Eos Promotes T_H2 Differentiation by Interacting with and Propagating the Activity of STAT5

This article is featured in Top Reads, p. 309

In response to viral, bacterial, or parasitic infection, CD4⁺ T cells differentiate into specialized effector “helper” subsets that coordinate pathogen-specific immune responses. In contrast with these beneficial roles, dysregulated responses of effector CD4⁺ T cell subsets have been implicated in the pathogenesis of allergic and autoimmune conditions. T_H2 cells are well-known for this dichotomy: although their effector cytokines, such as IL-4 and IL-13, are capable of orchestrating immune responses directed at extracellular pathogens, such as helminthic worms, they can also induce the pathogenesis of allergic asthma (1–5). Given this balance between benefit and harm, understanding the regulators of T_H2 differentiation is of key importance to the prevention and management of T_H2-mediated immune responses and diseases.

For CD4⁺ T helper cells to differentiate, naive CD4⁺ T cells receive three main signals: (1) TCR activation, (2) costimulation, and (3) environmental cytokine signals (6, 7). The latter lead to JAK-mediated tyrosine phosphorylation of STAT proteins, causing them to dimerize, translocate to the nucleus, and bind to target genes, including those for CD4⁺ T cell subset-specific lineage-defining transcription factors. For example, IL-4 and STAT6 drive expression of the T_H2 lineage-defining transcription factor Gata3 (8, 9). Beyond IL-4/STAT6 signaling, a second cytokine–STAT pathway, IL-2/STAT5, is also critical for the programming and effector functions of T_H2 cells (10–12). STAT5 collaborates with Gata3 at IL-4, IL-5, and IL-13 gene loci and induces the expression of CD25 (IL-2Rα), CD122 (IL-2Rβ), and CD124 (IL-4Rα) to amplify the T_H2 gene program (13–17). The IL-2/STAT5 pathway further supports T_H2 differentiation by directly suppressing genes associated with alternative T cell programs, including those important for T follicular helper (T_FH) and T_H17 cell differentiation (18–20).

In addition to STAT factors, Ikaros zinc-finger (ZF; IkZF) transcription factors have emerged as key regulators of CD4⁺ T cell differentiation (21–28). Members of the IkZF family have conserved N-terminal ZF DNA-binding domains that enable them to bind DNA in a sequence-specific fashion, as well as C-terminal ZF protein-interaction domains that can recruit coregulators (29). Although it is well established that IkZF factors recruit chromatin remodeling complexes to promote or repress gene expression, it is also becoming more recognized that IkZF factors engage in a mechanistic interplay with cytokine signaling pathways (23–25, 28, 30–32). One example comes from our prior work demonstrating that Aiolos (encoded by Ikzf3) interacts with STAT3 to cooperatively induce expression of Bcl-6, the lineage-defining transcription factor for T_FH cells (25). A key aspect of this study was identification of the conserved C-terminal ZF domain as a requirement for STAT3 interaction, suggesting that IkZF/STAT interactions could represent a conserved regulatory mechanism that functions across CD4⁺ T cell differentiation programs.

In addition to effector subsets, IkZF factors also have demonstrated roles in T regulatory (T_REG) cell populations. For example, studies have shown that Eos (encoded by Ikzf4) is a component of a transcriptional complex that includes the T_REG lineage-defining factor Foxp3 (33). This complex directly represses the expression of target gene loci, including Il2 (33–35). Eos enacts these changes by recruiting the chromatin modifier C-terminal Binding Protein-1, which results in increased DNA methylation and histone modifications associated with gene silencing (33). Consequently, T_REG cells with reduced or absent Eos expression lose their suppressive capabilities (33, 34, 36). Somewhat surprisingly, however, recent work also suggests that Eos may contribute to the differentiation of conventional, effector CD4⁺ T cell subsets (37, 38). Thus, the roles that Eos plays in effector and regulatory CD4⁺ T cell populations and whether these roles may depend on conserved or divergent mechanisms have remained enigmatic.

In this study, we demonstrate that Eos is a positive regulator of T_H2 cell programming. Data from both in vitro–generated T_H2 cells and an in vivo murine asthma model reveal disruptions to key T_H2 transcription factors, cytokine receptors, and effector cytokines in the absence of Eos. Mechanistically, we find that Eos and IL-2/STAT5 engage in a positive feedback loop, with STAT5 directly inducing Eos expression and Eos interacting with and propagating the tyrosine phosphorylation-mediated activation of STAT5. Finally, we find that Eos and STAT5 cooperate to induce Il4ra and Il2rb, which encode cytokine receptors important for driving T_H2 differentiation. Taken together, to our knowledge, our data reveal a novel mechanism whereby Eos interacts with and positively regulates STAT5 activity to promote T_H2 differentiation and effector cytokine production.

