Eos plays a critical role in Treg homeostasis and modulates the function of recirculating thymic Tregs in the control of Treg development

SUMMARY

Eos, a member of the Ikaros family of transcription factors, is expressed by T regulatory cells (Tregs) and has been postulated to play a role in Treg suppression and maintenance of Treg stability. We demonstrate that expression of Eos was limited to a subpopulation of thymus-derived, activated Tregs and is undetectable in resting or activated T conventional cells. Eos associates with Helios and Foxp3 and binds directly to the CD25 locus at a site identical to the Foxp3-binding site, thereby enhancing CD25 expression. Studies in heterozygous female mice demonstrate that Eos is critical for Treg survival and activation. Eos⁺ Tregs also represent the major population of recirculating thymic Tregs, in which Eos plays a critical role in regulating their migration and suppression of Treg precursors in the thymus by competing for IL-2 and depleting MHC II from thymic dendritic cells.

Graphical abstract

In brief

Xie et al. demonstrate that the transcription factor Eos associates with Helios and Foxp3 in regulatory T cells (Tregs) to control their migration, suppressive capacity, survival, and fitness. Eos mediates the enhanced expression of CD25 on recirculating thymic Tregs, thereby controlling the production of thymic Tregs.

INTRODUCTION

Regulatory T cells (Tregs) play a critical role in maintaining immune homeostasis and preventing autoimmune diseases by modulating immune responses.^1,2 Among the various subsets of Tregs, thymus-derived Tregs (tTregs) are crucial for establishing central tolerance, while peripherally induced Tregs (pTregs) arise from conventional CD4⁺ T cells in response to specific antigens.³ In addition to the function of the thymus in regulating the development of tTregs, a major subset of Tregs in the thymus is composed of Tregs that have recirculated from peripheral tissues, likely directed by Ccr6/Ccl20 and Cxcr4/Cxcl12 chemokine axes.^4,5 A number of studies have suggested that recirculating Tregs play a major role in restraining the development of Tregs in the thymus.^4,6 Thiault et al. added mature peripheral Tregs to thymic organ cultures and demonstrated that these peripheral Tregs can suppress the development of Tregs by ~30%–60%, but did not inhibit the development of T conventional cells (Tconvs).⁴ While the addition of mature Treg cells inhibited the development of CD4⁺Foxp3⁺ Tregs, they had no effect on the numbers of CD4⁺CD25⁺Foxp3⁻ precursor cells in the thymus. As the differentiation of precursor cells requires the presence of IL-2, these authors postulated that the recirculating Tregs inhibited Treg development in the thymus by absorbing IL-2.⁴ Additionally, Weist et al. used a thymic tissue slice culture model to study the differentiation of Tregs expressing a specific TCR transgene.⁶ They demonstrated a 50% reduction in Treg development when OT-II Foxp3⁻ thymocytes were added to thymic slices derived from OVA protein-injected IL-2-deficient (^−/−) mice. Partial reconstitution of Treg development was induced by the addition of IL-2-anti-IL-2 immune complexes to the slice. The failure of OT-II WT thymocytes to fully restore Treg development in IL-2^−/−slice cultures suggested that thymocytes were unlikely the important source of IL-2. The addition of OVA-loaded WT dendritic cells (DCs) to IL-2-deficient cultures restored Treg development, suggesting that antigen-bearing DCs could provide a local source of IL-2 needed for Treg development. Taken together, these studies demonstrate that the development of Tregs within the thymus is very sensitive to the local concentration of IL-2 in the thymus. One of the mechanisms by which recirculating thymic Tregs exert a regulatory influence on local thymic IL-2 concentrations is to compete for IL-2 availability. However, most of the studies were conducted in vitro or using TCR transgenic mice, leaving the relevance of this mechanism in vivo unclear.

The Ikaros family of zinc-finger transcription factors (TFs) comprises five members (Ikzf1-5: Ikaros, Helios, Aiolos, Eos, and Pegasus) essential for the development and function of hematopoietic and immune cells.⁷ Two distinct regions characterize Ikaros family members: N-terminal zinc finger domains that mediate DNA binding, and C-terminal zinc finger domains crucial for protein interactions (homo- and heterodimerization between Ikaros family members).⁸ Among the Ikaros family members, Helios was reported to be a potential marker of tTregs and to play a critical role in Treg fate determination.^9–11 Eos is a second member of the Ikaros family that is preferentially expressed in Tregs. Pan et al. initially showed that Foxp3 and Eos physically interacted in Treg cells, as evidenced by co-immunoprecipitation (coIP).¹² Furthermore, Tregs treated with Eos-specific siRNA were unable to suppress inflammatory bowel disease (IBD) induced by Tconvs, underscoring the essential role of Eos in Treg function.¹² Sharma et al. demonstrated that at sites of inflammation, the loss of Eos resulted in the reprogramming of Tregs into T helper (Th)-like T cells, loss of suppressor function, and acquisition of a pro-inflammatory phenotype.¹³ One major problem with both of these studies was that the specificity of the anti-Eos antibodies was not adequately confirmed using Tregs from Eos^−/− mice. Gokhale et al. (2019) used a genetic approach to define the role of Eos in Treg function by generating Eos^fl/fl × Foxp3-Cre (Eos cKO) mice, which developed loss of suppressor function in vivo accompanied by systemic- and organ-specific autoimmunity at the age of 3 months.¹⁴ However, global Eos-deficient mice (Eos KO) did not develop overt autoimmunity until approximately 1 year of age, and mice with T cell-specific Eos depletion (Eos^fl/fl × CD4-Cre) developed autoimmunity at 7–8 months of age, suggesting a potential regulatory role for Eos in Tconvs and possibly non-T cells.^14,15

In this study, we show that Eos associates with Helios and Foxp3 to form a complex regulating multiple aspects of Treg function. Importantly, Eos is preferentially expressed in recirculating thymic Tregs, where it regulates their migration and suppressive function through modulation of their chemotaxis, extracellular matrix, affinity for IL-2, and ability to capture MHC II from thymic DCs, thereby limiting thymic Treg generation and establishing immune homeostasis. Overall, our study demonstrates that Eos controls critical aspects of Treg biology, which may lead to advancement of potential therapeutic strategies for autoimmune diseases and other immune-related disorders.

RESULTS

Eos is preferentially expressed in a subpopulation of Helios⁺Foxp3⁺ Tregs

To investigate the function of Eos, we generated a monoclonal anti-Eos antibody (clone 18H2) by immunizing a hamster with aa 416–464 peptide of mouse Eos protein (Figure S1A). Binding specificity of this antibody was tested using splenic CD4⁺ T cells from Eos KO and wild-type (WT) mice (Figure S1B). Consistent with a previous report, we observed preferential expression of Eos in Foxp3⁺CD4⁺ Tregs from thymus, spleen and mesenteric lymph nodes (mLNs; Figures 1A and 1B).¹⁵ In young adult mice, approximately 50% of Foxp3⁺ T cells expressed Eos, and all Eos⁺ Tregs were Helios⁺. Very low numbers of Eos⁺Foxp3⁻ T cells could be detected. Even though Eos was predominantly expressed in a subpopulation of Helios⁺Foxp3⁺ Tregs, the expression of Foxp3 and Helios was not required for the expression of Eos. In hemizygous male scurfy mice, which lacked Foxp3-expressing T cells, significant numbers of Eos⁺CD4⁺ T cells were observed. Compared to WT mice, scurfy male mice showed an even higher percentage of Eos⁺CD4⁺ T cells in the spleen (Figure 1C). In Helios^fl/fl × CD4-Cre mice, which lacked Helios-expressing T cells and remained autoimmune-free, we also observed abundant Eos⁺CD4⁺ T cells.⁹ When compared to WT mice, the ratio of Eos-expressing CD4⁺ T cells and Eos MFI remained comparable, suggesting that the expression of Eos is independent of Helios (Figure 1D).

Figure 1. Eos is preferentially expressed in a subpopulation of Helios⁺Foxp3⁺ Tregs.

(A) Representative plots of Eos expression in total live cells, CD4⁺ T cells, and Tregs in the thymus, spleen, and mLNs from 8-week-old (wo) female B6 mice.

(B) Frequencies of Eos⁺ cells within indicated cell populations from the thymus, spleen, and mLNs of 8 wo female B6 mice.

(C) Representative plots of Eos and Foxp3 expression in CD4⁺ T cells in the thymus and spleen from 3 wo male WT mice, 3 wo male scurfy mice, and 3–6 wo male Eos KO mice. Frequencies of Eos⁺ cells within indicated cell populations.

(D) Representative plots of Eos and Helios expression in CD4⁺ T cells in the thymus and spleen from 12 wo WT mice, 8–13 wo Helios^fl/fl × CD4-Cre mice, and 12 wo male Eos KO × Foxp3-GFP mice. Frequencies of Eos⁺ cells and Eos MFI within indicated cell populations.

(E) Immunofluorescence microscopy imaging of Tregs purified from the spleen of B6, Helios^fl/fl × CD4-Cre and Eos KO mice, stained for Eos, Helios, Foxp3, and DAPI with indicated colors.

(F) Representative images of coIP between Eos, Helios, and Foxp3 using purified Tregs from the spleen of WT mice.

Data are representative of at least two independent experiments. p values are determined by Student’s t test (B) and one-way ANOVA (C and D). Mean ± SD. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Ikaros family members use their C-terminal zinc finger (ZF) domains to form homo- or heterodimers between themselves, and other Ikaros TFs were reported to directly interact with Hdac1, a core component of the NuRD complex, and Foxp3.^8,12,16,17 To validate if Eos could form heterodimers with Helios, we performed immunofluorescence (IF) microscopy and observed significant co-localization of Eos with Helios in the nucleus of Tregs from the spleen and thymus of WT mice (Figures 1E and S1C). The binding specificity of anti-Eos (clone 18H2) and anti-Helios (clone 22F6) antibodies under these experimental conditions was also confirmed using splenic Tregs from Eos KO and Helios^fl/fl × CD4-Cre mice (Figure 1E). Even though Foxp3 did not co-localize with Eos and Helios, coIP experiments using purified Tregs from WT mice revealed that the anti-Eos antibody precipitated both Helios and Foxp3, while Helios and Foxp3 also precipitated with each other, as previously reported (Figure 1F).^12,17 This finding is consistent with the possibility that Eos and Helios form heterodimers and physically interact with a subset of Foxp3 molecules.

Eos is preferentially expressed in tTregs

As Eos⁺ Tregs represented a subpopulation of Helios⁺ Tregs, and our previous studies demonstrated that Helios was selectively expressed in tTregs, we evaluated Eos expression in peripherally induced Tregs (pTregs) in vivo.⁹ In the gastrointestinal tract, a substantial population of pTregs is present, which plays an important role in maintaining tolerance to the gut microbiota and dietary antigens. Notably, these pTregs are characterized as RORγt⁺Helios⁻, distinguishing them from tTregs.^18–21 In the large intestine (LI), the percentage of Foxp3⁺ T cells expressing Eos was less than that seen in spleen. Importantly, all RORγt⁺ Treg cells were Eos⁻, indicating that most of the Eos⁺ Tregs in the gut were likely tTregs (Figure 2A).

Figure 2. Eos is preferentially expressed in tTregs.

(A) Representative plots of Eos, Foxp3, and RORγt expression in CD4 T cells and Tregs from the LI or spleen of 10 wo B6 mice at steady state. Quantifications of Eos⁺ cell frequencies within Tregs in the LI and spleen and RORγt⁺ cell frequencies within Eos⁺ or Eos⁻ Tregs in the LI.

