Ferritin heavy chain supports stability and function of the regulatory T cell lineage

Abstract

Regulatory T (TREG) cells develop via a program orchestrated by the transcription factor forkhead box protein P3 (FOXP3). Maintenance of the TREG cell lineage relies on sustained FOXP3 transcription via a mechanism involving demethylation of cytosine-phosphate-guanine (CpG)-rich elements at conserved non-coding sequences (CNS) in the FOXP3 locus. This cytosine demethylation is catalyzed by the ten–eleven translocation (TET) family of dioxygenases, and it involves a redox reaction that uses iron (Fe) as an essential cofactor. Here, we establish that human and mouse TREG cells express Fe-regulatory genes, including that encoding ferritin heavy chain (FTH), at relatively high levels compared to conventional T helper cells. We show that FTH expression in TREG cells is essential for immune homeostasis. Mechanistically, FTH supports TET-catalyzed demethylation of CpG-rich sequences CNS1 and 2 in the FOXP3 locus, thereby promoting FOXP3 transcription and TREG cell stability. This process, which is essential for TREG lineage stability and function, limits the severity of autoimmune neuroinflammation and infectious diseases, and favors tumor progression. These findings suggest that the regulation of intracellular iron by FTH is a stable property of TREG cells that supports immune homeostasis and limits the pathological outcomes of immune-mediated inflammation.

Synopsis

Sustained FOXP3 transcription is central to the development and maintenance of regulatory T (TREG) cells. This study shows that the regulation of iron metabolism by ferritin heavy chain (FTH) influences FOXP3 expression and TREG cell lineage stability under homeostatic and pathological conditions.

• Human and mouse TREG cells express FTH at high levels.

• FTH sustains TREG cell lineage stability.

• FTH regulates FOXP3 methylation and transcription through supporting the enzymatic activity of the TET family of iron-dependent dioxygenases.

• FTH expression in TREG cells limits inflammation and supports tumor progression.

Iron metabolism promotes FOXP3 transcription in regulatory T cells by supporting TET enzymatic activity.

Introduction

Identified and characterized (Powrie and Mason, 1990; Sakaguchi et al, 1982) originally on the basis of their critical involvement in maintaining peripheral immune tolerance (Coutinho et al, 1993), regulatory T (T_reg) cells partake in different aspects of immune homeostasis (Campbell and Rudensky, 2020; Dikiy and Rudensky, 2023; Josefowicz et al, 2012; Panduro et al, 2016). One of the main functions of T_reg cells, however, is most likely to restrain the breath of innate and adaptive immune responses against commensal microbes to prevent immunopathology (Belkaid, 2007; Demengeot et al, 2006). This evolutionarily conserved trait was probably co-opted through evolution to prevent peripheral self-reactive T and B cells from eliciting autoimmune diseases (Lafaille et al, 1994; Sakaguchi et al, 1995). As an evolutionary trade-off (Stearns and Medzhitov, 2015), T_reg cells are pathogenic, for example, when limiting immune-mediated inflammatory responses to pathogens to promote chronic infections (Belkaid, 2007; Demengeot et al, 2006) or when restraining anti-tumor immunity, to promote cancer progression (Curiel et al, 2004; Liu et al, 2016).

T_reg cell development and function are controlled by the X-chromosome-encoded transcription factor FOXP3 (Fontenot et al, 2003; Hori et al, 2003), together with auxiliary transcriptional regulators (Kanamori et al, 2016). The transcriptional program enforced by FOXP3 specifies T_reg cell lineage commitment in the thymus and in the periphery (Fontenot et al, 2003; Hori et al, 2003; Lee et al, 2012), generating thymic T_reg (tT_reg) cells and peripherally derived T_reg (pT_reg) cells, respectively (Chen et al, 2003). Sustained FOXP3 transcription maintains T_reg cell lineage stability (Williams and Rudensky, 2007), avoiding transdifferentiation towards pro-inflammatory T helper (T_H) cells (Gavin et al, 2007; Morikawa et al, 2014).

FOXP3 transcription is regulated by different signal transduction pathways, emanating from the T-cell receptor (TCR), interleukin (IL-2) receptor, and TGF-β receptor (Bennett et al, 2001; Brunkow et al, 2001; Hori and Sakaguchi, 2004), among others. Sustained FOXP3 transcription is enforced epigenetically (Gavin et al, 2007; Morikawa et al, 2014), in response to environmental cues (Chapman et al, 2020; Shi and Chi, 2019) that regulate different aspects of T_reg cell metabolism (Etchegaray and Mostoslavsky, 2016). These epigenetic modifications include the relative methylation status of cytosine-phosphate-guanine (CpG)-rich sequences in the FOXP3 conserved non-coding sequences (CNS) 1, 2, and 3 (Ohkura et al, 2012; Zheng et al, 2010), whereby cytosine methylation represses while demethylation sustains FOXP3 transcription (Ohkura et al, 2012; Zheng et al, 2010).

Cytosine methylation is catalyzed by DNA methyltransferase (DNMT) (Ohkura et al, 2012), while demethylation is catalyzed by the ten–eleven translocation (TET) family of dioxygenases (Wu and Zhang, 2017; Yue et al, 2016). Cytosine demethylation consists on redox-based reactions that oxidize 5-methylcytosine (5-mC) into 5-hydroxymethylcytosine (5-hmC), 5-formylcytosine (5-fC) and 5-carboxylcytosine (5caC) (Kohli and Zhang, 2013). TET dioxygenases catalyze cytosine demethylation at FOXP3 CNS1 and 2 (Ohkura et al, 2012), supporting T_reg cell lineage stability (Nakatsukasa et al, 2019; Wu and Zhang, 2017; Yue et al, 2019; Yue et al, 2016), and preventing T_reg cell transdifferentiating into inflammatory effector T_H cells, also referred as ex-T_reg cells (Duarte et al, 2009; Komatsu et al, 2009; Zhou et al, 2009).

TET dioxygenases use Fe as an essential cofactor and the intermediate metabolite α-ketoglutarate as an obligatory substrate (Huang and Rao, 2014; Pastor et al, 2013). This TET reliance on Fe availability entertained the hypothesis that regulation of cellular Fe metabolism acts upstream of TET dioxygenases to modulate T_reg cell lineage stability.

Several studies have shown that Fe metabolism impacts on immunity. For example, intracellular Fe availability and redox activity is essential to support B and T-cell development (Vanoaica et al, 2014), via a cytoprotective mechanism exerted by the ferritin H chain (FTH) (Berberat et al, 2003; Pham et al, 2004), likely involving the mitochondria (Blankenhaus et al, 2019; Vanoaica et al, 2014). Regulation of cellular Fe content and redox activity also modulate cytokine production by effector T_H cells, via a mechanism involving the PolyC-RNA-Binding Protein 1 (PCBP1) (Wang et al, 2018). Cellular Fe import, via the transferrin receptor 1 (TFR1/CD71), supports T_H type 1 (T_H1) cell immunity and its regulation by induced T_reg (iT_reg) cells (Voss et al, 2023) as well as antibody responses to vaccination (Frost et al, 2021; Jiang et al, 2019) and immunity against infection by pathogens such as Plasmodium, the causative agent of malaria (Wideman et al, 2023).

Here, we demonstrate that regulation of Fe metabolism by FTH operates upstream of TET dioxygenases to enforce cytosine demethylation at CpG-rich sequences in the CNS1 and 2 of the FOXP3 locus, sustaining FOXP3 transcription, expression and T_reg cell lineage identity. This cell-intrinsic property of T_reg cells is essential to maintain immune homeostasis while exerting a major impact on the outcome of immune-driven inflammation.

Results

T_reg cells express relatively high levels of FTH

In a previously unbiased proteomics analysis, we found that freshly isolated human naive CD45RA⁺CD25^hi T_reg (nT_reg) and memory CD45RA^-CD25^hi T_reg (mT_reg) cells expressed relatively higher levels of Fe-regulatory proteins, including FTH and ferritin L chain (FTL), when compared to CD45RA⁺CD25^- naive conventional (nTconv) cells or to CD45RA^-CD25^- activated/memory (mTconv) cells (Cuadrado et al, 2018). The relatively higher expression of the FTH and FTL components of the ferritin complex was maintained upon expansion of human CD4⁺CD127^–CD25⁺ T_reg cells in vitro, in comparison to T_conv cells CD4⁺CD127⁺CD25^– cells (Fig. 1A,B). Similarly, mouse CD4⁺Foxp3⁺ T_reg cells also expressed relatively higher levels of FTH protein, compared to CD4⁺Foxp3⁻CD44^lowCD62L^high naive T_H cells (T_N) or CD4⁺Foxp3⁻CD44^highCD62L^low memory T_H cells (T_M), as determined by western blot (Fig. 1C,D). This suggests that sustained and elevated levels of ferritin expression are a stable property of human and mouse T_reg cells. Of note, mouse T_reg cells express similar levels of Fth mRNA, compared to T_N cells, while (CD4^±Foxp3⁻CD44^highCD62L⁻) T_M cells express relatively higher levels of Fth mRNA, compared to T_reg cells (Appendix Fig. S1). This suggests that the relatively higher level of FTH protein expression in T_reg cells is enforced post-transcriptionally, similar to other cell types (Hentze et al, 1987; Meyron-Holtz et al, 2004; Muckenthaler et al, 2017; Rouault et al, 1988).

Figure 1. T_reg cell ferritin expression is a stable property of human and mouse T_reg cells.

(A) FTH and β-Actin protein expression, detected by western blot in whole-cell extracts from human T conventional (CD4⁺CD45RA⁺CD127⁺CD25^–; T_conv) and T_reg (CD4⁺CD127^–CD45RA⁺CD25^hi) cells after two weeks of expansion with anti-CD3, anti-CD28 mAb and IL-2. (B) Relative quantification of FTH protein expression, normalized to β-Actin, detected by western blot as in (A). n = 3 independent experiments. (C) FTH and histone H3 protein expression detected by western blot in whole-cell extracts from sorted mouse naive T cells (CD4⁺Foxp3^-CD44^lowCD62L^high; T_N), (CD4⁺Foxp3⁺GFP⁺) T_reg cells and memory T cells (CD4⁺Foxp3⁻CD44^highCD62L^low; T_M), by western blot. (D) Relative quantification of FTH, normalized to histone H3, protein expression, detected by western blot as in (C). Data were normalized to FTH expression in T_N cells, pooled from four independent experiments. (E) Schematic representation of the protocol used for the generation of iT_reg and representative flow cytometry dot plots of mouse iT_reg generated from sorted naive T cells (T_N), stimulated with anti-CD3 and anti-CD28 mAb plus IL-2 and TGFβ for 5 days. Control T_conv cells were subjected to the same experimental conditions, without TGFβ. (F) FTH and histone H3 protein expression, detected by western blot in whole-cell extracts from iT_reg and T_conv generated as depicted in (E). Data pooled from three independent experiments with similar trend. (G) The relative level of Fth mRNA expression, quantified by qRT-PCR, using Arbp0 as housekeeping gene. Data pooled from three independent experiments, with similar trend. (H–J) Schematic representation of the protocol used (top panels), representative flow cytometry dot plots (middle panels) and corresponding quantification of percentage (%) in CD4⁺ cells and cell number (Nbr.) (bottom panels) of live (TCRβ⁺CD4⁺ Foxp3⁺) GFP⁺ T_reg cells in (H) thymus, (I) spleen and (J) mesenteric LN (MLN). (H) Data from N = 8 mice per genotype, per organ, from two independent experiments, with similar trend. (I, J) Data from N = 12 mice per genotype, per organ, from three independent experiments, with similar trend. (K) Schematic representation of the experimental approach (top panel) used to monitor the expression of the GFP-hCre transgene in the thymus, spleen, and MLN of (CD4⁺GFP⁺) T_reg cells. Representative flow cytometry histograms of GFP-hCre (bottom left panel). Relative quantification of GFP-hCre expression (bottom right panel), represented as mean fluorescence intensity (MFI). Data from N = 8 mice per genotype, pooled from two independent experiments with similar trend. (L) Schematic representation of the experimental approach (left panel) used, representative flow cytometry dot plots (top panel) and corresponding quantification of percentage (%) (bottom left panel) and cell number (Nbr.) (bottom right panel) of live (TCRβ⁺CD4⁺ GFP⁺) Nrp1⁺ and Nrp1^-T_reg cells in MLN. Data from N = 8 mice per genotype, pooled from two independent experiments with similar trend. Data information: Data in (B, F, G) are presented as mean ± SD. Data in (D) are presented as mean ± SEM. Circles in (H–L) correspond to individual mice. P values in (B, F–J) determined using unpaired t test with Welch’s correction, in (D, H–J) using ordinary one-way ANOVA, and in (K, L) using Two-way ANOVA with Sidak’s multiple comparisons test. NS not significant (P > 0.05); *P < 0.05; **P < 0.01; ***P < 0.001. Source data are available online for this figure.

We monitored FTH expression in induced T_reg (iT_reg) cells, generated from mouse T_N cells activated in vitro with anti-CD3/CD28 mAb plus IL-2 and TGFβ (Chen et al, 2003) (Fig. 1E). To this aim we used Foxp3^GFP T_reg cell reporter mice, in which a green fluorescent protein (GFP) humanized Cre-recombinase (GFP-hCre) coding sequence is inserted downstream of the Foxp3 ATG translational start codon in a bacterial artificial chromosome (BAC) transgene carrying the intact Foxp3 promoter (Foxp3-GFP-hCre; referred herein as Foxp3^GFP) (Chen et al, 2003; Zhou et al, 2008). FTH protein expression was higher under culture conditions containing TGFβ and enriched in CD4⁺GFP⁺ iT_reg cells, compared to culture conditions lacking TGFβ and enriched in CD4⁺GFP^- T_conv cells (Fig. 1F). A similar trend was observed for Fth mRNA, which was upregulated in iT_reg cells (Fig. 1G). The relative level of Fth mRNA expression was similar in thymic, peripheral and iT_reg cells (Appendix Fig. S2).

FTH expression in T_reg cells is essential to maintain immune homeostasis

To determine the effect of regulation of intracellular Fe metabolism by FTH on T_reg cells, we introduced an additional loxP-flanked Fth allele (Fth^fl/fl) (Darshan et al, 2009) into Foxp3^GFP mice (Zhou et al, 2008), deleting Fth specifically in T_reg cells from Foxp3^{GFP-Fth∆/∆} vs. control Foxp3^GFP mice (Fig. EV1A). Fth deletion was associated with the accumulation of intracellular labile Fe²⁺ in T_reg cells, isolated from the mesenteric lymph nodes (MLN) of Foxp3^{GFP-Fth∆/∆} vs. control Foxp3^GFP mice (Fig. EV1B).

Figure EV1. FTH expression in T_REG cells alters systemic iron metabolism.

(A) Schematic representation of experimental approach (left panel) and relative quantification of Fth mRNA, by qRT-PCR, normalized to Arbp0 mRNA (right panel), of (CD4⁺ GFP⁺) T_reg cells sorted from mesenteric lymph nodes (MLN). Data from N = 3 mice per genotype. (B) Schematic representation of experimental approach used for relative quantification of intracellular Fe²⁺ in (CD4⁺GFP⁺) T_reg cells in mesenteric lymph nodes (MLN), using the FeRhoNox™-1 probe (left panel). Representative flow cytometry histograms (middle panel) and relative quantification (right panel) of mean fluorescence of intracellular Fe²⁺ intensity (MFI) N = 3 mice per genotype. (C) Schematic representation of experimental approach (left panel), representative flow cytometry dot plots (middle panel) and corresponding quantification of percentage (%) and cell number (Nbr.) (right panels) of live splenic follicular (CD4⁺GFP⁺CXCR5⁺PD1⁺) FT_reg cells. Data from N = 4 mice per genotype, from one experiment. (D, E) Representative flow cytometry dot plots (left panels) and number (right panel) of live activated (CD4⁺CD44^highCD62L^low) and (CD8⁺CD44^highCD62L^low) T cells in the spleen (D) and MLN (E). Data from N = 6–8 mice per genotype, pooled from four independent experiments, with similar trend. (F, G) Representative flow cytometry dot plots (left panels) and percentage (%) (right panel) of live activated (CD3⁺CD4⁺Foxp3⁻IFN-γ⁺; T_H1) T_H1 and (CD3⁺CD8⁺Foxp3⁻IFN-γ⁺) T_C in the spleen (F) and MLN (G). Data representative of N = 5 mice per genotype, pooled from two independent experiments, with similar trend. Data information: Data in (A–G) represented as mean ± SD, circles in (A, C–G) correspond to individual mice and red bars to mean values. P values in (A–C) calculated using unpaired t test with Welch’s correction, and in (D–G) using two-way ANOVA with Sidak’s multiple comparison test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Source data are available online for this figure.

