ILC2s regulate adaptive Th2 cell functions via PD-L1 checkpoint control

Abstract

Group 2 innate lymphoid cells (ILC2s) are important effector cells driving the initiation of type 2 immune responses leading to adaptive T helper 2 (Th2) immunity. Here we show that ILC2s dynamically express the checkpoint inhibitor molecule PD-L1 during type 2 pulmonary responses. Surprisingly, PD-L1:PD-1 interaction between ILC2s and CD4⁺ T cells did not inhibit the T cell response, but PD-L1-expressing ILC2s stimulated increased expression of GATA3 and production of IL-13 by Th2 cells both in vitro and in vivo. Conditional deletion of PD-L1 on ILC2s impaired early Th2 polarization and cytokine production, leading to delayed worm expulsion during infection with the gastrointestinal helminth Nippostrongylus brasiliensis Our results identify a novel PD-L1-controlled mechanism for type 2 polarization, with ILC2s mediating an innate checkpoint to control adaptive T helper responses, which has important implications for the treatment of type 2 inflammation.

Figure 1.

PD-L1 is expressed on ILC2s and is involved in the immune response against N. brasiliensis. (a) Lung lin⁻CD45⁺CD90⁺ST2⁺KLRG1⁺ ILC2 (for gating strategy, see Fig. S1) were analyzed for surface expression of MHC II and costimulatory molecules before (blue line) and 5 d after infection with N. brasiliensis (Nb, red line) and compared with fluorescence minus one (FMO; tinted gray). (b) Dot plots show expression of PD-L1–expressing lung ILC2s in WT C57BL/6 before and after infection with N. brasiliensis in comparison with FMO control and PD-L1^−/− control. (c) Graphs depict PD-L1 expression on lung ILC2s and PD-1–expressing lung CD4⁺ T cells on indicated days after N. brasiliensis infection in individual C57BL/6 (closed circles) and PD-L1^−/− (open circles) mice. Mean ± SEM from three experiments is depicted. (d) Bar graph shows fold up-regulation of the mean fluorescence intensity of PD-L1 on lung ILC2s and DCs from N. brasiliensis–infected mice in relation to untreated controls. Bar graph shows the mean + SEM of four mice per group from two experiments. (e) Intestinal worm burden in C57BL/6 (black bar) and PD-L1^−/− (open bar) mice 7 d after N. brasiliensis infection. Bar graphs show the mean + SEM of nine mice per group from three independent experiments. (f) Frequency of IL-4–, IL-5–, and IL-13–expressing ILC2s in the lung and mesenteric lymph nodes of naive (blue) and Nb-infected (red, day 5) WT (left) and PD-L1^−/− (right) mice. Bar graphs show the mean + SEM from six mice per group from two experiments. (g) Periodic acid–Schiff staining of fixed small intestinal tissue 5 d after N. brasiliensis infection. Bar graph shows the mean + SD of four mice, with 5 crypt-villus units counted per mouse. Bars, 1 mm. (h) PD-L1 expression on lung ILC2s in naive and day 5 N. brasiliensis–infected Rag1^−/− and PD-L1^−/−/Rag1^−/− mice. Bar graphs show the mean + SEM from four to six mice per group from two experiments. (i) PD-L1 expression on lung ILC2s in naive and day 5 N. brasiliensis–infected C57BL/6, MHC2^−/−, and PD-L1^−/− mice. Bar graphs show the mean + SD from three mice per group. (j) Small intestines from C57BL/6 mice were digested before and 5 d after infection with N. brasiliensis. ILC1s, ILC2s, and ILC3s were analyzed for PD-L1 expression. Bar graphs show the mean + SD from three individual mice per group. Student’s t test was used to analyze data, and p-values <0.05 were considered statistically significant. *, P < 0.05; ns, not significant.

Figure 2.