C57BL/6J mice were purchased from the Jackson Laboratory. Eos^KO (Ikzf4^−/−) mice were generously provided by Drs. Ethan Shevach (National Institutes of Health) and Bruce Morgan (Harvard Medical School) (37). To avoid unintentional bias in in vitro mouse studies, we used male and female mice with age- and sex-matched controls. Due to known estrogen-dependent variability in asthma pathogenesis, we exclusively used male mice for in vivo murine house dust mite (HDM) asthma studies (39–41). The Institutional Animal Care and Use Committees of The Ohio State University approved all experiments involving the use of mice. All methods were performed in accordance with the approved guidelines. EL4 thymoma T cells were acquired from the American Type Culture Collection (TIB-39) and cultured in complete RPMI (RPMI-1640, 10% FBS, 1% penicillin/streptomycin).

Naive CD4⁺ T cells were isolated from the spleens and lymph nodes of 5- to 8-wk-old mice using the BioLegend MojoSort naive CD4⁺ T cell isolation kit according to the manufacturer’s recommendations. Naive CD4⁺ T cell isolates were routinely checked via flow cytometry for purity (>96–98%). Harvested cells were plated at a density of 1.5 × 10⁵ cells/ml in 2 ml of complete IMDM (cIMDM; IMDM [Life Technologies], 10% FBS [26140079; Life Technologies], 1% penicillin-streptomycin [Life Technologies], and 0.05% [50 mM] 2-ME [Sigma-Aldrich]). Cells were stimulated on plate-bound anti-CD3 (clone 145-2C11; 5 mg/ml; BD Biosciences) and anti-CD28 (clone 37.51; 2 mg/ml; BD Biosciences) in the presence T_REG-, T_H2-, T_FH-like–, or T_H17-polarizing conditions (Table I). Neutralizing Abs were added at plating, and polarizing cytokines were added after 24 h of stimulation. For expansion of cells on day 3 into resting conditions, cells were plated at 2.5 × 10⁵ cells/ml in 2 ml of cIMDM with fresh polarizing cytokines and cultured for an additional 48 h.

Overexpression assays in EL4 T cells were performed using the Lonza 4D Nucleofection system (buffer SF, program CM-120). Expression vectors were made by cloning the indicated coding sequences into the pcDNA3.1/V5-His-TOPO vector (K480001; Life Technologies). Point mutations were generated using the QuikChange Site-Directed Mutagenesis Kit (200519; Agilent) according to the manufacturer’s instructions. For generation of constitutively active STAT5B (STAT5B^CA), point mutations were introduced into STAT5B (H298R and S715F), resulting in constitutive phosphorylation of the Y699 residue (42). For the Eos C-terminal protein–interaction domain mutant (Eos^ΔC), vector primers were designed to truncate the protein at the beginning of the two C-terminal ZFs. For the Eos N-terminal DNA-binding domain mutant (Eos^DBM), point mutations were introduced at the cysteine residues of N-terminal ZFs 1 and 2 to disrupt DNA binding. Wild-type (WT) and mutant coding sequences were transferred to the pEF1/V5-His vector (V92020; Life Technologies) for overexpression. EL4s were transfected for 22–24 h before downstream analyses. For the earlier experiments, overexpression of proteins was assessed via immunoblot using both V5 tag and protein-specific Abs, and alterations in transcript expression were assessed via quantitative RT-PCR (qRT-PCR) analysis.

Total RNA was isolated from the indicated cell populations using the Macherey-Nagel Nucleospin RNA Isolation kit as recommended by the manufacturer. cDNA was made using the Superscript IV First Strand Synthesis System (Thermo Fisher Scientific). qRT-PCR was performed with SYBR Select Mastermix for CFX (Thermo Fisher Scientific) using 6 ng of cDNA per reaction and primers for the appropriate genes (Table II). All qRT-PCR was performed on the CFX Connect (BioRad). Data were normalized to Rps18 and presented either relative to Rps18 or relative to a control sample, as indicated.