The frequency of thymic CD4⁺GFP⁺ T_reg cells was lower in Foxp3^{GFP-Fth∆/∆} vs. control Foxp3^GFP mice (Fig. 1H). This was not associated, however, with a concomitant reduction in the number of CD4⁺GFP⁺ T_reg cells (Fig. 1H). In contrast, there was a marked reduction of both the frequency and numbers of T_reg cells in the spleen (Fig. 1I) and in the MLN (Fig. 1J) of Foxp3^{GFP-Fth∆/∆} vs. control Foxp3^GFP mice. The frequency and number of CD4⁺GFP⁺CXCR5⁺PD1⁺ follicular T_reg (FT_reg) cells was also decreased in the spleen of Foxp3^{GFP-Fth∆/∆} vs. control Foxp3^GFP mice (Fig. EV1C). These observations suggest that regulation of intracellular Fe by FTH is required to sustain the number of circulating T_reg and FT_reg cells in the periphery, without interfering with thymic T_reg cell output.

We noticed that the relative level of GFP expression, reporting on Foxp3 transcription under the control of an intact Foxp3 locus (Chen et al, 2003; Zhou et al, 2008), was reduced in thymic, splenic and MLN T_reg cells from Foxp3^{GFP-Fth∆/∆} vs. control Foxp3^GFP mice (Fig. 1K). These observations suggest that FTH regulates Foxp3 transcription in the thymus as well as in circulating T_reg cells, which is not sufficient however, to interfere with thymic T_reg cell output.

Thymic and peripheral T_reg cell development give rise to tT_reg and iT_reg cells, expressing neuropilin1 (Nrp1) or not, respectively (Weiss et al, 2012; Yadav et al, 2012). The frequency and number of Nrp1⁺ tT_reg cells and Nrp1^- iT_reg cells was reduced, to the same extent, as assessed in the MLN (Fig. 1L) of Foxp3^{GFP-Fth∆/∆} vs. control Foxp3^GFP mice. This suggests that regulation of Fe metabolism by FTH is required for the maintenance of thymic-derived and peripherally induced T_reg cells.

FTH restrains T_reg cell transdifferentiation into inflammatory ex-T_reg cells

The reduction in T_reg cells imposed by Fth deletion was associated with an accumulation of activated CD4⁺CD44^highCD62L^low T_conv cells and CD8⁺CD44^highCD62L^low cytotoxic T (T_C) cells, in the spleen (Fig. EV1D) and in MLN of Foxp3^{GFP-Fth∆/∆} vs. control Fth^fl/fl mice (Fig. EV1E). Concomitantly, there was an increase in the frequency of interferon-γ (IFN γ)-expressing activated CD4⁺ T_H cells and CD8⁺ T_C cells in the spleen (Fig. EV1F) and MLN (Fig. EV1G). These observations suggest that regulation of Fe metabolism in T_reg cells is essential to maintain immune homeostasis, preventing the activation and accumulation of inflammatory CD4⁺ T_H cells and CD8⁺ T cells.

Secreted ferritin complexes can restrain human T-cell proliferation in vitro (Gray et al, 2001), entertaining the hypothesis that ferritin secretion supports the antiproliferative function of T_reg cells. However, T_reg cells from Foxp3^{GFP-Fth∆/∆} mice inhibited T-cell proliferation in vitro, to a similar extent as T_reg cells from Foxp3^GFP (Fig. EV2A). This suggests that FTH is not essential to support the antiproliferative function of T_reg cells, consistent with Foxp3^{GFP-Fth∆/∆} mice not developing overt autoimmune pathologic lesions, compared to control Fth^fl/fl mice (Fig. EV2B).

Figure EV2. FTH expression in T_REG cells prevents systemic cellular inflammation.

(A) Schematic representation of experimental approach (left panel) for (CD4⁺CD25⁺GFP⁺) T_reg cell sorting from the mesenteric lymph nodes (MLN) and coculture with conventional activated T cells (αCD3/28 + IL-2) to evaluate suppressive function of T_reg cells. Representative flow cytometry proliferation histograms (Cell Tracer Violet) of in vitro suppression assay of mouse T_N by different ratios of T_reg cells (middle panel). Inhibition of T_N cell proliferation quantified as percentage of undivided cells (right panel). Data from 1 out of 3 representative experiments, with similar trend. (B) Representative images of H&E-stained liver, lung, kidney, colon, and pancreas from N = 3–4 mice per genotype at 27–31 weeks after birth. (C) Schematic representation of the experimental approach (left panel) used to generate the Heatmap (right panel) of individual genes associated with T_H effector function programs, differentially expressed in (CD4⁺GFP⁺) T_reg cells sorted from Foxp3^GFP-FthΔ/Δ vs. Foxp3^GFP mice (same experiment as Fig. 2A,B). (D) Schematic representation of the experimental approach used (left panel) to evaluate the percentage (right panel) of (CD4⁺Foxp3⁺CD44^highCD62L^low) activated T_reg cells in the MLN and spleen. Data from N = 7–8 mice per genotype, pooled from three independent experiments, with similar trend. (E) Schematic representation of the experimental approach used (top panel), representative flow cytometry dot plots (bottom left panel) and percentage (bottom right panel) of splenic (CD4⁺Foxp3⁺) IFNγ-secreting T_reg. Data from N = 5 mice per genotype, pooled from two independent experiments, with similar trend. Data information: Data in (A, D, E) represented as mean ± SD. Circles in (A) represent individual wells, and red bars are mean values. Circles in (D, E) represent individual mice, and red bars are mean values. P values in (A, D) calculated using Two-way ANOVA with Sidak’s multiple comparison test and in (E) using unpaired t test with Welch’s correction. NS not significant (P > 0.05), *P < 0.05; ***P < 0.001. Source data are available online for this figure.

To gain further insight into the mechanism via which FTH modulates T_reg cell function in vivo, we performed RNA sequencing (RNAseq), to compare the gene expression profiles of CD4⁺ CD25⁺GFP⁺ T_reg cells sorted from the lymph nodes of Foxp3^{GFP-Fth∆/∆} vs. control Foxp3^GFP mice (Fig. 2A). T_reg cells from Foxp3^{GFP-Fth∆/∆} mice upregulated 1832 genes and downregulated 1340 genes, compared to T_reg cells from control Foxp3^GFP mice (Fig. 2B). Fth deletion was associated with dysregulation of Foxp3-dependent and -independent “T_reg transcriptional signature” (Hill et al, 2007), affecting at least 194 genes involved in T_reg cell function and lineage maintenance (Fig. 2C). Pathway enrichment analysis (Fig. 2D,E) showed that T_reg cells from Foxp3^{GFP-Fth∆/∆} mice presented T_H1 and type 2 (T_H2) transcriptional signatures (Fig. 2D), as illustrated by the induction of the transcriptional master regulators of T_H1 and T_H2 effector functions, Tbx21 (T-box transcription factor; T-bet) and Gata3 (GATA Binding Protein 3) as well as Runx3 (RUNX Family Transcription Factor 3), respectively (Fig. EV2C).

Figure 2. FTH expression in T_reg cells prevents transdifferentiation into inflammatory ex-T_reg cells.

(A) Schematic representation of experimental approach (left panel) and representative flow cytometry dot plots of (CD4⁺CD25⁺GFP⁺) T_reg cells from the lymph nodes before and after sorting. (B) Volcano plot representation of RNA sequencing data of genes overexpressed (red) or under-expressed (blue) in (CD4⁺CD25⁺GFP⁺) T_reg cells sorted from Foxp3^GFP-FthΔ/Δ and control Foxp3^GFP (N = 5 per genotype) mice. P values determined by Benjamini and Hochberg adjusted probabilities. (C) Heatmap representation of T_reg transcriptional signature genes differentially expressed in T_reg cells sorted from Foxp3^GFP-FthΔ/Δ vs. Foxp3^GFP mice, as illustrated in (A, B). (D, E) Pathway enrichment analysis of genetic programs overexpressed (D) or under-expressed (E) in T_reg cells sorted from Foxp3^GFP-FthΔ/Δ vs. Foxp3^GFP mice, as illustrated in (A, B). Data were analyzed using g:SCS multiple testing correction method with a significance threshold of 0.05. (F) Heatmap representation of individual genes associated with oxidative stress-responsive programs, differentially expressed in T_reg cells sorted from Foxp3^GFP-FthΔ/Δ vs. Foxp3^GFP mice, as illustrated in (A, B). Source data are available online for this figure.

Consistently, the percentage of activated CD4⁺GFP⁺CD44^highCD62L^low T_reg cells was higher in the spleen and MLN from Foxp3^{GFP-Fth∆/∆} vs. control Foxp3^GFP mice (Fig. EV2D). This was associated with an increase in the frequency of CD4⁺GFP⁺ T_reg cells expressing IFNγ in the spleen (Fig. EV2E). This suggests that FTH is essential to prevent the transdifferentiation of T_reg cells into inflammatory T_H cells.

RNAseq analysis also revealed that Fth deletion in T_reg cells led to the induction of the canonical oxidative stress response controlled by the transcription factor nuclear factor erythroid-derived 2-like 2 (NRF2) (Fig. 2D). This was associated with the activation of other canonical stress responses, including the cell cycle and DNA damage, unfolded protein, and hypoxic response (Fig. 2D) as well as an overall shutdown of eIF2, mTOR and eIF4/p70s6K signaling transduction pathways (Fig. 2E). Activation of the oxidative stress response regulated by NRF2 was characterized by the induction of Nqo1 and Hmox1, among several other NRF2-regulated genes (Fig. 2F). These observations suggest that FTH is essential to support a transcriptional profile that maintains T_reg cell redox homeostasis, similar to described in other cell types (Blankenhaus et al, 2019; Vanoaica et al, 2014).

FTH supports T_reg cell lineage maintenance

To establish whether FTH enforces T_reg cell lineage maintenance and prevents the transdifferentiation of T_reg cells into inflammatory T_H cells, an additional Rosa26-tandem dimer (td) Tomato-Flox-stop-Flox allele was introduced into Foxp3^{GFP-Fth∆/∆} mice, driving the expression of tdTomato (tdT) by Cre-driven excision of a Flox-stop-Flox cassette, under the control of Foxp3 regulatory regions (Fig. 3A). The resulting Foxp3^{GFP-Fth∆/∆-tdT} mice allow distinguishing CD4⁺GFP⁺tdT⁺ T_reg from CD4⁺GFP^-tdT⁺ ex-T_reg cells that at some point in their developmental history downregulated Foxp3 (i.e., GFP^-), while retaining TdT expression (Fig. 3A). Fth deletion in T_reg cells was associated with a progressive reduction in the frequency of circulating T_reg cells from Foxp3^{GFP-Fth∆/∆-tdT} vs. control Foxp3^GFP-tdT mice, as assessed from 2 to 24 weeks after birth (Figs. 3B and EV3A). Concomitantly, there was an increase in the frequency of circulating ex-T_reg cells (Figs. 3C and EV3A), with 60% of circulating T_reg cells becoming ex-T_reg cells in Foxp3^{GFP-Fth∆/∆-tdT}, 24 weeks after birth (Figs. 3D and EV3A). This relative enrichment in the proportion of ex-T_reg cells suggests that Fth deletion promotes the conversion of T_reg cells into ex-T_reg cells.

Figure 3. FTH enforces T_reg cell lineage stability.

(A) Schematic representation of Foxp3^{GFP-Fth∆/∆-tdT} mice used to monitor the transition of (CD4⁺GFP⁺tdT⁺) T_reg cells into (CD4⁺GFP^-tdT⁺) ex-T_reg cells that repressed GFP expression while retaining the expression of a tdT transgene. (B–D) Percentage of circulating: (B) T_reg and (C) ex-T_reg cells in Foxp3^{GFP-Fth∆/∆-tdT} and control Foxp3^GFP-tdT mice, and (D) Percentage of ex-T_reg cells among CD4⁺tdT⁺ cells, calculated as the ratio of CD4⁺GFP^-tdT⁺/CD4⁺tdT⁺ cells in the same mice as (B, C). Data from N = 4–6 mice per genotype was pooled from three independent experiments with a similar trend. (E–G) Percentage and number of splenic (E) T_reg cells, (F) ex-T_reg cells and (G) relative percentage of ex-T_REG cells over total CD4⁺tdT⁺ cells. Data from N = 4 mice per genotype, pooled from two independent experiments with similar trends. (H) Representative flow cytometry histograms of IFNγ expression by live activated (CD4⁺GFP⁺tdT⁺) T_reg and (CD4⁺GFP^-tdT⁺) ex-T_reg cells in lymph nodes and spleen from Foxp3^{GFP-Fth∆/∆-tdT} and control Foxp3^GFP-tdT mice (left panels) and corresponding quantification of the percentage of IFNγ expressing (CD4⁺GFP⁺tdT⁺) T_reg and (CD4⁺GFP^-tdT⁺) ex-T_reg cells (right panels). Expression of IFNγ was induced upon Phorbol-12-myristate-13-acetate (PMA) and Ionomycin re-activation in vitro. Data from N = 3 wells per genotype, in one experiment, representative of two independent experiments with similar trend. (I) Representative flow cytometry dot plots (left panels) and corresponding quantification (right panel) of the percentage of (CD4⁺GFP⁺tdT⁺) T_reg and (CD4⁺GFP^-tdT⁺) ex-T_reg cells expressing Ki67 and CD71 in the lymph nodes Foxp3^{GFP-Fth∆/∆-tdT} and control Foxp3^GFP-tdT mice. Data from N = 6 mice per genotype, pooled from two independent experiments, with similar trend. Data information: Data in (B–D) represented as mean ± SD. Circles correspond to mean values. Data in (E–I) circles correspond to individual mice and red bars to mean values. (H, I) represented as mean ± SD. P values in (B–D, H, I) calculated using Two-way ANOVA analysis with Sidak’s multiple comparisons test and in (E–G) with unpaired t test with Welch’s correction. NS not significant (P > 0.05); *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Source data are available online for this figure.

Figure EV3. FTH expression in T_REG cells prevents T_reg transdifferentiation into inflammatory T_reg cells.

(A) Representative flow cytometry dot plots of GFP and tdT expression in circulating CD4⁺ cells (same experiment as Fig. 3A–D). Numbers in quadrants correspond to percentages of positive cells at the indicated weeks after birth. (B) Relative quantification of Fth mRNA, by qRT-PCR, normalized to Arbp0 mRNA, in (CD4⁺GFP⁺tdT⁺) T_reg and (CD4⁺GFP^-tdT⁺) ex-T_reg cells sorted from the lymph nodes (LN). Data from N = 5–7 mice per genotype, pooled from two experiments with similar trend. (C) Mean fluorescence intensity (MFI) of IFNγ in activated (CD4⁺GFP⁺tdT⁺) T_reg cells and (CD4⁺GFP^-tdT⁺) ex-T_reg from the lymph nodes (LN) and spleen in the same experiment as (Fig. 3H). Data from N = 3 wells per genotype in one experiment, representative of 2 independent experiments with similar trend. (D) Representative flow cytometry dot plots (left panel) and corresponding percentage of Ki67⁺CD71⁺ (right panel) among splenic (CD4⁺GFP⁺TdT⁺) T_reg cells and (CD4⁺GFP^-TdT⁺) ex-T_reg cells. Data from N = 6 mice per genotype, pooled from two independent experiments, with similar trend. (E) Mean fluorescence intensity (MFI) of CD71 expression in Ki67⁺ T_reg and ex-T_reg cells from the lymph nodes (LN) and spleen, from the same experiments as in (D). (F) Representative flow cytometry dot plots and (G) corresponding percentages of CD45.1⁺ and CD45.2⁺ double negative (DN), double positive (DP) thymocytes, T_H cells, cytotoxic T cells and (CD4⁺Foxp3⁺) T_reg cells in the thymus from the same BM chimeric mice illustrated in (Fig. 4A–G). Data from N = 11–12 mice per genotype, pooled from two independent experiments with similar trend. Data information: Data in (B–E, G) are presented as mean ± SD, circles in (B, D, E) correspond to individual mice or individual wells (C) and red bars are mean values. P values in Panel (B–E, G) were calculated using two-way ANOVA with Sidak’s multiple comparison test. NS not significant (P > 0.05), *P < 0.05; **P < 0.01; ***P < 0.001. Source data are available online for this figure.

We reasoned that if FTH prevents T_reg cells from transdifferentiating into ex-T_reg cells, then Fth deletion in T_reg cells should be associated with an accumulation of ex-T_reg cells in the spleen and/or lymph nodes. As expected, the percentage and number of T_reg cells were reduced in the spleen from Foxp3^{GFP-Fth∆/∆-tdT} vs. control Foxp3^GFP-tdT mice, as assessed at 19–30 weeks after birth (Fig. 3E). The percentage and number of splenic ex-T_reg cells remained relatively stable (Fig. 3F), but the ratio of ex-T_reg over tdT⁺ cells was higher in the spleen from Foxp3^{GFP-Fth∆/∆-tdT} vs. control Foxp3^GFP-tdT mice (Fig. 3G), with over 80% of T_reg cells becoming ex-T_reg cells (Fig. 3G). Fth was deleted in ex-T_reg cells from Foxp3^{GFP-Fth∆/∆-tdT} mice, as determined by qRT-PCR (Fig. EV3B), confirming that the ex-T_reg cells in Foxp3^{GFP-Fth∆/∆-tdT} mice do originate from T_reg cells, in which the Fth allele was deleted under the control of the Foxp3 promoter.

