 production via ICOS in ILC2s. (A and B) Pulmonary aILC2s were isolated from WT and ICOS-KO mice subjected to i.n. IL-33 challenges for 3 days. Subsequently, RNA-Seq was performed on ILC2s following 18-hour culture. (B) Total RNA from WT and ICOS-KO mice was extracted to perform a bulk transcriptomic analysis. Volcano plots represent differentially expressed genes. (C and D) Representative plots of MAF (C) and NFIL3 (D) expression levels in WT and ICOS-KO ILC2s are shown. Corresponding quantitation is presented as MFI. n = 4. (E and F) aILC2s from ICOS-KO mice were cultured with or without Maf siRNA or Nfil3 siRNA. (F) Levels of IL-10 in the culture supernatant were measured. n = 3. (G–J) WT aILC2s were cultured with isotype control or anti-ICOS antibody. (H) Levels of IL-10 in the culture supernatant were measured. n = 4. (I and J) Representative plots of MAF (I) and NFIL3 (J) expression levels in each group are shown. Corresponding quantitation is presented as MFI. n = 4. (K–N) WT mice received i.p. injections of 500 μg anti-ICOS antibody or isotype control (ISO) on day 1 and were i.n. exposed to 0.5 μg of rmIL-33 or PBS for 3 days. On day 4, mice were euthanized. (L) Frequency of IL-10+ ILC2s in each group. n = 4. (M and N) Representative plots of MAF (M) and NFIL3 (N) expression levels in each group are shown. Corresponding quantitation is presented as MFI. n = 4. Data are presented as mean ± SEM and are representative of 4 independent experiments. Two-tailed Student’s t test or 1-way ANOVA followed by Tukey’s post hoc tests was employed for statistical analysis; *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Schematic images were created in Adobe Illustrator. FMO, fluorescence minus one.

ICOS regulates IL-10 production via cholesterol biosynthesis. Recent reports have indicated that the secretion of cytokines by ILC2s is subject to intracellular metabolic reprogramming (7, 9, 28). In our transcriptomic analysis, an Ingenuity Pathway Analysis (IPA) revealed alterations in gene sets associated with cholesterol biosynthetic process, steroid hormone biosynthesis, and cortisol biosynthesis (Figure 4A). We found that the RNA expression of Srebf2, a crucial gene in cholesterol biosynthesis (29), is significantly upregulated in ICOS-KO ILC2s (Figure 4B). In confirmation of these observations, we found that the protein expression of SREBP2, which is encoded Srebf2, was significantly increased in ICOS-KO ILC2s compared with WT ILC2s (Figure 4C). In addition, the expression levels of a specific set of genes related to cholesterol biosynthesis, which are located downstream of the cholesterol biosynthesis process, were found to be significantly elevated in ICOS-KO ILC2s in comparison with their WT counterparts (Figure 4D). To further investigate the hypothesis that SREBP2 regulates IL-10 production in ICOS-deficient ILC2s, we treated purified populations of activated lung ICOS-KO ILC2s with either vehicle or Fatostatin (30, 31), an SREBP2 inhibitor, for 24 hours (Figure 4E). Treatment with the SREBP2 inhibitor resulted in a significant reduction in IL-10 production compared with the vehicle group (Figure 4F). Similarly, expression of the key transcription factors MAF (Figure 4G) and NFIL3 (Figure 4H) was also decreased following SREBP2 inhibition. Importantly, annexin V staining confirmed that the SREBP2 inhibitor treatment did not induce toxicity (Supplemental Figure 5A). These findings suggest that ICOS may influence cholesterol biosynthesis and IL-10 production in ILC2s by modulating SREBP2 activity.

