Temporal and context-dependent requirements for the transcription factor Foxp3 expression in regulatory T cells

Hu, Wei; Dolsten, Gabriel A.; Wang, Eric Y.; Beroshvili, Giorgi; Wang, Zhong-Min; Ghelani, Aazam P.; Uhl, Lion F. K.; Bou-Puerto, Regina; Huang, Xiao; Michaels, Anthony J.; Hoyos, Beatrice E.; Jin, Wenjie; Pritykin, Yuri; Rudensky, Alexander Y.

doi:10.1038/s41590-025-02295-4

Download PDF

Article
Open access
Published: 08 October 2025

Temporal and context-dependent requirements for the transcription factor Foxp3 expression in regulatory T cells

Nature Immunology volume 26, pages 2059–2073 (2025) Cite this article

30k Accesses
21 Citations
26 Altmetric
Metrics details

Subjects

Abstract

Regulatory T (T_reg) cells, expressing the transcription factor Foxp3, are obligatory gatekeepers of immune responsiveness, yet the mechanisms by which Foxp3 governs the T_reg transcriptional network remain incompletely understood. Using a novel chemogenetic system of inducible Foxp3 protein degradation in vivo, we found that while Foxp3 was indispensable for the establishment of transcriptional and functional programs of newly generated T_reg cells, Foxp3 loss in mature T_reg cells resulted in minimal functional and transcriptional changes under steady state. This resilience of the Foxp3-dependent program in mature T_reg cells was acquired over an unexpectedly long timescale; however, in settings of severe inflammation, Foxp3 loss led to a pronounced perturbation of T_reg cell transcriptome and fitness. Furthermore, tumoral T_reg cells were uniquely sensitive to Foxp3 degradation, which led to impairment in their suppressive function and tumor shrinkage in the absence of pronounced adverse effects. These studies demonstrate a context-dependent differential requirement for Foxp3 for T_reg transcriptional and functional programs.

Foxp3 orchestrates reorganization of chromatin architecture to establish regulatory T cell identity

Article Open access 06 November 2023

Regulatory T cells function in established systemic inflammation and reverse fatal autoimmunity

Article 23 August 2021

The regulation and differentiation of regulatory T cells and their dysfunction in autoimmune diseases

Article 19 February 2024

Main

Regulatory T (T_reg) cells are requisite watchmen of the immune system, whose identity is distinguished by the expression of their X-chromosome-encoded lineage-specifying transcription factor Foxp3 (refs. ^1,2,3). Foxp3 plays a critical role in T_reg cell differentiation, conferring both suppressor function and fitness, largely by exploiting the pre-established epigenetic landscape in precursor cells before Foxp3 expression^4,5,6. In mice on a non-autoimmune prone genetic background showed that Foxp3 expression is stable in differentiated T_reg cells, whereas recently generated T_reg cells can lose Foxp3 expression^7,8,9. In inflammatory settings, T_reg cells upregulate Foxp3 expression and increase their suppressor function, yet both can become compromised in severe infections or autoimmunity, in particular in conjunction with genetic predispositions or interleukin (IL)-2 deprivation^7,10,11.

The indispensable role of Foxp3 in establishing T_reg cell functionality during their differentiation has been demonstrated by comparisons of mice expressing a Foxp3^GFPKO reporter-null versus a functional Foxp3^GFP allele^1,4,12. The currently prevailing notion of a requirement for continuous Foxp3 expression was suggested by Cre recombinase-induced ablation of a Foxp3 conditional allele in vitro in rapidly dividing T_reg cells followed by their adoptive transfers into lymphopenic hosts¹³; however, the confounders of these early studies left unresolved the question of whether the Foxp3-dependent T_reg functional program is intrinsically resilient or vulnerable. This uncertainty is particularly intriguing in light of the recent studies suggesting a model where Foxp3 is acting largely indirectly by inducing relatively modest changes in the expression of a few, yet to be defined, direct target genes, which in turn can act to establish the genome-wide transcriptional and functional program of T_reg cells¹². Thus, the role for Foxp3 in the maintenance of transcriptional and functional features of differentiated T_reg cells remains unknown despite major previous efforts.

Here we investigated the role of Foxp3 during T_reg cell differentiation, maintenance and turnover using a novel chemogenetic model, which enabled punctual inducible degradation of the Foxp3 protein in vivo. By analyzing the transcriptional and functional features of T_reg cells following short-term Foxp3 protein degradation, we found that Foxp3 was essential for the establishment of the gene expression program and T_reg cell function during thymic differentiation and in recently differentiated cells. In contrast with the complete loss of the T_reg cell-mediated suppression observed in developmental Foxp3 deficiency or T_reg cell ablation, Foxp3-degraded mature T_reg cells largely maintained their suppressive capacity, both in vivo and ex vivo. Accordingly, Foxp3 degradation led to minimal gene expression changes limited to a small group of genes that are likely enriched for direct Foxp3 targets; however, under severe inflammatory conditions, Foxp3 degradation led to exaggerated transcriptional changes. Finally, induced Foxp3 protein degradation preferentially destabilized intratumoral T_reg cells, leading to a loss of function and tumor rejection. These results highlight the varying roles of Foxp3 in T_reg cell phenotypic and functional features in different contexts, shedding light on its distinct mode of action as a transcription factor.

Results

Inducible Foxp3 protein degradation in vivo

Despite being central to T_reg cell biology, the role of Foxp3 protein expression in developing versus mature T_reg cells, particularly the mechanisms underlying the vulnerabilities and resilience of the Foxp3-dependent gene regulatory network in early-life versus adulthood, as well as in health versus disease, remains unknown. A major obstacle to gaining this insight has been the limitations of Foxp3 gene ablation strategies for dissecting gene regulatory programs, as the prolonged turnover of Foxp3 RNA and protein following gene deletion confounds the distinction between direct and indirect effects.

Therefore, we generated a new chemogenetic mouse model that enables rapid drug-inducible Foxp3 protein degradation in vivo (Fig. 1a,b). In this model, based on the auxin-sensing pathway in plants¹⁴, the endogenous Foxp3 allele encoding an auxin-inducible degron (AID)–Foxp3 fusion protein alongside a ZsGreen transcriptional reporter and Cre recombinase (Extended Data Fig. 1a,b) was combined with the ROSA26 (R26) allele harboring a plant-derived E3 ligase TIR1 and mCherry reporter preceded by a loxP-flanked STOP cassette (Extended Data Fig. 2a–c). The resulting Foxp3^AID mice exhibited the expected Foxp3 expression pattern (Extended Data Fig. 1c) limited to ZsGreen⁺ T_reg cells, whose suppressive capacity was similar to that of Foxp3^GFP T_reg cells¹⁵ (Extended Data Fig. 1d). Upon the addition of indole acetic acid (IAA) to TIR1-expressing Foxp3^AID T_reg cells in vitro, AID-fused Foxp3 underwent poly-ubiquitination and proteasomal degradation in a TIR1-dependent manner¹⁶ (Extended Data Figs. 1e and 2d–f). As we found the in vivo performance of the original AID-Foxp3-TIR1 protein degradation system to be suboptimal (Extended Data Fig. 2j,k), we mutated TIR1 phenylalanine 74 to a glycine in R26^TIR1 mice using CRISPR-mediated gene editing (Extended Data Fig. 2g–k). Instead of unmodified IAA, the mutant TIR1(F74G) protein recognizes 5-phenyl-IAA (5-ph-IAA)¹⁶. This improved degradation in Foxp3^AIDR26^TIR1(F74G) mice enabled rapid and near-complete in vivo degradation of the Foxp3 protein within 6 h upon 5-ph-IAA administration (Extended Data Fig. 2j,k). The effect persisted for at least 24 h after drug administration, ensuring continuous Foxp3 degradation upon once daily 5-ph-IAA administration (Fig. 1c).

**Fig. 1: Foxp3 degradation causes minimal immune activation in adult lymphoreplete mice.**

Foxp3 expression is largely dispensable for preventing autoimmune disease in adulthood

Using continuous in vivo Foxp3 degradation, we investigated its role in T_reg cell maintenance and function in adult mice using a side-by-side analysis of Foxp3^AIDR26^TIR1(F74G) treated with 5-ph-IAA and Foxp3^DTR mice subjected to T_reg cell ablation upon administration of diphtheria toxin (DT) (Fig. 1d). While continuous T_reg cell depletion led to flagrant splenomegaly and lymphadenopathy, these manifestations were unexpectedly mild following continuous Foxp3 degradation in Foxp3^AIDR26^TIR1(F74G) mice treated with 5-ph-IAA for the same duration (Fig. 1e). Accordingly, DT-mediated T_reg cell ablation induced pronounced T, B and myeloid cell activation, whereas Foxp3 degradation only had modest effects (Fig. 1f–h). Most T_reg cell-depleted Foxp3^DTR mice succumb to the resulting autoimmune syndrome within 2–3 weeks¹⁷. In contrast, continuous Foxp3 degradation for 4 weeks did not result in any noticeable clinical manifestations of autoimmune disease with only mildly increased state of immune cell activation observed (Fig. 1f–h and Extended Data Fig. 3a,b). In this regard, hepatitis and liver damage, associated with a marked immune infiltration, elevated serum alanine aminotransferase and diminished albumin in T_reg cell-depleted Foxp3^DTR mice, were undetectable in Foxp3^AIDR26^TIR1(F74G) mice after 4 weeks of continuous Foxp3 degradation (Fig. 1i–k). Notably, immune cell activation following Foxp3 degradation progressed at a slow pace as minimal changes were observed between 2 and 4 weeks of 5-ph-IAA treatment (Fig. 1f–k and Extended Data Fig. 3b–d).

While our Foxp3 degradation system enabled near-complete Foxp3 degradation, one potential caveat was that minute residual Foxp3 amounts remaining upon 5-ph-IAA treatment (Fig. 1c). To ensure that the latter does not account for the preserved T_reg cell functionality after Foxp3 degradation, we generated Cd4^creERT2Foxp3^fl/fl mice in which tamoxifen administration induced deletion of a conditional Foxp3^fl gene. Given the rapid turnover of Foxp3 protein, this approach resulted in a complete loss of its expression shortly after CreER-induced recombination, yet T_reg cells retained their identity and the function consistent with the Foxp3 degradation results (Extended Data Fig. 4). Thus, contrary to an absolute requirement of Foxp3^± T_reg cells for the restraint of fatal autoimmunity, Foxp3 protein in differentiated T_reg cells is largely dispensable for their suppressor function.

Foxp3 loss induces minimal transcriptional and functional changes in mature T_reg cells

To gain insights into the mechanisms of the observed retention of mature T_reg cell function upon Foxp3 loss, we analyzed its effect on T_reg gene expression using single-cell RNA sequencing (scRNA-seq) of fluorescence-activated cell sorting (FACS)-sorted ZsGreen⁺ cells from the secondary lymphoid organs (SLOs) of Foxp3^AID/WTR26^TIR1(F74G) mice and Foxp3^AID/WTR26^WT controls on days 3 and 7 of continuous 5-ph-IAA treatment (Fig. 2a). To avoid potential effects of mildly increased inflammation in hemizygous Foxp3^AID males following Foxp3 degradation, we performed the experiment in Foxp3^AID/WT heterozygous females, which harbor a mixed population of T_reg cells expressing either the Foxp3^WT or Foxp3^AID allele. Upon 5-ph-IAA treatment, the Foxp3^WT-expressing T_reg cells remain unaffected and maintain comparable immune tone in experimental and control mice. Our analysis of global ZsGreen⁺ cell transcriptomes, visualized by Uniform Manifold Approximation and Projection (UMAP), suggested minute differential gene expression on both day 3 and day 7 of induced Foxp3 degradation (Fig. 2b). Similarly, overlaying the UMAP plots of Foxp3^AIDR26^TIR1(F74G) T_reg cells on days 0, 3 and 7 of 5-ph-IAA treatment revealed minor changes (Extended Data Fig. 5a). To perform a more systematic, quantitative comparison and account for potential transcriptional changes within rare T_reg cell subpopulations, we performed Leiden clustering yielding 18 cell clusters (Fig. 2c and Extended Data Fig. 5b). Foxp3^AIDR26^WT and Foxp3^AIDR26^TIR1(F74G) T_reg cells from all three time points were similarly represented in most of these clusters indicative of minimal transcriptional changes across the entire T_r_eg cell population (Fig. 2d). Consistent with these findings, 5-ph-IAA-induced Foxp3 degradation in sorted Foxp3^AIDR26^TIR1(F74G) T_reg cells did not impact their ability to suppress CD4 T cell proliferation in vitro when compared to similarly treated Foxp3^AIDR26^WT T_reg cells (Fig. 2e).

Fig. 2: Foxp3 degradation induces minimal gene expression and functional changes in mature Treg cells. — **Fig. 2: Foxp3 degradation induces minimal gene expression and functional changes in mature T_reg cells.**

Because of its known limitation in capture efficiency, we complemented scRNA-seq analysis with bulk gene expression and chromatin accessibility analyses of resting and activated ZsGreen^± cells sorted from the SLOs of Foxp3^AID/WTR26^TIR1(F74G) and Foxp3^AID/WTR26^WT mice on day 7 of 5-ph-IAA treatment (Fig. 2f,g). As a comparison, we examined differential gene expression in cells expressing a Foxp3^GFPKO reporter-null allele⁴, which have never expressed Foxp3 protein, versus Foxp3-sufficient Foxp3^GFP T_reg cells¹² (Fig. 2h). Consistent with the scRNA-seq analysis, bulk RNA-seq revealed a dramatically reduced number of differentially expressed genes (DEGs) in both resting and activated T_reg cells upon Foxp3 protein degradation compared to those in Foxp3^GFPKO cells (Fig. 2i and Supplementary Table 1). Likewise, ATAC-seq analysis of Foxp3-degraded resting and activated Foxp3^AID T_reg cells revealed minimal changes in chromatin accessibility, unlike the substantial differences observed between resting and activated Foxp3^GFPKO T_reg cells and Foxp3-sufficient Foxp3^GFP T_reg cells¹² (Extended Data Fig. 5c).

Next, we performed additional analyses on the small group of genes whose expression was affected by Foxp3 degradation in our scRNA-seq dataset with activation states defined using the gene scores based on previously identified T_reg cell transcriptional signatures¹² (Fig. 3a and Extended Data Fig. 6a,b). We then performed pseudo-bulk differential gene expression analyses for the resting and activated T_reg cells. Gene expression changes on day 3 and day 7 post Foxp3 degradation were well correlated, with a larger fold change observed on day 7 indicating time-dependent augmentation of gene expression changes (Fig. 3b). We classified the latter into four groups: genes upregulated (‘resting TIR1-up’ and ‘activated TIR1-up’) and downregulated (‘resting TIR1-down’ and ‘activated TIR1-down’) upon Foxp3 degradation in resting and activated T_re_g cells, respectively. A closer examination of Foxp3-regulated transcripts in resting T_reg cells showed that the ‘resting TIR1-up’ genes repressed by Foxp3 were expressed more highly in activated versus resting T_reg cells, whereas ‘resting TIR1-down’ genes induced by Foxp3 showed the opposite pattern (Fig. 3c). This result suggests that in resting T_reg cells, Foxp3 enables a ‘goldilocks’ state of expression of T cell activation-associated genes.

Fig. 3: Foxp3 degradation in mature Treg cells induces expression changes in a small set of genes. — **Fig. 3: Foxp3 degradation in mature T_reg cells induces expression changes in a small set of genes.**

The gene set with statistically significant differential expression caused by Foxp3 degradation in either resting or activated T_reg cells included 32 TIR1-up and 38 TIR1-down genes repressed and activated by Foxp3, respectively (Fig. 3d). The latter group included the Foxp3 gene itself. Indeed, ZsGreen expression, reporting Foxp3 mRNA levels, showed a slight but statistically significant reduction in resting T_reg cells at day 7, consistent with the bulk RNA-seq analysis (Fig. 3e and Extended Data Fig. 6c). Flow cytometric analyses also showed a Foxp3 degradation-induced reduction in CD25(Il2ra) expression in both resting and activated T_reg cells (Fig. 3f and Extended Data Fig. 6d,e), whereas a reduction in CD122 (Il2rb), OX40 (Tnfrsf4), GITR (Tnfrsf18) and FR4 (Izumo1r) levels was limited to resting T_reg cells (Fig. 3f and Extended Data Fig. 6d,e). On the flip side, CD127 (Il7r) and TCF1 (Tcf7) protein expression was increased in resting T_reg cells following Foxp3 degradation, consistent with the observed changes in their transcript levels (Fig. 3g and Extended Data Fig. 6d,f).