We then asked whether the transdifferentiation of T_reg cells into ex-T_reg cells was associated with the expression of pro-inflammatory cytokine, which is a feature of T_H cells activation. In strong support of this hypothesis, a large percentage of T_reg and ex-T_reg cells in the LN and to a lesser extent in the spleen from Foxp3^{GFP-Fth∆/∆-tdT} mice expressed IFNγ, as compared to the lack of IFNγ expression in T_reg and ex-T_reg cells from control Foxp3^GFP-tdT mice (Figs. 3H and EV3C). Moreover, a significant proportion of T_reg and ex-T_reg cells in the LN (Fig. 3I) and spleen (Fig. EV3D) from Foxp3^{GFP-FthΔ/Δ-tdT} mice co-expressed the proliferation markers Ki67 and CD71 (i.e., transferrin receptor), as compared to T_reg and ex-T_reg cells from control Foxp3^GFP-tdT mice (Figs. 3I and EV3D). This was not associated, however, with changes in the relative levels of CD71 expression (Fig. EV3E). Taken together these observations suggest that FTH is essential to restrain T_reg cells from transdifferentiating into inflammatory ex-T_reg cells.

FTH supports T_reg lineage maintenance in a cell-autonomous manner

To disentangle cell-autonomous from systemic effects associated with Fth deletion in T_reg cells we used mixed bone marrow (BM) chimeric mice. Briefly, sub lethally irradiated lymphogenic Rag2-deficient (Rag2^-/-) mice were reconstituted with BM cells from CD45.2⁺ Foxp3^{GFP-Fth∆/∆-tdT} vs. Foxp3^GFP-tdT mice (50%), plus congenic BM cells (50%) from CD45.1⁺ C57BL/6 mice (Fig. 4A). The proportion of CD45.2⁺ vs. CD45.1⁺ T_reg cells in LN (Fig. 4A–C) was markedly reduced in BM chimeric mice, when reconstituted with BM cells from Foxp3^{GFP-Fth∆/∆-tdT} vs. Foxp3^GFP-tdT mice. In contrast, there were no differences in the relative proportion of CD45.2⁺ vs. CD45.1⁺ T_reg cells in the thymus of these BM chimeric mice (Fig. EV3F,G). This suggests that FTH sustains the number of circulating T_reg cells via a cell-autonomous mechanism that does not affect T_reg cell development in the thymus.

Figure 4. FTH enforces T_reg cell lineage stability via a cell-autonomous mechanism.

(A) Schematic representation of bone marrow (BM) chimeric mice used for flow cytometry analyzed 6 weeks after reconstitution. For reconstitution BM cells from C57BL/6 CD45.1⁺ mice were mixed with BM cells from congenic C57BL/6 CD45.2⁺ Foxp3^{GFP-Fth∆/∆-tdT} or control Foxp3^GFP-tdT mice at a 1:1 ratio and injected (i.v.; tail vein, 200 µL) into congenic C57BL/6-recipient Rag2-deficient (Rag2^−/−) female mice, 2 h after irradiation (600 Gys). (B) Representative flow cytometry dot plots and (C) Percentage of CD45.1⁺ vs. CD45.2⁺ CD4⁺ T cells, CD8⁺ T cells and CD4⁺Foxp3⁺ T_reg cells in the lymph nodes of mixed BM chimeric mice, from (A). Data from N = 10 mice per genotype. (D–G) Representative flow cytometry dot plots (D) and corresponding quantification (E, F) of the percentage and number of (CD45.2⁺CD4⁺GFP⁺tdT⁺) T_reg cells (E), (CD45.2⁺CD4⁺GFP⁻tdT⁺) ex-T_reg cells (F) and relative proportion of ex-T_reg cells over total CD45.2⁺CD4⁺tdT⁺ cells in lymph nodes (LN) (G) of mixed BM chimeric mice (as in A). Data from N = 9–10 mice per genotype. (H) Schematic representation of (CD45.2⁺CD4⁺GFP⁺tdT⁺) T_reg cells and (CD45.2⁺CD4⁺GFP^-tdT⁺) ex-T_reg cells FACS-sorting from the lymph nodes of mixed BM chimeric mice, used for RNA sequencing analysis. (I, J) Volcano plot representations of RNA sequencing data of genes overexpressed (red) or under-expressed (blue) in T_reg cells (I) and ex-T_reg cells (J) from lymph nodes of BM chimeric mice (as in H). Data from N = 3–4 mice per genotype. P values determined by Benjamini and Hochberg adjusted probabilities. (K) Euler plot of differentially regulated genes in (CD45.2⁺CD4⁺GFP⁺) T_reg cells sorted from the Foxp3^{GFP-Fth∆/∆} or control Foxp3^GFP T_reg cells (non-chimeric; blue; from analysis described in Fig. 2A,B) or from the mixed BM chimeric mice (chimeric; red; from analysis illustrated in H–J). (L) Dumbbell plot, showing the adjusted P value of the 134 overlapping differentially regulated genes (from analysis described in K). Gray bars connecting dots represent the difference in P values for the differentially regulated genes (from analysis described in K). (M) Functional enrichment analysis of the overlapping genes (N = 134) (from analysis described in K), considering five different functional categories: biological processes (Bio. Proc.); KEGG (Kyoto Encyclopedia of Genes and Genomes) database; reactome (Reac.); transcription factors (TF); and WikiPathways (WP). Data were analyzed using g:SCS multiple testing correction method with a significance threshold of 0.05. Data information: Data in (C) are represented as mean ± SD. Data in (E–G) are represented as mean, circles correspond to individual mice and red bars are mean values. P value in (C) was determined by two-way ANOVA with Sidak’s multiple comparisons test and in (E–G) by unpaired t test with Welch’s correction. *P < 0.05; **P < 0.01; ***P < 0.001. Source data are available online for this figure.

The proportion of CD45.2⁺ vs. CD45.1⁺ Foxp3^-CD3⁺CD4⁺ T_H cells and CD3⁺CD8⁺ T_C cells was indistinguishable in the LN (Fig. 4A–C) of mixed BM chimeras reconstituted with CD45.2⁺ BM cells from Foxp3^{GFP-Fth∆/∆-tdT} vs. control Foxp3^GFP-tdT mice. Moreover, the percentage and number of activated CD45.2⁺CD4⁺CD44^highCD62L^low T_conv cells and CD45.2⁺CD8⁺CD44^highCD62L^low T_C cells in the LN (Fig. EV4A–C) were also similar in these BM chimeric mice. This confirms that FTH sustains the number of circulating T_reg cells, via a cell-autonomous mechanism that acts irrespectively of the systemic inflammatory response associated with Fth deletion in T_reg cells from Foxp3^{GFP-Fth∆/∆} mice (Fig. EV1D–G).

Figure EV4. FTH acts in a non-cell-autonomous manner to prevent T_REG cells from transdifferentiating into inflammatory T_reg cells.

(A) Schematic representation of the experimental approach used for flow cytometry analysis from the lymph nodes of BM chimeric mice (same experiment as Fig. 4A). (B, C) Representative flow cytometry dot plots (B), quantification of percentage (left panel) and number (right panel) (C) of live activated (CD45.2⁺CD4⁺CD44^highCD62L^low) and (CD45.2⁺CD8⁺CD44^highCD62L^low) cells in the lymph nodes of BM chimeric mice from (A). Data in (C) from n = 11–12 mice per genotype, pooled from 2 independent experiments, with similar trend. (D) Schematic representation of cell sorting for adoptive transfers in the experiment illustrated in Fig. 5A. (E) Representative flow cytometry dot plots of (CD4⁺GFP⁺tdT⁺) T_reg cells and relative level of Fth mRNA expression in (CD4⁺GFP⁺tdT⁺) T_reg cells (right panel) used for adoptive transfers in the experiment illustrated in Fig. 5A. (F) Schematic representation of experimental approach used for in vitro generation of induced T_reg (iT_reg) cells and conventional T_H (T_CONV) cells from sorted naive T_H (T_N) cells, activated with anti-CD3 and anti-CD28 mAb plus IL-2 and TGFβ. (G) Representative flow cytometry dot plots of iT_reg and T_CONV cells, generated in (F). (H) Percentage (%) and (I) Number (Nbr.) of Foxp3⁺ T_CONV and iT_reg cells, generated as described in (F). N = 2–5 independent experiments with similar trend. Each experiment corresponds to the average of different wells. (J) Representative flow cytometry carboxyfluorescein succinimidyl ester (CFSE) staining (left panel) and quantification of percentage (%) (right panel) of proliferating (CD4⁺Foxp3⁺) iT_reg cells, generated as described in (F). Data from 3 to 6 technical replicates in 1 out of 3 independent experiments, with similar trend. Data information: Data in (C, E) are presented as mean ± SD, circles correspond to individual mice and red bars are mean values. Circles in (H–J) correspond to individual wells and red bars are mean values. P values in panels (C, H, I) were calculated using two-way ANOVA with Sidak’s multiple comparison test. P values in (E) were calculated using Mann–Whitney test. NS not significant, ***P < 0.001. Source data are available online for this figure.

We then asked whether FTH restrains the transdifferentiation of T_reg cells towards ex-T_reg cells via a cell-autonomous mechanism. In support of this notion, there was a marked reduction in the percentage and number of CD45.2⁺CD3⁺CD4⁺GFP⁺tdT⁺ T_reg cells (Fig. 4D,E) and CD45.2⁺CD3⁺CD4⁺GFP^-tdT⁺ ex-T_reg cells (Fig. 4D,F) in the LN of mixed BM chimeric mice reconstituted with BM cells from Foxp3^{GFP-Fth∆/∆-tdT} vs. Foxp3^GFP-tdT mice. The ratio of CD45.2⁺ ex-T_reg cells over tdT⁺ cells was increased in the LN from chimeric mice reconstituted with BM cells from Foxp3^{GFP-Fth∆/∆-tdT} vs. Foxp3^GFP-tdT mice (Fig. 4G). This suggests that FTH maintains peripheral T_reg cell lineage stability, via a cell-autonomous mechanism, irrespectively of the systemic inflammatory response associated with Fth deletion in T_reg cells from Foxp3^{GFP-Fth∆/∆} mice (Fig. EV1D–G).

FTH maintains T_reg cell redox homeostasis via a cell-autonomous mechanism

We then asked whether FTH regulates gene expression in T_reg cells, irrespective of the systemic inflammatory response associated with Fth deletion in T_reg cells from Foxp3^{GFP-Fth∆/∆} mice (Fig. EV1D–G). To test this hypothesis, the gene expression profile of T_reg cells sorted from the LN of mixed BM chimeras (Fig. 4H) was compared to that of T_reg cells sorted from the LN of non-chimeric mice (Fig. 2A). Analysis of RNAseq data from BM chimeric mice showed that Fth-deleted T_reg cells (CD45.2⁺GFP⁺tdT⁺) developing from the BM of Foxp3^{GFP-Fth∆/∆-tdT} mice, upregulated 149 genes and downregulated 96 genes, compared to Fth-competent T_reg cells from control Foxp3^GFP-tdT mice (Fig. 4I). In the same BM chimeric mice, Fth-deleted ex-T_reg cells (CD45.2⁺GFP^-tdT⁺) developing from the BM of Foxp3^{GFP-Fth∆/∆-tdT} mice upregulated 90 genes and downregulated 36 genes, compared to Fth-competent ex-T_reg cells from control Foxp3^GFP-tdT mice (Fig. 4J). The genes regulated in T_reg and ex-T_reg cells originating from the BM of Foxp3^{GFP-Fth∆/∆-tdT} vs. Foxp3^GFP-tdT mice in BM chimeric mice, were associated with the oxidative stress response regulated by NRF2 (Fig. 4I,J). This suggests that FTH exerts cell-autonomous control of T_reg cell redox homeostasis, irrespective of the systemic inflammatory response associated with Fth deletion in T_reg cells from Foxp3^{GFP-Fth∆/∆} mice (Fig. EV1D–G).

The inflammatory transcriptional signature of T_reg cells from Foxp3^{GFP-Fth∆/∆} vs. Foxp3^GFP mice (Fig. 2B–D) was not observed in T_reg cells from BM chimeric mice, originating from Foxp3^{GFP-Fth∆/∆-tdT} vs. Foxp3^GFP-tdT (Fig. 4I,J). Among the 245 differentially expressed genes in Fth-deficient (CD4⁺GFP⁺) T_reg cells from BM chimeric mice (Fig. 4I,J), 134 matched those differentially expressed in Fth-deficient T_reg cells from non-chimeric mice (Fig. 4K,L). Gene ontology analyzes of the overlapping genes, showed an enrichment for pathways related to oxidative stress response, comprising several NRF2-regulated genes (Fig. 4K–M). In contrast, Fth-deficient (CD4⁺GFP⁺) T_reg cells from BM chimeric mice did not show an enrichment for pathways associated with T_H cell activation (Fig. 4K–M). This suggests that FTH acts in a cell-autonomous manner to support T_reg cell redox homeostasis and restrain T_reg cell transdifferentiation into ex-T_reg cells (Fig. 4). In contrast, the inflammatory profile associated with the transition of Fth-deleted T_reg cells towards inflammatory ex-T_reg cells, observed in Foxp3^{GFP-Fth∆/∆} mice (Fig. 2) and Foxp3^{GFP-Fth∆/∆-tdT} mice (Fig. 2) requires, in addition, the development of systemic inflammation.

FTH acts in a cell-autonomous manner to support T_reg cell homeostatic expansion

We took advantage of the homeostatic expansion of T_reg cells, upon adoptive transfer into lymphopenic Rag2^−/− mice (Duarte et al, 2009), to compare the proliferative capacity of CD4⁺GFP⁺tdT⁺ T_reg cells from Foxp3^{GFP-Fth∆/∆-tdT} vs. control Foxp3^GFP-tdT mice. The number of CD4⁺GFP⁺tdT⁺ T_reg cells recovered from the LN was markedly reduced when Rag2^−/− mice received T_reg cells from Foxp3^{GFP-Fth∆/∆-tdT} vs. Foxp3^GFP-tdT mice (Fig. 5A–C). While there were no differences in the number of CD4⁺GFP^-tdT⁺ ex-T_reg, the ratio of T_reg over tdT⁺ cells (GFP^-tdT⁺/TdT + ) were higher in Rag2^−/− mice receiving T_reg cells from Foxp3^{GFP-Fth∆/∆-tdT} vs. Foxp3^GFP-tdT mice, albeit without statistical significance (Fig. 5A–C). Fth expression in CD4⁺GFP⁺tdT⁺ T_reg cells used in the adoptive transfer was confirmed by qRT-PCR (Fig. EV4D,E).

Figure 5. FTH is a T_reg cell-autonomous cytoprotectant.

(A) Schematic representation of flow cytometry analyses of (CD4⁺GFP⁺tdT⁺) T_reg cells, isolated from the lymph nodes of lymphopenic Rag2^−/− mice, 6 weeks after adoptive transfer of T_reg cells sorted from Foxp3^{GFP-Fth∆/∆-tdT} vs. control Foxp3^GFP-tdT mice. (B) Representative flow cytometry dot plots of (CD4⁺GFP⁺tdT⁺) T_reg cells and (CD4⁺GFP^-tdT⁺) ex-T_reg cells, analyzed, as illustrated in (A). (C) Number (CD4⁺GFP⁺tdT⁺) T_reg cells and (CD4⁺GFP^-tdT⁺) ex-T_reg (left panel), as well as the relative proportion of ex-T_reg cells over total CD4⁺tdT⁺ cells (right panel), 6 weeks after adoptive transfer into Rag2^-/- mice, as illustrated in (A). Data from N = 4 per genotype, in one out of two independent experiments with similar trend. (D) Schematic representation of the experimental approach (left panel) and number (Nbr.) of mitochondria in (CD4⁺GFP⁺) T_reg cells sorted from lymph nodes of the Foxp3^{GFP-Fth∆/∆} or control Foxp3^GFP mice (right panel), quantified by genomic quantitative PCR. Data from N = 3 mice per genotype in one experiment. (E) Schematic representation of the experimental approach used to quantify mitochondrial membrane potential in splenic mouse (CD4⁺GFP⁺) T_reg cells (left panel). Representative flow cytometry histograms of tetramethylrhodamine ethyl ester (TMRE) staining (middle panels). Mean fluorescence intensity (MFI) of TMRE (right panel). Data from N = 4–5 mice per genotype, pooled from two independent experiments, with similar trend. (F) Schematic representation of experimental approach (left panel) used to quantify oxygen consumption rate (OCR) in live splenic (CD4⁺GFP⁺) T_reg cells (right panel). Data pooled from N = 3 mice per genotype, represented as mean ± SD (N = 3–5 technical replicates) in one out of three independent experiments, with similar trend. Oligomycin (Oligo.), carbonilcyanide p-triflouromethoxyphenylhydrazone (FCCP), Antimycin A/Rotenone (A/A+Rot.). (G) Quantification of basal respiration, ATP production and spare respiratory capacity, from data represented in (F). (H) Extracellular acidification rate (ECAR) in live splenic (CD4⁺GFP⁺) T_reg cells represented as mean ± SD (N = 3–5 technical replicates) in one out of three independent experiments, with similar trend. (I) Schematic representation of the experimental approach (left panel), used to quantify the ratio of α-ketoglutarate to isocitrate and α-ketoglutarate to glutamate (right panels) in live splenic (CD4⁺GFP⁺) T_reg cells. Data from N = 4 mice per genotype in one out of two independent experiments with similar trend. Data information: Data in (C–E) circles correspond to individual value and red bars to mean values. Data are presented as mean ± SD. Data in (F, H) are presented as mean ± SD, circles correspond to technical replicates. Data in (G) are presented as mean ± SD of technical replicates. Data in (I) are presented as mean ± SD, circles correspond to individual values. P values in (C (left panel), G) were calculated using two-way ANOVA with Sidak’s multiple comparison test, P values in (F, H) were calculated using two-way ANOVA with Bonferroni’s multiple comparisons test. P values in (C, right panel), D, E, I) were calculated using unpaired t test with Welch’s correction. NS not significant (P > 0.05), *P < 0.05; **P < 0.01; ***P < 0.001. Source data are available online for this figure.

