Essential, dose-dependent role for the transcription factor Gata3 in the development of IL-5⁺ and IL-13⁺ type 2 innate lymphoid cells

ILC2s Express High Levels of Gata3 Protein and Require Gata3 Expression for Development in Vivo.

We first analyzed Gata3 protein expression in ILC2s present in different tissue sites (Fig. 1A). ILC2s were identified as lineage-negative cells that coexpress CD25, CD127, and IL-33R as described (3, 4). We found that bone marrow (BM) and lung ILC2s expressed Gata3 levels that exceeded by ∼10-fold the levels expressed by mature CD4⁺ T cells (Fig. 1A). It remains unclear whether ILC2s in the BM represent immature ILC2 precursors or a “reservoir” of peripheral ILC2s (2), although BM ILC2s clearly produced cytokines (see below) and had similar Gata3 levels as lung ILC2s (Fig. 1A). Gata3⁺ ILC2s were also identified in spleen, lymph nodes (LNs), and intestine (see below).

To assess the requirement for Gata3 in the development of ILC2s in vivo, we analyzed chimeric mice generated by using Gata3-deficient hematopoietic precursors cells as described (28, 29). To control for the resultant T-cell deficiency in these chimeras (30), we generated T-cell–deficient, Gata3-proficient chimeras using Cd3e^−/− hematopoietic precursors. Both Gata3^+/+ and Gata3^−/− hematopoietic precursors engrafted to a similar extent, leading to robust B, NK, and myeloid cell reconstitution (Fig. S1A and ref. 28).

Because Rag2^−/−Il2rg^−/− hosts lack ILC2s (3), all ILC2s in the chimeric mice were donor-derived, as expected (Fig. S1B). Both BM and lung ILC2s developed in Gata3^+/+ chimeras (Fig. 1 B and C), and these ILC2s normally expressed CD127, CD25, and IL-33R, as expected (5, 31). In Gata3^−/− chimeras, we found a similar number of BM Lin⁻CD127⁺ cells, although these cells failed to express characteristic surface markers of ILC2s (Fig. 1B). In the lung of Gata3^−/− chimeras, Lin⁻CD127⁺ cells were severely reduced, and ILC2s were essentially absent (Fig. 1C).

Analysis of IL-5 production by Lin⁻ cells in the lung and BM of Gata3^+/+ chimeras showed that all cells that produced IL-5 were ILC2s (Fig. 1D). In contrast, IL-5 production was undetectable in lung and BM cell preparations from Gata3^−/− chimeras (Fig. 1D), consistent with the absence of ILC2 in these tissues (Fig. 1 B and C).

GATA-3 Is Required for Characteristic Innate Responses to Exogenous IL-33 or -25.

Intranasal (i.n.) administration of IL-33 in vivo results in a dramatic increase in the absolute numbers of ILC2s in lung and broncho-alveolar lavage (BAL) fluid (2, 7–10, 32). i.n. IL-33 (Fig. 2A) triggered recruitment of Siglec-F⁺ eosinophils and Gr-1⁺ neutrophils in the BAL fluid in Gata3^+/+, but not in Gata3^−/−, chimeras (Fig. 2B). This inflammatory reaction was accompanied by a dramatic increase in ILC2 numbers within the lung tissue and BAL fluid of Gata3^+/+, but not Gata3^−/−, chimeras (Fig. 2 C and D). Gata3^+/+ ILC2s in the lung promptly produced IL-5 and -13 following stimulation, whereas cytokine production was not detected in Lin⁻ cells from Gata3^−/− chimeras (Fig. S2). We conclude that lung-resident IL-33–responsive ILC2s require Gata3 for their generation and/or function.

Because the expression of the IL-33R component T1/ST2 is regulated by GATA-3 (21), the failure of Gata3^−/− chimeras to respond to IL-33 in vivo may in part be explained by the lack of receptor expression by Gata3^−/− ILC2s (Figs. 1 and 2). Systemically administered IL-25 can trigger an inflammatory response from ILC2s that provokes multisystem myeloid cell recruitment and mucus production (33). IL-25 injection resulted in the expansion of ILC2s in the BM and lung of Gata3^+/+, but not Gata3^−/−, chimeras (Fig. 2E and Fig. S3A). Although we found that ILC2s were the major source of IL-5 produced by lung Lin⁻ lymphocytes, IL-5 was not detected in these cells in the absence of Gata3 (Fig. S3B). Collectively, these results demonstrate that characteristic inflammatory responses following triggering of ILC2s by IL-25 and -33 are not elicited in chimeric mice made with Gata3^−/− hematopoietic precursors.