Cells were harvested, counted, lysed directly in 1× SDS loading dye (50 mM Tris [pH 6.8], 100 mM DTT, 2% SDS, 0.1% bromophenol blue, 10% glycerol), and boiled for 15 min. Equal protein lysate amounts were loaded based on cell counts. Lysates were separated via SDS-PAGE and transferred to a 0.45-µm nitrocellulose membrane. Membranes were blocked with 2% nonfat dry milk in 1× TBST (10 mM Tris [pH 8], 150 mM NaCl, 0.05% Tween 20). The following Abs were used to detect proteins: anti-Eos (1:500, W15169A; BioLegend), anti–STAT5 tyrosine phosphorylation (pY-STAT5; 1:5000, 611964; BD Biosciences), anti-STAT5 (1:5000, 94205S; Cell Signaling), anti-STAT5B (1:5000, sc-1656; Santa Cruz), anti–pY-STAT6 (1:5000, 9361S; Cell Signaling), anti-STAT6 (1:5000, 9362S; Cell Signaling), anti–β-actin (1:15,000; GenScript), goat anti-mouse (1:5000–1:10,000; Jackson Immunoresearch), and mouse anti-rabbit (1:15,000; Santa Cruz). Immunoblot bands were quantified using ImageJ. For each protein row, the largest band of the row was framed, and mean gray value was measured using the same frame across the row. Background measurements were taken with the same frame measuring the area above or below bands on the image. Pixel densities were inverted, background values subtracted from sample bands and control bands, and a ratio of net protein bands to net loading control bands was calculated for protein quantification relative to the WT sample.

Coimmunoprecipitation (Co-IP) analyses were performed using primary murine in vitro–generated T_H2 cells, or EL4 cells overexpressing indicated proteins, as previously described (25, 43, 44). In brief, lysates were immunoprecipitated with anti-STAT5 (0.785 μg per immunoprecipitation [IP], 94205S; Cell Signaling) or an isotype control Ab overnight at 4°C. The following day, samples were incubated in the presence of protein A-Sepharose beads (Millipore) for 1–2 h, and IPed proteins were assessed via subsequent immunoblot analyses. Abs to detect IPed proteins were as follows: anti-Eos (1:500, W15169A; BioLegend), anti-STAT5 (1:5000, 74442X; Santa Cruz), and anti-V5 (1:20,000, R960-25; Invitrogen).

ChIP assays were performed as described previously (45). Resulting chromatin fragments were IPed with Abs against STAT5 (10 mg/IP, AF2168; R&D Systems) or IgG control (10 mg/IP, ab6709; Abcam; matched to experimental Ab). Enrichment of the indicated proteins was analyzed via quantitative PCR (Table III). Samples were normalized to total input DNA, and isotype control Abs were used to ensure lack of nonspecific background.

An Ikzf4 promoter–reporter construct (pGL3-Ikzf4) was generated by cloning the regulatory region of Ikzf4 (−1427 to +0 bp) into the pGL3-Basic vector (Promega) (Table IV). EL4 T cells were nucleofected with pGL3-Ikzf4 in combination with either STAT5B^CA, STAT5B^WT, or STAT3^CA expression vector, which was constructed as previously described (25). For each nucleofection, SV40-Renilla control was used to assess transfection efficiency. After 22–24 h of recovery, samples were harvested, and luciferase expression was analyzed using the Dual-Luciferase Reporter Assay System according to the manufacturer’s instructions (Promega).

On days 0 and 7, C57BL/6J mice were sensitized with 10 μg of HDM solution (Dermatophagoides pteronyssinus, 28,750 endotoxin units/vial; XPB82D3A2.5; Greer Laboratories) in 50 μl of 1× PBS (Thermo Fisher Scientific) via oropharyngeal aspiration. On days 14, 15, and 16, mice were challenged oropharyngeally with 2 μg of HDM before their lungs were harvested on day 17. Lung tissue was incubated with HBSS (Life Technologies) + 1.3 mM EDTA for 30 min at 37°C and processed in Collagenase IV–supplemented RPMI media + 4% FBS for 30 min at 37°C using the Miltenyi Biotec gentleMACS Dissociator. Samples were filtered through a 100-μm nylon mesh strainer and subsequently centrifuged with a Percoll density gradient to isolate lymphocytes. Erythrocytes were lysed, and cells were washed once in ice-cold FACS buffer (PBS + 4% FBS) before staining for flow cytometry analyses.