(B) Representative plots of CD45.1, CD45.2, Eos, Helios, and Foxp3 expression in CD4⁺ T cells, host CD4⁺ T cells (F1 CD4⁺), and adoptively transferred OT-II cells in the mLNs of host mice in the oral tolerance model. Quantification of Eos⁺ cell frequencies within indicated cell types, pTreg induction frequencies from transferred WT or Eos KO OT-II cells, and Eos⁺ and Helios⁺ cell frequencies within induced pTregs.

(C) Representative plots of Eos, Helios, and Foxp3 expression in adoptively transferred CD4⁺ TCRβ⁺ T cells from the mLNs of recipient mice in the IBD model. Quantification of Eos⁺ cell frequencies within IBD-induced pTregs from indicated cell populations, pTreg induction frequencies, and Eos⁺ cell frequencies within WT Helios⁺ or Helios⁻ pTregs.

Data are representative of at least two independent experiments. p values are determined by Student’s t test (A and C) or by mixed-effects analysis (B). Mean ± SD. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

To determine whether Eos could be induced in pTregs in vivo, we used an oral tolerance model in which naive WT or Eos KO OT-II CD4⁺ T cells were adoptively transferred into congenically marked WT recipient mice, which were then given OVA-supplemented water for 7 days (Figure 2B). While the induction of pTregs in the mLNs from WT and Eos KO OT-II cells occurred at a similar ratio, Eos expression in WT OT-II pTregs could not be detected (Figure 2C). It has been reported previously that pTregs in the lamina propria in the oral tolerance model failed to express Helios, but a substantial number of pTregs in the mLNs expressed Helios.⁹ This finding was also observed in our study (Figure 2B). This raises the possibility that Eos is more selectively expressed in tTregs than Helios. We also examined the expression of Eos in pTregs generated during the induction of IBD following transfer of naive CD4⁺Foxp3⁻ T cells to immunodeficient mice. In contrast to the results in the oral tolerance model, when purified naive CD4⁺Foxp3⁻ T cells from Foxp3-Cre or Eos^fl/fl × Foxp3-Cre mice were transferred to TCRα KO mice (Figure 2D), pTregs expressed high levels of Eos 4 weeks post transfer, and most of those Eos⁺ pTregs were Helios⁺ (Figure 2C). This suggests that under inflammatory conditions, Eos and Helios could be induced at high levels in pTregs.

Eos expression is markedly lower in Tconvs compared to Tregs

Our previous studies suggested that Eos was expressed in Tconvs and may play a role in their function, as mice with either global deletion of Eos or T cell-specific depletion of Eos developed autoimmunity much later in life than mice with Treg-specific deletion of Eos.^14,15 While Eos expression could not be detected in CD4⁺ Tconvs and CD8⁺ T cells in the steady state, we examined Eos expression in Th cells polarized in vitro. We purified naive CD4⁺ T cells from CD45.1 WT and Eos KO (CD45.2) mice, mixed the cells at a 1:1 ratio, and cultured them under Th-polarizing conditions and iTreg conditions. Eos expression could not be detected in in vitro-polarized Th1, Th2, and Th17 cells compared to iTregs (Figures S1D and S1E). We also examined cytokine production by the polarized cells upon stimulation with cell stimulation cocktail and found no differences between WT cells and Eos KO cells (Figures S1F and S1G).

Next, we examined the function of Eos in Tconvs in vivo. We transferred CD4⁺CD45RB^hiFoxp3⁻ cells from WT or Eos KO mice to Rag1 KO recipients to induce IBD (Figure S2A). Two weeks post transfer, we observed similar levels of IFN-γ, IL-10, and IL-2 production by transferred WT and Eos KO CD4⁺ T cells isolated from the spleen of recipient mice (Figures S2B and S2C). Next, we examined the potential function of Eos in Th2 cells in vivo by challenging WT or Eos^fl/fl × CD4-Cre mice with Alternaria alternata (A. alternata) extract intranasally (i.n.) twice (Figure S2D). 7 days post-second challenge, no differences were observed in the production of IL-13 and IL-4 by CD4⁺ T cells isolated from the lung between the two groups (Figures S2E and S2F). Lastly, we transferred WT or Eos KO naive OT-II T cells to congenically marked WT recipients and immunized them with OVA_323–339 in CFA subcutaneously (s.c.) to induce Th1 and Th17 cells. No differences in IFN-γ and IL-17 production by the two types of transferred OT-II cells were seen (Figures S2H and S2I). Thus, we were unable to define a role for Eos in Tconvs in these relatively short-term in vitro and in vivo assays, and the function of Eos in Tconvs, if any, requires further careful examination.

Eos is critical for the fitness and homeostasis of Tregs

We analyzed the expression of activation markers and effector Treg (eTreg)-associated markers on Tregs from WT mice and found that Eos⁺ Tregs exhibited higher expression levels of these markers compared to Eos⁻ Tregs in both the thymus and spleen, indicating that Eos⁺ Tregs were more activated than Eos⁻ Tregs (Figure 3A). Due to the development of autoimmunity in Eos^fl/fl × Foxp3-Cre mice, to further characterize the intrinsic functions of Eos in Tregs under normal conditions, we generated heterozygous female mice (Eos^fl/fl × Foxp3-YFPCre/RFP) by crossing Eos^fl/fl × Foxp3-RFP mice to Eos^fl/fl × Foxp3-Cre mice (Figure 3B). The ratio of Eos cKO Tregs to WT Tregs was 1:1 at 6 weeks of age in both the thymus and spleen, but this ratio decreased dramatically at 8 weeks and 12 weeks, indicating a critical role for Eos in the maintenance of Treg fitness (Figures 3C and 3D). We also observed that in 6 wo heterozygous females, both the percentage of CD25⁺ cells and the CD25 MFI were lower on Eos cKO Tregs than on WT Tregs, in both the thymus and spleen (Figures 3E and 3F).

Figure 3. Eos is critical for the fitness and homeostasis of Tregs.

(A) Quantifications of CD44, CD62L, PD-1, GITR, OX-40, Icos, and Tigit MFI of Eos⁺ and Eos⁻ Tregs from the thymus and spleen of 7–15 wo WT mice.

(B) Schematic representation of Eos^fl/fl × Foxp3-YFPCre/RFP heterozygous female mice having both WT Tregs and Eos cKO Tregs at the same time.

(C) Representative plots of WT Tregs (Foxp3-RFP⁺) and Eos cKO Tregs (Foxp3-YFP⁺) within CD4⁺ T cells in the thymus and spleen of heterozygous females at 6, 8, and 12 wo.

(D) Quantification of WT Treg (Foxp3-RFP⁺) frequencies and Eos cKO Treg (Foxp3-YFP⁺) frequencies within CD4⁺ T cells in the thymus and spleen of heterozygous females at 6, 8, and 12 wo.

(E) Representative histograms of CD25 in WT Tregs and Eos cKO Tregs from the thymus and spleen of heterozygous females at 6 wo.

(F) Quantifications of CD25⁺ cell frequencies and CD25 MFI of indicated Tregs from the thymus and spleen of heterozygous females at 6 wo.

p values are determined by Student’s t test (A, D, and F). Mean ± SD. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Eos is preferentially expressed in recirculating thymic Tregs

We analyzed the frequencies of Eos⁺ Tregs in the thymus and spleen of WT mice at different ages and found a logarithmic growth curve (Figure 4A), which closely resembled the reported curve of increasing recirculating thymic Tregs with aging.⁴ In previous studies, GFP⁻ thymic Tregs in Rag-GFP mice were defined as recirculating Tregs, and the majority of them expressed CD73.^22,23 Their homing to the thymus was postulated to be mediated by Cxcr4 and/or by the Ccr6/Ccl20 chemokine axes.⁵ We re-examined the expression of recirculating thymic Treg-related proteins on Eos⁺ thymic Tregs and found that Eos⁺ thymic Tregs were a subpopulation of CD73⁺ thymic Tregs. Furthermore, Eos expression on thymic Tregs was more closely correlated with both Ccr6 and Cxcr4 expression (Figure 4B). By contrast, the majority of thymic Tregs expressed Helios, as previously reported (Figure 4B).⁹ This finding aligns with the established role of Helios as a marker of tTregs and its early expression during T cell development. We speculated that Eos⁺ thymic Tregs were the actual recirculating thymic Tregs, while Eos⁻CD73⁺ thymic Tregs were mostly thymus-resident Tregs, which are mature Tregs that have never left the thymus.

Figure 4. Eos is preferentially expressed in recirculating thymic Tregs.

(A) Frequencies of Eos⁺ cells within Tregs in the thymus and spleen of WT mice of different ages.

(B) Representative plots of Eos, CD73, Ccr6, Cxcr4, and Helios expression on thymic Tregs from 12 wo WT mice.

(C) Uniform manifold approximation and projection (UMAP) visualization of 17 clusters from sorted WT and Eos cKO Tregs from the thymi of two 6 wo heterozygous females (F1 and F2).

(D) Violin plots visualizing the expression of Nt5e (CD73), Ccr6, Cxcr4, and Foxp3 within each cluster.

(E) UMAP visualization of sorted WT and Eos cKO Tregs from the thymi of two 6 wo heterozygous females (F1 and F2) re-grouped based on their expression of CD73, Ccr6, Cxcr4, and Foxp3.

(F) Pseudotime trajectories of sorted thymic Tregs from 6 wo heterozygous females. The arrowhead indicates the directions of cell pseudo-temporal trajectories.

(G) Alluvial plots visualizing overlapping of top 5 T cell clones from the 5 groups (20 clones in total) of thymic Tregs from 6 wo heterozygous females.

(H) Schematic representation of parabiotic WT mice with different congenic markers and recirculating cells from donor to host.

(I) Representative plots of Eos and Foxp3 expression on host CD4⁺ T cells and donor CD4⁺ T cells in the thymus and spleen of parabiotic mice.

(J) Representative plots of Ccr6, Cxcr4, and Foxp3 expression on host CD4⁺ T cells and donor CD4⁺ T cells in the thymus of parabiotic mice.

(K) Quantifications of Eos⁺ cell frequencies, Ccr6⁺ cell frequencies, and Cxcr4⁺ frequencies within indicated Tregs in the thymus or spleen of parabiotic mice.

p values are determined by Student’s t test (K). Mean ± SD. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

To better define subpopulations of thymic Tregs, we isolated WT and Eos cKO thymic Tregs from 6 wo heterozygous female mice and performed combined scRNA-seq and TCR sequencing. High-resolution clustering identified 17 distinct clusters from sorted cells (Figure 4C). We then examined the expression of Foxp3, Nt5e (CD73), Ccr6, and Cxcr4 in each cluster (Figure 4D) and regrouped the 17 clusters into 5 groups (Figure 4E). To determine the order in which cells change over time, we performed pseudotime analysis (Figure 4F), which allowed us to infer the progression of biological processes over time. During pseudotime analysis, we observed that certain small and biologically inconsistent clusters (mostly CD73⁺Ccr6⁺Cxcxr4⁻) interfered with the trajectory inference. These minor clusters, potentially representing rare cell states, outliers, or technical artifacts, introduced noise and disrupted the overall continuity of the inferred developmental paths. To improve the robustness and interpretability of the trajectories, we opted to exclude these low-population clusters prior to rerunning the pseudotime analysis. This refinement step facilitated a smoother and more biologically plausible reconstruction of cellular differentiation and transition trajectories. In the end, we obtained four trajectories, all originating from the Foxp3⁻ group. The Foxp3⁻ group exhibited transient expression of Nr4a1, Nr4a3, and Ccr8 along the trajectories, features that are typical of Treg precursors.²⁴ All four trajectories then transitioned to the CD73⁻ group, which was reported to contain mostly newly generated Tregs.^4,23 Subsequently, one trajectory terminated in the CD73⁻ group, while another led to the CD73⁺Ccr6⁻Cxcr4⁻ group, and two trajectories entered the CD73⁺Ccr6⁺Cxcr4⁺ group. These findings suggest a more mature status of CD73⁺ cells but also indicate distinct developmental pathways for the CD73⁺Ccr6⁻Cxcr4⁻ and CD73⁺Ccr6⁺Cxcr4⁺ groups.