ICOS is related with cholesterol biosynthesis. (A) Chord plot representing the differentially expressed genes from the most enriched metabolic pathways. Specific pathways are color-coded and represented in the right inner bands, where chords gather. Outer bands on the right depict the IPA −log10 (P value). The left inner bands represent the gene −log10 (P value). Outer bands on the left represent the gene log2(fold change). (B) The bar graphs represent the normalized counts of Srebf2. n = 3. (C) Representative histogram of protein expression of SREBP2. Corresponding quantitation is presented as MFI. n = 4. (D) Dot plot representing selected critical genes involved in cholesterol synthesis. Dot size is indicative of the total gene expression level. Gray histograms represent −log10 (P value), and the dotted line represents P < 0.05. (E–H) ICOS-KO aILC2s were cultured with vehicle or SREBP2 inhibitor. (F) Levels of IL-10 in the culture supernatant were measured. n = 4. (G and H) Representative plots of MAF (G) and NFIL3 (H) expression levels in each group are shown. Corresponding quantitation is presented as MFI. n = 4. Data are presented as mean ± SEM and are representative of 4 independent experiments. Two-tailed Student’s t test or 1-way ANOVA followed by Tukey’s post hoc tests was employed for statistical analysis; *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Schematic images were created in Adobe Illustrator. FMO, fluorescence minus one.

Next, we conducted a comparative analysis of intracellular cholesterol levels to measure the impact of ICOS on cholesterol biosynthesis and usage at the protein level. Activated lung ILC2s were isolated from WT and ICOS-KO mice and cultured for 24 hours. Subsequently, we used Filipin probe to measure the level of intracellular cholesterol, 22-(N-(7-Nitrobenz-2-Oxa-1,3-Diazol-4-yl)Amino)-23,24-Bisnor-5-Cholen-3β-Ol (NBD) cholesterol to measure cholesterol uptake, and BODIPY542/563 probes to measure cholesterol usage by flow cytometry (Figure 5A). We observed that the amount of intracellular cholesterol in ICOS-KO ILC2s was diminished in comparison with WT ILC2s, indicating enhanced cholesterol utilization in ICOS-KO ILC2s (Figure 5B). Subsequently, both cholesterol uptake (Figure 5C) and usage (Figure 5D) were upregulated in ICOS-KO ILC2s. These results suggest that ICOS-deficient ILC2s have decreased intracellular cholesterol levels due to increased cholesterol utilization, as indicated by increased NBD-labeled cholesterol levels and changes in BODIPY542/563 probes. We next investigated the impact of this higher cholesterol demand in ICOS-KO ILC2s on the production of IL-10 ex vivo using U18666A, a cholesterol transport inhibitor, and statin, a cholesterol synthesis inhibitor. Pure populations of activated lung WT and ICOS-KO ILC2s were cultured with U18666A or statin for 24 hours (Figure 5E). Remarkably, we found that both U18666A and statin reduced IL-10 production in ICOS-KO ILC2s compared with controls (Figure 5F). Similarly, both reagents reduced the expression of MAF (Figure 5G) and NFIL3 (Figure 5H) compared with controls. Of note, the absence of cellular toxicity for U18666A and statin was confirmed by an annexin V assay (Supplemental Figure 5B). Together, these results indicate that the demand for cholesterol is elevated in ICOS-KO and that suppressing cholesterol usage and biosynthesis modulates MAF and NFIL3 expression and ultimately decreases IL-10 production.

ICOS controls cholesterol biosynthesis in ILC2s for IL-10 production. (A) Pulmonary aILC2s were isolated from WT and ICOS-KO mice subjected to i.n. IL-33 challenge for 3 days and cultured. Subsequently, aILC2s were analyzed using Filipin, NBD cholesterol, or BODIPY542/563. (B) Representative histogram of cholesterol quantity assessed with Filipin fluorescent tracer. Corresponding quantitation is presented as MFI. n = 4. (C) Representative histogram of NBD cholesterol uptake assessed with a cholesterol uptake kit. Corresponding quantitation is presented as MFI. n = 4. (D) Representative histogram of lipid utilization assessed with BODIPY542/563. Corresponding percent change in MFI before and after incubation is presented. n = 4. (E–H) WT and ICOS-KO aILC2s were cultured with or without U18666A or statin. (F) Levels of IL-10 in the culture supernatant were measured. n = 4. (G and H) Representative plots of MAF (G) and NFIL3 (H) expression levels in each group are shown. Corresponding quantitation is presented as MFI. n = 4. Data are presented as mean ± SEM and are representative of 4 independent experiments. Two-tailed Student’s t test or 1-way ANOVA followed by Tukey’s post hoc tests was employed for statistical analysis; *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Schematic images were created in Adobe Illustrator. FMO, fluorescence minus one.