We considered the possibility of an increase in the ratio of ex-T_reg to T_reg cells associated with Fth deletion in T_reg cells reflecting, to some extent, a T_reg cell survival defect. Therefore, we asked whether FTH acts in a cell-autonomous manner to support T_reg viability and proliferation in vitro. The frequency and number of induced T_reg (iT_reg) cells generated from T_N cells activated in vitro with anti-CD3/CD28 mAb plus IL-2 and TGFβ, were indistinguishable regardless of whether T_N cells were sorted from Foxp3^{GFP-Fth∆/∆} or Foxp3^GFP mice (Fig. EV4F–I). There were also no changes in the iT_reg cell proliferation (Fig. EV4J), suggesting that FTH controls T_reg cells in vivo via a mechanism that acts beyond its cytoprotective effects (Berberat et al, 2003; Pham et al, 2004).

FTH regulates T_reg cell mitochondrial function and bioenergetics

T_reg cells rely on a core metabolic program whereby mitochondrial oxidative phosphorylation is the major source of ATP (Angelin et al, 2017; Weinberg et al, 2019). The observation that FTH regulates mitochondrial integrity in parenchyma cells (Blankenhaus et al, 2019) suggested that FTH also regulates the mitochondrial integrity of T_reg cells. However, the number of mitochondria per CD4⁺GFP⁺ T_reg cells in the LN of Foxp3^{GFP-Fth∆/∆} was similar to that of T_reg cells from Foxp3^GFP mice, corresponding to ~55 mitochondria per T_reg cell (Fig. 5D). This suggests that FTH is not essential to maintain the mitochondrial integrity of T_reg cells.

The mitochondrial membrane potential of Fth-deficient T_reg cells from Foxp3^{GFP-Fth∆/∆} mice was markedly reduced, compared to control T_reg cells from Foxp3^GFP mice (Fig. 5E). This suggests that FTH regulates the mitochondrial function of T_reg cells.

We then tested whether FTH regulates the mitochondrial energetic capacity of T_reg cells, by quantifying the relative increase in basal O₂ consumption rate (OCR), upon uncoupling of the mitochondrial respiratory chain by carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP). Spare mitochondrial respiratory capacity was higher in (CD4⁺GFP⁺) T_reg cells from Foxp3^{GFP-Fth∆/∆} vs. Foxp3^GFP mice (Fig. 5F,G). Basal OCR and mitochondrial ATP production were similar in T_reg cells from Foxp3^{GFP-Fth∆/∆} vs. Foxp3^GFP mice, as assessed by the relative decrease of OCR upon ATP synthase inhibition by Oligomycin (Fig. 5F,G). Non-mitochondrial OCR was also similar in T_reg cells from Foxp3^{GFP-Fth∆/∆} vs. Foxp3^GFP mice, as assessed by the inhibition of the electron transport chain complex-I and -III by Rotenone and Antimycin A, respectively (Fig. 5F,G).

The spare respiratory capacity of T_reg cells from control Foxp3^GFP mice was lower, compared to T_conv cells (Fig. EV5A,B). In contrast, the spare respiratory capacity of T_reg cells from Foxp3^{GFP-Fth∆/∆} mice was similar to that of T_conv cells (Fig. EV5C,D), while the extracellular acidification rate (ECAR), reflecting the rate of glycolysis, was higher in T_reg cells from Foxp3^{GFP-Fth∆/∆} vs. Foxp3^GFP mice (Fig. 5H). This suggests that FTH plays an essential role in supporting the metabolic and bioenergetic profile of T_reg cells, favoring mitochondrial oxidative OXPHOS over glycolysis (Angelin et al, 2017).

Figure EV5. FTH regulates mitochondrial energy metabolism and CpG methylation in T_reg cells.

(A) Oxygen consumption rate (OCR) in live splenic (CD4⁺GFP⁺) T_reg cells and (CD4⁺GFP^-) T_CONV cells from Foxp3^GFP mice. (B) Quantification of spare respiratory capacity, from data represented in (A). (C) Oxygen consumption rate (OCR) in live splenic (CD4⁺GFP⁺) T_reg cells and (CD4⁺GFP^-) T_CONV cells from Foxp3^{GFP-Fth∆/∆} mice. (D) Quantification of spare respiratory capacity, from data represented in (C). Data in (A–D) pooled from N = 3 mice per genotype, represented as mean ± SD. N = 3–5 technical replicates in 1 out of 3 independent experiments, with similar trend. Oligomycin (Oligo.), carbonilcyanide p-triflouromethoxyphenylhydrazone (FCCP), Antimycin A/Rotenone (A/A+Rot.). (E) Schematic representation of sorting of splenic (CD4⁺GFP⁺) T_reg cells used for targeted metabolomics (left panel). Quantification of intermediate metabolites from targeted metabolomics analyzes of splenic T_reg cells (right panel). Data from N = 3–4 mice per genotype in one experiment representative of 3 independent experiments with similar trend. (F) Schematic representation of the experimental approach used for flow cytometry analysis of tumor-infiltrating cells (left panel), representative flow cytometry dot plots (middle panel) and corresponding percentage and number (right panel) of live tumor-infiltrating (CD4⁺IFNγ⁺) T_H cells (CD8⁺IFNγ⁺) T_C cells, 3 weeks after tumor inoculation (2 × 10⁵ B16 cells). Data from N = 6 mice per genotype, pooled from two independent experiments, with similar trend. (G) Schematic representation of the experimental approach used for flow cytometry analysis of tumor-infiltrating cells (left panel), representative flow cytometry dot plots (middle panel) and corresponding percentage and number (right panels) of live tumor-infiltrating (CD8⁺GrzmB⁺) T cells, 3 weeks after tumor inoculation (2 × 10⁵ B16 cells). Data from N = 6 mice per genotype, pooled from two independent experiments, with similar trend. Data information: Circles and triangles in (A, C) correspond to mean values, circles, and triangles in (B, D) correspond to individual wells, and circles in (E–G) correspond to individual mice, and red bars are mean values. P values in (A, C) calculated using two-way ANOVA with Bonferroni’s (A, C) or Sidak´s (E, F) multiple comparisons test, in (B, D, G) using unpaired t test with Welch’s correction. NS, not significant (P > 0.05); *P < 0.05; **P < 0.01; ****P < 0.0001. Source data are available online for this figure.

To gain further insight regarding how FTH regulates T_reg cell bioenergetics, we performed targeted metabolomics analysis. The intracellular concentration of different TCA cycle metabolites in splenic Fth-deficient T_reg cells from Foxp3^{GFP-Fth∆/∆} mice was similar to that of control T_reg cells from Foxp3^GFP mice (Fig. EV5E). However, the ratio of α-ketoglutarate and isocitrate intracellular concentrations was lower in splenic Fth-deficient T_reg cells from Foxp3^{GFP-Fth∆/∆} mice vs. control T_reg cells from Foxp3^GFP mice (Fig. 5I). In contrast, the ratio of intracellular α-ketoglutarate and glutamate concentration was similar (Fig. 5I). This suggests that FTH modulates the rate of isocitrate to α-ketoglutarate conversion (Fendt et al, 2013), catalyzed by isocitrate dehydrogenase (IDH) in the mitochondrial TCA cycle. This occurs, most likely, without interfering with glutaminolysis, whereby α-ketoglutarate is generated via the conversion of glutamine into glutamate, catalyzed by glutamate dehydrogenases and glutaminase, respectively.

In contrast to other intermediate TCA cycle metabolites, the intracellular concentration of lactate was lower in splenic Fth-deficient T_reg cells from Foxp3^{GFP-Fth∆/∆} mice vs. control T_reg cells from Foxp3^GFP mice (Fig. EV5E). This is consistent with FTH modulating the metabolic and bioenergetics profile of T_reg cells, presumably favoring lactate production via glycolysis, similar to observed in hepatocytes (Weis et al, 2017).

FTH regulates cytosine methylation in T_reg cells

Maintenance of T_reg cell lineage relies on sustained demethylation of cytosines at CpG-rich sequences in the FOXP3 CNS1 and 2 (Ohkura et al, 2012; Zheng et al, 2010). Cytosine demethylation is catalyzed by the TET family of methylcytosine dioxygenases (Yue et al, 2019; Yue et al, 2016), via redox-based reactions that use Fe and α-ketoglutarate as an essential cofactor and obligate substrate, respectively (Kohli and Zhang, 2013; Pastor et al, 2013). Having established that FTH regulates intracellular catalytic Fe²⁺, cellular redox homeostasis and possibly the rate of α-ketoglutarate to isocitrate conversion, we asked whether FTH modulates cytosine demethylation in T_reg cells. To test this hypothesis, we performed a genome-wide methyl-sequencing (EM-seq) profiling of CD45.2⁺CD4⁺tdT⁺ cells, which include (GFP⁺) T_reg and (GFP^-) ex-T_reg cells sorted from the LN of mixed BM chimeras, reconstituted with CD45.1⁺ (50%) BM cells plus CD45.2⁺ (50%) BM cells from Foxp3^{GFP-Fth∆/∆-tdT} or control Foxp3^GFP-tdT mice (Fig. 6A). The Methylome of CD45.2⁺CD4⁺tdT⁺ cells, originating from the BM of Foxp3^{GFP-Fth∆/∆-tdT} and control Foxp3^GFP-tdT mice clustered independently, as assessed by principal component analysis (Fig. 6B), revealing that FTH regulates the Methylome of T_reg and ex-T_reg cells.

Figure 6. FTH regulates mitochondrial function and cytosine methylation in T_reg cells.

(A) Schematic representation of mixed bone marrow (BM) chimeric mice from which lymph node CD45.2⁺CD4⁺tdT⁺ cells (i.e., GFP⁺ T_reg cells and GFP⁻ ex-T_reg cells) were sorted for genome-wide EM-seq analyses. (B) Principal component analysis (PCA) of the methylome of lymph node T_reg and ex-T_reg cells in BM chimeric mice generated, as illustrated in (A). Data from N = 4 mice per genotype is represented as individual circles in one experiment. (C) Unsupervised heatmap representation of genome-wide EM-seq analyzes (i.e., 5-hmC) in CD45.2⁺CD4⁺tdT⁺ cells sorted from the mixed BM chimeric mice illustrated in (A). Statistical analysis for multiple testing correction was performed with Sliding Linear Model (SLIM). (D) The relative percentage of total methylated regions (i.e., hyper- and hypomethylated regions) (q < 0.01 and methylation difference >=10%) according to different genome regions (promoter, exon, intron and intergenic). (E) The number of hyper- and hypomethylation events (10% change in methylation and a q value of 1%) per chromosome, shown as a percent of the differential sites. (F) Schematic representation of the Foxp3 in enhancer regions CNS1 and CNS2. (G) Relative quantification of the methylation rate of CpG sequences in the Foxp3 CNS1 (left panel) and CNS2 (right panel) from CD45.2⁺CD4⁺tdT⁺ cells sorted from the lymph nodes of BM chimeric mice, illustrated in (A). Data from N = 4 mice per genotype in one experiment. (H) TET activity in nuclear extracts from HEK293T cells transiently transfected with human TET3-flag, FTH-flag, or FTH^mut-flag cDNAs. Data shows technical replicates, pooled from four independent experiments. (I) FTH-flag, FTH^mut-flag, TET3-flag, Lamin A/C and GAPDH protein expression, detected by western blot in nuclear and cytosol extracts from HEK293T cells transfected as described in (H). Relative quantification of TET3-flag, normalized to Lamin A/C (bottom left panel), and FTH-flag, normalized to GAPDH (bottom right panel). Data from one experiment, representative of three independent experiments with similar trend. Data information: Data in (G) are presented as mean ± SD, circles correspond to individual mice and red bars to mean values. Data in (H, I) are presented as mean ± SD, circles correspond to technical replicates. P values in (G) were calculated using Two-way ANOVA with Sidak’s multiple comparison test, and in (H, I) using one-way ANOVA using with Sidak’s multiple comparison test. NS not significant (P > 0.05), *P < 0.05; **P < 0.01, ***P < 0.001. Source data are available online for this figure.

Fth-deficient CD45.2⁺CD4⁺tdT⁺ cells, originating from the BM of Foxp3^{GFP-Fth∆/∆-tdT} mice, presented a discrete number of hypermethylated and hypomethylated CpG-rich sequences, compared to CD45.2⁺CD4⁺tdT⁺ cells originating from the BM of control Foxp3^GFP-tdT mice (Fig. 6C). These hyper and hypomethylated regions were located primarily in the promoter and intergenic regions and only to a lesser extent in introns and exons (Fig. 6D), from different chromosomes (Fig. 6E). This suggests that FTH regulates cytosine methylation in a mixed population of T_reg and ex-T_reg cells, via a cell-autonomous mechanism.

FTH regulates FOXP3 CNS1 and CNS2 methylation

Having established that FTH modulates the methylome of T_reg and ex-T_reg cells, we asked whether this was associated with changes in the methylation of CpG-rich sequences at the FOXP3 CNS1 and CNS2 (Fig. 6F), controlling T_reg cell lineage stability (Yue et al, 2019; Yue et al, 2016). Analyzes of the EM-seq data from mixed BM chimeric mice showed sustained hypermethylation of CpG-rich sequences in the FOXP3 CNS1 and CNS2 from splenic T_reg and ex-T_reg cells originating from Foxp3^{GFP-Fth∆/∆-tdT} vs. control Foxp3^GFP-tdT mice (Fig. 6G). This suggests that FTH acts in a cell-autonomous manner to sustain cytosine demethylation at FOXP3 CNS1 and CNS2 and presumably therefore regulate T_reg cell lineage maintenance.

The ferroxidase activity of FTH controls TET dioxygenase activity

We then questioned whether FTH modulates TET activity and tested this hypothesis in human HEK293 cells, transiently co-transfected with human TET3 plus FTH or an FTH mutant (FTH^mut) lacking ferroxidase activity (Broxmeyer et al, 1991). TET3 enzymatic activity was reduced in cells co-transfected with the FTH^mut, compared to control cells co-transfected with an empty vector (Fig. 6H). In contrast, FTH overexpression did not alter TET3 enzymatic activity, compared to controls. This suggests that FTH^mut, possibly acts as a dominant negative mutant failing to regulate catalytic Fe and likely therefore impairing TET activity. Expression of FTH or FTH^mut did not alter the level of TET3 protein expression, as assessed by western blot (Fig. 6I). Taken together, these observations establish, as a proof of principle, that FTH ferroxidase activity regulates TET methylcytosine dioxygenase activity.

FTH is required to sustain Foxp3 transcription and expression

Having established a functional link between FTH ferroxidase activity and TET methylcytosine dioxygenase activity, we questioned whether FTH regulates FOXP3 transcription and expression. Suppression of FTH expression in human T_reg cells transduced with shRNAs targeting FTH, was associated with a reduction of FOXP3 protein expression, compared to control T_reg cells transduced with non-targeting shRNA (Fig. 7A,B). This suggests that FTH acts in a cell-intrinsic manner to regulate the expression of FOXP3 in human T_reg cells.

Figure 7. FTH is required to sustain Foxp3 transcription and expression.