Next, we compared the proportions of the top 5 most abundant TCR clones from each of the 5 groups (20 clones in total) based on TCR sequencing data (Figure 4G). The CD73⁺Ccr6⁻Cxcr4⁻ group shared highly overlapping major clones with both the Treg precursor group and the newly generated Treg group, suggesting that the CD73⁺Ccr6⁻Cxcr4⁻ cells had never left the thymus. In contrast, the CD73⁺Ccr6⁺Cxcr4⁺ group contained very distinct major clones from all other four groups, suggesting that they are recirculating thymic Tregs whose TCR repertoire was shaped by antigens they encountered in the periphery.²⁵ Taking the pseudotime analysis into consideration, we suggest that the Foxp3⁻ group primarily comprised Treg precursors, the CD73⁻ group was predominantly composed of newly generated Tregs, the CD73⁺Ccr6⁻Cxcr4⁻ group was largely thymus-resident Tregs, and the CD73⁺Ccr6⁺Cxcr4⁺ group was predominantly recirculating thymic Tregs. Since Eos⁺ thymic Tregs were highly overlapping with Ccr6⁺ and Cxcr4⁺ thymic Tregs, this further confirmed that Eos was preferentially expressed in recirculating thymic Tregs.

We also analyzed differentially expressed genes between WT Tregs and Eos cKO Tregs from the thymus of 6 wo heterozygous female mice and observed reduced expression of eTreg-associated genes (Tigit and Icos; Figure S3A). We then performed a pathway enrichment analysis and observed a significant enrichment in pathways associated with T cell activation and migration in WT thymic Tregs (Figure S3B). Furthermore, we performed pathway enrichment analysis within each cluster and observed that in cluster 6 (the most mature recirculating thymic Treg cluster), multiple chemotaxis- and migration-related pathways were enriched in WT thymic Tregs (Figure S3C), confirming the important role of Eos in regulating the migration of recirculating thymic Tregs.

To more definitively address the preferential expression of Eos in recirculating thymic Tregs, we performed parabiosis experiments, in which two congenically marked WT mice were surgically joined for 1 month (Figure 4H). With congenic markers, we could easily distinguish between host cells and recirculating donor cells. Analysis of the frequency of Eos⁺ cells within host Tregs and donor Tregs in the thymus demonstrated a much higher frequency of Eos⁺ cells within donor Tregs but similar ratios of Eos⁺ cells in host and donor Tregs in the spleen (Figures 4I and 4K). Additionally, we also observed a similarly higher frequency of Ccr6⁺ and Cxcr4⁺ cells within donor Tregs in the thymus (Figures 4J and 4K), suggesting that the majority of recirculating thymic Tregs were Eos⁺, Ccr6⁺, and Cxcr4⁺. We did not observe a higher ratio of Eos⁺ cells within donor Tregs compared to host Tregs in the spleen, most likely secondary to non-selective circulation of both Eos⁺ Tregs and Eos⁻ Tregs to the spleen.

It has recently been reported that recirculating Tregs were responsible for the repair of thymic injury in association with the following genes: Areg, Tff1, and Penk.²⁶ We induced thymic injury with sublethal total body irradiation (SL-TBI) as described (Figure S4A).²⁶ The irradiation caused a dramatic decrease of thymocyte cellularity, but the total number of Tregs was unchanged, consistent with an influx of recirculating Tregs from the periphery (Figure S4C). The frequency of Eos⁺ Tregs increased significantly post SL-TBI, raising the possibility that the majority of the recirculating Tregs were Eos⁺ (Figures S4B and S4C). In contrast to what we observed in a normal mouse, where the majority of recirculating thymic Tregs were both Ccr6⁺ and Cxcr4⁺, the majority of the recirculating Tregs in the injury model were Ccr6⁺Cxcr4⁻ (Figures S4B and S4C). This study is consistent with the preferential expression of Eos in recirculating Tregs in the injury model but raises the possibility that different chemotaxis pathways are involved in their migration to the thymus. We also analyzed the expression of the three Treg reparative-function related genes across the 5 groups of thymic Tregs identified in our scRNA-seq analysis (Figure S3D). In WT samples, the highest expression levels of the three genes were observed in the CD73⁺Ccr6⁺Cxcr4⁺ subset, which represented the recirculating thymic Tregs in mice at steady state. Comparison of the mRNA levels of the three genes in the CD73⁺Ccr6⁺Cxcr4⁺ cells from WT cells and Eos cKO cells demonstrated a reduction in Eos cKO cells (Figure S3E), implicating the relevance of Eos in the reparative function of recirculating thymic Tregs.

Eos is involved in the migration of recirculating thymic Tregs

To further confirm the relationship between Eos expression and the migration capacity of recirculating thymic Tregs, we first performed bulk RNA-seq of sorted Eos⁺ Tregs, Eos⁻ Tregs, and Foxp3⁻ CD4SP (Tconv) cells from the thymus of WT mice using Ccr6 as a surrogate marker for Eos⁺ Tregs. Principal component analysis revealed that all three cell populations were very different from each other (Figure 5A). Analysis of genes differentially expressed between Eos⁺ and Eos⁻ thymic Tregs confirmed the higher expression of Eos and Ccr6 in the Ccr6⁺ thymic Treg sample (Figure 5B). Pathway enrichment analysis comparing Eos⁺ Tregs to Eos⁻ Tregs and Eos⁺ Tregs to Foxp3⁻CD4⁺ Tconvs demonstrated selective enrichment of pathways related to GPCR signaling and down-regulation of pathways related to extracellular matrix in Eos⁺ thymic Tregs (Figure 5C). Both GPCR and extracellular matrix are important for T cell migration, suggesting a differential migration capacity of Eos⁺ thymic Tregs.²⁷ To further confirm the involvement of Eos in Treg migration, we purified Ccr6⁺ and Ccr6⁻ thymic Tregs from WT and Eos cKO mice and performed an in vitro migration assay in 5-μm Transwell plates with the presence of Ccl20 (the ligand of Ccr6) in the bottom chamber (Figure 5D). We observed higher migration of Ccr6⁺ cells than Ccr6⁻ cells, suggesting that the migration system was set up successfully. While Ccr6 was expressed on the cell surface of both WT and Eos cKO Tregs, Eos cKO Tregs migrated less efficiently than WT thymic Tregs, suggesting that Eos regulated other genes that are important for the migration of Ccr6⁺ thymic Tregs. To prove the relevance of Eos in the migration of recirculating thymic Tregs in vivo, we injected 6 wo heterozygous female mice with anti-CD45-BUV737 antibody intravenously (i.v.), resulting in the instantaneous labeling of all CD45⁺ cells in the bloodstream. After 24 h, we harvested the thymus and spleen and identified cells that had migrated from the bloodstream to the thymus using CD45 labeling (Figure 5E). Among the migrated cells, a comparison of the ratio of WT (Foxp3-RFP⁺) and Eos cKO (Foxp3-YFP⁺) Tregs demonstrated a reduction of Eos cKO Tregs in the thymus but no reduction in the spleen. This result suggests that the lack of Eos expression impaired the capacity of Tregs to migrate from the bloodstream back to the thymus, whereas the migration of Tregs from the bloodstream to the spleen was unaffected (Figures 5F and 5G).

Figure 5. Eos is involved in the migration of recirculating thymic Tregs.

(A) Principal component analysis (PCA) to compare bulk RNA-seq samples of sorted Eos⁺ Tregs, Eos⁻ Tregs, and Foxp3⁻ CD4SP cells from the thymus of WT mice.

(B) Volcano plot showing genes differently expressed comparing bulk RNA-seq samples of Eos⁺ thymic Tregs to Eos⁻ thymic Tregs.

(C) Pathway enrichment analysis to show pathways down-regulated or up-regulated by comparing bulk RNA-seq samples of Eos⁺ thymic Tregs, Eos⁻ thymic Tregs, and Foxp3⁻ CD4SP Tconvs.

(D) In vitro migration assay using Ccr6⁺ or Ccr6⁻ thymic Tregs sorted from WT or Eos cKO mice cultured in the upper chamber of a 5-μm Transwell plate with the presence of Ccl20 in the bottom chamber. After 20 h, the number of cells that migrated to the bottom chamber was determined by flow cytometry.

(E) Schematic representation of the procedure for in vivo migration study of WT and Eos cKO Tregs in 6 wo heterozygous females following i.v. injection of α-CD45-BUV737 antibody.

(F) Representative plots of i.v.-labeled CD45⁺ cells, Foxp3-RFP⁺ cells (WT Tregs), and Foxp3-YFP⁺ cells (Eos cKO Tregs) among CD4⁺ T cells or CD45⁺ CD4⁺ T cells from the thymus and spleen of 6 wo heterozygous females injected with α-CD45-BUV737 antibody.

(G) Quantifications of WT or Eos cKO Treg frequencies within CD45⁺ CD4 T cells from the thymus and spleen of 6 wo heterozygous females injected with α-CD45-BUV737 antibody.

p values are determined by one-way ANOVA (D) and Student’s t test (G). Mean ± SD. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Eos regulates the responsiveness of Tregs to IL-2 via direct binding to the CD25 locus

During our analysis of 6 wo heterozygous female mice, we noted that the expression of CD25 was reduced on Eos cKO Tregs. To further explore the role of Eos in the IL-2 signaling pathway, we analyzed the relationship between Eos expression and CD25 expression in Tregs in WT mice at steady state. Eos⁺ Tregs expressed higher levels of CD25 than Eos⁻ Tregs in both the thymus and spleen (Figures 6A and 6B). We compared the expression of CD25 on recirculating thymic Tregs in young adult WT and Eos KO mice and observed that the expression of CD25 on Ccr6⁺ recirculating thymic Tregs was reduced in Eos KO mice (Figure 6C). As phosphorylation of Stat5 (pStat5) is a crucial step in the IL-2 signaling pathway, we stimulated WT thymic Tregs in vitro with IL-2 for 20 min. Almost 100% of Eos⁺ thymic Tregs became pStat5⁺, whereas only ~60% of Eos⁻ Tregs became pStat5⁺ (Figure 6D).²⁸ We also examined whether deletion of Eos would reduce the ratio of pStat5⁺ cells in recirculating thymic Tregs. However, the anti-Ccr6 antibody did not work under the staining conditions designed for pStat5. We therefore compared the frequency of pStat5⁺CD73⁺ cells (approximately half of which are recirculating thymic Tregs) among thymic Tregs between WT and Eos KO mice and observed a reduction in Eos KO mice (Figure 6E).

Figure 6. Eos regulates the responsiveness of Tregs to IL-2 via direct binding to the CD25 locus.

(A) Representative plots of Eos and CD25 expression on Tregs from the thymus or spleen of 7–9 wo WT mice.

(B) Quantifications of CD25 MFI and frequencies of CD25⁺ cells within Eos⁺ and Eos⁻ Tregs from the thymus or spleen of 7–9 wo WT mice.

(C) Representative plots of CD25 and Ccr6 expression on thymic Tregs from 8 wo WT and Eos KO mice. Quantification of CD25 MFI of Ccr6⁺ thymic Tregs from those mice.

(D) Representative plots of pStat5 and Eos expression in thymic Tregs from 7 wo WT mice after stimulation with IL-2 for 20 min in vitro. Frequencies of pStat5⁺ cells within Eos⁺ and Eos⁻ thymic Tregs.