Cortisol biosynthesis induces IL-10 production in ICOS-deficient ILC2s. Cholesterol serves as the essential precursor molecule for the biosynthesis of cortisol through a series of enzymatic reactions (32). In particular, the analysis of the transcriptome revealed that, among the genes related to steroid hormone biosynthesis, the Cyp11a1 gene, which plays an important role in cortisol biosynthesis (33), is increased in ILC2 lacking ICOS (Figure 6A). In addition, the results of the IPA depicted in Figure 6A demonstrated that the gene set involved in cortisol biosynthetic process was overexpressed in ICOS-KO ILC2s. Consequently, we examined the gene expression of enzymes associated with cholesterol biosynthesis and observed that the genes detected were also overexpressed (Figure 6B). This finding was corroborated by the elevated Cyp11a1 expression levels in ICOS-KO ILC2s at both the RNA and protein levels when compared with WT ILC2s (Figure 6, C and D). Furthermore, the expression of CYP11A1 was significantly upregulated in ICOS-KO ILC2s compared with WT ILC2s in the Alternaria stimulation model (Supplemental Figure 5, C and D). Based on these findings, we hypothesized that the biosynthesis of cholesterol is enhanced in ICOS-KO ILC2s and that cholesterol is used to biosynthesize cortisol. To investigate this hypothesis, we conducted an experiment to examine whether ICOS-KO or WT ILC2s produce cortisol using cholesterol inhibitors or cholesterol transport inhibitors. Pure populations of activated WT and ICOS-KO ILC2s were treated with or without U18666A or statin, and the cortisol level was measured in the culture supernatant by ELISA (Figure 6E). Strikingly, ICOS-KO ILC2s produced significantly more cortisol compared with controls, whereas U18666A and statin led to a decrease in cortisol production (Figure 6F). Subsequently, to ascertain the impact of cortisol on ILC2s, we examined the expression of GR, a receptor for cortisol, in ILC2s. This investigation revealed that GR is expressed in ILC2s, with significantly higher expression in ICOS-KO ILC2s compared with WT ILC2s (Figure 6G). Similar results were observed in the Alternaria stimulation model (Supplemental Figure 5, C and E). We next investigated the effects of cortisol on IL-10 production in ILC2s and exposed activated ICOS-KO ILC2s with cortisol ex vivo for 24 hours (Figure 6H). We found that cortisol induced IL-10 production in ILC2s (Figure 6I), accompanied by increased intranuclear MAF and NFIL3 expressions (Figure 6, J and K). To further validate this observation, ILC2ΔGR mice, which lack GR specifically in ILC2s, were generated by crossing Nmur1Cre+/– mice (ILC2WT mice) and GRfl/fl mice. We confirmed NMUR1 is specifically expressed in ILC2s in the lungs and not expressed on other pulmonary immune cells such as T cells or eosinophils in our context (Supplemental Figure 6A). To investigate the role of GR in ILC2 development and homeostasis, we quantified naive ILC and ILC2 populations under steady-state conditions and following IL-33–induced activation using ILC2WT and ILC2ΔGR mice. At steady state, no significant differences were observed in total ILC numbers (lineage–IL-7R+ cells), IL-7R expression on ILCs, or ILC2 frequencies between the 2 groups. In contrast, IL-33 stimulation resulted in a significantly greater expansion of ILC2s in ILC2ΔGR mice compared with ILC2WT mice, accompanied by increased ILC2 number and elevated IL-7R expression. These findings indicate that GR deficiency does not impair ILC2 development but enhances ILC2 activation in response to IL-33 (Supplemental Figure 6, B–D). Activated ILC2s were sorted from the lungs of ILC2WT and ILC2ΔGR mice and cultured with an anti-ICOS antibody and isotype control (Figure 6L). As expected, the frequency of IL-10 producing ILC2s was increased in WT ILC2s incubated with anti-ICOS. Remarkably however, IL-10 production drastically decreased in ILC2s isolated from ILC2ΔGR mice compared with controls, while no effects of anti-ICOS on IL-10 production were observed in these mice (Figure 6M). Correspondingly, MAF and NFIL3 intranuclear expressions also showed similar changes (Figure 6, N and O). Consistent with previous findings, GR-deficient ILC2s in the Alternaria stimulation model exhibited a similar downregulation of MAF and NFIL3 expression (Supplemental Figure 6, E–G). These observations were further confirmed using Mifepristone, a GR inhibitor (34) (Supplemental Figure 7, A–D). Together, these results suggest that cortisol promotes IL-10 production in ILC2s via GR expression located downstream of ICOS.