Foxp3 degradation-sensitive genes are enriched for Foxp3 binding

Previous studies of resting and activated Foxp3^GFP T_reg and Foxp3^GFPKO T_reg ‘wannabe’ cells identified the overall Foxp3-dependent gene set without distinguishing between potential direct and indirect Foxp3 targets^4,5. Given the short duration of Foxp3 degradation and the small number of genes impacted by it, we reasoned that ‘degradation-sensitive’ genes are likely enriched for direct Foxp3 transcriptional targets. To examine this, we grouped genes that were degradation sensitive (Foxp3^AIDR26^TIR1(F74G) versus Foxp3^AIDR26^WT) in both resting and activated T_reg cells based on their P values and examined the number of Foxp3 binding sites¹⁸ near the stratified genes in each group (Fig. 4a). As Foxp3 is known to bind predominantly to open chromatin regions and its global genome occupancy is not associated with Foxp3-dependent gene expression or chromatin accessibility changes, we normalized the number of Foxp3 peaks to the number of open chromatin regions surrounding each gene, using previously published Foxp3 CUT&RUN and T_reg ATAC-seq datasets¹². Notably, the top ‘TIR1-down’ degradation-sensitive genes in resting T_reg cells contained significantly more Foxp3 binding sites per gene (Fig. 4a). This observation suggests that genes in this group, such as Il2ra, Lrrc32 and Il2rb, known for their role in T_reg functionality^19,20, are extensively bound and likely directly regulated by Foxp3 (Extended Data Fig. 7a). Likewise, the top ‘TIR1-up’ degradation-sensitive genes in resting T_reg cells also contained more Foxp3 binding sites per gene (Fig. 4a). Genes in this group, including Tcf7, Id2 and Sox4, have been implicated in T_reg gene expression and function^{21,22,23,24,25} (Extended Data Fig. 7b). We refer to these gene groups as ‘Foxp3-activated’ and ‘Foxp3-repressed’ genes, respectively. Of note, when the overall Foxp3-dependent (Foxp3^GFPKO versus Foxp3^GFP) gene set was stratified and analyzed in the same fashion no enrichment for Foxp3 binding was observed, likely because a larger number of indirect Foxp3 target genes obscured small number of direct ones (Fig. 4b). In contrast, similar analyses of Foxp3 degradation-sensitive genes in their activated counterparts did not reveal significant enrichment of Foxp3 binding in any gene groups, suggesting that upon T_reg activation direct gene regulation by Foxp3 can be compensated by other T cell activation-dependent transcription factors (Fig. 4a,b). This observation was also consistent with the smaller number of DEGs identified by both bulk and scRNA-seq analyses of Foxp3 degradation in activated T_reg cells (Figs. 2i and 3d).

Fig. 4: Foxp3 degradation-sensitive genes in mature Treg cells are enriched for Foxp3 binding. — **Fig. 4: Foxp3 degradation-sensitive genes in mature T_reg cells are enriched for Foxp3 binding.**

By leveraging our published ChIP-seq data¹⁸, we explored additional features of the ‘Foxp3-activated’ and ‘Foxp3-repressed’ genes. Compared to genes not enriched for Foxp3 binding, ‘Foxp3-activated’ genes showed the highest level of the activating-H3K27Ac and the lowest level of the repressive-H3K27me3 histone modifications, whereas ‘Foxp3-repressed’ genes showed the opposite pattern (Fig. 4c). In addition, motif enrichment analysis of Foxp3-bound ATAC-seq peaks near ‘Foxp3-activated’ genes revealed a pronounced enrichment for STAT-binding motifs (Fig. 4d) consistent with the inclusion of both Il2ra and Il2rb within this gene set.

Foxp3 degradation effects are generally conserved across tissues

To assess Foxp3 degradation effects on the mature T_reg cell transcriptional program across both nonlymphoid versus lymphoid tissues under steady state, we performed scRNA-seq analysis of ZsGreen⁺ cells isolated from the SLOs, lung, liver and large intestine lamina propria (LILP) of Foxp3^AID/WTR26^TIR1(F74G) and Foxp3^AID/WTR26^WT mice on day 7 of continuous 5-ph-IAA treatment (Extended Data Fig. 8a). For all tissues, Foxp3^AIDR26^TIR1(F74G) and Foxp3^AIDR26^WT T_reg transcriptome distributions were highly overlapping when visualized using a UMAP embedding (Extended Data Fig. 8b). To account for varying activation states in different tissues, we further classified cells as resting or activated as in Fig. 3a (Extended Data Fig. 8c). Next, we identified significant DEGs in Foxp3^AIDR26^TIR1(F74G) versus Foxp3^AIDR26^WT T_reg cells across all tissues and correlated log fold changes of these DEGs between tissues with higher correlation corresponded to a higher degree of similarity of Foxp3 degradation-induced transcriptional effects. Consistent with our previous findings, resting T_reg cells had greater numbers of DEGs compared to activated ones in each tissue (Extended Data Fig. 8d); accordingly, hierarchical clustering of log fold change correlations revealed that resting and activated cells clustered separately irrespective of tissue origin (Extended Data Fig. 8e). Among resting cells, the effects of Foxp3 degradation were largely consistent between tissues as evidenced by highly correlated transcriptional changes (Pearson r = 0.83–0.91). Of note, DEGs in nonlymphoid tissue-activated T_reg cells were less correlated with those in SLOs (Pearson r = 0.53–0.69). K-means clustering of the DEGs from all tissues, based on log fold changes, revealed gene clusters with distinct patterns of regulation across tissues and activation states (Extended Data Fig. 8f,g). A few clusters exhibited tissue-specific alterations, most prominently in LILP-activated cells (clusters 5 and 7). Thus, Foxp3 degradation-induced transcriptional effects, while highly conserved across tissues in resting T_reg cells, were more variable across tissues in activated ones, particularly in the LILP, a tissue enriched for highly activated and devoid of resting T_reg cells. The lack of enrichment of Foxp3 binding sites near degradation-sensitive genes in activated cells (Fig. 4a) was consistent with the supposition that tissue-specific factors may contribute to distinct transcriptional effects of Foxp3 degradation in activated cells.

The LILP T_reg cell population encompasses peripherally (pT_reg) and thymically generated (tT_reg) T_reg cells²⁶. Subclustering of LILP cells identified a small actively proliferating cluster, a Rorc expressing ‘pT_reg’ cluster alongside Ikzf2^intermediate and Ikzf2^high ‘tT_reg’ clusters enriched for lymphoid tissue-associated genes and Gata3 expression, respectively (Extended Data Fig. 8h,i). We refer to these populations as pT_reg, lymphoid tissue tT_reg (LT-tT_reg) and nonlymphoid tissue tT_reg (NLT-tT_reg), respectively, consistent with published LILP T_reg transcriptomes²⁷. Comparing the DEGs induced by Foxp3 degradation within each cluster, the pT_reg cluster had the highest number of DEGs (Extended Data Fig. 8j); however, compared to the pT_reg cluster, the NLT-tT_reg cluster had a significantly larger magnitude of differential expression of degradation-sensitive Foxp3-repressed and activated genes (Extended Data Fig. 8k). Furthermore, the Foxp3 degradation-induced DEGs poorly correlated (Pearson r = 0.38) between the NLT-tT_reg and pT_reg cells (Extended Data Fig. 8l). These results are consistent with findings that colonic pT_reg cells maintain their fitness and exert some of their suppressive functions independently of Foxp3.

Long timescale of establishment of Foxp3-dependent gene program

To complement Foxp3-degradation-based ‘loss-of-function’ studies, we employed a ‘gain-of-function’ approach using a reversible Foxp3^{loxP-Thy1.1-STOP-loxP-GFP} reporter-null allele (Foxp3^LSL)⁷. The Foxp3^LSL allele harbors a loxP site-flanked Thy1.1 reporter followed by a STOP cassette and a Foxp3^GFP reporter. In Foxp3^LSL mice, Thy1.1 reporter marks T_reg ‘wannabe’ cells with the transcriptionally active Foxp3 locus yet lacking Foxp3 expression similar to the GFP⁺ cells in Foxp3^GFPKO mice. 4-hydroxytamoxifen (4-OHT) treatment of Cd4^creERT2Foxp3^LSL mice led to the excision of the STOP cassette and punctual induction of the Foxp3 protein, converting Foxp3⁻Thy1.1⁺ T_reg ‘wannabe’ cells into fully functional Foxp3-expressing GFP⁺ T_reg cells (Fig. 4e). Foxp3^LSL/WT female heterozygous mice are healthy as they harbor both functional Foxp3-sufficient T_reg cells and T_reg ‘wannabe’ cells expressing Foxp3^WT and Foxp3^LSL allele, respectively. Thus, we sought to investigate the temporal dynamics of the emerging Foxp3-dependent transcriptional features upon acquisition of Foxp3 expression by T_reg ‘wannabe’ cells in 4-OHT-treated female heterozygous Cd4^creERT2Foxp3^LSL/WT mice. Resting and activated Foxp3⁺GFP⁺ cells and control Foxp3⁻GFP⁻Thy1.1⁺ cells with a matching activation state were sorted from the Cd4^creERT2Foxp3^LSL/WT and Cd4^WTFoxp3^LSL/WT littermates, respectively, on days 3, 7, 14 and 28 after single 4-OHT administration and subjected to RNA-seq analysis (Fig. 4f and Extended Data Fig. 7c). The top Foxp3 degradation-sensitive ‘TIR1-down’ and ‘TIR1-up’ genes showed time-dependent increases and decreases in their expression in both resting and activated Foxp3⁺GFP⁺Thy1.1⁻ cells in comparison to time- and activation-state-matched Foxp3⁻GFP⁻Thy1.1⁺ ‘wannabe’ controls, respectively, consistent with a likely direct role for Foxp3 in regulating their expression (Fig. 4g). Further analysis of the pace at which Foxp3 installation drives transcriptional changes in T_reg ‘wannabe’ cells toward a bona fide T_reg transcriptome revealed an unexpectedly prolonged timeline of approximately 2 weeks. While it remains formally plausible that Foxp3-dependent transcriptional programs are established more rapidly in normally differentiating T_reg cells, these observations suggest that the establishment of the T_reg cell transcriptional program and functionality is critically dependent on Foxp3 during T_reg cell maturation. In contrast, fully differentiated T_reg cells in healthy adult mice may not rely on Foxp3 to the same extent. Moreover, the slow kinetics of the Foxp3-dependent gene program acquisition highlights the necessity for direct Foxp3 targets to act in trans to regulate downstream, indirect Foxp3-dependent genes.

A requirement for Foxp3 during T_reg cell maturation

To test whether Foxp3 is essential for the acquisition of the T_reg cell-specific transcriptional program and function during thymic T_reg cell differentiation and peripheral maturation, we employed two complementary approaches to assess the effects of induced Foxp3 protein degradation in developing T_reg cells. First, we investigated Foxp3 degradation-induced gene expression changes in developing T_reg cells in the thymus. Following 7 days of 5-ph-IAA-mediated Foxp3 degradation in Foxp3^AID/WTR26^TIR1(F74G) mice, we performed bulk RNA-seq analysis of developing ZsGreen⁺ CD73^low T_reg cells isolated from the thymus (Fig. 5a,b). Differential CD73 expression was used to discern recently generated nascent ZsGreen⁺CD73^low CD62L^high T_reg cells from recirculating CD73^high T_reg cells entering the thymus from the periphery²⁸. We then compared the expression of Foxp3 degradation-sensitive and Foxp3-dependent (Foxp3^GFPKO versus Foxp3^GFP) genes across T_reg developmental stages. The number of DEGs resulting from Foxp3 degradation was the highest in developing thymic T_reg cells followed by resting T_reg cells, whereas activated T_reg cells exhibited the lowest number (Fig. 5c). These data suggest that the transcriptional program of developing T_reg cells is markedly more vulnerable to Foxp3 loss in comparison to mature T_reg cells, with the latter becoming even less dependent on continuous Foxp3 expression as they become activated. This observation was consistent with the absence of significant enrichment for Foxp3 binding in Foxp3 degradation-sensitive gene loci in activated T_reg cells. Notably, Foxp3 degradation-sensitive genes and Foxp3-dependent genes exhibited the strongest correlation in developing thymic T_reg cells, with this correlation progressively decreasing as T_reg cells mature, reaching its lowest values in activated T_reg cells (Fig. 5d,e). This trend persisted even among Foxp3-bound genes (Extended Data Fig. 9a). The declining correlation suggests that while Foxp3 deficiency closely mirrors the effects of Foxp3 degradation in recently generated T_reg cells, sustained Foxp3 loss in mature T_reg cells, particularly those with a history of activation, may lead to secondary transcriptional effects beyond the primary Foxp3-regulated program.

Fig. 5: Foxp3 is preferentially required for regulation of gene expression during early Treg cell differentiation. — **Fig. 5: Foxp3 is preferentially required for regulation of gene expression during early T_reg cell differentiation.**

Meta-analysis of previously published scRNA-seq of thymic Foxp3⁺ T_reg cells and their progenitors²⁹ and our datasets showed that the ‘TIR1-up’ (Foxp3-repressed) and ‘TIR1-down’ (Foxp3-activated) degradation-sensitive gene signatures negatively and positively correlate with Foxp3 expression, respectively, in a dose-dependent manner (Fig. 5f). Notably, similar metacell analysis of Foxp3-degradation-induced transcriptional changes in resting and activated T_reg cells revealed a correlation between Foxp3 dosage and the ‘TIR1-down’ signature only (Extended Data Fig. 9b). In contrast, the ‘TIR1-up’ gene set showed no correlation with Foxp3 expression level. These findings further support a shift in the role of Foxp3 in gene regulation as T_reg cells mature, particularly in its function as a transcriptional repressor.

In adult mice, thymic output contributes minimally to the peripheral pool of differentiated T_reg cells, which is maintained by their self-renewal and whose transcriptional program and functionality we found resilient to the Foxp3 loss^7,9,30 (Figs. 1 and 2). In neonatal mice, Foxp3⁺ cells first appear among CD4SP thymocytes on days 2–3 after birth; thymic T_reg cell output continues to steadily increase until day 21 with recently generated T_reg cells accounting for the bulk of the T_reg peripheral pool¹⁵. Thus, we tested the requirement for Foxp3 expression in the suppressor function of early-life T_reg cells by treating neonatal Foxp3^AIDR26^TIR1(F74G) and Foxp3^AIDR26^WT control mice with 5-ph-IAA daily for 2 weeks, starting from day 1 after birth (Fig. 5g). Contrary to adults, Foxp3 degradation in neonates led to severe autoimmune disease featuring pronounced T cell activation (Fig. 5h and Extended Data Fig. 9c), myeloproliferation (Fig. 5i and Extended Data Fig. 9d), and tissue inflammation (Fig. 5j,k and Extended Data Fig. 9e) similar to those in Foxp3-deficient Foxp3^GFPKO mice indicative of a loss of T_reg function. The latter was confirmed by the lack of in vitro suppressor capacity of ZsGreen⁺ cells isolated from 5-ph-IAA-treated Foxp3^AIDR26^TIR1(F74G) neonates (Fig. 5l). Consistently, RNA-seq analysis of neonatal Foxp3^AIDR26^TIR1(F74G) and Foxp3^AIDR26^WT T_reg cells subjected to 7 days of in vivo Foxp3 degradation revealed hundreds of up- and downregulated genes far exceeding the number of DEGs resulting from Foxp3 degradation induced in mature T_reg cells in adults (Figs. 3 and 5m, Extended Data Fig. 9f and Supplementary Table 1). Foxp3 degradation-induced DEGs in early-life T_reg cells showed strong correlation with DEGs observed in Foxp3⁻ T_reg ‘wannabes’ from Foxp3^LSL neonates versus Foxp3⁺ T_reg cells from Foxp3^DTR controls, confirming that the loss of Foxp3 in recently generated T_reg cells phenocopies Foxp3 genetic deficiency (Fig. 5n). Among all maturation stages of adult T_reg cells, Foxp3 degradation-induced DEGs in adult thymic T_reg cells showed the highest concordance with those in neonates, suggesting a similarity in terms of Foxp3 dependence of their transcriptional programs. In contrast, Foxp3 degradation-induced DEGs in adult resting and activated T_reg cells showed no such similarity to the neonatal Foxp3-dependent gene expression features (Fig. 5o). These results suggest that in early life, persistent Foxp3 expression is required for the establishment of a stable gene regulatory network and functionality in recently generated T_reg cells, likely by acting on its few direct targets and through continuous enforcement of initially unstable feed-forward regulation of indirect targets via intermediates acting in trans¹².