(E) Representative plots of pStat5 and CD73 expression on thymic Tregs from 7 to 9 wo WT or Eos KO mice. Frequencies of pStat5⁺ cells within CD73⁺ thymic Tregs from WT and Eos KO mice.

(F) ChIP-seq analysis of the binding of Eos and Helios at the CD25 gene locus in WT or Eos cKO splenic Tregs.

(G) ChIP-qPCR analysis of the binding of Eos and Helios at the CD25 gene locus in WT or Eos cKO splenic Tregs.

(H) Representative pathways identified by Eos ChIP-seq using WT or Eos cKO splenic Tregs.

(I) Representative Helios-targeted pathways that are also regulated by Eos from Helios ChIP-seq using WT or Eos cKO splenic Tregs.

p values are determined by Student’s t test (B–E). Mean ± SD. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

As the total number of Tregs in the thymus is low, we performed anti-Eos and anti-Helios ChIP-seq on purified splenic Tregs from WT and Eos cKO mice. We observed very clear binding of Eos to the CD25 locus in WT cells compared to Eos cKO cells. Additionally, the binding of Helios to the CD25 locus was also markedly decreased in Eos cKO Tregs, suggesting that Eos was also important for the binding of Helios to the CD25 locus (Figure 6F). We confirmed this finding with ChIP-qPCR (Figure 6G). We also observed that the Eos-Helios complex binding site at the CD25 locus was identical to that of Foxp3.²⁹ Considering that Eos, Helios, and Foxp3 can physically interact with each other, it is possible that Eos, Helios, and Foxp3 collaborate to regulate CD25 expression directly. With peak calling, we identified all genes directly targeted by Eos and genes targeted by Helios that also require the assistance of Eos (Table S1). GO (Gene Ontology) enrichment analysis of Eos and Helios ChIP-seq identified multiple pathways directly targeted by Eos and pathways regulated by both Helios and Eos (Table S2), which included pathways related to IL-2 response, T cell activation, and T cell migration (Figures 6H and 6I). Taking into consideration that Eos and Helios are highly colocalized in the nucleus of Tregs and that the two TFs may potentially form heterodimers, it remains likely that the two TFs cooperate with each other and play a major role in controlling CD25 expression and Treg responsiveness to IL-2 by direct binding to the CD25 locus.

Eos mediates the suppression of Treg precursors by recirculating thymic Tregs via IL-2 deprivation and MHC II depletion from thymic DCs

Previous studies have claimed that the major function of recirculating thymic Tregs was to suppress the development of thymic Tregs via IL-2 deprivation.^4,6 As our studies strongly suggest that Eos regulates the expression of CD25 on recirculating thymic Tregs, we re-examined Treg frequencies in young adult Eos KO mice (before onset of autoimmunity). Both CD73⁻ newly generated thymic Tregs and CD73⁺Ccr6⁺ recirculating thymic Tregs were increased in Eos KO mice (Figures 7A and 7B). This increase was not due to defects in T cell development, because Eos was minimally expressed during T cell development (Figures S5A and S5B), and the cellularity of thymocytes, frequencies of DN, DP, CD4SP, and CD8SP cells were all normal in Eos KO mice (Figure S5C). This increase in newly generated and recirculating Tregs was also not due to lack of Eos in thymic DCs and thymic epithelial cells (TECs), as Eos was not expressed by these two cell types (Figure S5D). Even with more recirculating thymic Tregs present in Eos KO mice, the generation of new Tregs was still increased, suggesting reduced suppression of Treg precursors by recirculating thymic Tregs in Eos KO mice. We also examined 8 wo Eos^fl/fl × CD4-Cre mice, which do not exhibit an autoimmune phenotype until 7–8 months of age.¹⁴ While we observed a higher frequency of CD73⁻ newly generated Tregs, thymocyte cellularity was already reduced in these mice, most likely secondary to stress induced by slowly developing autoimmunity (Figures S5E and S5F). In addition, these mice exhibited increased frequencies of DN cells and decreased frequencies of DP cells, a phenotype consistent with thymic stress (Figures S5E and S5F). The development of autoimmune manifestations in Eos^fl/fl × Foxp3-Cre mice occurred even faster, precluding a detailed analysis of thymic subpopulations.¹⁴

Figure 7. Eos mediates the suppression of Treg precursors by recirculating thymic Tregs via IL-2 deprivation and via MHC II depletion from thymic DCs.

(A) Representative plots of Foxp3 and CD73 expression in thymic CD4⁺CD8⁻ cells from 12 wo WT and Eos KO mice.

(B) Frequencies of CD73⁻Foxp3⁺ and CD73⁺Ccr6⁺ cells within thymic CD4⁺CD8⁻ cells or thymic Tregs. Cell numbers of CD73⁻ thymic Tregs and CD73⁺Ccr6⁺ thymic Tregs from 12 wo WT and Eos KO mice.

(C) Representative plots of Foxp3 and CD25 expression in cells generated from Foxp3⁻CD25^hiGITR^hi Treg precursors cultured under the conditions indicated.

(D) Frequencies of CD25⁺Foxp3⁺ Tregs within cells generated from cultured Foxp3⁻CD25^hiGITR^hi Treg precursors with low IL-2 concentration under the conditions indicated.

(E) Representative histograms of MHC II on DCs from the thymus of 8 wo WT or Eos KO mice. Quantifications of DC cell numbers and MHC II MFI.

(F) Quantifications of mTEC MHC II MFI from 9 to 10 wo WT or Eos KO female mice.

(G) Representative plots of MHC II and Ccr6 levels on thymic Tregs from CD45.1 WT or Eos KO mice co-cultured with α-MHC II-AF647 (clone KH74) antibody pre-stained thymic DCs from WT mice.

(H) Frequencies of MHC II^hi cells within WT, Eos KO, Ccr6⁺ WT, and Ccr6⁻ WT thymic Tregs. MHC II MFI of WT and Eos KO thymic Tregs co-cultured with α-MHC II-AF647 (clone KH74) antibody pre-stained thymic DCs.

(I) Representative plots of Foxp3 and CD25 expression of cells generated from Foxp3⁻CD25⁻CD4⁺CD8⁻ thymocytes cultured under the conditions indicated.

(J) Frequencies of Foxp3⁺CD25⁺ cells within cells generated from Foxp3⁻CD25⁻CD4⁺CD8⁻ thymocytes cultured under the indicated conditions.

Data are representative of at least two independent experiments. p values are determined by Student’s t test (B, E, F, and H) or one-way ANOVA (D and J). Mean ± SD. *p < 0.05; **p < 0.01; ***p < 0.

The decreased expression of CD25 on Eos KO Tregs raised the possibility that Eos might play an important role in the regulation of recirculating thymic Tregs to suppress Treg generation via IL-2 deprivation. We purified Foxp3⁻CD25^hiGITR^hi CD4⁺CD8⁻ Treg precursors, which have the potential to differentiate into CD25⁺Foxp3⁺ Tregs in response to IL-2 in vitro, and co-cultured them with WT or Eos KO recirculating thymic Tregs for 24 h.²² In the presence of low levels of IL-2 (0.1 U/mL), the addition of WT recirculating thymic Tregs reduced the generation of Tregs. In contrast, the addition of Eos KO recirculating thymic Tregs resulted in much less inhibition of Treg induction (Figures 7C and 7D). When higher levels of IL-2 (10 U/mL) were added to the co-cultures, similar levels of Treg induction were observed in all 3 groups (Figure 7C, lower panel). These observations strongly suggest that Eos is important in mediating IL-2 deprivation of Treg precursors by Eos⁺ recirculating thymic Tregs.

One additional mechanism of Treg suppressor function is depletion of MHC II-peptide complexes from DCs.^30–32 As recognition of MHC II on thymic DCs contributes to Treg generation, we speculated that recirculating thymic Tregs may also regulate Treg generation in the thymus by depleting MHC II-peptide complexes from thymic DCs.³³ The absolute numbers of thymic DCs were similar in WT and Eos KO mice, but the expression of MHC II on thymic DCs from Eos KO mice was much higher than that observed on DCs from WT mice (Figure 7E). While mTECs also play a crucial role in inducing Tregs, no significant differences in MHC II expression were observed between WT and Eos KO mice (Figure 7F).³³ To directly assess whether the differences in MHC II expression on thymic DCs were related to the presence of Eos⁺ thymic Tregs, we co-cultured WT thymic Tregs and Eos KO thymic Tregs with WT thymic DCs that had been pre-stained with an anti-MHC II antibody (clone KH74, labeled with AF647) that does not interfere with the interactions between MHC II and the TCR.³² WT thymic Tregs exhibited enhanced uptake of the pre-labeled MHC II compared to Eos KO thymic Tregs (Figures 7G and 7H). Furthermore, among WT thymic Tregs, Ccr6⁺ (Eos⁺) thymic Tregs captured higher levels of MHC II than Ccr6⁻ thymic Tregs, suggesting a critical role of Eos in mediating the depletion of MHC II on thymic DCs by recirculating thymic Tregs (Figures 7G and 7H).

Lastly, to determine if the differences in the levels of MHC II on thymic DCs between WT and Eos KO mice were of functional importance, we purified Foxp3⁻CD25⁻CD4⁺ thymocytes, which contained cells that could eventually differentiate into Tregs in the presence of thymic DCs and IL-2 (or IL-7).³³ These cells were then co-cultured with thymic DCs from WT or Eos KO mice in the presence of IL-2. Thymic DCs from Eos KO mice were more potent inducers of Tregs than thymic DCs from WT mice (Figures 7I and 7J). Taken together, Eos plays a critical role in the suppressor function of recirculating thymic Tregs on Treg precursors, both via IL-2 deprivation of Treg precursors and MHC II depletion from thymic DCs.

DISCUSSION

Previous studies have suggested that Eos potentially plays a critical role in multiple aspects of Treg function. RNA interference studies demonstrated that Eos cooperated with Foxp3 directly in silencing Treg signature genes.¹² Other studies demonstrated that at sites of inflammation, Tregs can rapidly reprogram into helper-like cells, losing Eos expression but retaining Foxp3 expression.¹³ More importantly, the conditional deletion of Eos in Tregs led to the onset of both systemic- and organ-specific autoimmunity at 3 months of age, confirming the critical role of Eos in Treg function.¹²

The generation of a monoclonal anti-Eos antibody allowed us to more thoroughly investigate the function of Eos in Tregs. Surprisingly, in young adult mice at steady state, only ~50% of Foxp3⁺ T cells in the periphery expressed Eos. Eos expression was confined to cells expressing Helios and was absent from RORγt⁺Foxp3⁺ T cells in the gastrointestinal tract, consistent with the view that Eos is selectively expressed in tTregs in vivo. Under the steady-state and activation conditions tested here, Eos expression was barely detected in Tconvs. However, our previous work suggests that Eos may play a role in Tconvs or non-T cells to prevent Eos KO mice from developing autoimmune disease until they reach 1 year of age. Recent reports indicate the involvement of Eos in the function of Th1 and Th2 cells, although we could not detect obvious defects in the function of major Th cell subsets lacking Eos.^34,35 Thus, the function of Eos in Tconvs might be highly context dependent and requires further investigation.