Failure of ILC2 Differentiation from Gata3^−/− Hematopoietic Precursors in Vitro.

Notch-triggered differentiation of ILC2s from CLPs can be achieved in vitro by using stromal cells expressing delta-like ligand 4 in the presence of IL-7 and IL-33 (19). We found that culture of Gata3^+/+ fetal liver (FL) hematopoietic precursors generated abundant ILC2s using this approach (Fig. S4A), whereas cells expressing characteristic surface markers of ILC2s were markedly reduced in the absence of Gata3 (Fig. S4A). RT-PCR analysis of these FL cultures showed that in the absence of Gata3, Id2 and Rora expression were not induced (Fig. S5); these transcription factors are highly expressed in ILC2s and are critical for their development (3, 17, 34). Moreover, we could not detect IL-5 or -13 production in these ILC2 Gata3^−/− cultures, whereas ILC2 cultures of Gata3^+/+ fetal liver hematopoietic precursors gave rise to cells with characteristic IL-5 and -13 production (Fig. S4B).

Cell-Intrinsic Gata3 Is Required for ILC2 Development.

To exclude the possibility that cell-extrinsic effects caused by Gata3 deficiency underlie the observed defects in ILC2 development, we generated and analyzed BM chimeras generated by using 9:1 mixtures of CD45.2⁺ Gata3^−/− and CD45.1⁺ wild-type (WT) C57BL/6 hematopoietic precursors, respectively. Mice injected with 9:1 mixtures of CD45.2⁺ Gata3^+/− and CD45.1⁺ Gata3^+/+ precursors served as controls (Fig. 3A). Both sets of mixed chimeras reconstituted to a similar extent (Fig. S6A). We confirmed that B and NK cell development occurred normally from both Gata3^−/− and Gata3^+/+ precursors (Fig. 3B and Fig. S6B), and phenotypically mature ILC2s were generated in BM, lung, spleen, and small intestine (Fig. 3C and Fig. S6C). Further analysis revealed that all ILC2s in Gata3^−/− mixed BM chimeras expressed CD45.1⁺ (originating from Gata3^+/+ precursors), whereas ILC2s in Gata3^+/− mixed chimeras originated from both CD45.2⁺ Gata3^+/− and CD45.1⁺ Gata3^+/+ precursors (Fig. 3C and Fig. S6D). This result demonstrates that Gata3 is required in a cell-intrinsic fashion for ILC2 development. Interestingly, although analysis of the origin of B and NK cells recapitulated the ratio in which precursors were injected (i.e., 86–87% of the B and NK cells were CD45.2⁺ independent of Gata3 genotype), ILC2s derived from Gata3^+/− precursors were underrepresented relative to those derived from their Gata3^+/+ counterparts in control chimeras (Fig. 3C). This result suggests that the Gata3 gene dosage may influence overall ILC2 development.

GATA-3 Gene Copy Positively Correlates with ILC2 Development, Homeostasis, and Function.

We next compared the effects of modifying GATA3 gene copy number on ILC2 development and function. We characterized ILC2 homeostasis and function in three mouse strains: C57BL/6 mice with two functional Gata3 alleles, Gata3^lacZ/+ mice with one functional Gata3 allele (the second allele is disrupted by insertion of a beta-galactosidase reporter (35), and transgenic C57BL/6 mice that harbor a Gata3 transgene under the control of the human CD2 promoter (36). We found that Gata3 protein levels in BM and lung ILC2s were reduced in Gata3^lacZ/+ mice compared with C57BL/6 mice, indicating that both Gata3 alleles are apparently active in ILC2s (Fig. S7A). In contrast, ILC2s from CD2–Gata3 Tg mice expressed the same high levels of Gata3 protein as WT C57BL/6 mice (Fig. S7A). Because we detected increased Gata3 RNA expression in BM ILC2s (Table S1), it is possible that the intracellular FACS assay is not sensitive enough to detect differences in Gata3^hi WT and CD2–Gata3 Tg ILC2s.