Proliferation and severe inflammation sensitize T_reg cell transcriptome to Foxp3 loss

The observed dispensability of Foxp3 for the function of differentiated T_reg cells stood in a sharp contrast with our early finding of a loss of T_reg cell function upon ablation of a conditional Foxp3 allele in differentiated T_reg cells via Cre¹³. While T_reg cells residing in lymphoreplete healthy mice undergo a slow turnover, they undergo pronounced proliferation and activation during in vitro retroviral transduction and transfer into lymphopenic settings employed in these studies. Therefore, Foxp3 expression may be needed to maintain the T_reg-specific transcriptional program in robustly proliferative cells. To test this supposition, we first performed flow cytometric analysis of Foxp3^AID T_reg cells following 7 days of in vivo Foxp3 degradation, after parsing them into proliferative and nonproliferative cells on the basis of Ki67 expression (Fig. 6a). While the overall phenotypic shift in Foxp3-degraded versus -replete T_reg cells was modest, the proliferating Ki67⁺ subset accounted for most changes in CD25, GITR and CTLA4 protein levels, whereas the Ki67⁻ T_reg subset underwent little change (Fig. 6b), consistent with the above idea. Next, we performed phenotypic analysis of cell trace violet (CTV)-labeled Foxp3^AID T_reg cells stimulated to proliferate in vitro with anti-CD3 and anti-CD28 antibodies, alongside 5-ph-IAA treatment to induce Foxp3 degradation (Fig. 6c). The highly divided (CTV^low) cells showed a greater difference in T_reg cell markers encoded by Foxp3 degradation-sensitive genes such as CD153 (Tnfsf8) and GARP (Lrrc32) compared to their lowly divided (CTV^high) counterparts (Fig. 6d,e and Extended Data Fig. 6d). Expression of CD4, serving as a control, was unaffected by Foxp3 degradation regardless of cell division (Fig. 6d). Moreover, T_reg cells subjected to Foxp3 degradation for 72 h secreted more Foxp3-repressed proinflammatory cytokines, including IL-2, IL-4 and IL-13 (Fig. 6f and Extended Data Fig. 6d). To isolate the transcriptional effects of Foxp3 degradation in a highly proliferative context, we activated and expanded Foxp3^AIDR26^TIR1(F74G) and Foxp3^AIDR26^WT T_reg cells in vitro for 3 days, rested them in the absence of stimulation for 2 h, treated them with 5-ph-IAA for 7 h and performed bulk RNA-seq (Fig. 6g). Using this approach, we assessed the effects of Foxp3 loss in robustly dividing cells minimizing the effects of T cell receptor (TCR) stimulation and inflammatory cues, including Foxp3 degradation-induced inflammatory cytokine production. While some of the most highly significant DEGs included those identified in mature T_reg cells upon Foxp3 degradation in vivo, many were not identified in other experimental settings (Figs. 3a and 6h). Of note, as activated T_reg cells in other contexts consistently had fewer DEGs than resting cells (Figs. 2i and 3b and Extended Data Fig. 8d), the observed changes are unlikely to be driven solely by previous TCR-induced activation. Overall, these data are consistent with the notion of a heightened dependency of T_reg cells for the maintenance of their transcriptional features on Foxp3 during cell division. As neonatal T_reg cells were markedly more proliferative compared to adult cells (Extended Data Fig. 9g,h), their enhanced proliferation together with other potential factors could contribute to their heightened sensitivity to Foxp3 degradation.

Fig. 6: Proliferation and inflammation increase Treg cell sensitivity to Foxp3 degradation. — **Fig. 6: Proliferation and inflammation increase T_reg cell sensitivity to Foxp3 degradation.**

Given the upregulation of Foxp3 expression by T_reg cells and their proliferation in inflammatory settings⁷, we surmised that severe inflammation coupled to cell division may make mature T_reg cells vulnerable to the loss of Foxp3. Thus, we adoptively transferred Foxp3^AIDR26^TIR1(F74G) and Foxp3^AIDR26^WT T_reg cells mixed at a 1:1 ratio into Foxp3^DTR recipients subjected to DT-mediated host T_reg cell depletion, which creates severe polytypic inflammation. As a control, we assessed the transcriptional effects of Foxp3 degradation upon short-term transfers of the same T_reg cell mixture into T cell-deficient mice, which afford lymphopenia-induced T_reg cell proliferation in a minimal inflammatory setting. After 7 days of continuous Foxp3 degradation, sorted mCherry⁺ Foxp3^AIDR26^TIR1(F74G) and mCherry⁻ Foxp3^AIDR26^WT T_reg cells were subjected to scRNA-seq analysis (Fig. 6i,j). We observed severely compromised fitness of Foxp3-deprived Foxp3^AIDR26^TIR1(F74G) versus control Foxp3-sufficient Foxp3^AIDR26^WT T_reg cells in inflammatory settings reflected in their strongly biased ratios in the Foxp3^DTR recipients; however, in T cell-deficient recipients Foxp3^AIDR26^TIR1(F74G) T_reg cells were only mildly outcompeted (Fig. 6k). At the gene expression level, UMAP visualization showed that Foxp3^AIDR26^TIR1(F74G) and Foxp3^AIDR26^WT T_reg transcriptomes were largely overlapping (Fig. 6l). Fine-grained Leiden clustering showed similar representation of cells of each genotype from T cell-deficient recipients in all the clusters; however, the frequencies of Foxp3^AIDR26^TIR1(F74G) T_reg cells from Foxp3^DTR recipients were significantly reduced in cluster 4, whereas they were elevated in cluster 11 (Fig. 6m). Accordingly, many more DEGs in Foxp3^AIDR26^TIR1(F74G) versus Foxp3^AIDR26^WT cells were observed in severe inflammatory versus lymphopenic settings of Foxp3^DTR and Tcrb^−/−Tcrd^−/− recipients, respectively (Fig. 6n,o and Supplementary Table 1). These results suggest that the Foxp3-dependent gene regulatory network resilient to Foxp3 loss in mature T_reg cells loses its stability under severe inflammatory conditions.

Foxp3 degradation-induced tumor shrinkage with minimal adverse effects

Next, we asked whether the resilient state of T_reg cell transcriptional and functional program can be lost in a disease state associated with high T_reg turnover rates. Solid organ tumors are highly enriched for activated T_reg cells^31,32. Using both Ki67 staining and 5-ethynyl-2’-deoxyuridine (EdU) incorporation assay, we confirmed that tumoral T_reg cells were markedly more proliferative in comparison to their counterparts residing in tumor-draining lymph nodes (dLNs) (Extended Data Fig. 10a,b). To test whether Foxp3 degradation would compromise the tumoral T_reg function, we implanted B16-OVA melanoma cells in the flank of Foxp3^AIDR26^TIR1(F74G) and Foxp3^AIDR26^WT mice. On day 5 after tumor implantation, tumor-bearing mice were treated daily with 5-ph-IAA to induce Foxp3 degradation (Fig. 7a). While the tumors grow unabatedly in Foxp3^AIDR26^WT mice, tumors in Foxp3^AIDR26^TIR1(F74G) ceased to grow and underwent rapid shrinkage (Fig. 7b,c). CD8⁺ T cells and natural killer (NK) cells within the tumor exhibited heightened effector function, as evidenced by increased interferon (IFN)γ production (Fig. 7d and Extended Data Fig. 10c). Of note, ZsGreen⁻ effector CD4 T cells, instead of upregulating IFNγ, showed increased IL-4 expression (Fig. 7e and Extended Data Fig. 10d), which has been recently implicated in antitumor immunity^33,34. Notably, severe adverse effects, typically seen upon pan-T_reg cell ablation in cancer-bearing mice, were completely lacking, as there were no signs of body weight loss, hunched posture, skin lesions or tissue inflammation based on clinical or histological evaluations (Fig. 7f,g). Although Foxp3 degradation had no effect on the abundance of ZsGreen⁺ T_reg cells in the tumor, dLNs and nondraining lymph nodes (ndLNs), Foxp3 degradation-induced phenotypic changes were markedly more pronounced in tumoral T_reg cells. While increased TCF1 expression was observed in both nontumoral and tumoral T_reg cells, reduced CTLA4, GITR and CD39 expression was only observed in the latter (Fig. 7h). Thus, Foxp3 degradation boosts antitumor immunity with minimal immune related adverse effects due to a selective loss of intratumoral T_reg cell function.

**Fig. 7: Foxp3 degradation leads to tumor shrinkage with minimal adverse effects.**

scRNA-seq profiling of tumor-infiltrating T_reg cells isolated from Foxp3^AIDR26^WT and Foxp3^AIDR26^TIR1(F74G) mice following 15 days of continuous Foxp3 degradation showed profound and widespread transcriptional alterations within tumor T_reg cells visualized by UMAP (Fig. 7i). Characterization of the heterogeneity of Foxp3-degraded and control tumoral T_reg cells using coarse clustering identified four clusters: (1) Gata3^hi, T_H2-like cluster; (2) Cxcr3^hi cluster; (3) Tbx21^hi, T_H1-like cluster; and (4) Mki67^hi, proliferating cluster (Fig. 7j,k). Unlike steady-state T_reg cells, where both Foxp3-sufficient and Foxp3-degraded cells remained similarly distributed across clusters even under refined clustering conditions, intratumoral T_reg cells displayed a markedly uneven distribution depending on Foxp3 status, even at this level. Specifically, the T_H2-like (Gata3^hi) cluster was overwhelmingly populated by Foxp3-degraded T_reg cells from Foxp3^AIDR26^TIR1(F74G) mice, suggesting that Foxp3 loss preferentially skewed tumor T_reg cells toward a T_H2-like state. Conversely, the Cxcr3^hi cluster was predominantly composed of Foxp3-sufficient T_reg cells. The proliferating (Mki67^hi) cluster contained a slightly higher proportion of Foxp3-degraded T_reg cells, whereas the T_H1-like (Tbx21^hi) cluster maintained a comparable representation of T_reg cells of both genotypes (Fig. 7l). Given developmental stage-dependent requirements for Foxp3 in maintaining T_reg functionality observed in healthy mice, we sought to determine whether Foxp3 exerts cell state context-specific roles across distinct tumoral T_reg subsets. Thus, we performed pseudo-bulk differential gene expression analyses between Foxp3^AIDR26^WT and Foxp3^AIDR26^TIR1(F74G) tumor T_reg cells within each cluster. Notably, the proliferating cluster revealed the highest number of DEGs (Fig. 7m and Extended Data Fig. 10e). These results confirm a heightened dependence on Foxp3 for the maintenance of tumoral T_reg cells.

Discussion

Foxp3 binding, while sequence specific, is associated with transcriptional changes in only a few Foxp3-bound genes¹². The latter has been proposed to propagate Foxp3-dependent genome-wide features in a ‘relay-like’ manner¹². Such transcriptional ‘staging’ by Foxp3 would be expected to result in a readily reversible cell state especially when Foxp3 expression is decreased or lost. Indeed, it has been suggested that in inflammatory or hypoxic settings T_reg cells can lose Foxp3 expression and become proinflammatory cells, which may contribute to the disease^35,36,37. On the other hand, it has been proposed that Foxp3 instability is limited to newly generated T_reg cells and that after its transient loss, re-expression of Foxp3 restores T_reg suppressor function^8,9.

Our studies demonstrated that, contrary to a complete halt of functional T_reg development resulting from genetic Foxp3 deficiency, the function and fitness of differentiated T_reg cells in adult mice was remarkably durable upon the loss of Foxp3 at steady state. This conclusion is supported by our findings that Foxp3 degradation induced upon 5-ph-IAA treatment of adult Foxp3^AIDR26^TIR1(F74G) mice for up to 4 weeks did not result in clinical manifestations of autoimmune disease or wasting, whereas Foxp3^DTR mice of similar age subjected to DT-induced T_reg cell ablation succumbed to the disease within 2–3 weeks. The mild increase in the immune tone in the absence of any notable clinical manifestations observed upon continuous Foxp3 degradation for up to 4 weeks was in stark contrast to the rapid progression of fatal inflammatory disease commencing upon T_reg cell ablation¹⁷. Accordingly, instead of impacting the full set of direct and indirect Foxp3-dependent transcriptional features, Foxp3 degradation affected the expression of a much smaller gene set, likely enriched for Foxp3 direct target genes. These results suggest that after Foxp3 establishes the identity, fitness and functionality of T_reg cells during their differentiation, T_reg transcriptional and functional programs acquire resilience to its loss under physiological conditions.

Notably, our studies revealed that contrary to adult mice, Foxp3-dependent transcriptional and functional T_reg programs are absolutely dependent on Foxp3 in neonates, with its loss resulting in severe autoimmune inflammatory disease indistinguishable from that seen in mice with a congenital Foxp3 deficiency. It seems unlikely that the observed disease can be accounted for solely by the disruption of Foxp3 expression during thymic T_reg cell generation, because the expression of TIR1 is controlled by a Foxp3-driven Cre. In addition, small numbers of functional T_reg cells present in the periphery would be able to rescue the disease if they were resilient to Foxp3 loss. In this regard, adoptive transfer of small numbers of T_reg cells into neonatal Foxp3^null mice or temporally induced installation of Foxp3 protein expression in a small cohort Foxp3^LSL expressing T_reg ‘wannabe’ cells affords protection from the disease^1,7. Furthermore, Foxp3 protein degradation in both adult Foxp3-expressing thymocytes and early-life peripheral T_reg cells caused markedly more extensive transcriptional changes than those observed following Foxp3 degradation in fully differentiated adult peripheral T_reg cells. These findings suggest that stabilization of the Foxp3-dependent transcriptional program occurs at a relatively slow tempo. In support of this notion, it took approximately 2–3 weeks for the transcriptional program initiated by newly installed Foxp3 protein in T_reg ‘wannabe’ cells to approach that of bona fide T_reg cells. Together, these results support a model whereby the initially vulnerable Foxp3-dependent gene regulatory network in Foxp3⁺ thymocytes and early-life peripheral T_reg cells gradually matures into a stable, largely Foxp3 independent state. This transition seems to unfold over an unexpectedly long timescale, possibly reflecting the ‘relay-like’ propagation of Foxp3-mediated gene expression changes, whereby a small number of direct Foxp3 targets act in trans to modulate broader networks of gene expression¹². Additionally, a relatively modest scale of modulation of a number of genes by Foxp3 can contribute to the slow tempo of the acquisition of Foxp3-dependent network resilience. Finally, early-life peripheral T cells, almost exclusively made of recent thymic emigrants, produce limited amounts of IL-2, which can further compromise maintenance of T_reg cell transcriptional program upon removal of Foxp3 (ref. ³⁸). In line with this, we found that this program encompasses genes, whose cis-regulatory elements are enriched for STAT-binding motifs.

Developmentally established T_reg cell transcriptional and functional features can be further tuned by context-dependent interactions between Foxp3 and its protein partners activated in response extracellular cues such as TCR stimulation among others^{39,40,41,42,43,44}. Accordingly, transcriptional changes in T_reg cells observed across different tissues upon Foxp3 degradation were more variable in activated in comparison to resting T_reg cells, where they were highly correlated. Notably, Foxp3 degradation affected genes in resting T_reg cells were enriched for Foxp3 binding, whereas those in activated cells were not. These observations suggest that the indirect components of Foxp3-dependent transcriptional T_reg cell program can be influenced by environmental cues. Cell division presents a major challenge for inheritance of differentiated cell states which necessitates a wide range of epigenetic and genetic enforcement mechanisms. Fittingly, robust T_reg cell division in vitro and in vivo, especially in a severe inflammatory environment, compromised the resilience of the Foxp3-dependent transcriptional program with a wide range of Foxp3 degradation-induced transcriptional changes beyond the core ‘Foxp3-repressed’ and ‘Foxp3-activated’ genes identified under homeostatic conditions.

While the specific mechanisms behind the observed durability of the Foxp3-dependent transcriptional network and its vulnerability remain to be elucidated, their context dependence presents a unique therapeutic opportunity for selective targeting of T_reg cell functionality. In this regard, our observation of enhanced antitumor immunity and tumor growth control upon Foxp3 degradation in the absence of severe adverse effects typically observed upon wholesale T_reg cell ablation in experimental models of cancer^{45,46,47,48,49,50,51,52,53,54,55} offers a novel strategy for immunotherapy of tumors featuring highly activated and proliferative T_reg cells.

In conclusion, our studies suggest that the initially vulnerable Foxp3-dependent gene regulatory network and associated functionality of T_reg cells progress over time to a resilient state. Once this state is established, T_reg cell function, and much of the Foxp3-driven transcriptional program, except for a few genes likely enriched for Foxp3 direct targets, are maintained even after the loss of Foxp3 expression. Yet, the durability of Foxp3-dependent gene regulatory network in mature T_reg cells can be compromised in diseased tissue contexts such as the tumor microenvironment; this vulnerability can be leveraged for therapeutic disruption of intratumoral T_reg cell function. These findings have noteworthy implications for understanding T_reg cell function in health and disease.