Cortisol is a key regulator of ICOS-mediated IL-10 production in ILC2s. (A and B) Total RNA from WT and ICOS-KO mice was extracted to perform a bulk transcriptomic analysis. (A) Volcano plots represent differentially expressed genes involved in steroid biosynthesis. (B) Biosynthetic pathway of cortisol from cholesterol showing intermediate metabolites and key enzymes involved in each step. Changes in enzyme expression are indicated as log2(fold change) values in colored boxes. (C) The bar graph represents the normalized counts of Cyp11a1. n = 3. (D) Representative histogram of protein expression of CYP11A1. Corresponding quantitation is presented as MFI. n = 4. (E) WT and ICOS-KO aILC2s were isolated and cultured with or without U18666A or statin. (F) Levels of cortisol in the culture supernatant were measured by ELISA. n = 3. (G) Representative histogram of protein expression of GR. Corresponding quantitation is presented as MFI. n = 4. (H–K) ICOS-KO aILC2s were cultured with or without cortisol. (I) Levels of IL-10 in the culture supernatant were measured. n = 4. (J and K) Representative plots of MAF (J) and NFIL3 (K) expression levels in each group are shown. Corresponding quantitation is presented as MFI. n = 4. (L–O) Sorted aILC2s from NMUR1cre (ILC2WT mice) and NMUR1creGRfl/fl mice (ILC2ΔGR mice) were cultured with isotype control or anti-ICOS antibody. (M) Bar graph representing the frequency of IL-10+ ILC2s in each group. n = 4. (N and O) Representative plots of MAF (N) and NFIL3 (O) expression levels in each group are shown. Corresponding quantitation is presented as MFI. n = 4. Data are presented as mean ± SEM and are representative of 4 independent experiments. Two-tailed Student’s t test was employed for statistical analysis; *P < 0.05, **P < 0.01, and ***P < 0.001, and ****P < 0.0001. Schematic images were created in Adobe Illustrator. FMO, fluorescence minus one.

ILC2 conditional deletion of GR exacerbates the development of ILC2-dependent AHR. We next sought to investigate the role of GR and IL-10 production in the development of ILC2-driven AHR. ILC2WT and ILC2ΔGR mice were administered i.n. with IL-33 or PBS for 3 consecutive days. On the fourth day, lung resistance and dynamic compliance were measured by noninvasive plethysmography, followed by the analysis of BAL cellularity by flow cytometry and lung histology (Figure 7A). We found that ILC2ΔGR mice exhibited significantly higher pulmonary resistance compared with ILC2WT mice (Figure 7B), associated with a decrease dynamic compliance (Figure 7C). In line with our previous findings, the percentage of IL-10+ ILC2s isolated from the lungs of ILC2ΔGR mice was significantly decreased compared with ILC2WT mice (Figure 7D). The frequency of IL-5+ IL-13+ ILC2 was then assessed, and no statistically significant differences were found between the 2 groups (Supplemental Figure 8A). This observation was associated with increased inflammation, as indicated by the higher numbers of CD45+ cells, notably eosinophils (Figure 7, E and F), and the higher levels of IL-5 and IL-13 (Figure 7, G and H) in the BAL fluid of ILC2ΔGR mice. Histological analysis of lung tissue (Supplemental Figure 8B) corroborated these results, demonstrating that IL-33 challenges led to a substantial increase in epithelial thickness (Supplemental Figure 8C) and inflammatory cell count (Supplemental Figure 8D) in ILC2ΔGR mice compared with ILC2WT mice. Together, these findings support the notion that GR in ILC2s controls the magnitude of AHR, independently of other immune cells.

