wu-et-al-2024-ferritin-heavy-chain-supports-stability-and-function-of-the-regulatory-t-cell-lineage (1)-html.html

; Dikiy and Rudensky,

; Josefowicz et al,

; Panduro et al,

). 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,

; 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,

), T

REG

cells are pathogenic, for example, when

limiting immune-mediated in

ﬂ

ammatory responses to pathogens to

promote chronic infections (Belkaid,

; Demengeot et al,

2006

)

or when restraining anti-tumor immunity, to promote cancer
progression (Curiel et al,

; Liu et al,

REG

cell development and function are controlled by the X-

chromosome-encoded transcription factor FOXP3 (Fontenot et al,

; Hori et al,

mpsoares@igc.gulbenkian.pt

), together with auxiliary transcriptional

Instituto Gulbenkian de Ciência, Oeiras, Portugal.

International Institutes of Medicine, the Fourth Af

ﬁ

liated Hospital of Zhejiang University, School of Medicine, Yiwu, Zhejiang,

China.

Departamento de Biologia Animal, Centro de Ecologia, Evolução e Alterações Ambientais, Faculdade de Ciências, Universidade de Lisboa, Lisboa, Portugal.

Department

of Hematopoiesis and Department of Immunopathology, Sanquin Research and Landsteiner Laboratory, Amsterdam, The Netherlands.

Department of Applied Biomedical

Science, Faculty of Health Sciences, Mater Dei Hospital, and Centre for Molecular Medicine and Biobanking, University of Malta, Msida, Malta.

Genomic Core Facility, European

Molecular Biology Laboratory, Heidelberg, Germany.

Institute for Clinical Chemistry and Laboratory Medicine, University Clinic Carl Gustav Carus, TU Dresden, Dresden,

Germany.

Department for Anesthesiology and Intensive Care Medicine, Jena University Hospital, Friedrich-Schiller University, Jena, Germany.

Institute for Infectious Disease

and Infection Control, Jena University Hospital, Friedrich-Schiller University, Jena, Germany.

Leibniz Institute for Natural Product Research and Infection Biology, Hans-Knöll

Institute-HKI, Jena, Germany.

Leibniz Institute on Aging-Fritz Lipmann Institute, Jena, Germany.

Center for Research in Molecular Medicine and Chronic Diseases (CiMUS),

Universidade de Santiago de Compostela-Health Research Institute (IDIS), Santiago de Compostela, Spain.

The Second Af

ﬁ

liated Hospital, School of Public Health, Zhejiang

University School of Medicine, Hangzhou 310058, China.

Department of Experimental Immunology, Amsterdam UMC, University of Amsterdam, Amsterdam, The Netherlands.

These authors contributed equally: Qian Wu, Ana Rita Carlos, Faouzi Braza.

✉

E-mail:

1234567

890();,:

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

1445

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

regulators (Kanamori et al,

). The transcriptional program

enforced by FOXP3 speci

ﬁ

es T

REG

cell lineage commitment in the

thymus and in the periphery (Fontenot et al,

; Hori et al,

Lee et al,

), generating thymic T

REG

(tT

REG

) cells and

peripherally derived T

REG

(pT

REG

) cells, respectively (Chen et al,

). Sustained

FOXP3

transcription maintains T

REG

cell lineage

stability (Williams and Rudensky,

), avoiding transdifferentia-

tion towards pro-in

ﬂ

ammatory T helper (T

) cells (Gavin et al,

; Morikawa et al,

FOXP3

transcription is regulated by different signal transduc-

tion pathways, emanating from the T-cell receptor (TCR),
interleukin (IL-2) receptor, and TGF-