We next assessed the impact of alterations in Gata3 gene copy number on overall ILC2 homeostasis. Absolute numbers of mature ILC2s in the lung and BM were clearly reduced in Gata3^lacZ/+ mice, whereas CD2–Gata3 Tg mice had an approximately threefold increase in BM and lung ILC2s compared with Gata3^+/+ C57BL/6 mice (Fig. 4A). ILC2 function also positively correlated with Gata3 gene copy number: ILC2s from CD2–Gata3 Tg mice produced somewhat higher levels of IL-5 and -13 following stimulation compared with ILC2s from B6 WT mice, whereas cytokine production was reduced in ILC2s from Gata3^lacZ/+ mice having only one functional Gata3 allele (Fig. 4A and Fig. S7B). Moreover, CD2–Gata3 Tg ILC2s showed higher surface expression of IL-25R and IL-33R (Fig. S7C).

These results suggest that Gata3 expression levels can modify ILC2 generation from hematopoietic precursors and thereby influence ILC2 homeostasis. Whereas absolute numbers of CLPs were normal in CD2–Gata3 Tg mice, we found that Gata3 levels in CLPs were modestly but significantly increased in CLPs in CD2–Gata3 Tg mice compared with nontransgenic controls (Fig. 4B). This finding would be consistent with hCD2 Tg control elements being active in CLPs, a hypothesis that we confirmed using CD2–GFP mice (37) (Fig. 4C). To assess the influence of increased Gata3 levels in determining ILC2 potential from CLPs, we analyzed expression of ILC2-related genes (Rora, Id2, Il17rb, and Il1rl1) in CLPs from WT and CD2–Gata3 Tg mice. Whereas Gata3 expression was threefold increased, the levels of other ILC2-related genes were unaffected (Table S1). We next cultured single CLPs from Gata3^lacZ/+, Gata3^+/+, or CD2–Gata3 Tg mice on stromal cells in the presence of IL-7 and -33. ILC2s developed from Gata3^+/+ CLPs at the expected frequency (f = 1/5) (19), whereas CD2–Gata3 Tg CLPs were almost twice as efficient in generating ILC2s (f = 1/2.6). In contrast, Gata3^lacZ/+ CLPs showed a 2.6-fold reduction in ILC2 potential (f = 1/13) compared with their WT counterparts (Table 1, Fig. 4D, and Fig. S8). Collectively, these results suggest a model in which Gata3 levels in CLPs control initial commitment to the ILC2 lineage and confirm the important role for sustained Gata3 expression in maintenance of type 2 cytokine secretion from mature ILC2s (24).

Mice and Generation of Chimeric Mice.

Rag2–Il2rg double-deficient on the C57BL/6 background, Cd3e-deficient, Gata3–lacZ, CD2–Gata3, and CD2–GFP Tg mice were maintained at the animal facilities of the Institut Pasteur, Erasmus University Medical Center, and the Faculdade de Medicina de Lisboa. C57BL/6 mice and BALB/c Rag2–Il2rg mice were obtained from Harlan and Charles River, respectively. Mice were analyzed at 8–16 wk of age. Five-gray–irradiated Rag2–Il2rg double-deficient animals as hosts for hematopoietic stem cell grafts and were analyzed 5–10 wk after transplantation. All animals were kept under specific pathogen-free conditions and provided with food and water ad libitum. All experiments were approved and performed in accordance with local regulations (Institutional Animal Care and Use Committee, Institut Pasteur, Paris).

IL-25 and -33 Administration.

For IL-25–induced inflammation, mice received 2.5 µg of recombinant mouse IL-25 (BioLegend) in PBS for four subsequent days intraperitoneally (33). Three doses of 1 µg of recombinant mouse IL-33 (BioLegend) in PBS were administered i.n. to anesthetized mice to induce eosinophilic airway inflammation (9).

Phenotypic and Functional Analysis.

Lamina propria lymphocytes from small intestine were isolated as described (47). Single-cell suspensions from lungs, BAL, spleen, LNs, and BM were prepared and analyzed by FACS as described (9). CLPs (Lin⁻CD117^lowSca1^lowCD135⁺) were sorted from WT and CD2–Gata3 Tg mice for quantitative PCR analysis. Antibodies used for analysis of cell surface phenotype, intracellular cytokines, and transcription factors are described in SI Materials and Methods.