ILC2-specific GR deletion or inhibition in ICOS-KO mice exacerbates the ILC2-induced AHR. (A–H) NMUR1cre (ILC2WT mice) and NMUR1creGRfl/fl mice (ILC2ΔGR mice) were i.n. exposed to 0.5 μg of rmIL-33 or PBS for 3 days. On day 4, lung function, frequency of IL-10+ ILC2s in the lung, BAL cellularity, cytokine levels, and lung histology were analyzed. (B and C) Lung resistance (B) and dynamic compliance (C) in response to elevating doses of methacholine. n = 5. (D–F) The frequency of IL-10+ ILC2s in the lung (D), the total number of CD45+ cells (E), and CD45+, Gr-1–, CD11c–, and SiglecF+ eosinophils (F) in BAL fluid are demonstrated in bar graphs. n = 5. (G and H) Levels of IL-5 (G) and IL-13 (H) in the BAL fluid are shown in bar graphs. n = 5. (I–P) ICOS-KO mice were i.n. exposed to 0.5 μg of rmIL-33 or PBS with or without GR inhibitor (0.01nM) (anti-GR) for 3 days. On day 4, lung function, frequency of IL-10+ ILC2s in the lung, BAL cellularity, cytokine levels, and histology were analyzed. (J and K) Lung resistance (J) and dynamic compliance (K) in response to elevating doses of methacholine. n = 5. (L–N) The frequency of IL-10+ ILC2s in the lung (L), total number of CD45+ cells (M), and CD45+, Gr-1–, CD11c–, and SiglecF+ eosinophils (N) in BAL fluid are demonstrated in bar graphs. n = 5. (O and P) Levels of IL-5 (O) and IL-13 (P) in the BAL fluid are shown in bar graphs. n = 5. Data are presented as mean ± SD or SEM and are representative of 3 independent experiments. Two-tailed Student’s t test or 1-way ANOVA followed by Tukey’s post hoc tests was employed for statistical analysis; *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Schematic images were created in Adobe Illustrator.

GR depletion in ICOS-KO mice exacerbates ILC2-dependent AHR. To find a possible link between ICOS and GR effects on ILC2s, we conducted experiments in ICOS-KO mice using GR inhibitor (34). ICOS-KO mice were challenged i.n. with IL-33 or PBS, in the presence or absence of GR inhibitor (0.01 nM) (anti-GR) on days 1–3 (Figure 7I), and on day 4, lung resistance and dynamic compliance were directly measured, followed by the analysis of BAL cellularity by flow cytometry and lung histology. We found that lung resistance was significantly higher in mice treated with GR inhibitor compared with controls (Figure 7J), associated with the worst dynamic compliance (Figure 7K). In confirmation of our previous findings, blocking GR in ICOS-KO mice reduced the frequency of IL-10–producing ILC2s in the lungs (Figure 7L), associated with an increase in inflammation, as evidenced by the number of CD45+ cells (Figure 7M) and eosinophils (Figure 7N) as well as IL-5 (Figure 7O) and IL-13 (Figure 7P) levels in the BAL fluid. Histological analysis of lung tissue (Supplemental Figure 8E) corroborated these results, demonstrating that epithelial thickness (Supplemental Figure 8F) and inflammatory cell count (Supplemental Figure 8G) were higher in mice treated with GR inhibitor compared with vehicle mice. These results suggest that the inhibition of GR in ICOS-KO may promote AHR exacerbation in the presence of IL-33, a conclusion further supported by experiments using A. alternata, a common fungus associated with allergic disease (Supplemental Figure 9, A–K). These findings suggest that ICOS exerts a regulatory effect on IL-10 production by suppressing the activity of GR in ILC2, thereby modulating AHR.