Methods

Mice

All animal experiments in this study were approved by the Sloan Kettering Institute (SKI) Institutional Animal Care and Use Committee under protocol no. 08-10-023 or Yale University Institutional Animal Care and Use Committee under protocol no. 2023-20503. Mice were housed at the SKI or Yale University animal facility under specific-pathogen-free conditions on a 12-h light–dark cycle with free access to water and regular chow diet. The average ambient temperature is 21.5 °C and the average humidity is 48%. Foxp3^DTR, Foxp3^fl and Cd4^creERT2 mice used in this study have been previously described^1,17,56. Foxp3^AID, ROSA26^TIR1 and ROSA26^TIR1(F74G) mice were generated in this study. All control and experimental animals were age-matched and littermates were used as controls unless otherwise indicated.

Generation of Foxp3 ^AID, ROSA26 ^TIR1 and ROSA26 ^TIR1(F74G) mice

Gene targeting was carried out in 129/B6 F1 hybrid embryonic stem (ES) cells using a targeting vector spanning the Foxp3 locus. At the N terminus, the AID sequence was fused to Foxp3 via a seven-amino-acid flexible linker (GSHGGSG). An IRES-ZsGreen-T2A-iCre-Frt-neo-Frt cassette was inserted into the 3′ untranslated region (UTR) immediately downstream of the stop codon. ES cell clones with successful targeting were validated and injected into CD-1 tetraploid blastocysts to generate knock-in founders. These founders were then crossed with a Flp-deleter line to excise the neo-cassette. The resulting F1 progeny were backcrossed to the C57BL/6 background for at least three generations before in vivo experiments.

To generate the ROSA26^TIR1 strain, a targeting construct was assembled by cloning a TIR1 (wild-type; WT)-3xMyc-P2A-mCherry fragment into the FseI-linearized Ai32-targeting vector (Addgene, #34880), positioned between a loxP-Stop-loxP cassette and a WPRE (woodchuck hepatitis virus post-transcriptional regulatory element). The complete targeting vector included the ROSA26 left homology arm, the CAG promoter, a loxP-Stop-loxP cassette, WPRE, a bovine growth hormone (bGH) polyadenylation signal, an AttB-neo-AttP drug selection cassette and the ROSA26 right homology arm. The linearized construct was transfected into albino C57BL/6 ES cells. Neomycin-resistant clones were screened by Southern blot, followed by PCR to confirm correct targeting. Karyotypically normal ES cell clones were used to generate chimeras, which were subsequently bred with albino C57BL/6 mice. Germline-transmitted founders were crossed with a PhiC31-deleter strain to remove the neomycin resistance cassette.

To generate the ROSA26^TIR1(F74G) strain, CRISPR-mediated gene editing was used to introduce a point mutation in ROSA26^TIR1/+ zygotes, converting phenylalanine (F; TTC) to glycine (G; GGC) at position 74.

gRNA spacer sequence (synthesized by IDT):

GAAGCGGCTGAAGTTGGAGC

Homology-directed repair template sequence (synthesized by IDT as single-stranded DNA):

GCGGCGTCTTCGTGGGCAACTGCTACGCCGTGCGCGCCGGCCGCGTCGCCGCGCGGTTCCCCAACGTGCGGGCGCTCACGGTGAAGGGGAAGCCACACGGCGCCGACTTCAACCTCGTGCCCCCCGACTGGGGCGGCTACGCGGGGCCGTGGATCGAGGCGGCCGCGAGGGGATGCCACGGCCTGGAGGAGCTCAGGATG

In addition to the desired point mutation, a silent mutation (CCC to CCA) was included in the homology-directed repair template to prevent re-cutting by Cas9. Offspring were screened by PCR amplification followed by KasI restriction digest. Confirmed mutants were validated by Sanger sequencing and bred to C57BL/6 mice to establish germline-transmitting founders.

Treatment of mice with 5-ph-IAA and diphtheria toxin

DT (List Biological Laboratories, 150) was dissolved in phosphate-buffered saline (PBS) and administered intraperitoneally (i.p.) at a dose of 20 μg kg⁻¹ on day 0. This was followed by six subsequent injections of 5 μg kg⁻¹ every other day. Mice were killed for analysis on day 14. Then, 5-ph-IAA (MedChemExpress, HY-134653) was dissolved with 0.2 M NaOH, diluted in PBS and administered daily via i.p. injection at a dose of 10 mg kg⁻¹ per mouse.

Reagents and antibodies

The following antibodies and reagents were used in this study for flow cytometry, with clones, venders, catalog numbers and dilutions as indicated: anti-Siglec-F (E50-2440, BD, 562681, 1:400), anti-I-A/I-E (M5/114.15.2, Biosciences, 566086, 1:1,200), anti-NK1.1 (PK136, Thermo Fisher, 47-5941-82, 1:400), anti-CD45 (30-F11, BioLegend, 103136, 1:600), anti-CD11b (M1/70, BioLegend, 101257, 1:800), anti-CD11b (M1/70, BD Biosciences, 363-0112-82, 1:400), anti-CD3ε (17A2, BioLegend, 100237, 1:500), anti-γδTCR (GL3, BD Biosciences, 750410, 1:300), anti-Cd278 (C398.4A, BD Biosciences, 567918, 200), anti-TCR (H57-597, BD Biosciences, 748405, 1:300), anti-TCR β (H57-597, Thermo Fisher, 47-5961-82, 1:300), anti-TCR (H57-597, BioLegend, 109227, 1:200), anti-TCR β (H57-597, Thermo Fisher, 12-5961-83, 1:400), anti-CD153 (RM153, BD Bioscience, 741575, 1:400), anti-CD24 (M1/69, Thermo Fisher, 46-0242-82, 1:800), anti-CD304 (3E12, BioLegend, 145209, 1:300), anti-CD3 (IM7, BioLegend, 103049, 1:400), anti-CD44 (IM7, BD Biosciences, 563971, 1:400), anti-CD44 (IM7, BioLegend, 103026, 1:100), anti-ZsGreen (polyclonal, Frontier Institute Co., LTD, MSFR106470, 1:800), anti-KLRG1 (2F1, Thermo Fisher, 35-5893-82), anti-CD39 (24DMS1, Thermo Fisher, 25-0391-82, 1:400), anti-TCF1 (C63D9, Cell Signaling, 6709, 1:200), anti-IL-10 (JES5-16E3, BioLegend, 505021, 1:200), anti-CD4 (RM4-5, BD Biosciences, 414-0042-82, 1:400), anti-CD4 (RM4-5, BioLegend, 100536, 1:400), anti-CD4 (RM4-5, Thermo Fisher, 47-0042-82, 1:400), anti-CD4 (RM4-5, BioLegend, 100548, 1:400), anti-TNF (MP6-XT22, BioLegend, 506329, 1:400), anti-IFNg (XMG1.2, BioLegend, 505836, 1:200), anti-IL-22 (1H8PWSR, Thermo Fisher, 46-7221-80, 1:400), anti-IL-13 (eBio13A, Thermo Fisher, 12-7133-82, 1:400), anti-IL-4 (11B11, Thermo Fisher, 17-7041-82, 1:300), anti-CD11c (N418, Thermo Fisher, 48-0114-82, 1:200), anti-Ly6c (HK1.4, BioLegend, 128037, 1:1,200), anti-Ly6C (HK1.4, BioLegend, 128041, 1:1,000), anti-CD122 (TM-β1, BD Bioscience, 564763, 1:200), anti-GARP (YGIC86, Thermo Fisher, 25-9891, 1:200), anti-CD86 (GL1, Thermo Fisher, 12-0862-85, 1:400), anti-Ly6G (1A8, BioLegend, 127618, 1:500), anti-CD64 (X54-5/7.1, BioLegend, 139306, 1:200), anti-CD127 (A7R34, Tonbo Bioscience, 20-1271-U100, 1:200), anti-CD122 (5H4, Thermo Fisher, 13-1221-82, 1:200), anti-Guinea Pig (polyclonal, Thermo Fisher, SA5-10094, 1:1,000), anti-FR4 (12A5, BD Biosciences, 744121, 1:200), anti-FR4 (12A5, BD Biosciences, 560318, 1:200), anti-OX40 (OX-86, Thermo Fisher, 46-1341-82, 1:300), anti-CD120b (TR75-89, BD Bioscience, 564088, 1:200), anti-CD103 (M290, BD Biosciences, 566118, 1:300), anti-Ly-6C (HK1.4, BioLegend, 128037, 1:1,000), anti-CD90.2 (30-H12, BioLegend, 105320, 1:800), anti-CD90.2 (53-2.1, BD Biosciences, 564365, 1:1,500), anti-Foxp3 (FJK-16s, Thermo Fisher, 48-5773-82, 1:200), anti-Foxp3 (FJK-16s, Thermo Fisher, 17-5773-82, 1:200), anti-CD19 (6D5, BioLegend, 115510, 1:600), anti-F4|80 (BM8, BioLegend, 123133, 1:200), anti-CD4 (RM4-5, EBioscience, 564667, 1:400), anti-CD4 (RM4-5, BioLegend, 100553, 1:400), anti-CD8α (53-6.7, BioLegend, 100780, 1:600), anti-CD8α (53-6.7, BioLegend, 564297, 1:400), anti-CD8α (53-6.7, BioLegend, 100752, 1:500), anti-GITR (DTA-1, Thermo Fisher, 48-5874-82, 1:500), anti-CD73 (eBioTY/11.8, Thermo Fisher, 46-0731-82, 1:400), anti-CD73 (TY/11.8, BioLegend, 127208, 1:400), anti-CD62L (MEL-14, BioLegend, 104441, 1:100), anti-CD62L (MEL-14, BD Biosciences, 565213, 1:600), anti-CD62L (MEL-14, BD Biosciences, 741230, 1:800), anti-CD62L (MEL-14, BioLegend, 104441, 1:400), anti-CD62L (MEL-14, BioLegend, 104438, 1:1,600), anti-CTLA4 (UC10-4B9, BioLegend, 106323, 1:200), anti-CTLA4 (UC10-4B9, Thermo Fisher, 12-1522-82, 1:400), anti-Helios (22F6, BioLegend, 137216, 1:400), anti-Helios (22F6, BioLegend, 137236, 1:400), anti-Eos (ESB7C2, Thermo Fisher, 12-5758-82, 1:400), anti-Ki-67 (SolA15, Thermo Fisher, 61-5698, 1:2,000), anti-Ki67 (B56, BD Biosciences, 563757, 1:1,000), anti-Ki67 (SolA15, Fisher Scientific, 15-5698-82, 1:8,000), anti-CD25 (PC61, BD Biosciences, 564022, 1:300; Thermo Fisher, 17-0251-82, 1:400), anti-PD-1 (29 F.1A12, BioLegend, 135225, 1:400), anti-CD45 (30-F11, BioLegend, 103157, 1:1,000), anti-IL-2 (JES6-5H4, BioLegend, 503818, 1:400), streptavidin (Thermo Fisher, 46-4317-82, 1:1,000), Picolyl-Azide (Jena Bioscience, CLK-1288-5), CTV (Thermo Fisher, C34557), Zombie NIR dye (BioLegend, 423105, 1:1,000), Sytox Blue (Thermo Fisher, S34857) and anti-mouse CD16/32 (2.4G2, Tonbo, 70-0161-M001, 1:500).

The following capturing antibodies were used for ELISA: anti-mouse IL-13 (14-7133-68, Invitrogen, 88-7137-88), anti-mouse IL-4 (14-7041-68 A, Invitrogen, 88-7044-88), anti-mouse IL-2 (eBioscience, 14-7022-68), anti-mouse IgE (R35-72, BD Pharmingen, 553413), goat anti-mouse IgG1 (2794408, Southern Biotech, 1070-01), goat anti-mouse IgG3 (2794567, Southern Biotech, 1100-01), goat anti-mouse IgG2a (2794475, Southern Biotech, 1080-01), goat anti-mouse IgG2b (2794517, Southern Biotech, 1090-01), goat anti-mouse IgG2c (2794464, Southern Biotech, 1079-01), goat anti-mouse IgA (2314669, Southern Biotech, 1040-01) and goat anti-mouse IgM (2794197, Southern Biotech, 1020-01). The following detection antibodies were used for ELISA: biotin anti-mouse IL-13 (13-7135-68A, Invitrogen, 88-7137-88), anti-mouse IL-4 (13-7042-68C, Invitrogen, 88-7044-88), anti-mouse IL-2 (eBioscience, 33-7021-68), goat anti-mouse Ig (2728714, Southern Biotech, 1010-05) and biotin rat anti-mouse IgE (R35-118, BD Pharmingen, 553419).

The following reagents were used to generate ELISA standard curves: mouse IL-4 lyophilized standard (39-8041-60, Invitrogen, 88-7044-88), mouse IL-13 lyophilized standard (39-7137/2EB-60, Invitrogen, 88-7137-88), mouse IL-2 (Thermo Fisher 212-12-5UG), mouse IgG1, κ, isotype control (15H6, Southern Biotech, 0102-01), mouse IgG2a, κ, isotype control (UPC-10, Sigma, M5409), IgG2b isotype control (MOPC-141, Sigma, M5534), Mouse IgG2c (6.3, AB_2794064, Southern Biotech, 0122-01), purified mouse IgG3, κ, isotype control (A112-3, BD Pharmingen, 553486), Purified Mouse IgA, κ, isotype control (M18-254, BD Pharmingen, 553476), IgM isotype control from murine myeloma (MOPC 104E, Sigma, M5909), purified mouse IgE, κ, isotype control (C38-2, BD Pharmingen, 557079).

Enzyme-linked immunosorbent assay

ELISA experiments for IL-2, IL-4 and IL-13 were performed in the following way. Cells were isolated from pooled SLOs (peripheral lymph nodes (cervical, axillary, brachial and inguinal) and spleen) and cultured on a 96-well U-bottom plate with 5% CO₂ in 200 μl complete cell culture medium (RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS), 100 U ml⁻¹ penicillin–streptomycin, 2 mM L-glutamine, 10 mM HEPES and 50 µM β-mercaptoethanol) and recombinant human IL-2 (0.5 U µl⁻¹, Roche, C168121-01). Cells were treated with 5-ph-IAA (5 mM daily) in the presence of anti-CD3/CD28 activation beads (Thermo Fisher, 11452D). The culture was terminated after 3 days and the supernatant was used for the detection of the aforementioned cytokines. In brief, a 96-well flat-bottom plate was coated with capture antibody in Coating Buffer (00-0000-53, Invitrogen, 88-7044-88) and incubated overnight at 4 °C. The next day, the plate was washed and blocked with ELISA/ELISPOT diluent (00-4202-55, Invitrogen, 88-7044-88) at room temperature for 1 h. Then, the plate was washed and serial dilutions of standards were performed using the ELISA/ELISPOT diluent. Next, samples were added to the plate and incubated overnight at 4 °C. The next day, the plate was washed and the detection antibody was added and incubated for 1 h at room temperature. Next, the plate was washed and incubated with streptavidin–HRP for IL-2 and IL-4 (00-50050-68, Invitrogen, 88-7044-88) or avidin–HRP for IL-13 (00-4100-94, Invitrogen, 88-7137-88) for 30 min at room temperature. The plate was then washed and incubated with TMB solution (00-4201-56, Invitrogen, 88-7044-88) at room temperature for 5–15 min. Finally, 1 M H₃PO₄ (Sigma-Aldrich, P5811) was added to the plate to stop the colorimetric reaction.

Antibody ELISAs were conducted as previously described. In brief, mouse peripheral blood was collected via cardiac puncture immediately after killing into BD SST microcontainer tubes (02-675-185) and sera were collected after centrifugation. Flat-bottom 96-well plates were coated with capturing antibodies in 50 µl 0.1 M NaHCO₃ solution at pH 9.5 overnight at 4 °C. The plates were then emptied, blocked with 200 µl 1% bovine serum albumin (VWR, 97061-422) in PBS and washed three times with PBS containing 0.05% Tween-20 (Sigma-Aldrich, P1379). Then, 50 µl serum at appropriate dilutions was added and incubated overnight at 4 °C. The plate was then incubated with 50 µl biotinylated detection antibodies at 37 °C for 2–3 h, followed by 50 µl avidin–HRP (Thermo Fisher, 18-4100-51) at 37 °C for 30 min and 100 µl TMB solution (Thermo Fisher, 00-4201-56) at room temperature, with 3–4 washes with PBS–Tween in between each incubation step. The colorimetric reaction was stopped with 100 µl 1 M H₃PO₄ after 5–10 min.

Absorbance at 450 nm was measured with a Synergy HTX plate reader (BioTek). Concentrations of antigens were determined using standard curves constructed with purified recombinant proteins and calculated with Gen5 3.02.2 (BioTek).