While the rapid development of autoimmune manifestations in Eos cKO mice prohibited the analysis of Eos’s role in the steady state, heterozygous female mice that carry one WT allele and one Foxp3-Cre allele allowed us to study the intrinsic function of Eos under normal conditions. The rapid disappearance of Eos cKO Tregs in these mice demonstrated a critical requirement for Eos in maintaining Treg fitness and survival. A similar defect in Treg cell fitness was observed in heterozygous female mice with Treg conditional deletion of Helios, but the loss of the deficient cells occurred much more slowly over a period of 2–5 months.³⁶ We did not evaluate whether the Eos-deficient cells converted into ex-Tregs using fate-mapping techniques, and this would be of interest in future studies. Eos expression is acquired in the periphery following egress from the thymus, and it is thus curious that Eos-deficient Tregs (at least by staining) survive normally in WT mice. While we cannot exclude the possibility that very low levels are sufficient to maintain Treg fitness, an alternative possibility is that Eos is induced in peripheral Tregs during their transition from the resting state to the activated state, as Eos⁺ Tregs in WT mice express higher levels of CD25 and eTreg-related markers. Expression of Eos is thus required to prevent activation-induced death of Tregs when they differentiate from resting to activated Tregs. Unfortunately, despite several attempts, we have been unable to generate an Eos reporter mouse strain, which would enable cell transfer studies to address this issue.

It has been known for some time that Tregs recirculate from the periphery to the thymus and that the Ccr6/Ccl20 and Cxcr4/Cxcl12 axes mediate this process. Our parabiosis experiments and scRNA-seq analysis of thymic Tregs have directly demonstrated that Eos⁺ thymic Tregs are recirculating thymic Tregs and are marked by co-expression of Ccr6 and Cxcr4. The generation of Rag-GFP mice allowed researchers to distinguish between newly generated GFP⁺ Tregs and older GFP⁻ Tregs. It was suggested that the GFP⁻ thymic Tregs in these mice were recirculating Tregs that had migrated from the periphery back to the thymus.⁴ We believe that this compartmentalization of thymic Tregs is overly simplistic. Therefore, we propose a better classification of thymic Treg subpopulations based on their actual gene expression profiles and phenotypic characteristics. Newly generated thymic Tregs can also be identified by the lack of CD73 expression. Recirculating thymic Tregs, as shown above, are characterized as CD73⁺Ccr6⁺Cxcr4⁺Eos⁺ and express a unique gene signature with a distinct TCR repertoire. The third population, characterized as CD73⁺Ccr6⁻Cxcr4⁻, is hypothesized to represent thymus-resident Tregs whose significance and function remain unknown. A significant presence of Eos⁺ recirculating thymic Tregs involved in the reparative process within the thymus was also observed following sublethal irradiation. Unlike recirculating thymic Tregs in mice at steady state, these cells expressed Ccr6 but not Cxcr4, presumably secondary to the production of different chemokines after thymic injury

The regulation of Treg production in the thymus is a tightly controlled process, as enhanced production can lead to immunosuppression while defective production can result in autoimmunity. Thymic development of tTregs is a “TCR-instructive” process, where TCR plays a critical role in directing T cells into the Foxp3⁺ lineage but is limited by a small niche that can be saturated at low clonal frequencies.^37,38 The second major factor modulating Treg generation is the availability of IL-2. Studies using thymic organ cultures or thymic slice cultures have suggested that recirculating thymic Tregs mediate a rheostat-like function by competing for IL-2. Our studies at steady state, both in vivo and in vitro, strongly suggest that the Eos⁺ Treg population, due to its elevated expression of CD25, is the major mediator of this competition. While the cellular source of IL-2 in the thymus has been controversial, recent studies have shown that CD4⁺CD25⁺Foxp3⁻ thymocytes, the same population that contains Treg precursors, are the primary source of IL-2 in the thymus.³⁹ It is likely that the IL-2 producers, the Treg precursors, and the Eos⁺ recirculating Tregs exist in close physical space in the thymic medulla, creating a unique microenvironment permitting highly specific production, utilization, and competition for IL-2. Our demonstration by ChIP-seq that a complex containing Eos, Helios, and Foxp3 binds to the CD25 locus in Tregs and the observation that Eos positively regulates the expression of CD25 further confirm the importance of Eos in regulating IL-2 availability for Treg precursors.

Lastly, we noted that Eos⁺ recirculating thymic Tregs have an enhanced ability to capture MHC II-peptide complexes from thymic DCs but not mTECs. It has been reported that GITR^hiPD-1^hiCD25^hi Tregs are more self-reactive.⁴⁰ We also observed higher expression of these markers on Eos⁺ Tregs. This may help explain why Eos⁺ recirculating thymic Tregs have a higher uptake of MHC II from thymic DCs. As high levels of MHC II on thymic DCs likely contribute to the generation of Tregs in the thymus, modulation of MHC II levels on thymic DCs may provide a further safeguard against overproduction/underproduction of Tregs in the thymus, which cooperates with the role of Eos⁺ Tregs in limiting intra-thymic availability of IL-2 to Treg precursors.

Taken together, the TF Eos appears to play multiple unique but interrelated roles in the function of Tregs. Eos regulates both (1) the thymic homing of recirculating thymic Tregs (to limit de novo thymic Treg generation) and (2) the functional maturation of peripheral Tregs. The autoimmune phenotype in Eos cKO mice arises from severe functional defects in peripheral Tregs, which in turn cause thymic injury and dampen overall T cell development. Even though Eos cKO mice have higher frequencies of newly generated Tregs at an early age, these Tregs are less suppressive due to the lack of Eos and thus insufficient to suppress the autoimmune phenotype in these mice. Due to the absence of an Eos reporter mouse strain, we were unable to further investigate the role of Eos in peripheral Tregs. However, all evidence points to the dual functions of Eos in both recirculating thymic Tregs and peripheral Tregs. Future studies examining the involvement of Eos in peripheral Tregs are therefore required. Eos appears to be absolutely essential for Treg survival in a competitive environment. We believe this is secondary to Eos-mediated enhanced expression of CD25 in cooperation with Helios and Foxp3. Enhanced binding of IL-2 by Eos⁺ Tregs in the periphery may also be needed for the transition of resting Tregs to the activated stage. Similarly, Eos-mediated enhanced expression of CD25 on recirculating thymic Tregs is critical for controlling the production of thymic Tregs, in addition to its effects in controlling the migration of Tregs to the thymus. Multiple effects of Eos in controlling Treg function may contribute to the enhanced capacity of Eos⁺ thymic Tregs to capture MHC II from thymic DCs and thereby provide an additional level of control on Treg development in the thymus.

Limitations of the study

Our study demonstrated that Eos plays a critical role in the suppressor function of recirculating thymic Tregs on Treg precursors through IL-2 deprivation of Treg precursors and MHC II depletion from thymic DCs. It remains unclear to what extent the two mechanisms work in vivo. Eos could potentially function in non-Treg cells and thus affect the generation of Tregs as well. We examined the major components in Treg induction and observed no defects in T cell development in Eos KO mice. Based on the phenotypes of Eos KO, Eos^fl/fl × CD4-Cre, and Eos^fl/fl × Foxp3-Cre mice, Eos probably plays a role in Tconvs and/or non-T cells. However, we could not detect Eos expression in Tconv cells and we did not observe defects in the ability of Tconvs to differentiate into Th subsets or to produce effector cytokines. It remains possible that Eos is expressed at very low levels in Tconvs and plays important roles in other aspects of Tconv function, which will only become apparent in long-term studies. Lastly, our failure to generate an Eos-reporter mouse strain limited our ability to directly examine the differential properties of Eos⁺ and Eos⁻ Tregs in the periphery.

RESOURCE AVAILABILITY

Lead contact

Requests for further information, resources, and reagents should be directed to and will be fulfilled by the lead contact, Ethan M Shevach (eshevach@niaid.nih.gov).

Materials availability

Anti-Eos 18H2 antibody and Eos^fl/fl mouse strain generated in this study will be available upon request.

Data and code availability

• scRNA-seq, ChIP-seq, and bulk RNA-seq data have been deposited at GEO (accession numbers: GSE305537, GSE305408, and GSE305409) and are publicly available as of the date of publication.

• Original western blot images have been deposited at Mendeley at https://doi.org/10.17632/4s8sxvpp4j.1 and are publicly available as of the date of publication. Microscopy data reported in this paper will be shared by the lead contact upon request.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Mouse strains and cell preparation

B6 (C57BL/6), CD4-Cre, Foxp3-Cre, Foxp3-RFP mice were purchased from The Jackson Laboratory. Female heterozygous Scurfy mice were purchased from The Jackson Laboratory and bred to B6 WT male mice to generate hemizygous male Scurfy offspring. Foxp3-GFP, TCRα KO, Rag1 KO, OT-II, OT-I, CD45.1, CD45.1/CD45.2 mice were obtained from Taconic Biosciences under the National Institute of Allergy and Infectious Diseases (NIAID) contract. We previously generated Helios^fl/fl × CD4-Cre mice.⁹ CD4-Cre × Foxp3-GFP mice were generated by crossing CD4-Cre mice to Foxp3-GFP mice. Eos KO mice were generated by deleting the last three exons of the gene.¹⁵ Eos KO × Foxp3-GFP mice were generated by crossing between Eos KO and Foxp3-GFP mice. CD45.1 Foxp3-GFP mice were generated by crossing Foxp3-GFP and CD45.1 mice. OT-II mice (Rag2^+/+) used in this study were generated by crossing between CD4-Cre × Foxp3-GFP and OT-II mice. Eos KO OT-II (Rag2^+/+) mice were then generated by crossing Eos KO mice to OT-II mice. Eos^fl/fl Mice were generated by Genomic Research Section, Research Technologies Branch, NIAID, NIH (Figure S6A). Eos^fl/fl × Foxp3-Cre mice were generated by crossing Eos^fl/fl mice to Foxp3-Cre mice. Eos^fl/fl × Foxp3-RFP mice were generated by crossing Eos^fl/fl mice to Foxp3-RFP mice. Then Eos^fl/fl × Foxp3-YFPCre/RFP mice were generated by crossing Eos^fl/fl × Foxp3-Cre mice to Eos^fl/fl × Foxp3-RFP mice. Eos^fl/fl × CD4-cre mice were generated by crossing Eos^fl/fl mice to CD4-Cre × Foxp3-GFP mice. All studies were approved by the NIAID Animal Care and Use Committee (Protocol LISB 15E and LISB 8E).

For the isolation of lymphocytes, thymi, spleens or LNs were first ground with cell strainer pestles (Stellar Scientific) and pushed through a 70 μm filter in RPMI 1640 (Thermo Fisher Scientific) with 10% heat-inactivated FBS (Sigma-Aldrich). Spleen samples were treated with ACK Lysis Buffer (gibco) to lyse red blood cells. For autoMACS (Miltenyi Biotec) separations, cells were resuspended in autoMACS Running Buffer (Miltenyi Biotec). For purification using a FACSAria IIIu sorter (BD), cells were maintained in Sorting Buffer (RPMI 1640 without phenol red supplemented with 10% heat-inactivated FBS and 10 mM HEPES). To isolate splenic DCs, spleens were digested in PBS with 50 μg/mL Liberase and 200 μg/mL DNase I for 30 min at 37°C and pushed through a 70 μm filter, after which red blood cells were lysed with ACK Lysis Buffer (gibco). To isolate thymic DCs, thymi were chopped into small pieces and digested in PBS with 200 μg/mL Liberase and 40 μg/mL DNase I for 30 min at 37°C. Then cells were dissociated by pipetting with pipette tips and 27G needles and filtration with 100 μm filters. In the end cells were wash with autoMACS Running Buffer to dissociate T cells and DCs. For both splenic and thymic samples, DCs were labeled with CD11c MicroBeads (Miltenyi Biotec) and purified by autoMACS (Miltenyi Biotec) according to the manufacturer’s instructions.