The ICOS/GR pathway regulates IL-10 production and effector function in hILC2s. We then sought to determine whether the observations from the rodent studies were applicable to hILC2s. We isolated pure populations of hILC2s from PBMCs of 6 healthy individuals as CD45+, lineage–, CD127+, and CRTH2+ cells using flow cytometry (Figure 8A and Supplemental Figure 10). We previously reported that ICOS and ICOSL are expressed in both murine and hILC2s (17). We then cultured hILC2s isolated from each healthy subject with or without anti-human ICOSL antibody (anti-ICOSL) for blocking the ICOS–ICOSL interaction and examined the effects of ICOSL blockade on ILC2 activation, proliferation, and functional indices (Figure 8A). The administration of the anti-ICOSL led to a significant increase in IL-10 levels in the culture supernatant of all 6 healthy subjects. In addition, consistent with prior reports, hILC2s treated with anti-ICOSL showed a decrease in the levels of effector cytokines, including IL-4, IL-5, IL-6, and IL-13, as well as the intranuclear proteins Ki67, a measure of hILC2 proliferation, and GATA-3, a hallmark activation marker for hILC2s (Figure 8, B–H). We next sought to elucidate the role of ICOS, cholesterol, and cortisol on IL-10 production using anti-ICOSL, statin, or GR inhibitor (anti-GR) in hILC2 cultures. The expression of MAF (Figure 8I) and NFIL3 (Figure 8J) in hILC2 was increased by anti-ICOSL treatment, aligning with our observations from murine ILC2s. Furthermore, the genes associated with cortisol biosynthesis–related enzymes, as illustrated in Figure 6B, included the following: CYP11A, which catalyzes the conversion of cholesterol to pregnenolone; CYP17A1, which converts pregnenolone to 17α-hydroxy-pregnenolone; CYP21A2, which converts 17α-hydroxy-pregnenolone to 17α,21-dihydroxy-pregnenolone; HSD3B2, which converts 17α,21-dihydroxy-pregnenolone to 11-deoxycortisol; and CYP11B, which facilitates the transformation of 11-deoxycortisol to cortisol (Figure 9A). These enzymes were found to be significantly upregulated by anti-ICOSL treatment (Figure 9A). In addition, and consistent with our observations in murine models, the administration of GR inhibitor or statins led to a substantial decrease in IL-10 production (Figure 9B) and MAF (Figure 9C) and NFIL3 (Figure 9D) expression.

ICOS regulates IL-10 production in hILC2s. (A–H) hILC2s (CD45+, lineage–, CRTH2+, and CD127+) were freshly isolated from PBMCs of healthy donors and cultured with or without anti-ICOS ligand (anti-ICOSL). Right panel shows hILC2 purity after being sorted. (B–F) Levels of IL-10 (B), IL-4 (C), IL-5 (D), IL-6 (E), and IL-13 (F) in the culture supernatants following treatment with or without anti-ICOSL. n = 6. (G and H) The expression levels of intranuclear Ki67 (G) and GATA-3 (H) expression is presented as MFI. n = 6. (I and J) Representative histogram of MAF (I) and NFIL3 (J) protein expression levels. Corresponding quantitation is presented as MFI. n = 4. Data are presented as mean ± SEM. Two-tailed Student’s t test was employed for statistical analysis; *P < 0.05, **P < 0.01, and ***P < 0.001. Schematic images were created in Adobe Illustrator. FMO, fluorescence minus one.