Isolation of cells from lymphoid organs, lungs and tumors

For flow cytometry analyses, animals were killed and perfused with 20 ml PBS. Cells were isolated from lymphoid organs by meshing with syringe plunger through a 100-mm cell strainer (Corning, 07-201-432). Lungs and tumors were digested in RPMI 1640 with 2% FBS, 10 mM HEPES buffer, 100 U ml⁻¹ penicillin–streptomycin, 2 mM L-glutamate, 0.2 U ml⁻¹ collagenase A (Sigma, 11088793001) and 1 U ml⁻¹ DNase I (Sigma-Aldrich, 10104159001) for 45 min at 37 °C with vigorous shaking at 250 rpm. Then, 6.35-mm ceramic beads (MP Biomedicals, 116540034) were included to help with tissue dissociation. The digested lungs were filtered through 70-mm separation filters (Miltenyi Biotec, 130-095-823), washed and centrifuged in PBS-adjusted 40% Percoll (Sigma-Aldrich, 17-0891-01) to enrich for lymphocytes. Erythrocytes from spleen, lung and liver were lysed using ACK lysis buffer (150 mM NH₄Cl (Sigma-Aldrich, A9434), 10 mM KHCO₃ (Sigma-Aldrich, P7682) and 0.1 mM Na₂EDTA at pH 7.4).

For flow cytometry analysis, cells were stained with Zombie NIR dye in PBS for 10 min at 4 °C to identify the dead cells followed by staining with anti-mouse CD16/32 in staining buffer (PBS with 0.2% bovine serum albumin (BSA), 10 mM HEPES buffer and 2 mM EDTA) for 10 min at 4 °C to block the Fc receptors. Next, cells were stained with fluorescently conjugated antibodies detecting cell surface antigens for 30 min at 4 °C. To access the intracellular antigens, cells were fixed and permeabilized with eBioscience transcription factor staining buffer set (00-5523-00) according to the manufacturer’s instructions. Samples were recorded on Aurora cytometer (Cytek) by using of SpectroFlo software v.3.1.2 and analyzed in FlowJo v.10.10.0.

For cell sorting, cells isolated from pooled peripheral lymph nodes (cervical, axillary, brachial and inguinal) and spleen were enriched for CD4⁺ T cells using a mouse CD4⁺ T Cell Isolation kit (Miltenyi, 130-104-454) according to the manufacturer’s instructions. Next, samples were stained with antibodies, washed and resuspended in a Sytox Blue-containing (1:8,000) staining buffer to exclude dead cells. T_reg cells (CD4⁺TCRβ⁺ZsGreen⁺ from Foxp3^AIDROSA26^WT mice and CD4⁺TCRβ⁺ZsGreen⁺mCherry⁺ from Foxp3^AIDROSA26^TIR1(F74G) mice) and naive CD4⁺ T cells (CD4⁺TCRβ⁺ZsGreen⁻CD44^loCD62L^hi) were sorted into cell culture medium.

Flow cytometric analysis of cytokine production

To measure cytokine production following ex vivo stimulation, a single-cell suspension was incubated with 5% CO₂ at 37 °C for 4 h in cell culture medium (200 ml per well) supplied with 50 ng ml⁻¹ phorbol-12-myristate-13-acetate (Sigma-Aldrich, P8139), 500 ng ml⁻¹ ionomycin (Sigma-Aldrich, I0634), 2 μM monensin (Sigma-Aldrich, M5273) and 1 μg ml⁻¹ brefeldin A (Sigma-Aldrich, B6542). Cells were stained for flow cytometry as described above except for the fixation/permeabilization step and cytokine staining in which case BD Cytofix/Cytoperm Kit (BD Biosciences, 554715) was used according to the manufacturer’s instructions.

Ex vivo CTV labeling

Sorted T_reg cells were labeled with CTV (5 mM) and cultured with 5% CO₂ on a 48-well flat-bottom plate pre-coated with anti-CD3/CD28 antibody with concentration of 5 mg ml⁻¹ each of anti-CD3 (145-2C11, BioXcell, BE0001-1) and anti-CD28 (37.51, BioXcell, BE0015-1). The culture was maintained in cell culture medium (600 ml per well) supplied with a recombinant human IL-2 (0.5 U µl⁻¹). Cells were treated with 5-ph-IAA daily (5 mM). The first part of the culture was terminated after 16 h to capture undivided cells and the cells that divided once. Another part of the culture was terminated after 72 h to capture the cells dividing twice and more. Cells were stained for analysis as described above.

In vitro suppression assay

Cells were isolated from pooled SLOs and enriched for CD4⁺ T cells as described above. Resting T_reg cells (CD4⁺TCRβ⁺ZsGreen⁺CD62L^hiCD44^lo) were FACS sorted and 40,000 cells per well were plated on a 96-well U-bottom plate. Then, 40,000 sorted naive CD4⁺ T cells from CD45.1 Foxp3^DTR mice (CD4⁺TCRβ⁺GFP⁻CD62L^hiCD44^lo) and 100,000 red blood cell-lysed splenocytes from Tcrb^−/−Tcrd^−/− mice were added to the culture. Next, anti-CD3 (145-2C11, BioXcell, BE0001-1) antibody was added to each well to a final concentration of 1 mg ml⁻¹. Cells were treated with 5-ph-IAA daily (5 mM). Cells were incubated in cell culture medium (200 ml per well) for 72 h with 5% CO₂ and then prepared for flow cytometry as described above. CTV-labeled Naive CD4⁺ T cells that were divided more than four times were used to calculate T_reg cell-mediated suppression with the following formula: %Suppression (sample X) = (%divided (no T_reg) − %divided (sample X))/(%divided (no T_reg)), where sample X refers to any given sample.

Histology

Tissues were fixed in 10% neutral buffered formalin, transferred into 70% alcohol and sent to the HistoWiz for the downstream services. In brief, tissues were embedded in paraffin and sectioned into 5-mm slices followed by H&E staining. Lymphocytic infiltration was blindly scored with the criteria, 0, normal; 1, mild increase; 2, moderate increase; and 3, severe increase.

5-ethynyl-2’-deoxyuridine labeling

Mice were injected i.p. with EdU (5 mg per mouse, MedChemExpress, HY-118411) for three consecutive days. Cells were stained with surface and intracellular antibodies as described above. To stain for EdU, a click reaction was performed after staining with intracellular antibodies. In brief, cells were incubated in the Click Reaction Buffer (CuSO₄ 16.67 mM (Fisher Science Education,7758-98-7), BTTAA(2-(4-((bis((1-(tert-butyl)-1H-1,2,3-triazol-4-yl)methyl)amino)methyl)-1H-1,2,3-triazol-1-yl)acetic acid) 41.67 mM (Jena Bioscience, CLK-067-100), picolyl-azide Alexa Fluor 555 1.25 µM (Jena Bioscience, CLK-1288-5), sodium ascorbate 100 mM (Spectrum, 134-03-2), H₂O and PBS, pH 7.4) for 1 h at room temperature followed by extensive washing before recording data on the cytometer.

B16 melanoma model

B16-OVA melanoma cells⁵⁷ were cultured in RPMI medium supplemented with 10% FBS (Gibco), 100 U ml⁻¹ penicillin–streptomycin at 37 °C in a humidified incubator with 5% CO₂. Cells were maintained in logarithmic growth phase and collected at ~70–80% confluency using 0.05% trypsin-EDTA. Cells were washed twice with sterile PBS. For tumor implantation, 1.5 × 10⁶ cells in 100 μl sterile PBS were injected subcutaneously into the right flank of mice using a 27G needle. Starting from day 5 after tumor implantation, mice were injected daily with 5-ph-IAA.Mice were monitored daily for general health and tumor growth. Tumor dimensions were measured using calipers and the volume was calculated using the formula (length × width²)/2.

Bulk RNA-seq and data analysis

For bulk RNA-seq from adult female Foxp3^AID/+ROSA26^WT and Foxp3^AID/+ROSA26^TIR1(F74G) mice, single cell suspensions from pooled SLOs or thymus were enriched for CD4⁺ T cells as described above. CD4⁺TCRβ⁺ZsGreen⁺CD62L^hiCD44^lo resting T_reg cells and CD4⁺TCRβ⁺ZsGreen⁺CD62L^loCD44^hi activated T_reg cells from the SLOs, as well as CD4⁺TCRβ⁺ZsGreen⁺CD73⁻ nascent T_reg cells from the thymus, were double sorted into TRIzol Reagent (Thermo Fisher, 15596-018). RNA extraction was then performed according to the manufacturer’s instructions. In brief, phase separation in cells lysed in 1 ml TRIzol Reagent was induced with 200 µl chloroform. RNA was extracted from 350 µl of the aqueous phase using the miRNeasy Micro kit (QIAGEN cat. no. 217084) on the QIAcube Connect (QIAGEN) according to the manufacturer’s protocol. Samples were eluted in 15–18 µl RNase-free water.

After RiboGreen quantification and quality control by Agilent BioAnalyzer, 1–2 ng total RNA with RNA integrity numbers ranging from 5.5 to 9.3 underwent amplification using the SMART-Seq v.4 Ultra Low Input RNA kit (Clonetech cat. no. 63488), with 12 cycles of amplification. Subsequently, 4.5–8 ng of amplified cDNA was used to prepare libraries with the KAPA Hyper Prep Kit (Roche 07962363001) using eight cycles of PCR. Samples were barcoded and run on a NovaSeq 6000 in a PE100 run, using the NovaSeq 6000 S4 Reagent kit (200 cycles) (Illumina). An average of 58 million paired reads were generated per sample and the percent of mRNA bases per sample ranged from 79% to 87% and ribosomal reads averaged 0.35%.

Bulk RNA-seq was aligned to the mm39 reference using the STAR aligner v.2.7.10b run with default settings. Reads were counted using Rsubread FeatureCounts v.2.8.2 to create gene count matrices with the following settings:–primary -C -Q 30–countReadPairs -B -p.

For bulk RNA sequencing of in vitro activated and Foxp3-degraded cells (Fig. 6g), cells from Foxp3^AID/yROSA26^WT and Foxp3^AID/yROSA26^TIR1(F74G) were isolated from pooled SLOs and enriched for CD4⁺ T cells as described above. T_reg cells (CD4⁺TCRβ⁺ZsGreen⁺, mCherry⁻ for ROSA26^WT and mCherry⁺ for ROSA26^TIR1(F74G)) were FACS sorted and cultured at a density of 1 × 10⁶ cells per ml in complete RPMI (RPMI 1640 medium supplemented with 10% FBS, 100 U m⁻¹ penicillin–streptomycin, 2 mM L-glutamine, 10 mM HEPES and 50 µM β-mercaptoethanol), recombinant human IL-2 (2,000 U ml⁻¹) and Dynabeads Mouse T-Activator CD3/28 (2:1 bead-to-cell ratio, Gibco). T_reg cells were cultured at 37 °C in a humidified incubator with 5% CO₂ for 3 days. T_reg cells were washed with complete RPMI, resuspended at a density of 1 × 10⁶ cells per ml in complete RPMI, and rested in the incubator for 2 h, after which 5-ph-IAA (5 mM) was added to the medium. After 7 h, cells were lysed in RNA lysis buffer and RNA was extracted using the Quick-RNA Microprep kit (Zymo Research). Five technical replicates were performed for each genotype. RNA was quantified by NanoDrop. RNA-seq libraries were generated using the BRB-seq kit (Alithea Genomics) as per the manufacturer’s instructions. Then, 200 ng RNA per sample was used as input. Final sequencing libraries were run on an AVITI instrument (Element Biosciences) with the following read structure: index 1, eight cycles; index 2, eight cycles; read 1, 28 cycles; and read 2, 90 cycles. An average of 5.7 million reads and 3.9 million unique molecular identifiers (UMIs) were acquired per sample. FASTQ files were aligned to the GRCm39 reference genome and a counts matrix was generated using STARsolo⁵⁸ following commands provided in the Alithea Genomics BRB-seq user guide (v.0.8.1. rev A August 2024).

Single-cell RNA analysis of SLO T_reg cells

Foxp3^AID/+ROSA26^WT and Foxp3^AID/+ROSA26^TIR1(F74G) mice were treated with 5-ph-IAA for 0, 3 and 7 days. Enriched CD4⁺ T cells from the spleen and lymph nodes were barcoded using hashtag oligonucleotides (BioLegend). FACS-sorted Foxp3^AIDR26^WT and Foxp3^AIDR26^TIR1(F74G) T_reg cells (four biological replicates of each genotype) within each time point were multiplexed together, washed once with PBS containing 0.04% BSA and resuspended in PBS containing 0.04% BSA to a final concentration of 700–1,200 cells per μl. For each time point, RNA sequencing was performed in one lane on a Chromium instrument (10x Genomics) following the user guide manual for 3′ v.3.1. The viability of cells was above 80%, as confirmed with 0.2% (w/v) Trypan blue staining (Countess II). Approximately 30,000 cells were targeted. Following reverse transcription and cell barcoding in droplets, emulsions were broken and cDNA was purified using Dynabeads MyOne SILANE followed by PCR amplification as per the manual instructions. Final libraries were sequenced on Illumina NovaSeq 6000-S4 platform (R1, 28 cycles; i7, eight cycles; and R2, 90 cycles). Single-cell RNA-seq count matrices were generated using Cell Ranger v.6.1.2. Replicates from the same day of treatment (Tir1 or WT) were demultiplexed using hashtag oligonucleotides. This resulted in a filtered UMI matrix containing 45,314 cells and 32,285 genes, which was analyzed using Scanpy v.1.9.3. Cells were filtered based on mitochondrial counts and gene expression levels. Lowly expressed genes, nonprotein-coding genes, pseudogenes and genes corresponding to TCR/BCR were removed from the data. Gene expression was normalized using Pearson residual normalization with theta = 1. Principal-component analysis (PCA) was run with 100 principal components (PCs) and a k-nearest neighbors (k-NN) graph was established with k = 30. Leiden clustering was run with a resolution of 1. UMAP was run with default parameters. Clusters were scored using gene sets from ImmGen. Two clusters (342 cells) scoring high in pDC and NK gene signatures were filtered out. Together, this resulted in a filtered gene expression matrix of 39,256 cells and 9,155 genes. To remove the effect from genes that were differential due only to genotype (WT/Tir) and not the degradation of Foxp3, we identified differential genes between D0 Tir1 and D0 WT using DESeq2 and filtered these genes out from the data matrix, resulting in a matrix of 39,256 cells and 9,119 genes. Cells were then re-clustered according to the protocol (PCA, k-NN, Leiden and UMAP) described above.

Single-cell RNA-seq analysis of intratumoral T_reg cells

Single-cell preparation was performed using GEM-X Flex (10x Genomics). Foxp3^AID/yROSA26^WT and Foxp3^AID/yROSA26^TIR1(F74G) mice (three per group) were injected subcutaneously with 1.5 × 10⁶ B16-OVA cells in 100 μl sterile PBS. Starting on day 5 post-implantation, 200 μg 5-ph-IAA in 200 μl was injected i.p. daily. On day 14 post-implantation, tumors were processed and prepared for sequencing using the GEM-X Flex assay (10x Genomics). Following dissociation into single-cell suspensions, cells were stained with fluorescent and TotalSeq-C anti-mouse hashtag antibodies (BioLegend) and washed extensively. Cells from each mouse were stained with a unique hashtag antibody. Cells were pooled and fixed in Fixation Buffer B (10x Genomics) at 4 °C for 18 h. After quenching and washing cells as per the manufacturer’s instructions (10x Genomics, demonstrated protocol CG000782 Rev A), cells were resuspended in 500 μl Quenching Buffer B. Reporter-positive live/dead stain negative T_reg cells (CD45⁺CD3⁺TCRβ⁺CD8α⁻CD4⁺ZsGreen⁺mCherry⁺ for R26^TIR1(F74G) mice or mCherry⁻ for R26^WT mice) were isolated from fixed single-cell suspensions using FACS and sorted into PBS with 1% nuclease-free BSA Fraction V (Millipore Sigma, 126609) and 0.2 U μl⁻¹ RNase-inhibitor (Millipore Sigma, PN-3335399001). Following FACS isolation, cells were spun down, resuspended in 200 μl of Quenching Buffer B and pooled to normalize cell numbers between different mice. The remaining steps were performed as per the manufacturer’s instructions (10x Genomics, user guide CG000788 Rev A). In brief, cells were hybridized for 18 h, washed and encapsulated in one lane targeting 60,000 cells. Indexing PCR steps were performed using nine amplification cycles for both the gene expression and cell surface protein libraries. Final sequencing libraries were sequenced on an Illumina NovaSeq X Plus System (index 1, ten cycles; index 2, ten cycles; read 1, 28 cycles; and read 2, 88 cycles).