For panels requiring analysis of cytokine production, cells were first incubated in cIMDM with eBioscience PMA and ionomycin cell stimulation mixture plus protein transport inhibitors (PTIs; 00-4975-03; Invitrogen) for 4 h. For extracellular staining, samples were preincubated for 5 min at 4°C with TruStain FcX (anti-mouse CD16/32) Fc block (clone 93; 101320; BioLegend). Samples were then stained for extracellular markers in the presence of Fc block for 30 min at 4°C protected from light using the following Abs: anti-CD3 (allophycocyanin/Cy7; 1:300; clone 17A2; BioLegend), anti-CD4 (AF488; 1:300 [for cells not given PMA/ionomycin and PTI] or 1:100 [for cells incubated with PMA/ionomycin and PTI]; clone GK1.5; R&D Systems); anti-CD44 (V450; 1:300; clone IM7; BD Biosciences); anti-CD62L (allophycocyanin-eFluor 780; 1:300; clone MEL-14; Thermo Fisher Scientific); anti-CD25 (PE and allophycocyanin; 1:300; clone PC61.5; Thermo Fisher Scientific); anti-CD122 (PE/Cy7; 1:100, clone TM-b1; Thermo Fisher Scientific); and Ghost Dye (V510; 1:400; Tonbo Biosciences). Cells were then washed twice with FACS buffer before intracellular staining. For intracellular staining, cells were fixed and permeabilized using the eBioscience Foxp3 transcription factor staining kit (Thermo Fisher Scientific) for 30 min or overnight at 4°C. After fixation, samples were stained with the following Abs in 1× eBioscience permeabilization buffer (Thermo Fisher Scientific) for 1 h at room temperature protected from light: anti-Foxp3 (PerCP-Cy5.5; 1:100; clone FJK-16s; Thermo Fisher Scientific); anti-Gata3 (for restimulated panels: PE-Cy7; 1:20; clone TWAJ; Thermo Fisher Scientific); anti-Gata3 (for non-restimulated panels: allophycocyanin; 1:20; clone 16E10A23; BioLegend); anti–IL-4 (allophycocyanin; 1:50; clone 11B11; BioLegend); IL-13 (PE; 1:50; clone W17010B; BioLegend); anti–IL-2 (allophycocyanin/Cy7; 1:50, clone JES6-5H4; BD Biosciences); and Ki-67 (BV421; 1:50, clone 16A8; BioLegend). Cells were washed twice with 1× permeabilization buffer and resuspended in FACS buffer for analysis. Where appropriate, fluorescence-minus-one controls were used to account for nonspecific staining. Samples were run on a BD FACS Canto II flow cytometer and analyzed using FlowJo software (version 10.8.2). For Annexin V and propidium iodide staining, cells were harvested and washed twice with 500 μl FACS buffer. Cells were resuspended at 0.5 × 10⁶ cells/20 μl Annexin V Binding Buffer (BioLegend), and 1 μl of anti–Annexin V (FITC; BioLegend) and 2 μl propidium iodide (BioLegend) were added for 15 min at room temperature protected from light. Samples were then diluted with 80 μl Annexin V Binding Buffer and analyzed by flow cytometry.

Naive CD4⁺ T cells were cultured in T_H2-polarizing conditions for 3 d. Total RNA was isolated using the Macherey-Nagel Nucleospin kit. Samples were provided to Azenta Life Sciences for polyA selection, library preparation, sequencing, and DESeq2 analysis. Genes with an adjusted p ≤ 0.05 were considered differentially expressed. Genes were preranked by multiplying the sign of the fold-change by −log₁₀(adjusted p value) and analyzed using the Broad Institute Gene Set Enrichment Analysis (GSEA) software for comparison against “hallmark,” “gene ontology,” and “immunologic signature” gene sets. Heatmap generation and clustering (by Euclidean distance) were performed using normalized log2 counts from DESeq2 analysis and the Morpheus software (https://software.broadinstitute.org/morpheus). Volcano plots were generated using −log₁₀(adjusted p value) and log2 fold change values from DESeq2 analysis and VolcaNoseR software (https://huygens.science.uva.nl/VolcaNoseR/) (46).

RNA sequencing (RNA-seq) datasets have been deposited in the GEO repository under accession number GSE216737 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE216737). All other datasets and materials from this study will be made available on reasonable request. Requests should be sent to the corresponding author.

All statistical analyses were performed using the GraphPad Prism software (version 9.5.1). For single comparisons, unpaired Student t tests were performed. For multiple comparisons, one-way ANOVA with Tukey’s multiple comparisons test were performed. Error bars indicate the SEM. The p values <0.05 were considered statistically significant.

Our laboratory has previously shown that expression of the IkZF factor Aiolos is elevated in T_FH cells, and that it cooperates with STAT3 to promote expression of the T_FH lineage-defining transcription factor Bcl-6 (25, 47). As a follow-up to this prior study, we performed a screen for IkZF factors across several in vitro–generated effector CD4⁺ T subsets in efforts to (1) identify potential novel IkZF regulators of CD4⁺ T cell programming and (2) determine whether similar IkZF/STAT factor relationships may exist in other effector populations (Tables I, II). Consistent with the established literature, Ikaros was the most highly expressed IkZF factor across subsets (Supplemental Fig. 1A, 1B) (24, 25, 27). However, a comparison of Ikzf3 and Ikzf4 expression in T_H2 and T_FH-like cells revealed an inverse pattern of Aiolos and Eos gene expression, with Ikzf3 expression reduced in T_H2 cells relative to T_FH-like cells and Ikzf4 transcript elevated in T_H2 cells compared with T_FH-like cells (Fig. 1A, 1B, Supplemental Fig. 1B). This elevation in Eos expression in proinflammatory T_H2 cells was notable, because most studies to date have linked Eos with the immunosuppressive T_REG gene program (33, 36, 38, 48). Further comparison across T_REG, T_H2, T_FH-like, and T_H17 cells demonstrated that, as predicted, Eos expression was markedly elevated in T_REG cells (Fig. 1C, 1D, Supplemental Fig. 1B, 1C) (33, 36, 49). However, Eos expression was significantly increased in T_H2 cells compared with T_FH-like and T_H17 populations (Fig. 1C, 1D). These data suggested that, in addition to its established role in T_REG cells, Eos may also influence T_H2 cell programming.