PD-L1 is up-regulated on IL-33–activated ILC2s in vitro and in vivo. (a) ILC2s were sort-purified from lungs of C57BL/6 mice and cultured in cRPMI in the presence of indicated cytokines (10 ng/ml each) for 5 d. Bar graphs show the mean + SEM of five replicates per group from two independent experiments. Histogram overlay shows representative PD-L1 expression on ILC2s. (b) C57BL/6 mice were injected i.p. with PBS, IL-25 (500 µg/dose), or IL-33 (500 µg/dose) on three consecutive days, and lung ILC2s were analyzed for PD-L1 expression the next day. Bars show the mean + SD from three mice per group from one of two independent experiments. Histogram overlay shows representative PD-L1 expression on ILC2s. (c) WT, Il17br^−/−, and Il1rl1^−/− mice were infected with N. brasiliensis and analyzed for PD-L1–expressing lung ILC2s 5 d later. Bars show the mean + SEM from 5–10 mice per group from two independent experiments. Histogram overlay shows representative PD-L1 expression on ILC2s. (d) Rag1^−/− mice were injected with IL-2:anti–IL-2 complexes three times every other day, and lung ILC2s were analyzed for PD-L1 expression 2 d later. Bars show the mean + SD from at least two individual mice per group. Histogram overlay shows representative PD-L1 expression on ILC2s. (e) ILC2s were isolated from lungs of naive WT and PD-L1^−/− mice, expanded in vitro for 5 d (IL-2, IL-7, and IL-33; 10 ng/ml each), stimulated for the final 5 h with PMA/ionomycin in the presence of monensin, and stained intracellularly for IL-4, IL-5, and IL-13. Plots are gated on live lin⁻CD90⁺KLRG1⁺ST2⁺ cells. FMO, fluorescence minus one. (f) Bar graph shows the mean + SEM IL-4, IL-5, and IL-13 producers within the PD-L1^– and PD-L1⁺ population within the ILC2s from lungs of nine N. brasiliensis–infected WT mice from two experiments. Contour plots show representative gating of PD-L1^– and PD-L1⁺ cells and flow cytometric signal for intracellular IL-4, IL-5, and IL-13. Student’s t test was used to analyze data, and p-values <0.05 were considered statistically significant. *, P < 0.05; **, P < 0.01; ns, not significant.

Figure 3.

PD-L1:PD-1 interaction promotes Th2 polarization in CD4 T cells in vitro. (a) Sort-purified ILC2s from the lungs of day 5 N. brasiliensis–infected WT (PD-L1 expression: see Fig. 1 c, day 5, top) and PD-L1^−/− mice and naive WT CD3⁺CD4⁺CD44⁻ T cells (PD-1 expression: see Fig. 1 c, day 0, bottom) were co-cultured (or T cells only) for 5 d. GATA3 expression in T cells was analyzed by intranuclear staining. Cells were restimulated with PMA/ionomycin, and IL-13 was stained intracellularly and analyzed by flow cytometry. Bar graphs show the mean + SEM of 13 and four to eight independent replicates from three experiments, respectively. (b) Purified naive, CellTrace Violet–labeled OT-2 T cells were cultured for 5 d in the presence of IL-2 (20 ng/ml), anti-CD3/anti-CD28 (5 µg/ml/1 µg/ml; “pos.”), WT ILC2 + OVA (10 µg/ml), or PD-L1^−/− ILC2 + OVA (10 µg/ml). Histograms show the dilution of CellTrace Violet in CD4⁺ T cells. Bar graph shows the mean + SEM of six samples from two experiments. (c) WT or Pdcd1^−/− CD4⁺ T cells were cultured alone or together with WT ILCs as in a. Bar graph shows the mean + SEM of six replicates. (d) Cells were co-cultured as in a in Transwell or normal plates (“direct” cell–cell contact). Bar graph shows the mean + SEM of six samples from two experiments. (e) Purified naive CD4 T cells were cultured under nonpolarizing conditions for 24 h in the presence of anti-CD3 antibody (5 µg/ml) with recombinant mouse PD-L1-Fc (5 µg/ml; red bars) or control protein (5 µg/ml; blue bars). Cells were stained intranuclearly for indicated transcription factors. Bar graphs show the mean + SEM of four to nine replicates from three experiments. (f) T cells from IL-4Ra^−/− mice were cultured as in e. Bar graph shows the mean + SEM of five replicates from two experiments. (g) Naive T cells from WT and Pdcd1^−/− mice were cultured as described in e and analyzed for GATA3 expression 24 h later. Bars show the mean + SEM of five replicates. (h) WT CD4 T cells were cultured for 5 d under nonpolarizing and Th1-, Th2-, and Th17-polarizing conditions in the presence or absence of recombinant PD-L1 and analyzed for indicated transcription factors and cytokines. Heat maps depict means of six replicates. Bar graphs are shown in Fig. S2. (i) Naive WT CD4 T cells were cultured under nonpolarizing conditions in the presence of control recombinant hIgG-Fc, PD-L1-Fc, PD-L2-Fc, or PD-1-Fc (5 µg/ml each) and analyzed for GATA3 expression by flow cytometry 24 h later. Bars show the mean + SEM of six replicates. Student’s t test was used to analyze data, and p-values <0.05 were considered statistically significant. *, P < 0.05; **, P < 0.01; ns, not significant.