To isolate lymphocytes from the gut, large intestine was physically emptied, opened longitudinally, and cut into 2–3 cm pieces. Tissue pieces were incubated in RPMI 1640 (Thermo Fisher Scientific) containing 5 mM EDTA (Thermo Fisher Scientific) and 1mM DTT from NuPAGE Sample Reducing Agent (10x) (Thermo Fisher Scientific) for 20 min at 37°C with shaking, washed vigorously with PBS to remove remaining mucus, cut into even smaller pieces, and digested in RPMI 1640 (Thermo Fisher Scientific) containing 2.5 mg/mL Liberase (Millipore Sigma) and 40 μg/mL DNase I (Millipore Sigma) for 30 min at 37°C. The supernatant was filtered and the remaining tissue was smashed through a 70 μm filter. Red blood cells were lysed with ACK Lysis Buffer (gibco).

METHOD DETAILS

Antibodies and reagents

Antibodies and reagents used in this study are listed in key resources table.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-CD11c-Brilliant Violet 605™	BioLegend	Cat#: 117334; RRID:AB_2562415
Anti-CD25-PE-Cy7	Invitrogen	Cat#: 25-0251-82; RRID:AB_469608
Anti-CD25-PE	BD Biosciences	Cat#: 553866; RRID:AB_395101
Anti-CD25-Alexa Fluor™ 700	BioLegend	Cat#: 102024; RRID:AB_493709
Anti-CD28	BD Biosciences	Cat#: 553294; RRID:AB_394763
Anti-CD3	BD Biosciences	Cat#: 553057; RRID:AB_394590
Anti-CD4-BUV395	BD Biosciences	Cat#: 563790; RRID:AB_2738426
Anti-CD4-eFluor™ 450	Invitrogen	Cat#: 48-0042-82; RRID:AB_1272194
Anti-CD4-Alexa Fluor™ 700	BD Biosciences	Cat#: 557956; RRID:AB_396956
Anti-CD4-APC-eFluor™ 780	Invitrogen	Cat#: 47-0041-82; RRID:AB_11218896
Anti-Ccr6-BV421	BD Biosciences	Cat#: 564736; RRID:AB_2738926
Anti-Ccr6-BV510	BD Biosciences	Cat#: 747832; RRID:AB_2872295
Anti-Ccr6-BV711	BD Biosciences	Cat#: 740646; RRID:AB_2740335
Anti-CD44-Brilliant Violet 711™	BioLegend	Cat#: 103057; RRID:AB_2564214
Anti-CD44-Alexa Fluor™ 700	Invitrogen	Cat#: 56-0441-82; RRID:AB_494011
Anti-CD44-BV605	BD Biosciences	Cat#: 563058; RRID:AB_2737979
Anti-CD44-BUV737	BD Biosciences	Cat#: 612799; RRID:AB_2870126
Anti-CD44-BB700	BD Biosciences	Cat#: 566506; RRID:AB_2744396
Anti-CD45-PE-CF594	BD Biosciences	Cat#: 562420; RRID:AB_11154401
Anti-CD45-BUV737	BD Biosciences	Cat#: 568344; RRID:AB_3684199
Anti-CD45.1-BUV395	BD Biosciences	Cat#: 565212; RRID:AB_2722493
Anti-CD45.1-PE/Cyanine7	BioLegend	Cat#: 110729; RRID:AB_1134170
Anti-CD45.1-BV510	BD Biosciences	Cat#: 565278; RRID:AB_2739150
Anti-CD45.1-Brilliant Violet 650™	BioLegend	Cat#: 110736; RRID:AB_2562564
Anti-CD45.2-Alexa Fluor® 700	BioLegend	Cat#: 109822; RRID:AB_493731
Anti-CD45.2-PerCP/Cyanine5.5	BioLegend	Cat#: 109828; RRID:AB_893350
Anti-CD45.2-Brilliant Violet 785™	BioLegend	Cat#: 109839; RRID:AB_2562604
Anti-CD45.2-BUV395	BD Biosciences	Cat#: 564616; RRID:AB_2738867
Anti-CD45RB-PE	BD Bioscience	Cat#: 553101; RRID:AB_394627
Anti-CD62L-BUV737	BD Biosciences	Cat#: 612833; RRID:AB_2870155
Anti-CD62L-PE/Cyanine7	BioLegend	Cat#: 104418; RRID:AB_313103
Anti-CD73-Brilliant Violet 421™	BioLegend	Cat#: 127217; RRID:AB_2687251
Anti-CD73-Brilliant Violet 605™	BioLegend	Cat#: 127215; RRID:AB_2561528
Anti-CD73-PerCP/Cyanine5.5	BioLegend	Cat#: 127214; RRID:AB_11219403
Anti-CD8-BUV395	BD Biosciences	Cat#: 563786; RRID:AB_2732919
Anti-CD8-Alexa Fluor™ 700	Invitrogen	Cat#: 56-0081-82; RRID:AB_494005
Anti-CD8-APC	BD Biosciences	Cat#: 553035; RRID:AB_398527
Anti-CD8-APC-Cy™7	BD Biosciences	Cat#: 557654; RRID:AB_396769
Anti-Cxcr4-PE	BioLegend	Cat#: 146506; RRID:AB_2562783
Anti-Cxcr4-Brilliant Violet 421™	BioLegend	Cat#: 146511; RRID:AB_2562788
Anti-Eos 18H2	This study	Clone: 18H2
Anti-Eos 18H2-AF647	This study	Clone: 18H2
Anti-Ep-CAM-Brilliant Violet 711™	BioLegend	Cat#: 118233; RRID:AB_2632775
Anti-Foxp3-PE	BD Biosciences	Cat#: 560408; RRID:AB_1645251
Anti-Foxp3-PE-Cyanine5	Invitrogen	Cat#: 15-5773-82; RRID:AB_468806
Anti-Foxp3-eFluor™ 450	Invitrogen	Cat#: 48-5773-82; RRID:AB_1518812
Anti-Foxp3-Alexa Fluor™ 488	Invitrogen	Cat#: 53-5773-82; RRID:AB_763537
Anti-Foxp3-Alexa Fluor® 594	Novus Biologicals	Cat#: NB100-39002AF594; RRID:AB_3157022
Anti-Foxp3-Alexa Fluor® 700	BioLegend	Cat#: 126421; RRID:AB_2750492
Anti-Foxp3 FJK-16s	Invitrogen	Cat#: 14-5773-82; RRID:AB_467576
Anti-Foxp3 polyclonal	Invitrogen	Cat#: PA1-46126; RRID:AB_2278500
Anti-GITR-PE	Invitrogen	Cat#: 12-5874-82; RRID:AB_465986
Anti-Helios-PE	BioLegend	Cat#: 137216; RRID:AB_10660749
Anti-Helios-PE-eFluor™ 610	Invitrogen	Cat#: 61-9883-42; RRID:AB_2574682
Anti-Helios-Alexa Fluor® 594	Novus Biologicals	Cat#: NBP2-37723AF594; RRID:AB_3296989
Anti-Helios 22F6	BioLegend	Cat#: 137202; RRID:AB_10900638
Anti-Helios D8W4X	Cell Signaling	Cat#: 42427S; RRID:AB_2799221
Anti-MHC II-PE-Cyanine5	Invitrogen	Cat#: 15-5321-82; RRID:AB_468800
Anti-MHC II-eFluor™ 450	Invitrogen	Cat#: 48-5321-82; RRID:AB_1272204
Anti-MHC II KH74-AF647	BioLegend	Cat#: 115310; RRID:AB_492826
Anti-OX-40-BV711	BioLegend	Cat#: 119421; RRID:AB_2687176
Anti-PD-1-BV421	BD Biosciences	Cat#: 748268; RRID:AB_2872696
Anti-IFN-γ	Bio X cell	Cat#: BE0055; RRID:AB_1107694
Anti-IFN-γ-PE-CF594	BD Biosciences	Cat#: 562303; RRID:AB_11153140
Anti-IL-12	Bio X cell	Cat#: BE0233; RRID:AB_2687715
Anti-IL-2	Bio X cell	Cat#: BE0043; RRID:AB_1107702
Anti-IL-13-PE-Cy7	Invitrogen	Cat#: 25-7133-82; RRID:AB_2573530
Anti-IL-17-BUV395	BD Biosciences	Cat#: 565246; RRID:AB_2722575
Anti-IL-4	Bio X cell	Cat#: BE0045; RRID:AB_1107707
Anti-IL-4-PE	BioLegend	Cat#: 504104; RRID:AB_315318
Anti-pStat5-PE	Invitrogen	Cat#: 12-9010-42; RRID:AB_2572671
Anti-CD16/32	BD Biosciences	Cat#: 553142; RRID:AB_394657
Chemicals, peptides, and recombinant proteins
Ulex Europaeus Agglutinin I (UEA I)-DyLight® 649	Vector Laboratories	Cat#: DL-1068
Ulex Europaeus Agglutinin I (UEA I)-DyLight® 594	Vector Laboratories	Cat#: DL-1067
Albumin from chicken egg white (OVA)	Millipore Sigma	Cat#: A5378-25G
OVA 323-339	InvivoGen	Cat#: vac-isq
Freund′s Adjuvant, Complete	Millipore Sigma	Cat#: F5881-6X10ML
A. Alternata extract	Greer Laboratories	Cat#: XPM1D3A2.5
RPMI 1640	Thermo Fisher Scientific	Cat#: 21870092
RPMI 1640 without phenol red	Thermo Fisher Scientific	Cat#: 32404014
Iscove’s Modified Dulbecco’s Medium (IMDM)	ATCC	Cat#: 30-2005
Fetal Bovine Serum	Sigma-Aldrich	Cat#: 23E082
Fetal Bovine Serum	R&D Systems	Cat#: S11550
Penicillin-Streptomycin Solution	Fisher Scientific	Cat#: MT30002CI
2-mercaptoethanol	Sigma-Aldrich	Cat#: M3148
Sodium Pyruvate	Lonza	Cat#: 13-115E
L-glutamine	Fisher Scientific	Cat#: MT25005CI
NEAA (non-essential amino acids)	Fisher Scientific	Cat#: 11-140-076
2-mercaptoethanol	Millipore Sigma	Cat#: M3148
HEPES	Thermo Fisher Scientific	Cat#: 15630130
PBS, pH 7.4	Thermo Fisher Scientific	Cat#: 10010049
EDTA (0.5 M), pH 8.0, RNase-free	Thermo Fisher Scientific	Cat#: AM9260G
autoMACS Running Buffer	Miltenyi Biotec	Cat#: 130-091-221
autoMACS Washing Solution	Miltenyi Biotec	Cat#: 130-092-987
Recombinant human IL-2	BRB Preclinical Biologics Repository	N/A
Recombinant Mouse IL-1b	R&D Systems	Cat#: 401-ML
Recombinant Mouse IL-4	R&D Systems	Cat#: 404-ML
Recombinant Mouse IL-6	PeproTech	Cat#: 216-16
Recombinant Mouse IL-12	PeproTech	Cat#: 210-12
Recombinant Mouse TGF-β	BioLegend	Cat#: 763104
Recombinant Mouse Ccl20	BioLegend	Cat#: 582304
Dimethyl sulfoxide	Sigma-Aldrich	Cat#: D2650
Paraformaldehyde 16% Solution, EM Grade	Electron Microscopy Sciences	Cat#: 15710
Formaldehyde solution	Millipore Sigma	Cat#: F8775
Glycine	Millipore Sigma	Cat#: G8898
TRIS-buffered saline (TBS, 10X) pH 7.4	Thermo Fisher Scientific	Cat#: J60764.K2
TRIzol™ Reagent	Thermo Fisher Scientific	Cat#: 15596026
Liberase™ TL Research Grade	Millipore Sigma	Cat#: 5401020001
DNase I	Millipore Sigma	Cat#: 11284932001
DAPI ready made solution	Millipore Sigma	Cat#: MBD0015-1ML
ProLong™ Gold Antifade Mountant	Thermo Fisher Scientific	Cat#: P36934
Sucrose, Macron Fine Chemicals™	Avantor	Cat#: 8360-03
Triton™ X-100	Fisher Scientific	Cat#: BP151-1008
Bovine Serum Albumin (BSA)	GeminiBio	Cat#: 700-100P
ACK Lysis Buffer	gibco	Cat#: A10492-01
Critical commercial assays
Pierce™ MS-Compatible Magnetic IP Kits	Thermo Fisher Scientific	Cat#: 90409
RNeasy Plus Micro Kit	QIAGEN	Cat#: 74034
Halt™ Protease and Phosphatase Inhibitor Cocktail (100X)	ThermoFisher Scientific	Cat#: 78442
ChIP DNA Clean & Concentrator	Zymo Research	Cat#: D5205
NEBNext Ultra II End repair/dA-Tailing Module	New England Biolabs	Cat#: NEB #E7546S/L
NEBNext Ultra II Ligation Module	New England Biolabs	Cat#: NEB #E7595S/L
Fixable Viability Stain 440UV	BD Biosciences	Cat#: 566332
Foxp3/Transcription Factor Staining Buffer Set	Thermo Fisher Scientific	Cat#: 00-5523-00
BD Cytofix Fixation Buffer	BD Biosciences	Cat#: 554655
BD Phosflow™ Perm Buffer III	BD Biosciences	Cat#: 558050
Cell Stimulation Cocktail (plus protein transport inhibitors) (500X)	Thermo Fisher Scientific	Cat#: 00-4975-03
NuPAGE™ 4 to 12%, Bis-Tris, 1.0-1.5 mm, Mini Protein Gels	Thermo Fisher Scientific	Cat#: NP0335PK2
NuPAGE™ LDS Sample Buffer (4X)	Thermo Fisher Scientific	Cat#: NP0007
NuPAGE™ Sample Reducing Agent (10X)	Thermo Fisher Scientific	Cat#: NP0009
NuPAGE™ MES SDS Running Buffer (20X)	Thermo Fisher Scientific	Cat#: NP0002
NuPAGE™ Transfer Buffer (20X)	Thermo Fisher Scientific	Cat#: NP00061
SuperBlock™ (TBS) Blocking Buffer	Thermo Fisher Scientific	Cat#: 37535
SuperSignal™ West Atto Ultimate Sensitivity Substrate	Thermo Fisher Scientific	Cat#: A38555
SuperSignal™ West Pico PLUS Chemiluminescent Substrate	Thermo Fisher Scientific	Cat#: 34577
PVDF/Filter Paper Sandwiches, 0.45 μm (for mini gels)	Thermo Fisher Scientific	Cat#: LC2005
CD8a (Ly-2) MicroBeads, mouse	Miltenyi Biotec	Cat#: 130-117-044
CD4+CD25+ Regulatory T cell Isolation Kit, mouse	Miltenyi Biotec	Cat#: 130-091-041
Naive CD4 T cell Isolation Kit, mouse	Miltenyi Biotec	Cat#: 130-104-453
CD4 (L3T4) MicroBeads, mouse	Miltenyi Biotec	Cat#: 130-117-043
CD4+ T cell Isolation Kit, mouse	Miltenyi Biotec	Cat#: 130-104-454
CD11c MicroBeads UltraPure, mouse	Miltenyi Biotec	Cat#: 130-125-835
Anti-PE MicroBeads	Miltenyi Biotec	Cat#: 130-048-801
NuPAGE™ Antioxidant	Thermo Fisher Scientific	Cat#: NP0005
Precision Count Beads™	BioLegend	Cat#: 424902
Transwell™ −96 Well Permeable Support System 5μm	fisher scientific	Cat#: 09-761-83
Deposited data
Single cell RNA Sequencing data	This study	GEO: GSE305537
ChIP-seq data	This study	GEO: GSE305408
Bulk RNA-seq data	This study	GEO: GSE305409
Experimental models: Organisms/strains
B6 (C57BL/6)	The Jackson Laboratory	RRID:IMSR_JAX:000664
Scurfy (B6.Cg-Foxp3sf/J)	The Jackson Laboratory	RRID:IMSR_JAX:004088
CD4-Cre	The Jackson Laboratory	RRID:IMSR_JAX:022071
Foxp3-Cre	The Jackson Laboratory	RRID:IMSR_JAX:016959
Foxp3-RFP	The Jackson Laboratory	RRID:IMSR_JAX:008374
Foxp3-GFP	Taconic Biosciences	Line: 382
TCRα KO	Taconic Biosciences	Line: 98
Rag1 KO	Taconic Biosciences	Line: 165
OT-II	Taconic Biosciences	Line: 4234
OT-I	Taconic Biosciences	Line: 300
CD45.1	Taconic Biosciences	Line: 7
CD45.1/CD45.2	Taconic Biosciences	Line: 8422
CD4-Cre × Foxp3-GFP	This study	N/A
CD45.1 Foxp3-GFP	This study	N/A
Eos KO	Rieder et al.15	N/A
Heliosfl/fl × CD4-Cre	Thornton et al.9	N/A
Eos KO × Foxp3-GFP	This study	N/A
OT-II (Rag2+/+)	This study	N/A
Eos KO OT-II	This study	N/A
Eosfl/fl	This study	N/A
Eosfl/fl × Foxp3-Cre	This study	N/A
Eosfl/fl × Foxp3-RFP	This study	N/A
Eosfl/fl × Foxp3-YFPCre/RFP	This study	N/A
Eosfl/fl × CD4-Cre	This study	N/A
Oligonucleotides
Gapdh qPCR probe	Thermo Fisher Scientific	Mm99999915_g1
CD25 qPCR primer forward	IDT	5′-CCACAACAGACATGCAGAAGCC-3′
CD25 qPCR primer reverse	IDT	5′-GCAGGACCTCTCTGTAGAGCCTTG-3′
Software and algorithms
FlowJo (10.9.0)	Flowjo	https://www.flowjo.com
Prism (9.5.1; 10.4.1)	GraphPad Software	https://www.graphpad.com/scientific-software/prism
Fiji (2.9.0/1.53t)	Open source image processing software	http://imagej.net/Contributors
Adobe Illustrator (28.7.3)	Adobe	https://www.adobe.com/products/illustrator
Adobe Fresco (6.2.0)	Adobe	https://www.adobe.com/products/fresco
Procreate (5.3.14)	Savage Interactive Pty Ltd	https://procreate.com
Other
autoMACS Pro Separator	Miltenyi Biotec	N/A
FACSAria IIIu sorter	BD Biosciences	N/A
FACSymphony A3	BD Biosciences	N/A
LSRFortessa	BD Biosciences	N/A
XCell II™ Blot Module	Thermo Fisher Scientific	Cat#: EI9051
Digital Sonicator	Diagenode	N/A