(A) FTH protein detected by western blot in whole-cell extracts from HEK293T cells infected with recombinant lentiviruses coding shRNAs targeting FTH (FTH⁴²⁹ and FTH⁴³²) or control (Ctrl.) recombinant lentiviruses non-targeting shRNA. (B) Mean fluorescence intensity (MFI) of FOXP3 expression, detected by flow cytometry in human (CD4⁺CD45RA⁺CD25⁺) T_reg cells infected with the same recombinant lentiviruses as in (A). Data from N = 6 samples per experimental group. (C) Schematic representation of the experimental approach (left panel) used to monitor GFP transgene expression in the mesenteric LN (MLN) of (CD4⁺GFP⁺Nrp1⁺) tT_reg cells and (CD4⁺GFP⁺Nrp1⁻) pT_reg cells. Representative flow cytometry histogram (middle panel) and quantification of relative GFP expression (right panel), shown as mean fluorescence intensity (MFI). Data from N = 8 mice per genotype, pooled from two independent experiments with similar trend. (D) Schematic representation of the experimental approach (left panel) used to monitor Foxp3 expression by flow cytometry in mouse spleen and MLN (CD4⁺Foxp3⁺) T_reg cells. Representative flow cytometry staining of Foxp3 (middle panel). Relative quantification of Foxp3 expression (right panel), represented as mean of fluorescence intensity (MFI). Data from N = 9 mice per genotype, pooled from two to three independent experiments with similar trend. (E) Schematic representation of the experimental approach (left panel) used to monitor Foxp3 expression in (CD45.2⁺CD4⁺GFP⁺tdT⁺) T_reg cells isolated from the spleen and LN of BM chimeras. Representative flow cytometry of Foxp3 staining (middle panel). Relative quantification of Foxp3 expression (right panel), shown as mean fluorescence intensity (MFI). Data in (E) from N = 5–6 mice per genotype, representative of two independent experiments with similar trend. (F) Schematic representation of the experimental approach (left panel) used to monitor GFP expression in the spleen and LN of CD45.2⁺CD4⁺tdT⁺ cells (T_reg+ ex-T_reg) from BM chimeras. Representative flow cytometry of GFP (F) staining (middle panel). Relative quantification of GFP expression (right panel), represented as mean fluorescence intensity (MFI). Data from N = 10 mice per genotype, pooled from two independent experiments with similar trend. (G) Schematic representation of the experimental approach (left panel), where (CD4⁺tdT⁺) cells were sorted from the spleen and LN for qRT-PCR (G, left panel). Relative expression of Foxp3 (right panel). (H) Gfp, Fth, and tdT mRNA expression normalized to Arbp0 of cells sorted as in (G). Data in (G, H) from N = 3–4 mice per genotype from one experiment. Data information: Data in (B) are presented as mean ± SD, circles correspond to individual wells and red bars to mean values. Data in (C–H) are presented as mean ± SD, circles correspond to individual mice and red bars to mean values. P values in (B) were calculated using the Fiedman test with Dunn’s multiple comparison test, in (C–H) using two-way ANOVA with Sidak’s multiple comparison test. NS not significant (P > 0.05), *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Source data are available online for this figure.

Having established that regulation of Fe metabolism by FTH regulates Foxp3 transcription in the thymus as well as in circulating T_reg cells (Fig. 1K) we compared the relative level of Foxp3-GFP expression in Nrp1⁺ tT_reg and Nrp1^- pT_reg cells. GFP expression was reduced in both tT_reg and pT_reg cells, as assessed in the MLN from Foxp3^{GFP-Fth∆/∆} vs. control Foxp3^GFP mice (Fig. 7C). Considering that GFP is expressed under the control of a bacterial artificial chromosome (BAC) transgene carrying the intact Foxp3 promoter (Chen et al, 2003; Zhou et al, 2008), these observations suggest that FTH is essential to sustain Foxp3 transcription in tT_reg and pT_reg cells.

The relative level of Foxp3 protein expression was also reduced when Fth was deleted, as assessed in the spleen and MLN from Foxp3^{GFP-Fth∆/∆} vs. control Foxp3^GFP mice (Fig. 7D). This was also observed in BM chimeric mice, that is, the relative levels of Foxp3 protein expression were lower in Fth-deficient GFP⁺tdT⁺ cells (T_reg cells) originating from Foxp3^{GFP-Fth∆/∆-tdT} vs. control Foxp3^GFP-tdT mice (Fig. 7E). Moreover, GFP expression was lower in Fth-deficient tdT⁺ cells (ex-T_reg+T_reg cells) originating from Foxp3^{GFP-Fth∆/∆-tdT} vs. control Foxp3^GFP-tdT mice (Fig. 7F). This suggests that FTH acts in a cell-intrinsic manner to regulate Foxp3 transcription and expression in mouse T_reg cells.

To establish whether FTH regulates endogenous Foxp3 transcription, we asked whether Fth deletion was associated with a concomitant reduction of Foxp3 and Gfp mRNA expression. The relative level of Foxp3 (Fig. 7G) and Gfp (Fig. 7H) mRNA expression were lower in Fth-deficient CD4⁺tdT⁺ cells, including T_reg (GFP⁺tdT⁺) and ex-T_reg (GFP^-tdT⁺) cells, from Foxp3^{GFP-Fth∆/∆-tdT} vs. control Foxp3^GFP-tdT mice. Fth deletion in CD4⁺tdT⁺ cells from Foxp3^{GFP-Fth∆/∆-tdT} vs. Foxp3^GFP-tdT mice was confirmed by qRT-PCR (Fig. 7H). While tdT mRNA expression was not altered (Fig. 7H). These observations suggest that FTH is essential to sustain FOXP3 transcription.

FTH expression in T_reg cells limits the pathologic outcome of autoimmune neuroinflammation

Given the central role of T_reg cells in preventing the onset of autoimmune diseases (Kohm et al, 2002; Lafaille et al, 1994), we tested whether FTH expression in T_reg cells impacts on the pathogenesis of experimental autoimmune encephalomyelitis (EAE) (Fig. 8A). Foxp3^{GFP-Fth∆/∆} mice had an increase in EAE incidence (Fig. 8B) and severity (Fig. 8C), in response to immunization with myelin oligodendrocyte glycoprotein (MOG)-derived peptide 35-55 (MOG_35-55) emulsified in complete Freund’s adjuvant compared to control immunized Fth^fl/fl mice. Of note, the immunization protocol used was “sub-optimal”, as suggested by the relatively low disease scores in control immunized Fth^fl/fl mice, likely favoring the increase in EAE severity observed in Foxp3^{GFP-Fth∆/∆} mice.

Figure 8. FTH expression in T_reg cells controls the pathologic outcome of experimental immune-driven inflammatory conditions.

(A) Schematic representation of the induction experimental autoimmune encephalomyelitis (EAE) in response to MOG_35-55 immunization. (B) EAE incidence (percentage) and (C) EAE severity in MOG_35-55 immunized mice. Data from N = 18–20 mice per genotype, pooled from three independent experiments, with similar trend. (D) Experimental approach (right panel), representative flow cytometry dot plots (middle panel) and corresponding quantification (left panel) of the relative percentage of activated T_H1 (CD3⁺CD4⁺IFN-γ⁺), T_H17 (CD3⁺CD4⁺IL-17A⁺); and double positive IFN-γ⁺IL-17A⁺ T_H cells in the spinal cord, 22 days after MOG_35-55 immunization. Data from N = 3–4 mice per genotype. (E) Survival (left panel) and number of circulating Plasmodium chabaudi chabaudi (Pcc)-infected red blood cells (iRBC) per µL of whole blood (i.e., parasite burden) (right panel). N = 9 mice per genotype, pooled from two independent experiments, with similar trend. (F, G) Representative flow cytometry dot plot (left panels) and corresponding percentage and cell numbers (right panels) of splenic (CD4⁺Foxp3-GFP⁺) T_reg cells (F) and IFNγ⁺CD4⁺ activated T_H cells (G), 7 days after Pcc infection. Data from N = 8–9 mice per genotype, pooled from two independent experiments, with similar trend. (H) Relative tumor (B16-F10-luc2) size, 13–19 days after inoculation (2 × 10⁵ cells). Data from N = 7–11 mice per genotype, pooled from 3 independent experiments, with similar trend. (I, J) Representative flow cytometry dot plots (left panels) and corresponding percentage (right panels) of live tumor-infiltrating (CD4⁺Foxp3⁺) T_reg cells (I) and (CD4⁺Foxp3^-CD25⁺) effector T_H cells (J), 3 weeks after tumor inoculation. N = 7 mice per genotype, pooled from three independent experiments, with similar trend). Data information: Circles in (D, F, G, I, J) correspond to individual mice and red bars to mean values. Data in (C, H) are presented as mean ± SEM. Data in (E, right panel) are presented as mean ± SD. P values in (C), (E, right panel), and (H) were determined using Holm–Sidak method (multiple t tests), with alpha = 0.05 under the assumption that both genotypes have similar SEM, in (B, E) by log-rank (Mantel–Cox) test, and in (D, F, G, I, J) by unpaired t test with Welch’s correction. NS not significant (P > 0.05); *P < 0.05; **P < 0.01; ***P < 0.001. Source data are available online for this figure.

The frequency of T_H cells expressing IFNγ, IL-17A and IL17⁺IFNγ⁺ T_H cells was higher in the spinal cord of MOG_35-55-immunized Foxp3^{GFP-Fth∆/∆} vs. control Fth^fl/fl mice (Fig. 8D). This suggests that FTH expression in T_reg cells limits the activation, proliferation and/or infiltration of self-reactive T_H type 1 (T_H1) and T_H type 17 (T_H17) cells as well as pathogenic IL17⁺IFNγ⁺ T_H cells (Duhen et al, 2013) into the central nervous system, in response to MOG_35-55 immunization.

FTH expression in T_reg cells limits malaria severity

T_reg cells constraint the extent of immunopathology associated with infectious diseases (Arpaia et al, 2015), supporting the hypothesis that regulation of T_reg cell lineage stability by FTH impacts on the severity of infectious diseases (Soares et al, 2017). We tested this hypothesis for severe malaria, an often-lethal outcome of Plasmodium spp. infection. Foxp3^{GFP-Fth∆/∆} mice succumbed to Plasmodium chabaudi chabaudi (Pcc) AS infection, as compared to Foxp3^GFP mice that survived and cleared the parasite (Fig. 8E). However, the number of circulating infected red blood cells (i.e., parasite burden) was lower at the peak of infection, in Foxp3^{GFP-Fth∆/∆} vs. control Foxp3^GFP mice (Fig. 8E). The lethal outcome of Pcc infection in Foxp3^{GFP-Fth∆/∆} mice was associated with a reduction in the frequency (but not the number) of splenic T_reg cells (Fig. 8F), as well as with an increase in the number (but not the frequency) of IFNγ-expressing T_H cells in the spleen (Fig. 8G). This suggests that FTH expression in T_reg cells is essential to restrain immune-driven pathology underlying the development of severe presentation of malaria, emphasizing the critical involvement of T_reg cells in the control of the pathogenesis of severe malaria (Walther et al, 2005). Moreover, these findings illustrate how dysregulation of Fe metabolism promotes the pathogenesis of severe malaria (Ferreira et al, 2008; Gozzelino et al, 2012; Ramos et al, 2019; Seixas et al, 2009; Wu et al, 2023).

FTH expression in T_reg cells favors tumor progression

T_reg cells are pathogenic, for example, when restraining anti-tumor immunity (Curiel et al, 2004; Liu et al, 2016), suggesting that FTH expression in T_reg cells favors tumor progression. In support of this hypothesis, the relative growth of syngeneic B16 melanoma cells was reduced in Foxp3^{GFP-Fth∆/∆} vs. control Fth^fl/fl mice (Fig. 8H). This was associated with lower frequency of tumor-infiltrating T_reg cells (Fig. 8I) and higher frequency of activated Foxp3^-CD4⁺CD25⁺ effector T_H cells (Fig. 8J). The frequency and number of activated CD4⁺IFNγ⁺ T_H cells isolated from B16 melanomas was similar in Foxp3^{GFP-Fth∆/∆} vs. control Foxp3^GFP mice (Fig. EV5F). In contrast, the number of activated CD8⁺IFNγ⁺ T_C cells was higher in B16 melanomas from Foxp3^{GFP-Fth∆/∆} vs. control Foxp3^GFP mice (Fig. EV5F). This tendency was also observed for CD8⁺granzymeB⁺ T_C cells, albeit not statistically significant (Fig. EV5G). This suggests that regulation of Fe metabolism by FTH in T_reg cells supports tumor progression, via a mechanism that hinders anti-tumor immunity.

Discussion

T_reg cells respond to variations in the relative levels of nutrients, vitamins and metabolites in their environment (Chapman et al, 2020; Shi and Chi, 2019), via dedicated transporter-receptor coupled sensors that modulate T_reg cell function and lineage stability (Kempkes et al, 2019; Shi and Chi, 2019). In keeping with Fe-regulatory genes being a core property of T_reg cells (Cuadrado et al, 2018), we found that the Fe-regulatory protein FTH is essential to support T_reg cell lineage maintenance in vivo (Figs. 1–4) and support immune homeostasis (Figs. 2,  EV1D–G, and EV2C).

FTH regulates T_reg cell lineage stability (Figs. 3 and 4A–G) without interfering with the antiproliferative function of T_reg cells (Fig. EV2A). Instead, FTH targets the intracellular pool of redox-active Fe²⁺ (Fig. EV1B) to regulate T_reg cell redox homeostasis (Figs. 2F and 4K–M) in a manner that controls T_reg cell: (i) energy metabolism (Figs. 5F–I and EV5A–E), (ii) TET activity (Fig. 6H), (iii) cytosine demethylation at CpG-rich sequences at CNS1 and CNS2 in the FOXP3 locus (Fig. 6G), and (iv) FOXP3 transcription/expression (Figs. 1K,L and 7F–H). As the latter is essential to maintain the transcriptional program supporting T_reg cell lineage stability (Williams and Rudensky, 2007) (Nakatsukasa et al, 2019; Ohkura et al, 2012; Yue et al, 2019) Fth deletion is associated with a decrease of T_reg cells, including tT_reg cells and pT_reg cells, without interfering with thymic T_reg cell output (Fig. 1H,L). These observations are consistent with FTH acting upstream of TET methylcytosine dioxygenases, to control redox-based cytosine demethylation in T_reg cells (Fig. 6) (Huang and Rao, 2014; Pastor et al, 2013).

That FTH targets the intracellular pool of redox-active Fe²⁺ to control TET enzymatic activity is suggested by the observation that overexpression of a ferroxidase deficient FTH^m compromised TET methylcytosine dioxygenase activity (Fig. 6H, I). Several non-mutually exclusive mechanisms might explain this observation. First, FTH could act “directly” via sequestration of catalytic Fe²⁺ (Fig. EV1B), controlling the availability of this essential cofactor of TET enzymatic activity (Kohli and Zhang, 2013; Pastor et al, 2013). Second, FTH could act “indirectly” via the regulation of cellular redox homeostasis (Figs. 2D,F and 4K–M), preventing catalytic Fe²⁺ from catalyzing oxidative stress, which compromises TET activity (Niu et al, 2015). Moreover, this should restrain NRF2 activation (Figs. 2D,F and 4K–M) from repressing Foxp3 expression and impair T_reg cell function (Klemm et al, 2020).

Consistent with our findings, intracellular Fe mobilization via lysosome-mediated ferritinophagy, was recently shown to regulate TET-driven (de)methylation of the peroxisome proliferator-activated receptor γ (PPARγ) locus, the master regulators of adipocyte development (Suzuki et al, 2023). This suggest that FTH controls Fe-responsive epigenetic programs defining different cellular developmental programs.

FTH regulates T_reg cell mitochondrial TCA cycle and OXPHOS (Figs. 5F–I and EV5A–E), consistent with similar findings in other cell types (Blankenhaus et al, 2019; Oexle et al, 1999). While FTH does not modulate the intracellular concentration of intermediate TCA cycle metabolites (Fig. EV5E), including α-ketoglutarate (Fig. 5I), it does appear to modulate the rate of isocitrate conversion into α-ketoglutarate, likely acting irrespectively of α-ketoglutarate generation via glutaminolysis. It is possible therefore that FTH regulates the production of this obligatory substrate of TET cytosine dioxygenases (Kohli and Zhang, 2013; Pastor et al, 2013) via regulation of the TCA cycle. This interpretation is consistent with other signal transduction pathways regulating the TCA cycle, such as the one triggered by the programmed cell death 1 ligand 2 (PD-L2), regulating cytosine methylation at CpG sequences in the T_reg-specific demethylation region, compromising Foxp3 stability in T_reg cells (Hurrell et al, 2022).

The metabolic program orchestrated by FOXP3, supports T_reg cell antiproliferative function and lineage stability, via a mechanism that relies on the suppression of c-Myc, a transcription factor that represses mitochondrial OXPHOS and promotes glycolysis (Angelin et al, 2017). Importantly, c-Myc can repress FTH transcription and translation in other cell types, the later occurring via a mechanism involving the Fe-regulatory protein 2 (IRP2) (Wu et al, 1999). This suggests that the FOXP3-driven genetic program encompasses the induction of FTH via a mechanism that could involve the repression of c-Myc. Moreover, the FTH promoter contains at least one Foxp3 DNA binding site (Appendix Figs. S3 and 4) and as such it is possible that Foxp3 would regulate FTH expression directly. This would argue for a positive feedback loop in which FTH enforces Foxp3 expression, via the regulation of TET dioxygenases, and the later enforces the FTH expression transcriptionally. This hypothesis remains however to be tested experimentally.

While FTH prevents the transdifferentiation of T_reg cells toward inflammatory ex-T_reg cells (Figs. 3 and EV2C–E), via a non-cell-autonomous mechanism that relies on systemic inflammation (Fig. EV1D–G), this is not associated with the accumulation of inflammatory ex-T_reg cells in Foxp3^{GFP-Fth∆/∆-tdT} vs. control Foxp3^GFP-tdT mice (Fig. 3E–G). The same is true in mixed BM chimeric mice reconstituted with BM cells from Foxp3^{GFP-Fth∆/∆-tdT} vs. control Foxp3^GFP-tdT mice (Fig. 4A–G). One possible interpretation is that FTH is essential not only to restrain T_reg cells from transdifferentiating into inflammatory ex-T_reg cells but also to support the viability of highly proliferating T_reg and ex-T_reg cells. This is consistent with the lack of autoimmune lesions in Foxp3^GFP-FthΔ/Δ mice (Fig. EV2B). We note, however, that Fth deletion in T_reg cells does not compromise the T_reg cell thymic output (Fig. 1H) nor the generation and proliferation of iT_reg cells in vitro (Fig. EV4F–J).