Figure 4.

PD-L1–expressing ILC2s control the PD-1 checkpoint toward Th2 polarization during helminth infection. (a) Rag2^−/−Il2rg^−/− mice received naive CD4 T cells alone or together with WT or PD-L1^−/− ILC2s and were infected with N. brasiliensis the next day. Lungs were analyzed for GATA3⁺ CD4 T cells and eosinophils (b) by flow cytometry. Bar graphs show the mean + SEM of five individual mice per group from two independent experiments. (c) Intestinal worm burden and fecal egg counts (d) 7 d after infection in mice receiving CD4 T cells alone (light gray), WT CD4 T cells and PD-L1^+/+ (black bars) or PD-L1^−/− (open bars) ILC2s, or Pdcd1^−/− CD4 T cells and WT ILC2s (dark gray). Bars show the mean + SEM of four to nine mice per group from three experiments. (e) Representative dot plots showing the PD-L1 expression on lung ILC2s (gated on live lin⁻CD45⁺CD90⁺ST2⁺KLRG1⁺ cells) from infected WT, PD-L1^−/− and PDL1^fl/-IL7Rα^Cre/+ mice 5 d after infection. FMO, fluorescence minus one. (f) PD-L1 expression on lung ILC2s in WT, PD-L1^fl/fl, PD-L1^−/−, and PDL1^fl/-IL7Rα^Cre/+ mice. Bar graphs show the mean + SD of at least three mice per group. (g) Control and PDL1^fl/-IL7Rα^Cre/+ mice were infected with N. brasiliensis, and worm burden was analyzed 7 d after infection. Bar graph shows the mean + SEM of nine mice from three experiments. (h) Th2 cell IL-4-, IL-5-, and IL-13 production (i) in lung CD4 T cells or ILC2s (j) and lung eosinophil infiltration (k) were analyzed by flow cytometry 5 d after infection. Bar graphs show the mean + SEM of six to 10 mice from three independently performed experiments. (l) Periodic acid–Schiff staining of fixed small intestinal tissue 5 d after N. brasiliensis infection. Bar graph shows the mean + SD of three mice, with 5 crypt-villus units counted per mouse. Bars, 1 mm. (m) WT or PDL1^fl/flID2^CreERT2 mice were left untreated or injected i.p. with tamoxifen before infection with N. brasiliensis. Bar graphs show the mean + SEM of PD-L1 expression on lung ILC2s (left) and CD4 T cells (middle) and worm burden 7 d after infection (right) from at least three mice per group from two experiments. Student’s t test was used to analyze data, and p-values <0.05 were considered statistically significant. *, P < 0.05, ns, not significant. In c, d, and m, ANOVA was used to determine statistical significance. #, P < 0.01.

Figure 5.

Working model. Helminth- or allergen-inflicted epithelial damage leads to the release of IL-33, which activates ILC2s via ST2. ILCs up-regulate PD-L1 and interact with CD4 T cells on site to promote increased GATA3 expression and thereby commit them to IL-13–producing Th2 cells. IL-13 released from both cell types then activates the epithelium to promote worm expulsion.