Flow cytometry

Cells were first incubated with anti-CD16/32 antibody (BD Biosciences) then stained with Live/Dead cell viability dye (BD Biosciences), and antibodies for surface markers. Cells were subsequently fixed and permeabilized with a Foxp3/Transcription Factor Staining Buffer Set (Thermo Fisher Scientific) according to the manufacturer’s instructions. Then cells were stained with antibodies for intracellular proteins. For intracellular cytokine staining, cells were incubated with Cell Stimulation Cocktail (plus protein transport inhibitors) (Thermo Fisher Scientific) for 4–5 h before staining. After surface staining, cells were fixed with 2% PFA for 15 min at RT before proceeding with fixation and permeabilization using Foxp3/Transcription Factor Staining Buffer Set. Then cells were stained with antibodies for cytokines and intracellular proteins. For pStat5 staining, cells were first stimulated with IL-2 (2 U/mL) for 20 min at 37°C. Stimulated cells were then subjected to BD Cytofix Fixation Buffer (BD Biosciences) and BD Phosflow Perm Buffer III (BD Biosciences) according to the manufacturer’s instructions. Finally, cells were stained with antibodies for pStat5 and other marker proteins. Data were acquired through a FACSymphony A3 or LSRFortessa (BD Biosciences) and analyzed with FlowJo software (FlowJo).

Immunofluorescence microcopy

For imaging of Tregs, single cell suspensions were prepared from spleen or thymus of WT mice. CD4 T cells were enriched using CD4 (L3T4) MicroBeads, mouse (Miltenyi Biotec) or Tregs were isolated by FACS sorting. Purified cells were fixed and permeabilized with a Foxp3/Transcription Factor Staining Buffer Set (Thermo Fisher Scientific), blocked with α-CD16/32 antibody and stained with DAPI (Millipore Sigma) and antibodies for Helios, Eos and Foxp3 in 1x Perm Buffer (Thermo Fisher Scientific). In the end, cells in a small volume of suspension were transferred to a microslide and mounted with Prolong Gold Antifade Mountant (Thermo Fisher Scientific). Confocal imaging was performed using a Leica SP-8 inverted microscope (Leica Microsystems). Images were processed with Fiji (Open source image processing software).

Co-immunoprecipitation (Co-IP)

For Co-IP using primary mouse cells, Tregs were purified using CD4⁺CD25⁺ Regulatory T cell Isolation Kit (Miltenyi Biotec). Co-IP was performed with 5 μg anti-Eos 18H2 (this study), anti-Helios 22F6 (BioLegend) and anti-Foxp3 (Invitrogen) antibodies using Pierce MS-Compatible Magnetic IP Kit (Thermo Fisher Scientific) according to manufacturer’s instructions. Protein samples from whole cell lysates (Input) or Co-IP were analyzed by Immunoblot analysis.

Immunoblot (IB)

Input and Co-IP samples were boiled at 70°C for 10 min in 1x NuPAGE LDS Sample Buffer (Thermo Fisher Scientific) and separated using NuPAGE 4 to 12% Bis-Tris protein gels (Thermo Fisher Scientific) under reduced conditions and then transferred to a PVDF membrane (Thermo Fisher Scientific) using XCell II Blot Module (Thermo Fisher Scientific) under reduced conditions. The membranes were blocked with SuperBlock (TBS) Blocking Buffer (Thermo Fisher Scientific) and probed with the following primary antibodies: anti-Helios D8W4X (Cell Signaling) and anti-Foxp3 FJK-16s (Invitrogen).

In vitro T cell differentiation

Naive CD4 T cells were purified from the spleens of CD45.1 or Eos KO mice using the Naive CD4 T cell Isolation Kit (Miltenyi Biotec) according to the manufacturer’s instruction and mixed at a 1:1 ratio. Enriched cells were cultured for 3 days under Th1 polarizing conditions (α-CD3 1 μg/mL, α-IL-4 10 μg/mL, IL-12 10 ng/mL, IL-2 50 U/mL), Th2 polarizing conditions (α-CD3 1 μg/mL, α-IFN-γ 10 μg/mL, α-IL-12 10 μg/mL, IL-4 5000 U/mL, IL-2 100 U/mL), Th17 polarizing conditions (α-CD3 1 μg/mL, α-IL-4 10 μg/mL, α-IFN-γ 10 μg/mL, α-IL-12 10 μg/mL, TGF-β 1 ng/mL, IL-6 10 ng/mL, IL-1b 10 ng/mL) and iTreg conditions (α-CD3 1 μg/mL, α-IL-4 10 μg/mL, α-IFN-γ 10 μg/mL, α-IL-12 10 μg/mL, TGF-β 5 ng/mL, IL-2 100 U/mL). DCs were isolated from CD45.1/CD45.2 mice as described above and added to the culture at 1/5 the concentration of CD4 T cells. Then cells were rested in cRPMI with IL-2 (no IL-2 for Th17 cells) for 2 days. For cytokine expression analysis, differentiated cells were stimulated with Cell Stimulation Cocktail (plus protein transport inhibitors) (Thermo Fisher Scientific) for 4–5 h.