We next wanted to define the signals that may be responsible for driving Eos expression in T_H2 cells. We considered the commonalities that exist between T_REG and T_H2 cells and noted that the IL-2/STAT5 cytokine pathway supports their differentiation and conversely represses that of T_FH and T_H17 gene programs (50). To determine whether IL-2/STAT5 may be responsible for inducing Eos expression, we first disrupted IL-2 signaling by culturing T_H2 cells with both anti-CD25 (anti–IL-2Rα) and anti-CD122 (anti–IL-2Rβ) blocking Abs and found that inhibition of IL-2 signaling resulted in significantly reduced Eos transcript and protein expression (Fig. 2A). Published ChIP-seq data demonstrate that STAT5B can bind to the Ikzf4 promoter (51). To determine whether STAT5 associates with the Ikzf4 promoter in T_H2 cells, we performed ChIP analysis and found that STAT5 enrichment was significantly increased at the Ikzf4 promoter compared with an IgG control Ab, as well as a negative control region with no STAT5 binding in CD4⁺ T cells (Fig. 2B, 2E, Table III). To investigate whether STAT5 is capable of modulating Ikzf4 promoter activity, we generated a firefly luciferase gene reporter construct containing the portion of the Eos (Ikzf4) promoter harboring the STAT5 enrichment site. On overexpression of STAT5B^CA, the dominant STAT5 isoform in effector and regulatory T populations, we observed significantly increased Ikzf4 promoter activity (Fig. 2C, 2E, Table IV) (52). This induction was specific, because we did not observe an increase in Ikzf4 promoter activity in response to overexpression of STAT3^CA or nonconstitutively active STAT5B (STAT5B^WT) (Fig. 2C, 2E). We considered the possibility that other cytokine signaling pathways, such as IL-4/STAT6, which is important to T_H2 differentiation, could upregulate Eos expression. However, manipulation of this pathway with either anti–IL-4–blocking Abs or addition of exogenous IL-4 to activated T cell cultures did not impact the levels of Eos expression (Fig. 2D). Collectively, these findings indicate that IL-2/STAT5 signaling specifically promotes Eos expression in T_H2 cells.

Given the elevated Eos expression in T_H2 cells, we next sought to determine whether loss of Eos would result in defects to the T_H2 gene program. We began by performing RNA-seq analyses of WT and Eos^KO naive CD4⁺ T cells that had been cultured under T_H2-polarizing conditions (Fig. 3A). We observed global transcriptomic changes in Eos-deficient samples, including significant downregulation of T_H2 genes, such as those encoding transcription factors (Gata3 and Prdm1), effector cytokines (Il4, Il5, and Il13), and cytokine and chemokine receptors (Il4ra, Il2ra, Il10ra, and Ccr8) (Fig. 3B, 3C, Supplemental Fig. 2A). Conversely, genes associated with the T_FH and T_H17 gene programs, including Bcl6 and Rorc, respectively, were upregulated in the absence of Eos (Fig. 3B, 3C). GSEA of “immunologic signature” gene sets further revealed that genes upregulated in IL-4–treated naive CD4⁺ T cells were downregulated in Eos^KO T_H2 cells (Fig. 3D). In contrast, “immunologic signature” gene sets associated with T_H17- and T_FH-like–polarized cells were upregulated (Supplemental Fig. 2B). To determine whether the disruptions to the T_H2 gene program had a functional consequence, we analyzed the in vitro–generated WT and Eos^KO T_H2 cells by flow cytometry and found the Eos^KO T_H2 cells had significantly reduced protein expression of Gata3, IL-4, and IL-13 compared with their WT counterparts (Fig. 3E, 3F). Taken together, these data indicate that the T_H2 gene and functional programs are compromised in Eos-deficient CD4⁺ T cells.