ICOS regulates IL-10 production in hILC2s via controlling cholesterol and cortisol biosynthesis. (A) qRT-PCR results show CYP11A1, CYP17A1, CYP21A2, HSD3B2, and CYP11B1 expression in each group. n = 3. (B) Levels of IL-10 in the culture supernatants following treatment with or without GR inhibitor (anti-GR) or statin. n = 4. (C and D) Representative histogram plots of intranuclear MAF (C) and NFIL3 (D) expression levels and corresponding quantitation presented as MFI. n = 4. Data presented as mean ± SEM. Two-tailed Student’s t test or 1-way ANOVA followed by Tukey’s post hoc tests was employed for statistical analysis; *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Schematic images were created in by Adobe Illustrator. FMO, fluorescence minus one.

Taken together, these results align with those previously observed in murine models, suggesting a conserved mechanism through which ICOS regulates IL-10 production in hILC2s by modulating cholesterol and cortisol biosynthesis.

The central finding of this study is that in pulmonary ILC2s, ICOS negatively controls the production of the anti-inflammatory cytokine IL-10 by regulating cholesterol and cortisol biosynthesis. We found that the anti-inflammatory cytokine IL-10 was upregulated in the absence of ICOS in ILC2s ex vivo, as we used an anti–IL-10R antibody to specifically demonstrate the immunomodulatory role of IL-10 on the function of ILC2s and development of AHR in multiple mouse models. Remarkably, in the absence of ICOS signaling, ILC2s accumulated increased amounts of cholesterol ex vivo, as measured by Filipin and NBD cholesterol assays. A combination of transcriptomic and protein analysis further showed that ILC2s lacking ICOS increased CYP11A1, an enzyme involved in the production of cortisol from cholesterol, which led to higher amounts of intracellular cortisol in ILC2s and an increase in cortisol receptor GR expression, an effect neutralized by statins. We further found that MAF and NFIL3, 2 transcription factors known to positively control IL-10 production, were similarly upregulated in the absence of ICOS in ILC2s. To confirm our findings in an in vivo setting, we generated mice with an ILC2-specific deletion of GR. We found that ILC2s lacking GR were unable to induce IL-10 upon ICOS inhibition, associated with the failure to upregulate MAF and NFIL3. In support of these findings, mice lacking ICOS and treated with an anti-GR antibody showed similarly increased development of AHR compared with controls, which was associated with lower IL-10 production by ILC2s in the lungs in multiple mouse models. To the best of our knowledge, this is the first report showing that ICOS regulates IL-10 production by controlling steroid metabolism in ILC2s.

ICOS is a costimulatory molecule expressed primarily on immune cells, particularly Tregs (35). ICOS molecules possess intracellular signals that modulate immune cell function, and the induction of ICOS signaling can have distinct effects depending on the immune cells and the context. For instance, in Th2 cells, ICOS enhances allergic responses by promoting the production of Th2 cytokines (15). In contrast, in Tregs, ICOS has been shown to promote the production of IL-10, an anti-inflammatory cytokine, thereby facilitating immune tolerance (14). Notably, recent work by O’Brien et al. further illustrated this context specificity, demonstrating that ICOS can differentially regulate IL-10 production in T cells depending on the immune environment, such as during chronic infection, highlighting its intricate role in modulating immune responses (36). Collectively, these observations underscore the critical and nuanced regulatory functions of ICOS in orchestrating immune allergic responses within the immune system. In our previous report, we demonstrated the presence of ICOS and ICOSL in ILC2s, and the interaction between ICOS and ICOSL led to the production of Th2 cytokines such as IL-5 and IL-13. In this study, we demonstrated that the suppression of ICOS in pulmonary ILC2s resulted in the secretion of IL-10, a cytokine that is recognized for its anti-inflammatory properties. This finding is further corroborated by reports of reduced ICOS expression in T cells and enhanced IL-10 production in chronic inflammation (37, 38).