FASTQ files were processed using Cell Ranger v.9.0 and reads were aligned to the Chromium Mouse Transcriptome Probe Set v.1.1.0 GRCm39-2024-A (10x Genomics). Cells containing fewer than 1,000 UMI counts or 1,000 unique genes were filtered out. Cells containing more than 5% mitochondria-derived transcripts were filtered out. Genes that were expressed in >5 cells were retained for further analysis. Hashtag antibody data were demultiplexed using HashSolo with the following prior probabilities (negative, 0.01; singlet, 0.81; and doublet, 0.18). Cells called doublet or negative were filtered out. The resulting count matrix consisted of 35,367 cells × 19,088 genes and was normalized to median UMI counts. The normalized data were then ln(1 + counts)-transformed for downstream analysis.

Single-cell RNA-seq analysis of tissue T_reg cells

Single-cell preparation was performed using GEM-X Flex (10x Genomics). Foxp3^AID/+ROSA26^WT and Foxp3^AID/+ROSA26^TIR1(F74G) mice were treated with 5-ph-IAA for 7 days (four per group) as described above. Animals were killed and perfused with 20 ml PBS. SLOs, lung and liver were collected and digested as described above. LILP was processed as follows. Dissected large intestines (from cecum to anus) were placed in 3 ml wash medium (RPMI 1640, 2% FBS, 10 mM HEPES buffer, 1% penicillin–streptomycin and 2 mM L-glutamine) on ice until they were ready to be processed. Samples were centrifuged at 700g for 4 min at 4 °C unless mentioned otherwise. Each large intestine was placed in a 50-ml screw-cap tube with 25 ml wash medium supplemented with 5 mM EDTA and 1 mM dithiothreitol and shaken horizontally at 250 rpm for 15–20 min at 37 °C to remove the epithelial layer. After a brief vortex, the suspension was filtered through a tea strainer. Intact tissue from the filter was quickly rinsed in PBS and placed into new 50-ml tubes and digested by shaking in 25 ml wash medium supplemented with 0.2 U ml⁻¹ collagenase A and 1 U ml⁻¹ DNase I, along with four ceramic beads, at 250 rpm for 45 min at 37 °C. The suspension was then passed through a 100-μm strainer, centrifuged and the pellets were washed by centrifugation in 40% Percoll in wash medium. Percoll-washed cell pellets were washed once with staining buffer (PBS with 2% FBS and 2 mM EDTA) and used for downstream analysis.

Following dissociation into single-cell suspensions, cells were stained with fluorescent and TotalSeq-C anti-mouse hashtag antibodies (BioLegend) and washed extensively. For each genotype, cells from individual mice and tissues were stained with a unique hashtag antibody. Cells from each genotype were pooled and fixed in Fixation Buffer B (10x Genomics) at 4 °C for 24 h. After quenching and washing of cells per manufacturer instructions (10x Genomics, demonstrated protocol CG000782 Rev A), cells were resuspended in 500 μl Quenching Buffer B. Reporter-positive live/dead stain negative T_reg cells (CD45⁺CD3⁺TCRβ⁺CD8α⁻CD4⁺ZsGreen⁺mCherry⁺ for R26^TIR1(F74G) mice or mCherry⁻ for R26^WT mice) were isolated from fixed single-cell suspensions using FACS and sorted into PBS with 1% nuclease-free BSA Fraction V (Millipore Sigma, 126609) and 0.2 U μl⁻¹ RNase-inhibitor (Millipore Sigma, PN-3335399001). Following FACS isolation, cells were spun down, resuspended in 500 μl Quenching Buffer B, 50 μl Enhancer and 137.5 μl 50% glycerol, and stored at –80 °C until further processing. On the day of processing, cells were thawed and washed as per the manufacturer’s instructions. Probe hybridization was performed as per the manufacturer’s instructions. Cells from ROSA26^TIR1(F74G) and ROSA26^WT genotypes were hybridized with separately barcoded probes. The remaining steps were performed as per the manufacturer’s instructions (10x Genomics, user guide CG000788 Rev A). In brief, cells were hybridized for 18 h and washed. Cells from this experiment and the Foxp3^DTR versus Tcrbd KO experiment, which were hybridized with unique barcoded probes, were pooled and encapsulated in one lane targeting 80,000 cells. Indexing PCR steps were performed using nine amplification cycles for both the gene expression and cell-surface protein libraries. Final sequencing libraries were sequenced on an Element Biosciences AVITI24 system (index 1, ten cycles; index 2, ten cycles; read 1, 28 cycles; and read 2, 90 cycles).

FASTQ files were processed using Cell Ranger v.9.0 and reads were aligned to the Chromium Mouse Transcriptome Probe Set v.1.1.0 GRCm39-2024-A (10x Genomics). Cells containing fewer than 800 UMI counts or 800 unique genes were filtered out. Cells containing more than 5% mitochondrial derived transcripts were filtered out. Genes that were expressed in >10 cells were retained for further analysis. Hashtag antibody data were demultiplexed using HashSolo with the following prior probabilities (negative, 0.01; singlet, 0.91; and doublet, 0.08). Cells called doublet or negative were filtered out. The resulting count matrix consisted of 32,709 cells × 19,087 genes and was normalized to median UMI counts. The normalized data were then ln(1 + counts)-transformed for downstream analysis.

Single-cell RNA-seq analysis of adoptively transferred T_reg cells

Single-cell preparation was performed using GEM-X Flex (10x Genomics). On the day of transfer (day 0), cells from Foxp3^AID/yROSA26^WT and Foxp3^AID/yROSA26^TIR1(F74G) were isolated from pooled SLOs and enriched for CD4⁺ T cells as described above. T_reg cells (CD4⁺TCRβ⁺ZsGreen⁺, mCherry⁻ for ROSA26^WT and mCherry⁺ for ROSA26^TIR1(F74G)) were FACS sorted. Similarly, CD4⁺TCRβ⁺CD44⁻CD62L⁺CD25⁻ (T_conv) cells were FACS sorted from pooled SLOs of CD45.1 mice. Cells were pooled in appropriate ratios. First, a 1:1 mixture of ROSA26^WT and ROSA26^TIR1(F74G) T_reg cells was created. A portion of this mixture was further pooled with an equal number of T_conv cells. Distribution of cells in these pools was verified by flow cytometric analysis. Cells were washed twice with sterile PBS and resuspended at an appropriate concentration. For transfer into Foxp3^DTRCD45.1 mice, 1 × 10⁶ cells from the T_reg cell-only pool (0.5 × 10⁶ ROSA26^WT, 0.5 × 10⁶ ROSA26^TIR1(F74G)) were injected retro-orbitally in 200 μl sterile PBS (four mice total). For transfer into Tcrbd KO mice, 4 × 10⁶ cells from the T_reg + T_conv pool (1 × 10⁶ ROSA26^WT, 1 × 10⁶ ROSA26^TIR1(F74G), 2 × 10⁶ T_conv) were injected retro-orbitally in 200 μl sterile PBS (four mice total). For continuous endogenous T_reg depletion, Foxp3^DTRCD45.1 were injected i.p. with 1,000 ng DT in 200 μl sterile PBS on day –1 and day 0 followed by 500 ng on day 3 and day 6. Continuous Foxp3 degradation was maintained by daily 5-ph-IAA injections as described above. On day 7, animals were killed. Pooled SLOs were collected from each mouse and enriched for CD4⁺ T cells as described above. Following dissociation into single-cell suspensions, cells were stained with fluorescent and TotalSeq-C anti-mouse hashtag antibodies (BioLegend) and washed extensively. Cells from each mouse were stained with a unique hashtag antibody. Cells from individual Foxp3^DTR mice were pooled and cells from individual Tcrbd KO mice were pooled. Each pool was fixed in Fixation Buffer B (10x Genomics) at 4 °C for 20 h. After quenching and washing cells as per the manufacturer’s instructions (10x Genomics, demonstrated protocol CG000782 Rev A), cells were resuspended in 500 μl Quenching Buffer B. Reporter-positive live/dead stain negative T_reg cells (TCRβ⁺CD8α⁻CD4⁺ZsGreen⁺mCherry⁺ or mCherry⁻) were isolated from fixed single-cell suspensions using FACS and sorted into PBS with 1% nuclease-free BSA Fraction V (Millipore Sigma, 126609) and 0.2 U μl⁻¹ RNase inhibitor (Millipore Sigma, PN-3335399001). mCherry⁺ and mCherry⁻ cells were sorted into separate tubes for each condition. All sorted mCherry⁺ cells were pooled and all sorted mCherry⁻ cells were pooled. Probe hybridization was performed as per the manufacturer’s instructions. Cells from mCherry⁺ and mCherry⁻ pools were hybridized with separately barcoded probes. Single-cell encapsulation and library preparation were performed together with the cross-tissue experiment as described above.

FASTQ files were processed using Cell Ranger v.9.0 and reads were aligned to the Chromium Mouse Transcriptome Probe Set v.1.1.0 GRCm39-2024-A (10x Genomics). Cells containing fewer than 800 UMI counts or 800 unique genes were filtered out. Cells containing more than 5% mitochondria-derived transcripts were filtered out. Genes that were expressed in >10 cells were retained for further analysis. Hashtag antibody data were demultiplexed using HashSolo with the following prior probabilities (negative, 0.01; singlet, 0.91; and doublet, 0.08). Cells called doublet or negative were filtered out. The resulting counts matrix consisted of 8,449 cells × 19,085 genes and was normalized to median UMI counts. The normalized data were then ln(1 + counts)-transformed for downstream analysis.

Cluster frequency testing

Cluster enrichment was tested by calculating the fraction of cells from each sample belonging to each cluster. These fractions were calculated for all samples and a t-test was undertaken between conditions. P values were false discovery rate (FDR)-corrected using a Benjamini–Hochberg correction. When single-cell samples for WT and TIR1 were matched (taken from the same mouse), a paired t-test was used.

Computational analysis of resting and activated T_reg cells

Active and resting cells were identified using bulk resting and activated T_reg cell gene signatures from Van der Veeken et al.¹² (aTreg/rT_reg, logFC > 0.5; FDR < 0.05). Scoring was performed using the sc.tl.score_genes function. Then, thresholds were used to categorize cells as resting, active or neither. For the D3/D7 SLO scRNA-seq dataset, cells with an active score 0.5 × s.d. above the mean were categorized as active (n = 6,731). Non-active cells with a resting gene score higher than 0.25 × s.d. below the mean were categorized as resting (n = 24,211). All other cells were labeled as ‘neither’. For the multi-tissue scRNA-seq dataset, dividing cells were first filtered out to account for different proportions among different tissues. Then, to account for differences across tissues, tissue-specific scoring was performed. Then, a cutoff was selected for each tissue to select the upper mode of cells after plotting the distribution of active scores. Specifically, the following cutoffs were used. For active cells, active was defined as cells with an active gene score >0.075 (LILP; n = 7,882), 0.02 (lung; n = 1,850), –0.01 (SLO; n = 677) and –0.01 (liver; n = 1055). For resting cells, resting was defined as non-active cells with a resting gene score > –0.075 (LILP), 0.05 (lung), 0.05 (SLO) and 0.1 (liver), and an active gene score <–0.12 (LILP), –0.12 (lung), –0.15 (SLO) and –0.15 (liver), resulting in the following number of resting cells: 25 (LILP), 4,202 (lung), 4,150 (SLO) and 2,581 (liver).

Differential gene expression analysis

Differential gene expression for bulk RNA-seq data was carried out using DESeq2 v.1.42.0. Genes were filtered with a mean read count of 10. DESeq2 was run with the following formula after subsetting to samples from the same cell-type or condition: ~genotype. Differential genes were taken using an FDR cutoff of 0.05, except where otherwise stated. For single-cell data, raw read counts were pseudobulked by replicate and activation status, or for LILP data, by Leiden cluster. Genes with a mean expression per cell >5 × 10⁻² were considered. Then, raw read counts were passed into DESeq2 to identify differential genes at an FDR of 0.05. For the TCRbd/DTR transfer scRNA-seq dataset, cells from WT and TIR1 were paired (taken from the same mouse). To account for this effect, DESeq2 was run with the following formula: ~genotype + mouse_id.

Foxp3 ChIP-seq and ATAC-seq data analysis

Foxp3 peaks were analyzed from Konopacki et al.¹⁸ and lifted over to mm39 using pyliftover v.1.1.17. Foxp3 peaks were annotated to the nearest transcription start site (TSS) within 1 Mb using mm39 GENCODE GTF annotations. Only TSSs corresponding to genes with sufficient expression to be included in bulk RNA-seq analysis (mean read count >10) were included.

ATAC-seq reads were aligned using bowtie2. A peak atlas was generated by running MACS2 v.2.2.7.1. MACS2 was run with a P value cutoff of 0.05, –nomodel, –shift = –100 and –extsize of 200. Irreproducible discovery rate was run on replicates with a threshold of 0.2. Reproducible peaks were merged into a single peak atlas and peak summits overlapping these peaks (n = 248,805) were used for quantification. Accessibility was quantified using Rsubread FeatureCounts v.2.8.2 and differential accessibility was calculated using DESeq2 v.1.42.0 with an FDR cutoff of 0.05.

Motif analysis

The JASPAR2020 Mouse and Human Motif Databases were used to identify motifs at Foxp3 and ATAC-seq peaks using FIMO v.5.2.2 with a default P value cutoff of 1 × 10⁻⁴. Statistical enrichment for motifs was calculated using a Fisher exact test for differential enrichment within one peak set compared to a control peak set.

FC–FC plots

For FC–FC plots between bulk RNA-seq experiments in the context of degradation and Foxp3 KO (Fig. 2), genes were labeled as shared if they shared the same sign in the log fold change and FDR < 0.05. For FC–FC plots between scRNA-seq experiments in the context of TCRbd/DTR or different organs, a more promiscuous threshold was applied to define ‘shared’ DEGs to account for the differences in statistical power between experimental conditions. First, all DEGs with an FDR < 0.05 across either condition were taken. A DEG (FDR < 0.05) in condition 1 was considered ‘shared’ in condition 2 if it met either of the following conditions in condition 2: (1) FDR < 0.2 with the same sign as in condition 1 or (2) |log fold change| > 0.5 with the same sign as in condition 1. The same conditions were applied for condition 2 DEGs with condition 1. This defined the set of all ‘shared’ DEGs. DEGs that were not shared were labeled as condition specific.

Statistics

Statistical significance was determined using tests indicated in the respective figure legends. P values for t-tests and analyses of variance were calculated with GraphPad Prism v.7 and were corrected for multiple hypothesis testing.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

RNA-seq and ATAC-seq data generated in this study have been deposited in the Gene Expression Omnibus and are publicly available. Accession numbers are GSE297451 (bulk RNA-seq), GSE297597 (bulk ATAC-seq), GSE297598 (tumor scRNA-seq) and GSE297622 (tissue T_reg cell and adoptive transfer scRNA-seq). Source data are provided with this paper.

Materials availability

The corresponding authors can provide reagents and materials generated in this study upon request.

Code availability

The code used for computational analysis is available at https://github.com/pritykinlab/scTir1_Analysis. The corresponding authors can provide additional information required to reanalyze the data reported in this paper upon request.