One cannot exclude that the highly methylated status of the Foxp3 locus from ex-T_reg cells (Komatsu et al, 2014; Miyao et al, 2012; Zhou et al, 2009) together with the increase frequency of ex-T_reg cells among CD45.2⁺CD4⁺tdT⁺ cells (Fig. 4A–G), contributes to the observed increase in the methylation of the Foxp3 locus in Fth-deleted T_reg cells (Fig. 6A–G). This does not invalidate however, that Fth deletion acts in a cell-autonomous manner to decrease Foxp3 transcription/expression (Figs. 1K,L and 7F–H), therefore increasing the frequency at which T_reg cells transdifferentiate into ex-T_reg cells (Fig. 4A–G).

In contrast with genetic deletion of Tet2 and Tet3 in T_reg cells, which promotes the transdifferentiation of T_reg cells into inflammatory ex-T_reg cells and the development of autoimmunity (Nakatsukasa et al, 2019; Ohkura et al, 2012; Yue et al, 2019), Fth deletion in T_reg cells is not associated with overt autoimmunity (Fig. EV2BG). This is consistent with FTH exerting additional effects, beyond the regulation of TET dioxygenases, preventing cellular stress from compromising the viability of the inflammatory ex-T_reg cells that would otherwise elicit autoimmunity.

Fth deletion in T_reg cells led to an increase in EAE susceptibility and severity (Fig. 8A–C), induced by an immunization protocol leading to low-grade disease severity in control mice (Fig. 8C). This was associated with a higher accumulation of activated T_H1 and T_H17 cells as well as pathogenic IFNγ⁺IL-17A⁺ T_H cells (Duhen et al, 2013) in the central nervous system (Fig. 8D), likely originating from auto-reactive T_N cells and/from ex-T_reg cells that lost Foxp3 expression to become inflammatory and presumably pathogenic (Bailey-Bucktrout et al, 2013). This is consistent with regulation of Fe metabolism modulating the incidence and severity of autoimmune conditions, as demonstrated for systemic lupus erythematosus (Gao et al, 2022a). Consistent with our findings, this was linked to modulation of T_reg (Gao et al, 2022a), T_H17 (Teh et al, 2021), and FT_H (Gao et al, 2022b) cells and was associated with regulation of DNA demethylation (Gao et al, 2022b; Teh et al, 2021). However, whether regulation of Fe metabolism in T_reg cells affects systemic lupus erythematosus was, to the best of our knowledge, not established (Gao et al, 2022a).

Fth deletion in T_reg cells increased susceptibility to malaria (Fig. 8E), consistent with dysregulation of Fe metabolism promoting malaria lethality (Ferreira et al, 2008; Ramos et al, 2022; Ramos et al, 2019; Wu et al, 2023). This was associated with an increase in host-parasite burden (Fig. 8E), in keeping with T_reg cells being essential to counter the pathogenesis of severe presentations of malaria while limiting immune-driven resistance mechanisms driving parasite clearance (Hisaeda et al, 2004; Kurup et al, 2017; Walther et al, 2005). We infer that the protective effect of T_reg cells against malaria acts via a mechanism that is not associated with a reduction of the host-pathogen burden, a defense strategy known as disease tolerance (Medzhitov et al, 2012; Soares et al, 2017). Presumably, the mechanism(s) via which FTH acts in T_reg cells to establish disease tolerance to malaria is multifactorial, restraining unfettered immune activation to prevent the pathogenesis of severe forms of malaria.

Dysregulation of Fe metabolism in Fth-deleted T_reg cells was associated with better control of tumor progression (Fig. 8H), consistent with a relative reduction of tumor-infiltrating T_reg cells (Fig. 8I) and a more pronounced activation and/or infiltration of activated T effector cells (Fig. 8J), including CD8⁺IFNγ⁺ T_C cells (Fig. EV5F). While the mechanism via which FTH expression in T_reg cells promotes tumor progression is not clear, these observations are consistent with the regulation of Fe metabolism in the tumor microenvironment impacting on tumor progression (Alaluf et al, 2020; Consonni et al, 2021).

In conclusion, regulation of intracellular Fe metabolism by FTH is essential to maintain T_reg cell lineage and function in vivo, reflecting how intracellular catalytic Fe controls the activity of TET dioxygenases that sustain FOXP3 transcription and support T_reg cell lineage identity. Moreover, FTH might regulate other iron-dependent mechanisms supporting T_reg function, for example, by enforcing the expression of c-Maf in T_reg cells (Zhu et al, 2023) that control immunological tolerance to the microbiota (Xu et al, 2018). We propose that targeting Fe metabolism pharmacologically maybe considered when manipulating T_reg cells for therapeutic purposes, either to enhance T_reg cell function in the context of immune-mediated inflammatory diseases or to dampen T_reg cell function as in the context of cancer therapies.

Methods

Reagents and tools table

Reagent/resource	Reference or source	Identifier or catalog number
Experimental models
Human: HEK293T	ATCC	ATCC® CRL-3216™
Mouse: Tumor cells B16-F10-luc2 (B16)	CaliperLS	B16-F10-luc2
Mouse: B6.C57BL/6 Fthfl/fl	Lukas Kuhn, ETH, Switzerland	(Darshan et al, 2009)
Mouse: B6129S-Tg(Foxp3 EGFP/icre)1aJbs/J backcrossed into B6.C57BL/6 background	Jackson Laboratory	JAX stock: 023161
Mouse: B6.Cg-Gt(ROSA)26Sortm9(CAG-tdTomato)Hze/J	Jackson Laboratory	JAX stock: 007909
Mouse: RAG2 -/- (B6.129S6-Rag2<tm1Fwa>N12)	Taconic	Taconic # RAGN12
Plasmodium chabaudi chabaudi strains: PcAS clone AJ4916	Reece & Thompson, 2008	N/A
Blood samples from anonymized healthy male donors were obtained in accordance with guidelines established by the Sanquin Medical Ethical Committee.	This paper	NA
Recombinant DNA
psPAX	Addgene	Cat#12260
pMD2.G	Addgene	Cat#12259
pCMV-FTH-3tag3a	This paper
pCMV-FTHmut-3tag3a	This paper
pCMV-3×FLAG-TET3(human)-Neo	miaolingBio	P45302
Antibodies
PE anti-human CD25 (Clone 2A3)	BD Biosciences	Cat#341011 (RRID: AB_2783790)
PE-Cy7 anti-human CD45RA (Clone HI100)	BD Biosciences	Cat#560675 (RRID: AB_1727498)
Brilliant Violet 421 Anti-human CD127 (Clone A019D5)	Biolegend	Cat#351309 (RRID: AB_10898326)
PE-Cy7 anti-Human FOXP3 (Clone 236A/E7)	eBioscience	Cat#25-4777-42 (RRID: AB_2573450)
Anti-Ferritin Heavy Chain	Abcam	ab65080 (RRID:AB_10564857)
CD45 APC-eFluor780	eBioscience	30-F11, 47-0451-82 (RRID:AB_1548781)
TCR-β BV421	BioLegend	H57-597, 109229 (RRID:AB_10933263)
TCR-β BV711	BioLegend	H57-597, 109243 (RRID:AB_2629564)
CD4 PE-Cy7	eBioscience	RM4-5, 25-0042-82 (RRID:AB_469578)
CD4 APC-eFluor780	eBioscience	GK1.5, 47-0041-82 (RRID:AB_11218896)
CD4 BV421	BioLegend	GK1.5, 100438 (RRID:AB_10900241)
CD8 PercpCy5.5	eBioscience	53-6.7, 45-0081-82 (RRID:AB_1107004)
CD8 APC/Fire 750	BioLegend	53-6.7, 100766 (RRID:AB_2572113)
CD44 eFluor450	eBioscience	IM7, 48-0441-82 (RRID:AB_1272246)
CD62L Pe-Cy7	eBioscience	MEL-14, 25-0621-82 (RRID:AB_469633)
CD304 (NRP1) PE	BioLegend	3E12, 145204 (RRID:AB_2561928)
PD1-PE	eBioscience	RMP1-30, 12-9981-82 (RRID:AB_466290)
CXCR5-biotin	BD Biosciences	2G8, 551960 RRID: AB_394301)
Alexa Fluor® 647 streptavidin	BioLegend	405237
CD25 PE-Cy7	BioLegend	PC61, 102016 (RRID:AB_312865)
Foxp3 PE	eBioscience	FJK-16s, 12-5773-82 (RRID:AB_465936)
Foxp3 FITC	eBioscience	FJK-16s, 11-5773-82 (RRID:AB_465243)
Foxp3 eF450	eBioscience	FJK-16s, 48-5773-82 (RRID:AB_1518812)
CD71	BioLegend	RI7217, 113811 (RRID:AB_2203383)
Ki67	eBioscience	SolA15, 50-5698-82 (RRID:AB_2896285)
IL-17A	eBioscience	eBio17B7, 50-7177-82 (RRID:AB_11220280)
IFN-gamma	eBioscience	XMG1.2,12-7311-81 (RRID:AB_466192)
CD45RA	BD Biosciences	HI100 (RRID: AB_1727498)
CD45.1 Pacific blue	Produced at IGC	A20
CD45.2 AF647	Produced at IGC	104.2
CD4 MicroBeads, human	Miltenyi Biotec	130-045-101 (RRID:AB_2889919)
Thy1.1	Produced at IGC	19E12
Thy1.2	Produced at IGC	30H12
anti-CD3 mAb (Clone 1XE)	Pelicluster	M1654 (RRID:AB_10553652)
anti-CD28 mAb (Clone CD28.2)	eBioscience	16-0289-85 (RRID:AB_468927)
HO-1	Enzo Life Sciences	ADI-SPA-896 (RRID:AB_10614948)
CD16/CD32	BioLegend	93, 101331
Oligonucleotides and other sequence-based reagents
Human FTH1 shRNA: FTH429; TRCN0000029429; target sequence: GCCTCGGGCTAATTTCCCATA	GPP Web Portal, Broad Institute	RHS3979-9596837
Human FTH1 shRNA: FTH432; TRCN0000029432; target sequence: CCTGTCCATGTCTTACTACTT	GPP Web Portal, Broad Institute	RHS3979-9596840
Primers for RT-qPCR, see Table 1	See Table 1	N/A
Chemicals, enzymes, and other reagents
TMRE-Mitochondrial Membrane Potential Assay Kit	Abcam	ab113852
FerroFarRed™	Goryo Chemical	GC903-01
LIVE/DEAD™ Fixable Aqua Stain	ThermoFisher Scientific	L34957
eBioscience™ Cell Stimulation Cocktail	ThermoFisher Scientific	00-4970-93
Protein Transport Inhibitor Cocktail	ThermoFisher Scientific	00-4980
CFSE	ThermoFisher Scientific	C34554
Solid Phase Reversible Immobilization (SPRI) beads	Beckman Coulter
Software Include version where applicable
ImageJ	Schneider et al, 2012	https://imagej.nih.gov/ij/
Flowjo	BD Sciences	Version 10.8.1
R	R Core Team 2014, Vienna, Austria	Version 3.5.1
FastQC method	Babraham bioinform.	Version 0.11.5
Python (Linux/UNIX)	Python	Version 2.7.12
Trimmomatic	Bolger et al, 2014	Version 0.36
HISAT2	Kim et al, 2015	Version 2.1.0
HTseq	Anders et al, 2015	Version 0.6.1p1
DESeq2	Love et al, 2014	Version 1.26.0
Other
QIAamp DNA Micro Kit	QIAGEN	56304
NucleoSpin RNA XS	Macherey-Nagel	740902
ChIP DNA Clean & Concentrator columns	Zymo Research	D5205
MagniSort™ Mouse CD4 T-cell Enrichment Kit	ThermoFisher Scientific	8804-6821-74
NEBNext® Enzymatic Methyl-seq Kit	New England Biolabs	E7120
Qubit HS dsDNA kit	ThermoFisher Scientific	Q32851
High Sensitivity DNA Bioanalyzer kit	Agilent	5067-4626
Seahorse XF Cell Mito Stress Test Kit	Agilent Technologies	103015-100

Animals

Mice were bred and maintained under specific pathogen-free (SPF) conditions at the Instituto Gulbenkian de Ciência (IGC). All experimental protocols were approved by the Ethics Committee of the IGC, the “Órgão Responsável pelo Bem-estar dos Animais” (ORBEA) (license numbers A001-2017, A003-2017) and the Portuguese National Entity (Direcção Geral de Alimentação e Veterinária) (notification numbers 003722, 008830). Experimental procedures were performed according to the Portuguese (Portaria no. 1005/92, Decreto-Lei no. 113/2013 and Decreto-lei no.1/2019) and European (Directive 2010/63/EU) legislations, concerning housing, husbandry, and animal welfare. Foxp3^{GFP-Fth∆/∆} mice were generated by crossing Foxp3^GFP (i.e., B6129S-Tg(Foxp3 EGFP/icre)1aJbs/J) mice (Chen et al, 2003; Zhou et al, 2008) with Fth^fl/fl mice (Darshan et al, 2009). Mouse progeny was genotyped for the presence of the Cre allele. The Foxp3^GFP mice express a humanized Cre-recombinase (GFP-hCre) from a Foxp3 ATG translational start codon, inserted in a bacterial artificial chromosome (BAC) transgene (Chen et al, 2003; Zhou et al, 2008). As Cre expression is not sex-dependent, this allows for Fth deletion in T_reg cells from male and female Foxp3^{GFP-Fth∆/∆} mice, using Foxp3^GFP and Fth^fl/fl mice as controls. Foxp3^{GFP-Fth∆/∆-tdT} mice were generated by crossing Foxp3^{GFP-Fth∆/∆} mice with Ai9 (RCL-tdT) mice. Control Foxp3^GFP-tdT mice were generated by crossing Foxp3^GFP mice with C57BL/6 Ai9 (RCL-tdT) mice, similar to described above. Progeny was genotyped for the presence of the humanized Cre-recombinase (Gfp-hCre) allele. The genetic background of the mouse strains used was C57BL/6J, including Foxp3^GFP mice, backcrossed into C57BL/6/J background for over 10 generations. Rag2^−/− mice used as recipients of BM precursor cells were in C57BL/6NTac background.

Experimental autoimmune encephalomyelitis (EAE)

C57BL/6 mice were immunized with the MOG_35–55 peptide (s.c.; 200 µg), emulsified in Complete Freund’s Adjuvant (CFA) containing Mycobacterium tuberculosis (4 mg/mL; BD Diagnostics). Mice received 200 ng of Pertussis toxin (i.v.; Sigma-Aldrich) at the time of immunization and 2 days thereafter. Clinical EAE severity scores were evaluated daily as follows: 0, normal; 1, limp tail; 2, partial paralysis of the hind limbs; 3, complete paralysis of the hind limbs; 4, hind-limb paralysis and forelimb weakness; 5, moribund or deceased, essentially as described (Chora et al, 2007).

Plasmodium infection (malaria)

Mice were infected by the inoculation of blood isolated from mice infected with a Plasmodium chabaudi chabaudi (Pcc) AS strain [i.p.; 2 × 10⁶ infected red blood cells (iRBC) per mouse]. Mice were monitored daily for parasitemia, weight, temperature, RBC number, and survival, essentially as described (Seixas et al, 2009).

Tumor model

Tumor cells B16-F10-luc2 (B16) (CaliperLS) were cultured at 37 °C in RPMI 1640 (Life Technologies) supplemented with 10% Fetal Bovine Serum (Biowest), 1% penicillin–streptomycin (Life Technologies), 50 µg/mL Gentamicin (Life Technologies), and 50 µM 2-Mercaptoethanol (Life Technologies). After trypsin (Life Technologies) treatment, single-cell suspensions were resuspended in ice cold calcium-free and magnesium-free HBSS (Life Technologies). Mice were injected subcutaneously in the right flank with 2 × 10⁵ B16 cells in 100 µL volume. Tumor size was measured with a caliper every 2 or 3 days, from day 8 post injection, and the tumor diameter (TD) was calculated as TD = (L + W)/2. Mice were sacrificed when TD ≥ 20 mm. By the end of each experiment, tumor clearance (TD ≤ 5 mm) was confirmed upon dissection.

Human peripheral blood mononuclear cells (PBMC)

Blood samples from anonymized healthy male donors were obtained in accordance with guidelines established by the Sanquin Medical Ethical Committee. Briefly, PBMC was isolated from fresh buffy coats using Ficoll-Paque Plus (GEHealthcare) gradient centrifugation. Next, CD4⁺ T cells were isolated using magnetic sorting with CD4 microbeads (Miltenyi Biotec) and viable cells were separated using flow cytometric sorting based on the expression of CD25, CD45RA, and CD127 on a FACS Aria III (BD Biosciences).