We next assessed whether T_H2 differentiation and effector cytokine production may be impacted by the loss of Eos in the context of an in vivo immune response. To do this, we used a murine HDM allergic asthma model, a robust and well-established model of T_H2-induced asthma, and harvested lung tissue for flow cytometry analyses (Fig. 4A) (53). Compared with their WT counterparts, the Eos^KO mice exhibited a significantly lower frequency of Gata3⁺ cells in the lungs (Fig. 4B, Supplemental Fig. 3A). In addition, Eos-deficient cells produced significantly lower levels of IL-4 and IL-13 relative to their WT counterparts (Fig. 4C, 4D). We considered the possibility that Eos deficiency may result in disrupted immunosuppressive T_REG cells. However, we did not observe a decline in the percentage of Foxp3⁺ T cells in this setting (Supplemental Fig. 3B). Overall, our findings suggest that Eos deficiency results in a significant reduction of T_H2 differentiation and effector responses.

We next wanted to address the Eos-dependent mechanisms that result in the dysregulation of T_H2 differentiation in the context of Eos deficiency. Intriguingly, GSEA analysis revealed a downregulation of “hallmark” genes associated with the IL-2/STAT5 signaling pathway in Eos-deficient T_H2 cells (Fig. 5A). To further examine the impact of Eos deficiency on the IL-2 pathway, we performed flow cytometry analyses for CD25 and CD122 and found that both proteins were significantly reduced in the absence of Eos (Fig. 5B). Because IL-2 signaling is known to augment CD25 and CD122 expression, we investigated whether Eos deficiency results in alterations to IL-2 production. However, IL-2 protein expression was upregulated in the absence of Eos, suggesting that the downregulation of CD25 and CD122 was not due to decreased IL-2 production (Fig. 5C). Because the IL-2/STAT5 pathway promotes both T cell survival and proliferation, we also sought to determine the effects of Eos deficiency on cell death and expansion. We analyzed in vitro–generated WT and Eos^KO T_H2 cells using Annexin V, propidium iodide, and Ki-67 staining and found no significant differences in viability, apoptosis, or proliferation between WT and Eos^KO T_H2 cells (Supplemental Fig. 4A, 4B). Furthermore, in the HDM model, we analyzed surface expression of CD25 by flow cytometry and observed a significant reduction in the percentage of Foxp3⁻CD25⁺ cells in the absence of Eos (Fig. 5D). Curiously, this defect in CD25 surface expression was also observed in the Foxp3⁺ T_REG compartment, further supporting at least a partially conserved mechanism for Eos in regulating the IL-2/STAT5 pathway across T cell subsets (Supplemental Fig. 4C).

In addition, GSEAs also revealed that the “gene ontology” gene set associated with pY-STAT—modifications most closely associated with STAT dimerization, nuclear translocation, and transcriptional activity—was also downregulated in the absence of Eos (Fig. 5E) (54, 55). To determine whether loss of Eos resulted in reduced pY-STAT5–dependent activation, we performed immunoblot analyses of WT versus Eos^KO in vitro–generated T_H2 and T_REG cells. Indeed, we observed a significant reduction in the levels of pY-STAT5 in the absence of Eos in both populations (Fig. 5F, Supplemental Fig. 4D). Together, these data demonstrate that Eos deficiency is associated with a downregulation of IL-2R expression and reduced pY-STAT5.

Our prior study demonstrated that another IkZF family member, Aiolos, interacts and cooperates with STAT3 to induce expression of the transcriptional repressor Bcl-6 (25). Because of the high degree of homology between members of the IkZF and STAT families, as well as our GSEA data indicating that Eos deficiency disrupts IL-2/STAT5 signaling (Fig. 5A, 5E), we hypothesized that Eos may impact T_H2 differentiation, at least in part, via interaction and cooperation with STAT5. Co-IP assays detected interactions between Eos and STAT5 in primary T_H2 cells (Fig. 6A). Again, consistent with a conserved role for Eos in regulation of the IL-2/STAT5 pathway, we similarly detected interactions in in vitro–generated T_REG cells (Supplemental Fig. 4E).

We next overexpressed Eos and STAT5B^CA in murine thymoma EL4 T cells and likewise detected interactions via Co-IP (Fig. 6B). The IkZF–STAT5 interaction was specific to Eos, because Ikaros and Aiolos did not coimmunoprecipitate with STAT5 (Fig. 6C). To determine the functional significance of Eos and STAT5 cooperativity, we individually expressed or coexpressed each protein and assessed the impact on expression of Il4ra, an early T_H2 gene identified in the RNA-seq analyses that is upregulated downstream of IL-2/STAT5 signaling (Figs. 3B, 6D) (16). Indeed, we found that expression of Il4ra was highest when Eos and STAT5B^CA were coexpressed compared with when either protein was overexpressed alone (Fig. 6D). We extended our findings by assessing expression of Il2rb, which is required for the functional dimeric and trimeric forms of the IL-2R complex needed to initiate T_H2 differentiation (56). Similar to our findings with Il4ra, Il2rb expression was most significantly induced in the presence of both Eos and STAT5B^CA, suggesting that Eos and STAT5 work in concert to promote at least a subset of STAT5 target genes (Fig. 6E).