References

Fontenot, J. D., Gavin, M. A. & Rudensky, A. Y. Foxp3 programs the development and function of CD4⁺CD25⁺ regulatory T cells. Nat. Immunol. 4, 330–336 (2003).
CAS PubMed Google Scholar
Hori, S., Nomura, T. & Sakaguchi, S. Control of regulatory T cell development by the transcription factor Foxp3. Science 299, 1057–1061 (2003).
CAS PubMed Google Scholar
Khattri, R., Cox, T., Yasayko, S. A. & Ramsdell, F. An essential role for Scurfin in CD4+CD25+ T regulatory cells. Nat. Immunol. 4, 337–342 (2003).
CAS PubMed Google Scholar
Gavin, M. A. et al. Foxp3-dependent programme of regulatory T-cell differentiation. Nature 445, 771–775 (2007).
CAS PubMed Google Scholar
Lin, W. et al. Regulatory T cell development in the absence of functional Foxp3. Nat. Immunol. 8, 359–368 (2007).
CAS PubMed Google Scholar
Samstein, R. M. et al. Foxp3 exploits a pre-existent enhancer landscape for regulatory T cell lineage specification. Cell 151, 153–166 (2012).
CAS PubMed PubMed Central Google Scholar
Hu, W. et al. Regulatory T cells function in established systemic inflammation and reverse fatal autoimmunity. Nat. Immunol. 22, 1163–1174 (2021).
CAS PubMed PubMed Central Google Scholar
Miyao, T. et al. Plasticity of Foxp3(+) T cells reflects promiscuous Foxp3 expression in conventional T cells but not reprogramming of regulatory T cells. Immunity 36, 262–275 (2012).
CAS PubMed Google Scholar
Rubtsov, Y. P. et al. Stability of the regulatory T cell lineage in vivo. Science 329, 1667–1671 (2010).
CAS PubMed PubMed Central Google Scholar
Oldenhove, G. et al. Decrease of Foxp3⁺ T_reg cell number and acquisition of effector cell phenotype during lethal infection. Immunity 31, 772–786 (2009).
CAS PubMed PubMed Central Google Scholar
Sumida, T. S., Cheru, N. T. & Hafler, D. A. The regulation and differentiation of regulatory T cells and their dysfunction in autoimmune diseases. Nat. Rev. Immunol. 24, 503–517 (2024).
CAS PubMed PubMed Central Google Scholar
van der Veeken, J. et al. The transcription factor Foxp3 shapes regulatory T cell identity by tuning the activity of trans-acting intermediaries. Immunity 53, 971–984 e975 (2020).
PubMed PubMed Central Google Scholar
Williams, L. M. & Rudensky, A. Y. Maintenance of the Foxp3-dependent developmental program in mature regulatory T cells requires continued expression of Foxp3. Nat. Immunol. 8, 277–284 (2007).
CAS PubMed Google Scholar
Holland, A. J., Fachinetti, D., Han, J. S. & Cleveland, D. W. Inducible, reversible system for the rapid and complete degradation of proteins in mammalian cells. Proc. Natl Acad. Sci. USA 109, E3350–E3357 (2012).
CAS PubMed PubMed Central Google Scholar
Fontenot, J. D., Dooley, J. L., Farr, A. G. & Rudensky, A. Y. Developmental regulation of Foxp3 expression during ontogeny. J. Exp. Med. 202, 901–906 (2005).
CAS PubMed PubMed Central Google Scholar
Yesbolatova, A. et al. The auxin-inducible degron 2 technology provides sharp degradation control in yeast, mammalian cells, and mice. Nat. Commun. 11, 5701 (2020).
CAS PubMed PubMed Central Google Scholar
Kim, J. M., Rasmussen, J. P. & Rudensky, A. Y. Regulatory T cells prevent catastrophic autoimmunity throughout the lifespan of mice. Nat. Immunol. 8, 191–197 (2007).
CAS PubMed Google Scholar
Konopacki, C., Pritykin, Y., Rubtsov, Y., Leslie, C. S. & Rudensky, A. Y. Transcription factor Foxp1 regulates Foxp3 chromatin binding and coordinates regulatory T cell function. Nat. Immunol. 20, 232–242 (2019).
CAS PubMed PubMed Central Google Scholar
Nasrallah, R. et al. A distal enhancer at risk locus 11q13.5 promotes suppression of colitis by T_reg cells. Nature 583, 447–452 (2020).
CAS PubMed PubMed Central Google Scholar
Chinen, T. et al. An essential role for the IL-2 receptor in T_reg cell function. Nat. Immunol. 17, 1322–1333 (2016).
CAS PubMed PubMed Central Google Scholar
Xing, S. et al. Tcf1 and Lef1 are required for the immunosuppressive function of regulatory T cells. J. Exp. Med. 216, 847–866 (2019).
CAS PubMed PubMed Central Google Scholar
Yang, B. H. et al. TCF1 and LEF1 control T_reg competitive survival and Tfr development to prevent autoimmune diseases. Cell Rep. 27, 3629–3645 e3626 (2019).
Hwang, S. M. et al. Inflammation-induced Id2 promotes plasticity in regulatory T cells. Nat. Commun. 9, 4736 (2018).
PubMed PubMed Central Google Scholar
Miyazaki, M. et al. Id2 and Id3 maintain the regulatory T cell pool to suppress inflammatory disease. Nat. Immunol. 15, 767–776 (2014).
CAS PubMed PubMed Central Google Scholar
Gerner, M. C. et al. The TGF-b/SOX4 axis and ROS-driven autophagy co-mediate CD39 expression in regulatory T-cells. FASEB J. 34, 8367–8384 (2020).
CAS PubMed Google Scholar
Tanoue, T., Atarashi, K. & Honda, K. Development and maintenance of intestinal regulatory T cells. Nat. Rev. Immunol. 16, 295–309 (2016).
CAS PubMed Google Scholar
Miragaia, R. J. et al. Single-cell transcriptomics of regulatory T cells reveals trajectories of tissue adaptation. Immunity 50, 493–504 e497 (2019).
CAS PubMed PubMed Central Google Scholar
Owen, D. L. et al. Thymic regulatory T cells arise via two distinct developmental programs. Nat. Immunol. 20, 195–205 (2019).
CAS PubMed PubMed Central Google Scholar
Owen, D. L., La Rue, R. S., Munro, S. A. & Farrar, M. A. Tracking regulatory T cell development in the thymus using single-cell RNA sequencing/TCR sequencing. J. Immunol. 209, 1300–1313 (2022).
CAS PubMed Google Scholar
Yang, S., Fujikado, N., Kolodin, D., Benoist, C. & Mathis, D. Immune tolerance. Regulatory T cells generated early in life play a distinct role in maintaining self-tolerance. Science 348, 589–594 (2015).
CAS PubMed PubMed Central Google Scholar
Pacella, I. et al. Fatty acid metabolism complements glycolysis in the selective regulatory T cell expansion during tumor growth. Proc. Natl Acad. Sci. USA 115, E6546–E6555 (2018).
CAS PubMed PubMed Central Google Scholar
Deng, G. Tumor-infiltrating regulatory T cells: origins and features. Am. J. Clin. Exp. Immunol. 7, 81–87 (2018).
PubMed PubMed Central Google Scholar
Feng, B. et al. The type 2 cytokine Fc–IL-4 revitalizes exhausted CD8⁺ T cells against cancer. Nature 634, 712–720 (2024).
CAS PubMed PubMed Central Google Scholar
Liu, M. et al. TGF-β suppresses type 2 immunity to cancer. Nature 587, 115–120 (2020).
CAS PubMed PubMed Central Google Scholar
Zhou, X. et al. Instability of the transcription factor Foxp3 leads to the generation of pathogenic memory T cells in vivo. Nat. Immunol. 10, 1000–1007 (2009).
CAS PubMed PubMed Central Google Scholar
Bailey-Bucktrout, S. L. et al. Self-antigen-driven activation induces instability of regulatory T cells during an inflammatory autoimmune response. Immunity 39, 949–962 (2013).
CAS PubMed PubMed Central Google Scholar
Freuchet, A. et al. Identification of human exT_reg cells as CD16⁺CD56⁺ cytotoxic CD4⁺ T cells. Nat. Immunol. 24, 1748–1761 (2023).
CAS PubMed PubMed Central Google Scholar
Boursalian, T. E., Golob, J., Soper, D. M., Cooper, C. J. & Fink, P. J. Continued maturation of thymic emigrants in the periphery. Nat. Immunol. 5, 418–425 (2004).
CAS PubMed Google Scholar
He, M. et al. Dynamic Foxp3-chromatin interaction controls tunable T_reg cell function. J. Exp. Med. 221, e20232068 (2024).
CAS PubMed PubMed Central Google Scholar
Kwon, H. K., Chen, H. M., Mathis, D. & Benoist, C. Different molecular complexes that mediate transcriptional induction and repression by FoxP3. Nat. Immunol. 18, 1238–1248 (2017).
CAS PubMed PubMed Central Google Scholar
Rudra, D. et al. Transcription factor Foxp3 and its protein partners form a complex regulatory network. Nat. Immunol. 13, 1010–1019 (2012).
CAS PubMed PubMed Central Google Scholar
Zheng, Y. et al. Regulatory T-cell suppressor program co-opts transcription factor IRF4 to control T(H)2 responses. Nature 458, 351–356 (2009).
CAS PubMed PubMed Central Google Scholar
Marson, A. et al. Foxp3 occupancy and regulation of key target genes during T-cell stimulation. Nature 445, 931–935 (2007).
CAS PubMed PubMed Central Google Scholar
Chaudhry, A. et al. CD4+ regulatory T cells control TH17 responses in a Stat3-dependent manner. Science 326, 986–991 (2009).
CAS PubMed PubMed Central Google Scholar
Onizuka, S. et al. Tumor rejection by in vivo administration of anti-CD25 (interleukin-2 receptor α) monoclonal antibody. Cancer Res. 59, 3128–3133 (1999).
CAS PubMed Google Scholar
Yu, P. et al. Intratumor depletion of CD4⁺ cells unmasks tumor immunogenicity leading to the rejection of late-stage tumors. J. Exp. Med. 201, 779–791 (2005).
CAS PubMed PubMed Central Google Scholar
Tawara, I., Take, Y., Uenaka, A., Noguchi, Y. & Nakayama, E. Sequential involvement of two distinct CD4⁺ regulatory T cells during the course of transplantable tumor growth and protection from 3-methylcholanthrene-induced tumorigenesis by CD25-depletion. Jpn J. Cancer Res. 93, 911–916 (2002).
CAS PubMed PubMed Central Google Scholar
Bos, P. D., Plitas, G., Rudra, D., Lee, S. Y. & Rudensky, A. Y. Transient regulatory T cell ablation deters oncogene-driven breast cancer and enhances radiotherapy. J. Exp. Med. 210, 2435–2466 (2013).
PubMed PubMed Central Google Scholar
Klages, K. et al. Selective depletion of Foxp3+ regulatory T cells improves effective therapeutic vaccination against established melanoma. Cancer Res. 70, 7788–7799 (2010).
CAS PubMed Google Scholar
Ganesan, A. P. et al. Tumor-infiltrating regulatory T cells inhibit endogenous cytotoxic T cell responses to lung adenocarcinoma. J. Immunol. 191, 2009–2017 (2013).
CAS PubMed Google Scholar
Joshi, N. S. et al. Regulatory T cells in tumor-associated tertiary lymphoid structures suppress anti-tumor T cell responses. Immunity 43, 579–590 (2015).
CAS PubMed PubMed Central Google Scholar
Jang, J. E. et al. Crosstalk between regulatory T cells and tumor-associated dendritic cells negates anti-tumor immunity in pancreatic cancer. Cell Rep. 20, 558–571 (2017).
CAS PubMed PubMed Central Google Scholar
Amoozgar, Z. et al. Targeting T_reg cells with GITR activation alleviates resistance to immunotherapy in murine glioblastomas. Nat. Commun. 12, 2582 (2021).
CAS PubMed PubMed Central Google Scholar
Teng, M. W. et al. Multiple antitumor mechanisms downstream of prophylactic regulatory T-cell depletion. Cancer Res. 70, 2665–2674 (2010).
CAS PubMed Google Scholar
Golgher, D., Jones, E., Powrie, F., Elliott, T. & Gallimore, A. Depletion of CD25⁺ regulatory cells uncovers immune responses to shared murine tumor rejection antigens. Eur. J. Immunol. 32, 3267–3275 (2002).
CAS PubMed Google Scholar
Sledzinska, A. et al. TGF-β signalling is required for CD4⁺ T cell homeostasis but dispensable for regulatory T cell function. PLoS Biol. 11, e1001674 (2013).
CAS PubMed PubMed Central Google Scholar
Curran, M. A., Montalvo, W., Yagita, H. & Allison, J. P. PD-1 and CTLA-4 combination blockade expands infiltrating T cells and reduces regulatory T and myeloid cells within B16 melanoma tumors. Proc. Natl Acad. Sci. USA 107, 4275–4280 (2010).
CAS PubMed PubMed Central Google Scholar
Kaminow, B., Yunusov, D. & Dobin, A. STARsolo: accurate, fast and versatile mapping/quantification of single-cell and single-nucleus RNA-seq data. Preprint at bioRxiv https://doi.org/10.1101/2021.05.05.442755 (2021).

Download references

Acknowledgements

We acknowledge the use of the Integrated Genomics Operation Core at the Memorial Sloan Kettering Cancer Center, funded by the National Cancer Institute Cancer Center Support Grant (P30 CA08748), Cycle for Survival and the Marie-Josée and Henry R. Kravis Center for Molecular Oncology. This work was supported by the National Institutes of Health (NIH)/National Cancer Institute Cancer Center Support Grant P30 CA008748 (A.Y.R.), NIH grant R01 AI034206 (A.Y.R.), NIH/National Institute of Allergy and Infectious Diseases grant DP2AI171161 and the Ludwig Institute for Cancer Research (Y.P.). A.Y.R. is an investigator with the Howard Hughes Medical Institute. E.Y.W. was supported by a Medical Scientist Training Program grant from the National Institute of General Medical Sciences of the NIH under award no. T32GM152349 to the Weill Cornell/Rockefeller/Sloan Kettering Tri-Institutional MD–PhD Program.

Author information

Zhong-Min Wang
Present address: Immunology Discovery, Genentech Inc., South San Francisco, CA, USA
These authors contributed equally: Wei Hu, Gabriel A. Dolsten, Eric Y. Wang, Giorgi Beroshvili.
These authors jointly supervised this work: Wei Hu, Yuri Pritykin, Alexander Y. Rudensky.

Authors and Affiliations

Howard Hughes Medical Institute and Immunology Program, Sloan Kettering Institute, Memorial Sloan Kettering Cancer Center, New York, NY, USA
Wei Hu, Eric Y. Wang, Giorgi Beroshvili, Zhong-Min Wang, Aazam P. Ghelani, Lion F. K. Uhl, Regina Bou-Puerto, Xiao Huang, Anthony J. Michaels, Beatrice E. Hoyos & Alexander Y. Rudensky
Department of Immunobiology and Institute of Biomolecular Design and Discovery, Yale University, West Haven, CT, USA
Wei Hu & Wenjie Jin
Lewis-Sigler Institute for Integrative Genomics, Princeton University, Princeton, NJ, USA
Gabriel A. Dolsten & Yuri Pritykin
Immunology and Microbial Pathogenesis Program, Weill Cornell Graduate School of Medical Sciences, New York, NY, USA
Eric Y. Wang, Giorgi Beroshvili, Aazam P. Ghelani, Regina Bou-Puerto, Anthony J. Michaels & Alexander Y. Rudensky
Tri-Institutional MD–PhD Program, Weill Cornell Medicine, The Rockefeller University and Memorial Sloan Kettering Cancer Center, New York, NY, USA
Eric Y. Wang
Gerstner Sloan Kettering Graduate School of Biomedical Sciences, Memorial Sloan Kettering Cancer Center, New York, NY, USA
Zhong-Min Wang & Alexander Y. Rudensky
Department of Computer Science, Princeton University, Princeton, NJ, USA
Yuri Pritykin

Authors

Wei Hu
View author publications
Search author on:PubMed Google Scholar
Gabriel A. Dolsten
View author publications
Search author on:PubMed Google Scholar
Eric Y. Wang
View author publications
Search author on:PubMed Google Scholar
Giorgi Beroshvili
View author publications
Search author on:PubMed Google Scholar
Zhong-Min Wang
View author publications
Search author on:PubMed Google Scholar
Aazam P. Ghelani
View author publications
Search author on:PubMed Google Scholar
Lion F. K. Uhl
View author publications
Search author on:PubMed Google Scholar
Regina Bou-Puerto
View author publications
Search author on:PubMed Google Scholar
Xiao Huang
View author publications
Search author on:PubMed Google Scholar
Anthony J. Michaels
View author publications
Search author on:PubMed Google Scholar
Beatrice E. Hoyos
View author publications
Search author on:PubMed Google Scholar
Wenjie Jin
View author publications
Search author on:PubMed Google Scholar
Yuri Pritykin
View author publications
Search author on:PubMed Google Scholar
Alexander Y. Rudensky
View author publications
Search author on:PubMed Google Scholar

Contributions

W.H. and A.Y.R. conceived the project, interpreted the results and wrote the paper. Y.P., G.A.D. and E.Y.W. performed computational analyses of the sequencing data and wrote the relevant sections of the paper, with input from W.H. and A.Y.R. W.H., E.Y.W., G.B., Z.-M.W. and R.B.-P. conducted wet laboratory experiments and analyzed the data. A.P.G., L.F.K.U., R.B.-P., X.H., A.J.M., B.E.H. and W.J. assisted with wet laboratory experiments.

Corresponding authors

Correspondence to Wei Hu, Yuri Pritykin or Alexander Y. Rudensky.

Ethics declarations

Competing interests

A.Y.R. is a Scientific Advisory Board member and has equity in Sonoma Biotherapeutics, RAPT Therapeutics, Coherus BioSciences, Santa Ana Bio, Odyssey Therapeutics, Nilo Therapeutics and Vedanta Biosciences; he is also a Scientific Advisory Board member of BioInvent and Amgen and a co-inventor of a CCR8⁺ T_reg cell depletion IP licensed to Takeda, which is unrelated to the content of this publication. Z.-M.W. is an employee of Genentech, which is unrelated to the content of this publication. The other authors declare no competing interests.

Peer review

Peer review information

Nature Immunology thanks R. Allan, J. Geginat and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available. Primary Handling Editor: L. A. Dempsey, in collaboration with the Nature Immunology team.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 Generation of Foxp3^AID mice.