Mouse leukocyte isolation

For isolation of leukocytes, spleen and lymph were harvested, disrupted, passed through a cell strainer (70 μm) in PBS (3% FBS, 1 mM EDTA), pelleted (300 × g, 4 °C, 10 min), and RBC were lyzed (5 mL RBC lysis buffer; 5 min, RT). Lysis was stopped by adding 5 mL of medium, and cells were passed through a 40-μm cell strainer, centrifuged (300 × g, 4 °C, 10 min) and resuspended in PBS containing 3% FBS and 1 mM EDTA.

Cell sorting

Mice were sacrificed, and LN (i.e., inguinal, brachial, axillary, mandibular, superficial cervical, mesenteric, pancreatic, renal, and lumbar) or spleen were collected, and leukocytes isolated as described above. The negative fraction from CD4⁺ T cells enrichment (MagniSort™ Mouse CD4 T-cell Enrichment Kit) was recovered, centrifuged, and stained with the following antibody panel: anti-CD11b A647, anti-B220 A647, anti-CD8 A647, anti-CD4 PerCPCy5.5, anti-CD62L Pe-Cy7, and anti-CD44 eF450. Foxp3⁺ cells were sorted based on endogenous Foxp3^EGFP/icre expression (FACS Aria; BD Biosciences). When indicated, naive T cells (CD11b/B220/CD8^-CD4⁺Foxp3^-CD44^lowCD62L^high), memory/activated T cells (CD11b/B220/CD8^-CD4⁺Foxp3^-CD44 ^highCD62L^low) and T_reg cells (CD11b/B220/CD8^-CD4⁺Foxp3⁺) were sorted and recovered. A similar procedure was followed for sorting ex-T_reg cells, based on endogenous expression of Tomato within GFP⁺ (CD4⁺GFP⁺TdT⁺) and GFP^- (CD4⁺GFP⁻TdT⁺) cell populations.

Immunophenotyping

Cells isolated as described in “Leukocyte isolation” were stained for flow cytometry analysis. For surface staining, cells were incubated with Fc block together with LIVE/DEAD™ Fixable Aqua Stain in PBS, followed by incubation with antibodies against the following surface markers: CD8, CD4, CD62L, CD44, CD25, TCRβ, CD11b, and CD304 (Nrp1). Intracellular Foxp3 staining was performed using Foxp3/Transcription Factor Staining Buffer Set. Briefly, upon surface staining, cells were fixed, washed with permeabilization buffer, and incubated with anti-Foxp3 antibody in permeabilization buffer. For T_reg cells (CD4⁺GFP⁺tdT⁺) and ex-T_reg cells (CD4⁺ GFP^-tdT⁺) staining, cells were fixed and incubated with Fc block, followed by incubation with antibodies directed against the following surface markers: CD4, CD62L, CD44, CD3, TCRβ. T_reg cells and ex-T_reg cells were distinguished based on endogenous Foxp3^EGFP/icre and Tomato expression. Follicular T cells were fixed and incubated with Fc block, followed by incubation with antibodies against surface markers: CD4, CD3, TCRβ, CXCR5, and PD1. Foxp3⁺ cells were selected based on endogenous Foxp3^EGFP/icre expression. For T_reg cells lineage maintenance analysis comparing T_reg cells (CD4⁺GFP⁺tdT⁺) and ex-T_reg cells (CD4⁺ GFP^-tdT⁺), cells were incubated with Fc block together with LIVE/DEAD™ Fixable Aqua Stain in PBS, followed by incubation with antibodies against surface markers: CD4, CD3, CD71, and fixation with intracellular staining for the proliferation marker Ki67. Cell acquisition was performed using a CYTEK Aurora (Cytek Biosciences) flow cytometer, and data was analyzed using FlowJo software Version 10.8.1.

Cytokine staining

Cells were isolated as described in “Leukocyte isolation“ and were stimulated using Cell Stimulation Cocktail together with Protein Transport Inhibitor Cocktail (4 h; 37 °C) in complete RPMI (10% FBS, 100 U/mL Penicillin and 100 µg/mL Streptomycin). For surface staining, cells were incubated with Fc block together with LIVE/DEAD™ Fixable Aqua Stain, followed by incubation with antibodies directed against the following surface markers: CD8, CD4, and TCRβ. Intracellular Foxp3, IFNγ, and IL-17 staining were performed using Foxp3/Transcription Factor Staining Buffer Set. Briefly, were fixed upon surface staining cells, washed with permeabilization buffer, and incubated with anti-Foxp3, anti-IFNγ, and anti-IL-17 antibodies in permeabilization buffer. Cell acquisition was performed using BD LSRFortessa X-20 (BD Biosciences) flow cytometer. Alternatively, cells were incubated with Fc block together with LIVE/DEAD™ Fixable Yellow Stain, followed by incubation with antibodies against the following surface markers: CD8, CD4 and TCRβ. Intracellular Foxp3, IFNγ, IL-17 and IL10 staining were performed using Foxp3/Transcription Factor Staining Buffer Set. Briefly, upon surface staining cells were fixed, washed with permeabilization buffer, and incubated with anti-Foxp3, anti-IFNγ, anti-IL-17 and anti-IL-10 antibodies in permeabilization buffer. Cell acquisition was performed using a CYTEK Aurora (Cytek Biosciences) flow cytometer. Data were analyzed with FlowJo software Version 10.8.1.

Leukocyte staining with fluorescent probes

Cells were isolated as described in “Leukocytes isolation”. To evaluate mitochondrial membrane potential, cells were incubated with (tetramethylrhodamine, ethyl ester) TMRE-Mitochondrial Membrane Potential probe (20 nM) in RPMI without serum (20 min 37 °C). Control samples were pre-incubated with the ionophore uncoupler of oxidative phosphorylation FCCP (carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone; 20 μM, 10 min, 37 °C), to eliminate mitochondrial membrane potential and positive TMRE staining. To detect intracellular labile Fe²⁺, cells were incubated with FerroFarRed (5 μM; 1 h 37 °C) in RPMI without serum. After incubation with the fluorescent probes, cells were stained with Fc block together with LIVE/DEAD™ Fixable Aqua Stain, followed by incubation antibodies against the following surface markers: CD4, CD44, CD62L, and TCRβ. Foxp3⁺ cells were selected based on endogenous Foxp3^GFP expression. Cell acquisition was performed using BD LSRFortessa X-20 (BD Biosciences) flow cytometer and data were analyzed using FlowJo software Version 10.8.1.

Bone marrow transplants

Bone marrow (BM) cells were harvested by flushing the femurs and tibias of donor mice, and T cells were depleted by antibody-mediated complement killing. The rabbit complement was prepared fresh, via incubation on ice (30 min), centrifugation (300 × g, 10 min, 4 °C) and filtering (0.22 µm). BM cell suspensions (1 × 10⁷/mL in PBS) were incubated with an anti-Thy1.2 mAb (0.5 µg/mL) and mixed gently (every 15 min) with rabbit complement (LowTox-M, CL3051, CEDARLANE) at a ratio of 100 µL per mL of BM cell suspension (37 °C, 1 h). Complement activity was neutralized (FBS, 200 µL/mL), cell suspensions were filtered (70-μm mesh, cell strainer), washed (2x in PBS, 2% FBS and 1× in PBS without serum), and cell numbers adjusted (10⁸/mL) in PBS. BM cells from C57BL/6 CD45.1⁺ mice were mixed with BM cells from congenic C57BL/6 CD45.2⁺ Foxp3^{GFP-Fth∆/∆-tdT} or control Foxp3^GFP-tdT mice at a 1:1 ratio and injected (i.v.; tail vein, 200 µL) into congenic C57BL/6-recipient Rag2-deficient (Rag2^−/−) female mice, 2 h after irradiation (600 Gys). Hematopoietic chimerism was monitored by immunophenotyping, 6 weeks after bone marrow reconstitution and thereafter.

T_reg cell in vivo homeostatic expansion

LN (i.e., superficial, cervical, axillary, brachial, mesenteric, inguinal, lumbar, renal, caudal, and popliteal) and spleen were collected from Foxp3^GFP-tdT and Foxp3^{GFP-Fth∆/∆-tdT} mice and gently disrupted in 70-μm mesh tissue to isolate leukocytes. Cell suspensions were washed in cold PBS, red blood cells were lyzed (i.e., ammonium chloride), passed through 40-μm cell strainer and CD4⁺GFP⁺TdT⁺ T_reg cells were sorted upon surface marker staining, as described in “Cell sorting”. To test T_reg cell stability in vivo, CD4⁺GFP⁺TdT⁺ T_reg cells from Foxp3^GFP-tdT and Foxp3^{GFP-Fth∆/∆-tdT} mice were adoptively transferred (i.v., 1 × 10⁵ cells) to Rag2^-/- mice. LN and spleen were collected six weeks later, disrupted, and RBC was lysed. The number of CD4⁺GFP⁺TdT⁺ T_reg cells and CD4⁺GFP^-TdT⁺ ex-T_reg cells was quantified by flow cytometry, as described in “Immunophenotyping”. An additional aliquot of spleen cells was used to sort CD4⁺GFP⁺TdT⁺ T_reg cells, as described in “Cell sorting”. These were used to extract mRNA and monitor Fth mRNA expression in the CD4⁺GFP⁺TdT⁺ T_reg cells used for the adoptive transfer.

T_reg cell proliferation suppression assay

Naive T cells were sorted, as described in “Cell sorting”, washed with PBS, and incubated with Cell Tracer Violet (CTV) (Thermofisher; 1/1000 in PBS without serum) at RT in the dark for 15 min. Staining was stopped by adding five volumes of complete media (containing 10% FBS). Sort-purified T_reg cells were plated and serially twofold diluted, starting at 2.5 × 10⁴ cells/well in round-bottom 96-well plates with 1 µg/mL soluble anti-CD3 mAb and 5 × 10⁴ irradiated splenocytes. CTV-labeled T_N cells were plated (2.5 × 10⁴ cells/well), resulting in T_reg:T_NAIVE ratio ranging from 1:1 to 1:64. Cultures were set in triplicates in a final volume of 200 µL. On day 3 of culture, CTV intensity was measured in responder T cells defined as Thy1.1⁺ Thy1.2^-TCRb⁺CD4⁺, live lymphocytes.

In vitro induction of T_reg cells and flow cytometry analysis

Naive T cells sorted, as described in “Cell sorting”, were cultured for 5 days on a maxisorb 96-well plate pre-coated with anti-CD3 mAb (1 µg/mL; 100 µL/well in PBS) in RPMI complete media (10% FBS, 100 U/mL Penicillin and 100 µg/mL Streptomycin), supplemented with anti-CD28 mAb (1 µg/mL), mouse recombinant IL-2 (20 ng/mL) and TGFβ (5 ng/mL) (iT_reg cell differentiation medium). Alternatively, naive T cells were cultured in the same media, without TGFβ (conventional T-cell differentiation medium). For cell surface staining, cells were incubated with Fc block together with LIVE/DEAD™ Fixable Aqua Stain, followed by incubation with anti-CD4 antibody. For intracellular Foxp3 staining cells were fixed after surface staining, washed with permeabilization buffer, and incubated with anti-Foxp3 antibody in permeabilization buffer, according to the Foxp3/Transcription Factor Staining Buffer Set. Cells were acquired in a BD LSRFortessa X-20 (BD Biosciences) flow cytometer and analyzed using FlowJo software Version 10.8.1. For analysis at day 12 after T_reg induction, induced iT_reg cells were re-plated at day 5 in RPMI media supplemented with IL-2 (100 ng/mL) with or without Fe sulfate (20 µM) and cultured for 7 days. To determine cell proliferation, naive T cells (5 × 10⁶/mL) were stained with CFSE (5 µM, 20 min RT), washed with complete medium to stop the reaction and re-cultured in T_reg or conventional T cells medium.

Lentiviral transduction

Lentivirus was produced by transfecting confluent human HEK293T cells with packaging (psPAX2) and envelope plasmids (pMD2.G) with pLKO.1. HEK293T cells were cultured in DMEM with HEPES (Life Technologies) supplemented with 10% fetal calf serum and 1% penicillin/streptomycin. Polyethylenimine (Polysciences, Hirschberg an der Bergstrasse, Germany) was used as a transfection reagent. After 24 h, the cultures were refreshed with medium with 2% FCS and 24 h later, lentiviral particles were concentrated and purified by ultracentrifugation at 50,000 × g, 2.5 h, 8 °C. Naive T_conv (CD4⁺, CD127⁺, CD25⁻, CD45RA⁺) and naive T_reg (CD4⁺, CD127^-, CD25⁺, CD45RA⁺) cells were isolated by FACS sorting (FACS Aria III, BD Biosciences) as described above. The cells were then cultured in presence of 0.1 µg/mL of anti-CD3 mAb (M1654, clone 1XE, PeliCluster) and anti-CD28 mAb (16-0289-85, clone CD28.2, eBioscience) for 5 days in IMDM containing 10% FCS and 300 U/mL IL-2 and restimulated one day prior to transduction. Cells were then infected in RetroNectin® (Clontech) coated plates for 24 h. After that, the cells were transferred into tissue culture-treated plates with medium containing 100 U/mL IL-2 and puromycin. After 5 days, the cells were directly used for FACS analysis or lysed for western blot assays.

Western blot

Human T conventional (CD4⁺CD45RA⁺CD127⁺CD25^–), T_reg (CD4⁺CD127^–CD45RA⁺CD25^hi) cells or HEK293T were washed (2× in PBS) and directly lysed in RIPA buffer. Cell lysates containing equal amounts of protein were boiled in a sample buffer prior to gel electrophoresis. SDS-PAGE gel electrophoresis was performed using the NuPAGE electrophoresis system (Novex, Life Technologies). Proteins were transferred using the iBlot system (Thermo Scientific) and analyzed using the corresponding antibodies. ECL signals on Western blots were developed using the Pierce ECL substrate kit (Pierce) followed by autoradiographic detection on film (Fuji Medical). Sorted cells mouse T_reg, T_M, and T_N cells were directly lysed in 2× SDS-page sample buffer (20% glycerol, 4% SDS, 100 mM Tris pH 6.8, 0.002% bromophenol blue, 100 mM dithiothreitol). Samples were then sonicated or treated with Benzonase to degrade DNA, heated (10 min; 70 °C), and centrifuged. The supernatant was collected, and the protein was quantified using NanoDrop™ 1000. Anti-FTH1 (1:1000), anti-Histone H3 (1:1000) were detected using peroxidase-conjugated secondary antibodies (1 h; RT) and developed with SuperSignal™ West Pico PLUS Chemiluminescent Substrate (ThermoFisher Scientific). ECL signal was developed using Pierce ECL substrate kit followed by autoradiographic detection on film (Fuji Medical). Alternatively, western blots were developed using Amersham Imager 680 (GEHealthcare), equipped with a Peltier-cooled Fujifilm Super CCD. Densitometry analysis was performed with ImageJ using images without saturated pixels.

qRT-PCR

RNA was isolated from cells using NucleoSpin RNA XS kit (Macherey-Nagel) according to the manufacturer’s instructions. cDNA was transcribed from total RNA with transcriptor first strand cDNA synthesis kit (Roche) or Xpert cDNA Synthesis Kit (GRiSP). Quantitative real-time PCR (qRT-PCR) was performed using 1 μg cDNA and SYBR Green Master Mix (Applied Biosystems, Foster City, CA, USA) in duplicate on a 7500 Fast Real-Time PCR System (Applied Biosystems) under the following conditions: 95 °C/10 min, 40 cycles/95 °C/15 s, annealing at 60 °C/30 s, and elongation 72 °C/30 s. Primers listed in Table 1 were designed using Primer Blast (Ye et al, 2012).

Table 1.

RT-qPCR primers.

Oligonucleotides	Sequences	Reference
Arbp0 Fwd	5′-CTTTGGGCATCACCACGAA-3′	Blankenhaus et al, 2019
Arbp0 Rev	5′-GCTGGCTCCCACCTTGTCT-3′	Blankenhaus et al, 2019
Fth Fwd	5′-CCATCAACCGCCAGATCAAC-3′	Blankenhaus et al, 2019
Fth Rev	5′-GCCACATCATCTCGGTCAAA-3′	Blankenhaus et al, 2019
Ftl Fwd	5’-AAGATGGGCAACCATCTGAC-3’	This work
Ftl Rev	5’-GCCTCCTAGTCGTGCTTGAG-3’	This work
Hk2 Fwd	5′-GCCAGCCTCTCCTGATTTTAGTGT-3′	Quiros et al, 2017
Hk2 Rev	5′-GGGAACACAAAAGACCTCTTCTGG-3′	Quiros et al, 2017
Nd1 Fwd	5′-CTAGCAGAAACAAACCGGGC-3′	Quiros et al, 2017
Nd1 Rev	5′-CCGGCTGCGTATTCTACGTT-3′	Quiros et al, 2017
GFP Fwd	5′-CGACGTAAACGGCCACAAGTTCAG-3′	Liu et al, 2012
GFP Rev	5′-CCGTAGGTCAGGGTGGTCACGAG-3′	Liu et al, 2012
Tdt Fwd	5′-GCCGACATCCCCGATTACAAGA-3′	Wienert et al, 2015
Tdt Rev	5′-CGATGGTGTAGTCCTCGTTGTGG-3′	Wienert et al, 2015
Foxp3 Fwd	5′-GGCCCTTCTCCAGGACAGA-3′	Fontenot et al, 2003
Foxp3 Rev	5′-GCTGATCATGGCTGGGTTGT-3′	Fontenot et al, 2003

Serology

Mice were euthanized using CO₂ inhalation. Whole blood was collected by cardiac puncture and transferred into an EDTA for hemogram analyses or heparin tubes for serology (Iron, Transferrin, and Transferrin saturation). Analysis was performed by DNAtech (Lisbon).