In vitro migration assay

To test the importance of Eos in regulating the migration of recirculating thymic Tregs, Ccr6⁺ (CD4⁺Foxp3⁺CD73⁺Ccr6⁺) or Ccr6⁻ (CD4⁺Foxp3⁺CD73⁻Ccr6⁻) thymic Tregs were sorted from WT or Eos cKO (Eos^fl/fl × CD4-Cre Foxp3-GFP) mice. Cells were resuspended in cRPMI at a concentration of 5 × 10⁵ cells/mL and 100 μL/well cells were plated in the apical chambers of 96-well Transwell (5-μm, fisher scientific). 150 μL/well cRPMI containing 200 ng/mL Ccl20 (BioLegend) was added to the bottom chambers. 20 h post culture, cells from the bottom chamber were collected and mixed with Precision Count Beads (BioLegend). Cells were then stained with surface markers in PBS for 20 min at RT and analyzed using flow cytometry.

In vitro Treg precursor suppression assay

Foxp3⁻CD25^hiGITR^hi CD4SP Treg precursors were FACS-sorted from thymus of CD45.1 Foxp3-GFP mice. CD73⁺Ccr6⁺ recirculating thymic Tregs were FACS-sorted from CD4-Cre × Foxp3-GFP or Eos KO mice. In 96-well plate, 1 × 10⁴ Treg precursors and 1 × 10⁵ recirculating thymic Tregs were co-cultured with IL-2 for 24 h. Thymic Treg suppression was assessed by analyzing Treg induction from Treg precursors using flow cytometry.

In vitro thymic Treg induction

Thymic DCs were isolated from B6 or Eos KO mice as described above. Foxp3⁻CD25⁻ thymic CD4SP cells (containing Treg precursors) were FACS-sorted from CD45.1 Foxp3-GFP mice. In a round-bottom 96-well plate, 5 × 10³ thymic DCs and 1 × 10⁴ sorted Foxp3⁻CD25⁻ thymic CD4SP cells were cultured together with 10 U/mL IL-2 for 4 days. Thymic Treg induction was assessed by staining for CD45.1, CD4, CD25, and Foxp3.

In vitro MHC II capture assay

Thymocytes were isolated from thymus of CD45.1 WT mice or Eos KO mice. Then CD8⁺ cells were depleted using CD8a (Ly-2) MicroBeads (Miltenyi Biotec) and CD25⁺ cells were enriched using α-CD25-PE antibody (BD Biosciences) and Anti-PE MicroBeads (Miltenyi Biotec). Thymic DCs were isolated from B6 mice and pre-stained with α-MHC II KH74-AF647 antibody (BioLegend) at 4°C for 20 min. Purified CD45.1 WT thymocytes, Eos KO thymocytes and pre-stained DCs were co-cultured for 18 h at 37°C at a 1:1:1 ratio. Then thymocytes and DCs were dissociated by washing with autoMACS Running Buffer containing 2mM EDTA. Cell suspensions were then stained with antibodies for CD45.1, CD45.2, CD4, CD8, CD25, CD11c, CD73, Ccr6 and Foxp3 and analyzed by flow cytometry.

Oral tolerance model

Naive CD4⁺Foxp3⁻CD44^loCD45RB^hi T cells were FACS-sorted from the spleens of WT OT-II or Eos KO OT-II mice and 1 × 10⁶ cells were injected retro-orbitally into CD45.1/CD45.2 recipients. Then mice were given 1% OVA (Millipore Sigma) supplemented water for 7 days, changing the water every 2 days.

IBD model

Naive CD4 T cells were purified from the spleens of Foxp3-Cre or Eos^fl/fl × Foxp3-Cre mice using the Naive CD4 T cell Isolation Kit (Miltenyi Biotec) according to the manufacturer’s instruction. 1 × 10⁶ or 2 × 10⁶ cells were transferred by retro-orbital injection into TCRα KO or Rag1 KO recipient mice. In some experiments, naive CD4 T cells were purified from B6 or Eos KO mice. Mice were monitored for weight loss and analyzed after 2–4 weeks.

Parabiosis

Female B6 and CD45.1 mice underwent surgery to establish parabiosis, pairs of mice expressing either CD45.1 or CD45.2 were cohoused for 1 month. To distinguish host and donor derived cells, the congenic markers CD45.1 and CD45.2 were used by flow cytometry.

SL-TBI

To induce thymic damage, mice were given 550 centi-gray (cGy) sublethal total body irradiation.

In vivo Th2 model

WT or Eos^fl/fl × CD4-Cre mice were challenged i.n. with A. Alternata extract on day 1 and day 7. On day 14, lungs were harvested, and cells were analyzed with flow cytometry.

In vivo Th1/Th17 model

Naive CD4⁺Foxp3⁻CD44^loCD45RB^hi T cells were FACS-sorted from the spleens of WT OT-II or Eos KO OT-II mice and 2 × 10⁶ cells were injected retro-orbitally into CD45.1/CD45.2 recipients. The following day, recipient mice were immunized s.c. with 50 μg of OVA_323–339 peptide in CFA to each flank. Draining LNs (dLNs) were harvested on day 7 post immunization and cells were analyzed with flow cytometry.

ChIP

Tregs were purified using CD4⁺CD25⁺ Regulatory T cell Isolation Kit (Miltenyi Biotec) from the spleens of WT or Eos^fl/fl × Foxp3-Cre (Eos cKO) mice and fixed for 10 min at room temperature with 1% Formaldehyde (Millipore Sigma). Cross-linking reaction was stopped by adding 125 mM Glycine (Millipore Sigma). Cells were washed twice in ice-cold PBS and cell pellets were flash-frozen and stored at −80°C. Cross-linked DNA was fragmentated by sonication using Digital Sonicator (Diagenode). The lysates were incubated overnight at 4°C with 50 μL Protein A/G magnetic beads (Thermo Fisher Scientific) that had been pre-incubated with 6 μg appropriate antibodies. Samples were washed, eluted, reverse cross-linked at 65°C for 24 h, and purified using ChIP DNA Clean & Concentrator (Zymo Research). For each ChIP analysis, 5 × 10⁶ – 10 × 10⁶ cells were used. Antibodies used were α-Eos 18H2 (this study) and α-Helios D8W4X (Cell Signaling).

For ChIP-seq, library was prepared using NEBNext Ultra II End repair/dA-Tailing Module (New England Biolabs) and NEBNext Ultra II Ligation Module (NEB#E7595S/L) according to manufacturer’s instructions. All samples were pooled and sequenced on NextSeq 2000 P2 using paired-end sequencing. Reads were processed using chrom-seek (1.2.0) pipeline (https://github.com/OpenOmics/chrom-seek). Reads were trimmed with Cutadapt v4.4 and then aligned to mm10 GRCm38.p6 using BWA v0.7.1. All reads aligning to the mm10 v2 blacklist regions were identified and removed with Picard SamToFastq. Reads with a mapQ score less than 6 were removed with SAMtools v1.17 and PCR duplicates were removed with Picard MarkDuplicates. Data was converted into bigwigs for viewing and normalized by reads per genomic content (RPGC) using deepTools v3.5.1. Coverage plots were created using karyoploteR and GenomicRanges. Narrow peaks were called using macs v 2.2.7.1 treating matching Eos cKO samples as controls. Peaks were annotated using Uropa v4.0.2 and the Gencode reference M18. Specifically, peaks were associated with the nearest protein coding gene’s TSS. Pathway enrichment analysis was accomplished using ShinyGO v0.82. This work utilized the computational resources of the NIH HPC Biowulf cluster.

For ChIP-qPCR, enrichment of binding to CD25 locus was analyzed using qPCR with probes or primers in key resources table. Reads were normalized to Gaphd and the sample of Eos ChIP using Eos cKO Tregs.

scRNA-seq

scRNA-seq was performed on FACS-sorted WT (RFP⁺) or Eos cKO (YFP⁺) Tregs from the thymus of 6 wo Eos^fl/fl × Foxp3-YFPCre/RFP mice. Cells were sorted into RPMI (without phenol red) (Thermo Fisher Scientific) medium supplemented with 10% heat-inactivated FBS (R&D Systems) and 10 mM HEPES (Thermo Fisher Scientific). Subsequently cells were washed with PBS supplemented with 10% heat-inactivated FBS (R&D Systems). At the Genomic Research Section (RTB, NIAID, NIH), libraries were prepared using the 10X Genomics Chromium GEM-X Universal 5′ Gene Expression v3 workflow for gene expression, cell hashing, and full-length V(D)J T cell receptor sequencing. Sequencing was performed on an Illumina NovaSeq X Plus and CellRanger 8.0.1 was used to process the sequencing data. The data was initially processed using Seurat 5.2.1 in combination with scRepetiore v2.3.2 to integrate the gene expression and V(D)J data, using the standard pipeline. Samples were integrated based on the gene expression data using Harmony v1.2.0. Treg type was identified based on the expression of known Treg markers. Differentially expressed markers between conditions were identified using a pseudobulk analysis and DeSeq2 within Seurat. scRepetiore was then used to identify differences in surface marker complement and diversity between groups.

Bulk RNA-seq

For Bulk RNA-seq analysis, Eos⁺ Tregs (CD4⁺Foxp3⁺Ccr6⁺Cxcr4⁺), Eos⁻ Tregs (CD4⁺Foxp3⁺Ccr6⁻Cxcr4⁻) and Tconvs (CD4⁺Foxp3⁻) were sorted from thymus of CD4-Cre × Foxp3-GFP mice. RNA was extracted from sorted cells using TRIzol (Thermo Fisher Scientific) and RNeasy Plus Micro Kit (QIAGEN) according to the manufacturer’s instruction. RNA-sequencing was performed at the Sequencing Facility, Illumina (CCR) (NCI, NIH). The SMARTer Ultra Low Input RNA Kit was used to generate high-quality cDNA directly from 1 ng of Total RNA using their patented SMART (Switching Mechanism at 5′ End of RNA Template) technology. The cDNA was then made into a sequencing ready library using the Nextera XT DNA Sample Preparation Kit from Illumina. Samples were sequenced on NextSeq 2000 P2 using paired-end sequencing. The instrument run was processed by using Real Time Analysis software (RTA 3.10,30). The Illumina bcl2fastq2.20 was used to demultiplex and convert binary base calls and quality scores to fastq format. The sequencing reads were trimmed of adapters and low-quality bases using Cutadapt (version 1.18). The trimmed reads were mapped to mouse reference genome (mm10) and Gencode annotation M21 using STAR aligner (version 2.7.0f) with two-pass alignment option. RSEM (version 1.3.1) was used for gene and transcript quantification based on GENCODE annotation file. Principal component analysis was performed using iDEP (2.0). Differential gene expression analyses were performed using the DESeq2 package. Pathway enrichment analysis was performed using GSEA enrichment analysis.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analyses were calculated with Prism software. Normal distribution was assumed a priori for all samples. p values of less than 0.05 were considered significant (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001).

Supplementary Material

Supplemental information can be found online at https://doi.org/10.1016/j.celrep.2025.116838.

Highlights.

• Eos is expressed preferentially in activated, thymus-derived Foxp3⁺Helios⁺ Tregs

• Eos forms complexes with Helios and Foxp3 and directly binds the CD25 locus

• Eos is critical for the survival, activation, migration, and suppression of Tregs

• Eos⁺ Tregs are the major recirculating thymic Tregs and suppress Treg precursors