(a) Gene targeting strategy. The auxin-inducible degron (AID) sequence was fused to the N-terminus of Foxp3 via a seven-amino-acid flexible linker. An IRES-ZsGreen-T2A-iCre-Frt-neo-Frt cassette was inserted into the 3’ UTR. Arrows indicate the locations of PCR primers used to distinguish Foxp3^WT and Foxp3^AID alleles. IRES: internal ribosome entry sequence. iCre: codon improved Cre recombinase. PGK: PGK promoter. Neo: neomycin-resistant gene. pA: bGH polyA sequence. (b) PCR validation of the knock-in allele. Data is representative of over 200 mice from each genotype. (c) Expression pattern of the ZsGreen reporter in Foxp3^AID/WT heterozygous females. ZsGreen expression was restricted to CD25⁺ CD4 T cells, consistent with Foxp3 expression. (d) ZsGreen⁺ Foxp3^AID T_reg cells suppressed naïve CD4⁺ T cell proliferation comparably to Foxp3^GFP T_reg cells in vitro. Line graph represents mean ± SEM of two biological replicates. (e) Naïve CD4⁺ T cells from Foxp3^AID mice were cultured under T_reg inducing conditions and transduced with either a TIR1-encoding retrovirus or the empty vector control. AID-tagged Foxp3 protein was selectively degraded in TIR1-transduced induced T_reg (iT_reg) cells upon indole acetic acid (IAA) treatment. Scatter plots represents mean ± SEM. Each point represents cells from a unique mouse.

Source data

Extended Data Fig. 2 Generation of ROSA26^TIR1 and ROSA26^TIR1(F74G) mice.

(a) Gene targeting strategy. WPRE: Woodchuck hepatitis virus post-transcriptional regulatory element; DTA: Diphtheria toxin fragment A). (b) Southern blot validation of heterozygous ROSA26^TIR1/+ mice using the hybridization probe shown in (a). Data is representative of two independent F1 mice. (c) PCR validation of ROSA26^TIR1/+ mice. Data are representative of over 100 mice from each genoype. (d–f) Naïve CD4⁺ T cells from ROSA26^TIR1/+ mice were co-transduced with retroviruses expressing Cre and AID-Foxp3 (d). TIR1 expression was induced in a Cre-dependent manner (e), resulting in AID-Foxp3 degradation upon IAA treatment (f). Data are representative of two independent experiments. (g) Guide RNA (gRNA) design for CRISPR-mediated F74-to-G mutation in TIR1. The gRNA seed sequence is shown in gray; the PAM sequence is in pink. The F74G mutation creates a KasI restriction site. (h–i) Validation of the F74G mutation by KasI digestion (h) and Sanger sequencing (i). (j–k) The TIR1 F74G mutation enables in vivo protein degradation in response to 5-ph-IAA.

Source data

Extended Data Fig. 3 Foxp3 protein degradation in adult lymphoreplete mice induces minimal immune activation.

(a) Size of the spleen and lymph nodes after 28 days of Foxp3 degradation. (b) Serum antibody levels following T_reg ablation or Foxp3 degradation. (c,d) Representative H&E stain (c) and histology scores (d) of the skin following T_reg ablation or Foxp3 degradation. Each point represents a unique mouse. Data are pooled from two independent experiments. Scatter blots represent mean ± SEM. One-way ANOVA.

Source data

Extended Data Fig. 4 Inducible Foxp3 gene knockout causes minimal immune activation in adult lymphoreplete mice.

(a) Experimental design. (b-c) Representative plot (b) and combined data (c) showing the efficiency of Foxp3 gene knockout. (d) Size of the spleen and lymph nodes after 14 days of inducible Foxp3 gene knockout. (e-g) T cell activation (e), serum antibody levels (f), and myeloid cell expansion (g) following 14 days of inducible Foxp3 gene knockout. (h) Representative H&E stain (left) and histology scores (right) of the skin, liver, and lung following 14 days of T_reg ablation or Foxp3 degradation. Each point represents a unique mouse. Data are pooled from two independent experiments. Scatter blots represent mean ± SEM. One-way ANOVA.

Source data

Extended Data Fig. 5 Foxp3 degradation in developed T_reg cells induces minimal gene expression and chromatin accessibility changes.

(a) UMAP visualization of scRNA-seq data from Foxp3^AIDR26^WT and Foxp3^AIDR26^TIR1(F74G) T_reg cells on days 0, 3 and 7 of 5-ph-IAA-induced Foxp3 degradation. (b) UMAP visualization of representative genes from the single-cell RNA-seq dataset colored by expression level. (c) MA plot showing differentially accessible ATAC-seq peaks induced by Foxp3 degradation (left) and Foxp3 gene deficiency (right). Red points represent differentially accessible regions.

Extended Data Fig. 6 Foxp3 degradation in developed T_reg cells altered the expression of a small group of genes.

(a) The composition of Foxp3^AIDR26^WT and Foxp3^AIDR26^WT(F74G) T_reg cells within resting and activated subsets from day 0, day 3, and day 7 of Foxp3 degradation. (b) Number of resting and activated T_reg cells utilized for the differential gene expression analysis in Fig. 3. (c) Flow cytometry analysis of Foxp3 protein and mRNA levels (reported by ZsGreen) in Foxp3^WT T_reg cells from heterozygous Foxp3^AID/WTR26^WT and Foxp3^AID/WTR26^TIR1(F74G) females after 7 days of Foxp3 degradation. Scatter plots represent mean ± SEM. Data are pooled from two independent experiments. Two-way ANOVA. (d) Bulk RNA-seq read counts of indicated genes in Foxp3^AID T_reg cells from heterozygous Foxp3^AID/WTR26^WT and Foxp3^AID/WTR26^TIR1(F74G) females after 7 days of Foxp3 degradation. Each point represents a unique mouse. Bar graphs represent mean ± SEM. (e) Flow cytometry analysis of CD25, CD122, OX40, GITR, and FR4 protein levels in Foxp3^WT T_reg cells from heterozygous Foxp3^AID/WTR26^WT and Foxp3^AID/WTR26^TIR1(F74G) females after 7 days of Foxp3 degradation. (f) Flow cytometry analysis of CD127 and TCF1 protein levels in Foxp3^AID T_reg cells from heterozygous Foxp3^AID/WTR26^WT and Foxp3^AID/WTR26^TIR1(F74G) females after 7 days of Foxp3 degradation. (e-f) Scatter plots represent mean ± SEM. Each point represents a unique mouse. Data are pooled from two independent experiments. Multiple t-tests.

Source data

Extended Data Fig. 7 Genes sensitive to Foxp3 degradation in mature T_regs are potentially enriched for direct Foxp3 targets.

(a–b) Representative tracks showing Foxp3 ChIP, T_reg H3K27Ac, T_reg H3K27me3, and RNA-seq profiles in activated, resting, and nascent thymic T_reg cells for candidate Foxp3-activated (a) and Foxp3-repressed (b) genes. (c) Representative flow cytometry plots and gating strategy for RNA-seq analysis of T_reg “wannabe” cells with or without Foxp3 induction at the indicated time points following 4-OHT administration.

Extended Data Fig. 8 Conservation of Foxp3 degradation effects across different tissues.

(a) Experimental design of single-cell RNA-seq experiment. Each genotype consisted of four independent biological replicates. (b) Density contour plots of Foxp3^AIDR26^WT and Foxp3^AIDR26^TIR1(F74G) T_reg cells from SLO, liver, lung and LILP overlayed on UMAP embeddings. (c) UMAP visualization of T_reg cells from depicted tissues colored by “resting” and “activated” gene signature scores based on the scoring strategy from Fig. 3a. (d) Bar plots showing the number of DEGs between R26^WT and R26^TIR1(F74G) in resting and activated T_reg cells from indicated tissues, colored by up- or downregulation. (e) Pairwise Pearson correlation coefficients for log₂ fold changes of DEGs between activated and resting T_reg cells from each tissue. Rows and columns are hierarchically clustered. (f) Heatmap of log₂ fold changes in Foxp3 degradation DEGs separated by T_reg cell activation status and tissue of origin. Genes are clustered by k-means clustering and cluster numbers are indicated by the color bars on the left. (g) Heatmap of average log₂ fold changes for each k-means cluster in T_reg cells separated by tissue and activation status. Rows and columns are hierarchically clustered. (h) Leiden clustering results visualized by UMAP for LILP T_reg cells. Clusters are labeled as peripheral T_reg cell (pT_reg), dividing T_reg cell, lymphoid tissue thymic T_reg cell (LT-tT_reg) and non-lymphoid tissue thymic T_reg cell (NLT-tT_reg) based on differences in their gene expression profile. (i) Heatmap of scaled mean UMI-normalized expression values of indicated genes in each cluster. (j) Number of DEGs between R26^WT and R26^TIR1(F74G) T_reg cells in pT_reg, LT-tT_reg and NLT-tT_reg cells, colored by up- or downregulation. (k) Log₂ fold changes of “Foxp3-repressed” and “Foxp3-activated” genes in pT_reg, LT-tT_reg and NLT-tT_reg cells. Each point represents a gene from the Foxp3-repressed or Foxp3-activated gene set identified in Fig. 2. Horizontal line marks the median and box marks the interquartile range. Mann-Whitney U test. (l) Scatter-plot of log₂ fold changes of DEGs between pT_reg and NLT-tT_reg cell clusters (left-top), pT_reg and LT-tT_reg cell clusters (left-bottom) and LT-tT_reg and NLT-tT_reg cell clusters (right-top). Points are colored by their direction and populations in which they are altered. Correlations were calculated with Pearson correlation over all DEGs.

Extended Data Fig. 9 Foxp3 is preferentially required during T_reg cell development.

(a) Pearson correlation between Foxp3 degradation-induced and Foxp3-dependent DEGs in thymic, resting, and activated T_reg cells, limited to Foxp3-bound genes. (b) Meta-cell analysis of resting and activated T_reg scRNA-seq data from secondary lymphoid organs following Foxp3 degradation, correlating Foxp3 expression levels with “TIR1-up” and “TIR1-down” gene signatures identified in Fig. 3. Dashed red line depicts line of best fit. Correlations and corresponding p-values were calculated with Pearson correlation over all genes. (c) Cytokine production by CD4⁺ T cells from neonatal Foxp3^AID mice after 14 days of Foxp3 degradation, in comparison to age-matched Foxp3^WT and Foxp3^GFPKO mice. (d) Neutrophil expansion in adult and neonatal Foxp3^AID mice following 14 days of Foxp3 degradation. Age-matched Foxp3^WT and Foxp3^GFPKO mice serve as controls for neonatal Foxp3^AID mice. (e) Representative H&E staining and histology scores of skin inflammation in neonatal Foxp3^AID mice after 14 days of Foxp3 degradation, in comparison to age-matched Foxp3^WT and Foxp3^GFPKO mice. (c–e) Scatter plots represent mean ± SEM. Each point represents a unique mouse. Data are pooled from two independent experiments. One-way ANOVA. (f) Scatter plots of DESeq2 normalized counts of select DEGs from the neonatal Foxp3 degradation RNA-seq experiment (Fig. 5m). Each point represents a unique mouse. Two-sided t-test. (g-h) Representative plots (g) and combined data (h) showing T_reg proliferation in 8-week-old adult versus 7-day-old neonatal mice measured by Ki67 positivity. Scatter-plot represents mean ± SEM. Data are combined from two independent experiments. Two-tailed t-test.

Source data

Extended Data Fig. 10 Foxp3 degradation leads to tumor shrinkage with minimal adverse effects.

(a–b) Representative flow cytometry plots (a) and combined data (b) of Ki67 expression and EdU incorporation in T_reg cells from the tumor and tumor-draining lymph node (dLN). Scatter-plot shows mean ± SEM. Each point represents a unique mouse. Data are representative of two independent experiments. Two-tailed t-test. (c) Representative flow cytometry plots (left) and combined data (right) of IFN-γ production by tumor-infiltrating NK cells. Scatter-plot shows mean ± SEM. Each point represents a unique mouse. Data are pooled from two independent experiments. Two-tailed t-test. (d) Representative flow cytometry plots (left) and quantification (right) of IFN-γ production by tumor-infiltrating ZsGreen⁻ CD4 T cells. Scatter-plot shows mean ± SEM. Each point represents a unique mouse. Data are pooled from two independent experiments. Two-tailed t-test. (e) Volcano plts showing the number of genes up- and downregulated in Foxp3^AIDR26^TIR1(F74G) tumor T_reg cells within each clustered defined in Fig. 7(l).

Source data

Supplementary information

Reporting Summary (download PDF )

Peer Review File (download PDF )

Supplementary Table 1 (download XLSX )

List of DEGs upon Foxp3 degradation across all analyses.

Source data

Source Data Fig. 1 (download XLSX )

Statistical source data.

Source Data Fig. 2 (download XLSX )

Statistical source data.

Source Data Fig. 3 (download XLSX )

Statistical source data.

Source Data Fig. 5 (download XLSX )

Statistical source data.

Source Data Fig. 6 (download XLSX )

Statistical source data.

Source Data Fig. 7 (download XLSX )

Statistical source data.

Source Data Extended Data Fig. 1 (download XLSX )

Statistical source data.

Source Data Extended Data Fig. 1 (download TIF )

Unprocessed gel image for Extended Data Fig. 1b.

Source Data Extended Data Fig. 2 (download PDF )

Unprocessed gel image for Extended Data Fig. 2.

Source Data Extended Data Fig. 3 (download XLSX )

Statistical source data.

Source Data Extended Data Fig. 4 (download XLSX )

Statistical source data.

Source Data Extended Data Fig. 6 (download XLSX )

Statistical source data.

Source Data Extended Data Fig. 9 (download XLSX )

Statistical source data.

Source Data Extended Data Fig. 10 (download XLSX )

Statistical source data.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Hu, W., Dolsten, G.A., Wang, E.Y. et al. Temporal and context-dependent requirements for the transcription factor Foxp3 expression in regulatory T cells. Nat Immunol 26, 2059–2073 (2025). https://doi.org/10.1038/s41590-025-02295-4

Download citation

Received: 05 May 2025
Accepted: 03 September 2025
Published: 08 October 2025
Version of record: 08 October 2025
Issue date: November 2025
DOI: https://doi.org/10.1038/s41590-025-02295-4

Subjects

Abstract

Similar content being viewed by others

Main

Results

Inducible Foxp3 protein degradation in vivo

Foxp3 expression is largely dispensable for preventing autoimmune disease in adulthood

Foxp3 loss induces minimal transcriptional and functional changes in mature Treg cells

Foxp3 degradation-sensitive genes are enriched for Foxp3 binding

Foxp3 degradation effects are generally conserved across tissues

Long timescale of establishment of Foxp3-dependent gene program

A requirement for Foxp3 during Treg cell maturation

Proliferation and severe inflammation sensitize Treg cell transcriptome to Foxp3 loss

Foxp3 degradation-induced tumor shrinkage with minimal adverse effects

Discussion

Methods

Mice

Generation of Foxp3 AID, ROSA26 TIR1 and ROSA26 TIR1(F74G) mice

Treatment of mice with 5-ph-IAA and diphtheria toxin

Reagents and antibodies

Enzyme-linked immunosorbent assay

Isolation of cells from lymphoid organs, lungs and tumors

Flow cytometric analysis of cytokine production

Ex vivo CTV labeling

In vitro suppression assay

Histology

5-ethynyl-2’-deoxyuridine labeling

B16 melanoma model

Bulk RNA-seq and data analysis

Single-cell RNA analysis of SLO Treg cells

Single-cell RNA-seq analysis of intratumoral Treg cells

Single-cell RNA-seq analysis of tissue Treg cells

Single-cell RNA-seq analysis of adoptively transferred Treg cells

Cluster frequency testing

Computational analysis of resting and activated Treg cells

Differential gene expression analysis

Foxp3 ChIP-seq and ATAC-seq data analysis

Motif analysis

FC–FC plots

Statistics

Reporting summary

Data availability

Materials availability

Code availability

References

Acknowledgements

Author information

Authors and Affiliations

Contributions

Corresponding authors

Ethics declarations

Competing interests

Peer review

Peer review information

Additional information

Extended data

Supplementary information

Source data

Rights and permissions

About this article

Cite this article

Share this article

Search

Quick links

Foxp3 loss induces minimal transcriptional and functional changes in mature T_reg cells

A requirement for Foxp3 during T_reg cell maturation

Proliferation and severe inflammation sensitize T_reg cell transcriptome to Foxp3 loss

Generation of Foxp3 ^AID, ROSA26 ^TIR1 and ROSA26 ^TIR1(F74G) mice

Single-cell RNA analysis of SLO T_reg cells

Single-cell RNA-seq analysis of intratumoral T_reg cells

Single-cell RNA-seq analysis of tissue T_reg cells

Single-cell RNA-seq analysis of adoptively transferred T_reg cells

Computational analysis of resting and activated T_reg cells