Through the course of these overexpression studies, we monitored expression of Eos, STAT5, and pY-STAT5 at the protein level. Due to the constitutively active nature of the STAT5B protein used, we expected to observe similar levels of pY-STAT5 regardless of whether it was expressed alone or in combination with Eos. Strikingly, we found that when Eos and STAT5B^CA were coexpressed, there was a marked increase in the amount of pY-STAT5 compared with expression of STAT5B^CA alone (Fig. 6F). This suggested that the presence of Eos bolsters or sustains pY-STAT5 levels. Notably, because these experiments were carried out in the absence of environmental cytokines (e.g., IL-2), these results indicate that pY-STAT5 relies on interactions with Eos rather than Eos-dependent effects on cytokine signaling, such as an induction of IL-2R subunits (57, 58). Further supporting this, we found that overexpression of Eos, but not Ikaros or Aiolos, resulted in increased pY-STAT5 (Fig. 6G). Collectively, these data indicated that Eos augments the expression of at least a subset of STAT5 target genes, and that this is likely due to enhanced pY-STAT5 in the presence of Eos.

We previously found that the C-terminal dimerization domain of Aiolos was necessary for interaction with STAT3 (25). To determine whether the C-terminal domain of Eos was similarly required to mediate interactions with STAT5, we created Eos mutant proteins with disruptions to the C-terminal protein–interaction domain (Eos^ΔC), as well the conserved N-terminal DNA-binding domain (Eos^DBM), and performed Co-IP analyses (Fig. 7A). Indeed, we found that the C-terminal protein–interaction domain, but not the DNA-binding domain, was required for Eos to interact with STAT5 (Fig. 7B). We further wanted to determine whether the Eos–STAT5 interaction was required for pY-STAT5. Indeed, we observed a marked reduction in pY-STAT5 when STAT5B^CA was coexpressed with Eos^ΔC (Fig. 7C). Likewise, coexpression of STAT5B^CA with the interacting Eos or Eos^DBM proteins resulted in elevated pY-STAT5. Collectively, these data suggest that the ability of Eos to augment pY-STAT5 depends on interaction with STAT5, and not its DNA-binding properties. Finally, we tested whether Eos–STAT5 interactions were required to induce Il4ra or Il2rb. Strikingly, Il4ra and Il2rb induction was substantially reduced when STAT5B^CA was coexpressed with Eos^ΔC, relative to Eos^WT (Fig. 7D, 7E). In contrast, we did not observe a significant decrease in Il4ra or Il2rb expression when STAT5B^CA was coexpressed with Eos or Eos^DBM, indicating that the STAT5-interacting and pY-STAT5–enhancing properties of Eos are paramount to its ability to bind DNA for the induction of these gene targets (Fig. 7D, 7E). Taken together, these data demonstrate that Eos and STAT5 interact and are capable of engaging in cooperative mechanisms to promote T_H2 cell programming.

To date, Eos has been categorized primarily as a transcription factor important for maintaining both CD4⁺ T_REG cell identity and associated suppressive functions (33, 34, 36, 38, 48, 59). More recent work has suggested that Eos plays additional roles in regulating effector T cell responses (37, 38). However, the precise role of Eos in these processes has remained elusive. To this end, we now demonstrate that Eos is, to our knowledge, a novel regulator of T_H2 cell differentiation and effector cytokine production. Our data demonstrate that Eos^KO mice have reduced T_H2 responses in a murine HDM asthma model, and that there are global defects in the expression of the T_H2 gene program in Eos-deficient CD4⁺ T cells. Mechanistically, we found that Eos promotes IL-2/STAT5 signaling, and that this occurs, at least in part, via an interaction with STAT5 that potentiates pY-STAT5. Thus, our findings implicate Eos as an important regulator of T_H2 differentiation and suggest that augmentation of STAT5 activity represents at least one mechanism by which Eos promotes the T_H2 programming.

Previous work from our laboratory demonstrated that the IkZF family member Aiolos physically interacts and cooperates with STAT3 to positively regulate Bcl-6 expression in CD4⁺ T cell populations (25). In this study, we show that Eos and STAT5 interact in T_H2 cells, and that this interaction is dependent on the C-terminal ZF domain of Eos (29, 60). This is consistent with our previous findings for Aiolos–STAT3 interactions, establishing the C-terminal ZF domain of Ikaros family members as a conserved regulatory feature that enables the formation of IkZF–STAT factor modules (25). It is interesting to note that distinct cytokine signatures appear to drive the formation and function of the Eos–STAT5 and Aiolos–STAT3 modules in T_H2 and T_FH-like cells, respectively (25). Notably, however, we demonstrate in this article that IL-2/STAT5, and not IL-4/STAT6, is the primary driver of Eos expression. Nevertheless, it is tempting to speculate that other cytokine–STAT pathways directing the differentiation of other Th effector subsets may promote the formation and function of additional distinct IkZF–STAT regulatory complexes.