Histology

Organs were harvested, fixed (10% formalin), embedded in paraffin, sectioned (3-μm-thick sections), and stained with Hematoxylin and Eosin (H&E). Whole sections were analyzed, and images acquired with a Leica DMLB2 microscope (Leica) and NanoZoomer-SQ Digital slide scanner (Hamamatsu).

mtDNA/nDNA qPCR

Total isolated DNA was used to perform the quantification of mitochondrial DNA (mtDNA) in comparison to nuclear DNA (nDNA) using a qRT-PCR-based method (Quiros et al, 2017). Briefly, qRT-PCR was performed form using 20 ng of DNA and SYBR Green Master Mix (Bio-Rad), in duplicate on a 7500 Fast Real-Time PCR System (Applied Biosystems), under the following conditions: 50 °C/2 min and 95 °C/5 min (Hold stage), 45 cycles/95 °C/10 s, annealing at 60 °C/30 s, and elongation 72 °C/20 s, followed by melting curve: 95 °C/15 s, 60 °C/1 min, and gradual increase in temperature up to 95 °C. Primers for NADH-ubiquinone oxidoreductase chain 1 encoded by the mitochondrial gene MT-Nd1 (Nd1) and for the nuclear-encoded hexokinase 2 gene (Hk2) (Quiros et al, 2017) are listed in Table 1. Mitochondria number per cell was calculated according to the ratio of mRNA expression of the single copy mitochondrial gene Nd1 and the single copy nuclear gene Hk2.

Seahorse assays

Oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were measured using a Seahorse XFe96 analyzer (Agilent Tech.) and the Seahorse XF Cell Mito Stress Test Kit according to instructions from the manufacturer. Specifically sorted TREG and TCONV cells were plated on poly-L lysine-coated Seahorse XF96 plates (15 × 10³ cells/well) XF medium (10 mM glucose, 1 mM sodium pyruvate, 2 mM l-glutamine, pH 7.4), and centrifuged (400 × g, for 5 min) to promote cell adhesion. Cells were incubated in a non-CO₂ incubator (37 °C; 1 h) prior to the assay. The analyzer was programmed to calibrate and equalize samples, followed by three baseline measurements (3 min each) and mixing (2 min) between measurements prior to inhibitor injection. The inhibitors were injected in the following order: Oligomycin (1 μM); FCCP (2 μM); Antimycin A/Rotenone (1 μM); and three measurements (3 min each) were made following each injection with 2 min of mixing between measurements.

RNA sequencing

RNA was extracted from (CD4 + GFP + ) T_REG cells sorted from the LN of Foxp3^{GFP-Fth∆/∆} vs. control Foxp3^GFP mice (see Fig. 2) or from (CD45.2 + CD3 + CD4 + GFP+tdT + ) T_REG and (CD45.2 + CD3 + CD4 + GFP-tdT + ) ex-T_REG cells sorted from LN of BM chimeric mice (see Fig. 4). RNA sequencing data from non-chimeric mice was analyzed as follows: RNA sequencing was performed using Illumina’s next-generation sequencing (Bentley et al, 2008). Quality check and quantification of total RNA was done using the Agilent Bioanalyzer 2100 in combination with the RNA 6000 pico kit (Agilent Technologies). Library preparation was done using SMARTer Stranded Total RNA-Seq Kit v2-Pico Input Mammalian (Takara) following the manufacturer’s description. Library quantification and quality check was done using the Agilent Bioanalyzer 2100 in combination with the DNA 7500 kit. Libraries were sequenced on a HiSeq2500 running in 51cycle/single-end/rapid mode. All libraries were pooled and sequenced in two lanes. Sequence information was extracted in FastQ format using Illumina’s bcl2fastq v.2.19.1.403. Sequencing resulted in around 29mio reads per sample. RNA sequencing data from cells extracted from mixed BM chimeric mice was analyzed as follows: RNA was extracted and quality assessed using Agilent Bioanalyzer 2100 using the RNA 6000 pico kit (Agilent Technologies). Full-length cDNAs and sequencing libraries were generated following the SMART-Seq2 protocol⁶². Library preparation, including cDNA “tagmentation”, PCR-mediated adaptor addition and library amplification was performed using the Nextera library preparation protocol (Nextera XT DNA Library Preparation kit, Illumina). Libraries were sequenced (NextSeq 500, Illumina) using High Output kit v2.5 (75 cycles). Sequencing data was extracted in FastQ format, using Illumina’s bcl2fastq v.2.19.1.403, producing 30.14 × 10⁶ reads per sample on average. Library preparation and next-generation sequencing were performed at the IGC Genomics Unit. Fastq reads were aligned against the mouse reference genome GRCm39 using the GENCODE vM27 annotation to extract splice junction information (STAR; v.2.5.2a)⁶⁴. Read summarization was performed by assigning uniquely mapped reads to genomic features using FeatureCounts (subread package v.1.5.0-p1). Gene expression tables were imported into the R environment (v.4.1.0) to perform differential gene expression, functional enrichment analyses, and data visualization (R Core-Team, 2021). Differential gene expression analysis was performed using the DESeq2 R package (v.1.32). Gene expression was modeled by genotype. Genes not expressed or with fewer than 10 counts across the samples were removed. We subsequently ran the function DESeq to estimate the size factors (by estimateSizeFactors), dispersion (by estimateDispersions) and fit a binomial GLM fitting for βi coefficient and Wald statistics (by nbinomWaldTest). Pairwise comparisons were performed with the function results (alpha = 0.05), and the log₂ fold change for each pairwise comparison was shrunken with the function lfcShrink using the algorithm ashr (v.2.2-47)⁶⁵. Differentially expressed genes were considered as genes with an adjusted P value < 0.05 and an absolute log₂ fold change >0. Normalized gene expression counts were obtained with the function counts using the option normalized = TRUE. Regularized log-transformed gene expression counts were obtained with rlog, using the option blind = TRUE. Ensembl gene ids were converted into gene symbols from Ensembl (v.104 - May 2021—https://may2021.archive.ensembl.org) by using the mouse reference (GRCm39) database with biomaRt R package (v.2.48.2). All plots were created using the ggplot2 R package (v.3.3.5). Heatmaps were created with pHeatmap (v.1.0.12), using Euclidean distance and Ward.D2 methods for clustering estimation. For hierarchical clustering, gene expression counts were scaled (Z-score) with the function scale. Functional enrichment analysis was performed using the gprofiler2 R package (v.0.2.1). Enrichment was performed with the function gost based on the list of up- or downregulated genes, between each pairwise comparison, against annotated genes (domain_scope = “annotated”) of the organism Mus musculus (organism = “mmusculus”). Gene lists were sorted based on adjusted p value (ordered_query = TRUE) to generate GSEA (Gene Set Enrichment Analysis) style p values. Only statistically significant (user_threshold=0.05) enriched functions are returned (significant=TRUE) after multiple testing corrections with the default method g:SCS (correction_method = “analytical”). Gprofiler2 queries were run against the default functional databases for mouse which include Gene Ontology (GO:MF, GO:BP, GO:CC), KEGG (KEGG), Reactome (REAC), TRANSFAC (TF), miRTarBase (MIRNA), Human phenotype ontology (HP), WikiPathways (WP), and CORUM (CORUM). Gprofiler2 was performed using database versions Ensembl 104, and Ensembl gene 51 (database updated on 07/05/2021).

RNA-sequencing data analysis

Sequence read quality was assessed by means of the FastQC method (v0.11.5; http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Trimmomatic version 0.36 was used to trim Illumina adapters and poor-quality bases (trimmomatic parameters: leading=3, trailing=3, sliding window=4:15, minimum length=40). The remaining high-quality reads were used to align against the Genome Reference Consortium mouse genome build 38 (GRCm38). Mapping was performed by HISAT2 version 2.1.0 with parameters as default. Count data were generated by means of the HTSeq method and analyzed using the DESeq2 method in the R statistical computing environment (R Core Team 2014. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria). Statistically significant differences were defined by Benjamini & Hochberg adjusted probabilities <0.05. Canonical signaling pathways and biofunctions were generated by Ingenuity Pathway Analysis (IPA; QIAGEN) specifying mouse species and ingenuity database as reference. Benjamini & Hochberg adjusted probabilities <0.05 demarcated significance. Functional enrichment analysis was performed with gProfiler (Kolberg et al, 2023). Data were analyzed using g:SCS multiple testing correction method with a significance threshold of 0.05.

Targeted metabolomics

Targeted metabolomics analyses of cell extracts for intermediates of the tricarboxylic acid cycle (TCA cycle; citrate, isocitrate, α-ketoglutarate, succinate, fumarate, malate, cis-aconitate) and additional metabolites closely linked to the TCA cycle (pyruvate, lactate, DL-2-hydroxyglutarate, itaconate, and the amino acids aspartate, glutamate, glutamine) were performed by liquid chromatography-tandem mass spectrometry (LC-MS/MS), as described elsewhere (Richter et al, 2019). For metabolite quantification, ratios of analyte peak areas to peak areas of respective stable isotope labeled internal standards determined in cell extracts were compared to those in calibrators.

Genome-wide methyl-sequencing (EM-seq)

The initial genomic DNA (gDNA) was quantified using the Invitrogen Qubit 4 as per the manufacturer’s protocol (1 µL in 199 µL of Qubit working solution). For library preparation, NEB enzymatic Methy-Seq kit was used, following the manufacturer’s protocol for large insert libraries. Based on the quality assessment, samples were standardized to 10 ng of DNA in a volume of 50 µL, including a spike-in of pUC and Lambda DNA provided with the kit, as a control for methylation efficiency as per the manufacturer’s protocol. Samples were fragmented using the Covaris S2 system, with settings to achieve an average fragment size of 350-400 bp and individual barcoded during the PCR, using eight PCR cycles as per the manufacturer’s protocol. Libraries were quantified using the Qubit HS DNA assay as per the manufacturer’s protocol (1 µL of sample in 199 µL of Qubit working solution). The quality and molarity of the libraries were assessed using Agilent Bioanalyzer with the DNA HS Assay kit as per the manufacturer’s protocol. Molarity was used to equimolarly combine individual libraries into one pool, loaded and sequenced on an Illumina NextSeq 2000 platform (Illumina, San Diego, CA, USA) using a P3 300 cycle kit and a read-length of 155 bp paired-end reads. Raw sequencing reads were deposited to ENA under the accession number: Sequencing reads were processed by the methylseq (v1.6.1) (https://zenodo.org/record/2555454/export/xd) nf-core (Ewels et al, 2020). Default parameters were used with BWA-meth as the aligner, GRCm38 as the reference genome. The --em-seq and --methyl_kit options were also used for downstream analysis. R methylkit package (v1.22.0) (Akalin et al, 2012) was used to load the methylation calls, calculate the methylation rate, and generate plots presented in this paper. Sliding Linear Model (SLIM) multiple testing correction (Wang et al, 2011) was used to extract the significant methylated sites.

TET activity assay

TET enzymatic activity was monitored in human HEK293T cells, transiently transfected with human TET3 (pCMV-3×FLAG-TET3(human)-Neo; MiaolingBio P45302) plus human FTH (pCMV-FTH-3Tag3a) or FTH mutant (FTH^mut; pCMV-FTH^mut-3Tag3a; Glu 62 and His 65 and Lys86 were mutated as Lys, Gly, Gln, respectively, to achieve fully ablation of ferroxidase activity) lacking ferroxidase mutant FTH (FTH^mut), similar to previously described (Broxmeyer et al, 1991), followed by nuclear protein extraction and ELISA-based TET activity measurement. Briefly, HEK293T cells, at 60% confluency in six-well plates, were transiently transfected with the human TET3 (800 ng/well) together with FTH (200 ng/well), FTH^mut (200 ng/well) cDNA expressing vectors or empty vector (pCMV-3Tag3a; 200 ng/well) using transfection reagent (YEASEN, 40802ES03). Nuclear proteins were extracted 48 h after transfection, using a commercial kit (EpiQuik, OP-0002-1) and quantified (BCA assay; Meilunbio MA0082-2). TET activity was measured by ELISA-based the manufacturer’s instructions (EpiQuik, P-3087). Expression of flag-tagged TET3, FTH and FTH^mut was monitored by western blot. Briefly, cytosol and nuclear extracts were loaded on a 4–20% gradient SDS-PAGE precast-Gel (ACE Biotechnology, ET15420LGel) and proteins were transferred on PVDF membranes. These were blocked (5% skimmed milk in TBST; 1 h) and incubated (4 °C; overnight) with an HRP-labeled anti-FLAG-HRP antibody (Sigma, A8592). Membranes were washed (3 × 10 min; TBST) and HRP activity was detected by ECL (Thermo Scientific, 32106). Images were capture by Molecular Imager^® ChemiDoc^TM XRS+ with Image Lab^TM Software (BIO-RAD). Membranes were further incubated (1.5 h, RT) with anti-Lamin A/C (Proteintech, 10298-1-AP) and anti-GAPDH (Abclonal, AC033) antibodies, used as reference cytosolic and nuclear proteins, respectively. Membranes were washed (3 × 10 min; TBST), incubated with goat anti-rabbit (Abclonal, AS014) and goat anti-mouse (Abclonal, AS003), secondary antibodies, respectively, washed (3 × 10min in TBST), and HRP signal was detected by ECL to blot references proteins for cytosol and nuclear fractions, respectively, as above.

Quantification and statistical analysis

Statistical analysis was conducted using GraphPad Prism 9 software. All distributed data are displayed as means ± standard deviation of the mean (SD) unless otherwise noted. Measurements between two groups were performed with unpaired t test with Welch’s correction, paired t test, or Mann–Whitney test. Groups of three or more were analyzed by one-way or two-way analysis of variance (ANOVA). Survival was assessed using a log-rank (Mantel–Cox) test. Statistical parameters for each experiment can be found within the corresponding figure legends.

Author contributions

Qian Wu: Conceptualization; Resources; Data curation; Formal analysis; Funding acquisition; Validation; Investigation; Methodology; Writing—review and editing. Ana Rita Carlos: Conceptualization; Resources; Data curation; Formal analysis; Funding acquisition; Validation; Investigation; Methodology; Writing—review and editing. Faouzi Braza: Conceptualization; Data curation; Formal analysis; Investigation; Visualization; Methodology. Marie-Louise Bergman: Formal analysis; Investigation; Methodology. Jamil Z Kitoko: Formal analysis; Methodology. Patricia Bastos-Amador: Conceptualization; Formal analysis; Investigation; Methodology. Eloy Cuadrado: Formal analysis; Investigation; Methodology. Rui Martins: Software; Formal analysis; Investigation; Visualization; Methodology. Bruna Sabino Oliveira: Data curation; Investigation; Methodology. Vera C Martins: Data curation; Formal analysis; Investigation; Methodology; Writing—review and editing. Brendon P Scicluna: Software; Formal analysis; Visualization; Methodology. Jonathan JM Landry: Software; Formal analysis; Visualization; Methodology. Ferris E Jung: Software; Formal analysis; Investigation. Temitope W Ademolue: Data curation; Investigation; Methodology. Mirko Peitzsch: Software; Formal analysis; Investigation; Methodology. Jose Almeida-Santos: Data curation; Investigation; Methodology. Jessica Thompson: Formal analysis; Investigation; Methodology. Silvia Cardoso: Resources; Methodology. Pedro Ventura: Data curation; Investigation; Methodology. Manon Slot: Data curation; Investigation; Methodology. Stamatia Rontogianni: Data curation; Investigation; Methodology. Vanessa Ribeiro: Data curation; Investigation; Methodology. Vital Da Silva Domingues: Data curation; Investigation; Methodology. Inês A Cabral: Data curation; Investigation; Methodology. Sebastian Weis: Data curation; Software; Investigation; Methodology. Marco Groth: Software; Formal analysis; Investigation; Methodology. Cristina Ameneiro: Data curation; Investigation; Methodology. Miguel Fidalgo: Data curation; Formal analysis; Investigation; Methodology. Fudi Wang: Resources. Jocelyne Demengeot: Conceptualization; Writing—review and editing. Derk Amsen: Data curation; Formal analysis; Investigation; Methodology. Miguel P Soares: Conceptualization; Resources; Data curation; Formal analysis; Supervision; Funding acquisition; Validation; Visualization; Writing—original draft; Project administration; Writing—review and editing.

Data availability

Data from RNA sequencing studies are available at GEO database GSE173181 (Fig. 3B) and GSE226032 (Fig. 5I,J). The methylation data for this study have been deposited in the European Nucleotide Archive (ENA) at EMBL-EBI under accession number PRJEB59884 (Fig. 6).

Disclosure and competing interests statement

MPS is a consultant to the New York Blood Center (NYBC), NYC, USA. The remaining authors declare no competing interests.

Supplementary information

Expanded view data, supplementary information, appendices are available for this paper at 10.1038/s44318-024-00064-x.