The precise mechanisms by which Eos mediates target gene expression are incompletely defined. Current evidence indicates that IkZF factors primarily recruit remodeling complexes to alter chromatin structure (29, 30, 61). However, to our knowledge, our data in T_H2 cells point to a novel mechanism whereby Eos works in a feed-forward fashion to promote STAT5 activation and subsequent regulation of target gene expression. Although how Eos modulates STAT5 phosphorylation is currently unclear, we hypothesize that there are several nonmutually exclusive possibilities. First, and most clearly, our data indicate that Eos regulates expression of the IL-2R subunits CD25 and CD122, through which STAT5 is activated in the T_H2 cell population. Indeed, a prior study described Eos-dependent regulation of CD25 in undifferentiated, activated conventional CD4⁺ T cell populations (37). Although much of the data we present in T_H2 cells are consistent with this mechanism, it is important to note that these overexpression studies were performed with a STAT5B^CA, the dominant STAT5 isoform in effector and regulatory T cell responses, in the absence of signals from IL-2 (52). Thus, we speculate that the ability of Eos to modulate STAT5 activity may be because of effects on kinases or phosphatases that act on STAT5 (54). This includes the possibility that the physical interaction between Eos and STAT5 enhances STAT5 protein stability and/or protects it from phosphatase activity. Ultimately, future studies will be necessary to comprehensively evaluate the mechanism(s) by which Eos influences STAT5 phosphorylation.

Our findings also offer the opportunity to compare roles for Eos in T_REG and effector T_H2 cell populations. In T_REG cells, Eos functions as a transcriptional repressor that works in concert with Foxp3 to direct T_REG cell–specific expression patterns (33, 34, 36). Notably, a known Eos target in T_REG cells is the Il2 locus (33). However, findings from our study suggest that Eos-dependent repression of IL-2 is not conserved in T_H2 cells. Although we cannot definitively rule out a repressive role for Eos in T_H2 cells, our findings suggest that Eos plays an important role in positively regulating the T_H2 gene program by potentiating STAT5 activity. Curiously, Eos is expressed at much higher levels in T_REG versus effector T cell populations (37). As such, our findings suggest that Eos may operate as a molecular rheostat controlling the amplitude of the IL-2/STAT5 signaling pathway in different Th cell subsets. Functionally, this may be particularly advantageous for T_REG cells, because it may provide a mechanism to maintain high levels of active STAT5 in low IL-2 environments (62, 63). In addition, our data further demonstrate that the Eos–STAT5 complex exists in T_REG cells and is associated with increased pY-STAT5 in this population. This suggests that the Eos–STAT5 transcriptional mechanisms may be at least partially conserved across different CD4⁺ T subsets. To our knowledge, this novel Eos–STAT5 relationship also is likely to be clinically relevant, because therapies employing low doses of IL-2 have been used in efforts to preferentially enhance T_REG cell populations for the treatment of a variety of autoimmune disorders (64, 65). A comprehensive understanding of the conserved and/or divergent roles for Eos in driving the transcriptional programs of T_REG and T_H2 cell populations awaits additional study, including those that will probe other CD4⁺ T cell subsets dependent on IL-2 signaling for their differentiation (e.g., T_H1 or T_H9 cells) (16, 66–68).

Finally, it will be important for future studies to assess whether Eos–STAT5 regulatory modules function downstream of cytokine signals other than IL-2. Indeed, STAT5 is also activated downstream of signals from the cytokines IL-7 and IL-15, which impart diverse functions across several immune cell populations (69, 70). As such, it will be of interest to assess potential cooperative roles for Eos and STAT5 in effector and memory CD8⁺ T, B, NK, and innate lymphoid cell populations (71). Ultimately, these studies will generate important insights into general mechanisms by which Ikaros family members regulate the differentiation and function of immune cell populations, which may be leveraged in targeted immunotherapy approaches to treat human disease.

The authors have no financial conflicts of interest.

We thank all members of the Oestreich Laboratory, as well as colleagues in the Department of Microbial Infection and Immunity, for constructive feedback. We would also like to thank Drs. Ethan Shevach and Bruce Morgan for providing the Eos^KO (Ikzf4^−/−) mice for these studies; Dr. Greg Whitehead for guidance for HDM dosing; and Dr. Fernanda Novais and Erin Fowler for sharing Annexin V, propidium iodide, and Ki-67 Abs for flow cytometry analyses.