receptor (Bennett et al,

2001

; Brunkow et al,

2001

; Hori and Sakaguchi,

), among

others. Sustained

FOXP3

transcription is enforced epigenetically

(Gavin et al,

; Morikawa et al,

), in response to

environmental cues (Chapman et al,

; Shi and Chi,

) that

regulate different aspects of T

REG

cell metabolism (Etchegaray and

Mostoslavsky,

). These epigenetic modi

ﬁ

cations 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,

; Zheng et al,

2010

), whereby

cytosine methylation represses while demethylation sustains

FOXP3

transcription (Ohkura et al,

; Zheng et al,

2010

Cytosine methylation is catalyzed by DNA methyltransferase

(DNMT) (Ohkura et al,

), while demethylation is catalyzed by

the ten

–

eleven translocation (TET) family of dioxygenases (Wu and

Zhang,

; Yue et al,

). 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,

). TET dioxy-

genases catalyze cytosine demethylation at

FOXP3

CNS1 and 2

(Ohkura et al,

), supporting T

REG

cell lineage stability

(Nakatsukasa et al,

; Wu and Zhang,

; Yue et al,

Yue et al,

), and preventing T

REG

cell transdifferentiating into

ﬂ

ammatory effector T

cells, also referred as ex-T

REG

cells (Duarte

et al,

; Komatsu et al,

; Zhou et al,

TET dioxygenases use Fe as an essential cofactor and the

intermediate metabolite

-ketoglutarate as an obligatory substrate

(Huang and Rao,

; Pastor et al,

). 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,

), via a cytoprotective mechanism exerted by the ferritin

H chain (FTH) (Berberat et al,

; Pham et al,

), likely

involving the mitochondria (Blankenhaus et al,

; Vanoaica

et al,

). Regulation of cellular Fe content and redox activity also

modulate cytokine production by effector T

cells, via a mechanism

involving the PolyC-RNA-Binding Protein 1 (PCBP1) (Wang et al,

). Cellular Fe import, via the transferrin receptor 1 (

TFR1/

CD71

), supports T

type 1 (T

1) cell immunity and its regulation

by induced T

REG

(iT

REG

) cells (Voss et al,

) as well as antibody

responses to vaccination (Frost et al,

; Jiang et al,

) and

immunity against infection by pathogens such as

Plasmodium

, the

causative agent of malaria (Wideman et al,

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 in

ﬂ

ammation.

Results

REG

cells express relatively high levels of FTH

In a previously unbiased proteomics analysis, we found that freshly
isolated human naive CD45RA

CD25

REG

(nT

REG

) and memory

CD45RA

CD25

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,

). The relatively higher expression of the FTH and FTL

components of the ferritin complex was maintained upon expansion of
human CD4

CD127

–

CD25

REG

cells in vitro, in comparison to T

CONV

cells CD4

CD127

CD25

–

cells (Fig.

A,B). Similarly, mouse

CD4

Foxp3

REG

cells also expressed relatively higher levels of FTH

protein, compared to CD4

Foxp3

−

CD44

low

CD62L

high

naive T

cells

) or CD4

Foxp3

−

CD44

high

CD62L

low

memory T

cells (T

), as

determined by western blot (Fig.

C,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

cells, while

(CD4

Foxp3

−

CD44

high

CD62L

−

) T

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,

; Muckenthaler et al,

; Rouault et al,

1988

We monitored FTH expression in induced T

REG

(iT

REG

) cells,

generated from mouse T

cells activated in vitro with anti-CD3/

CD28 mAb plus IL-2 and TGF

(Chen et al,

) (Fig.

E). To this

aim we used

Foxp3

GFP

REG

cell reporter mice, in which a green

ﬂ

uorescent protein (GFP) humanized Cre-recombinase (GFP-

hCre) coding sequence is inserted downstream of the

Foxp3

ATG

translational start codon in a bacterial arti

ﬁ

cial chromosome (BAC)

transgene carrying the intact

Foxp3

promoter (Foxp3-GFP-hCre;

referred herein as

Foxp3

GFP

) (Chen et al,

; Zhou et al,

FTH protein expression was higher under culture conditions
containing TGF

and enriched in CD4

GFP

REG

cells, compared

to culture conditions lacking TGF

and enriched in CD4

GFP

CONV

cells (Fig.

F). A similar trend was observed for

Fth

mRNA,

which was upregulated in iT

REG

cells (Fig.

G). 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

ﬂ

anked

Fth

allele (

Fth

ﬂ

) (Darshan et al,

) into

Foxp3

GFP

mice (Zhou et al,

), deleting

Fth

speci

ﬁ

cally in T

REG

cells from

Foxp3

GFP-Fth

vs.

control

Foxp3

GFP

mice (Fig. EV1A).

Fth

deletion was associated with

The EMBO Journal

Qian Wu et al

1446

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

CONV

REG

FTH

-Actin

0 .0 2

0 .0 4

0 .0 6

0 .0 8

0 .1

FTH/

-Actin

CONV

REG

KDa

CD4

Foxp3

GFP-Fth∆/∆

1 10 10

12,4%

1 10 10

5,71%

1 10 10

2,65%

13,3%

***

1 10 10

Fth/H3

Fth

REG

KDa

REG

Spleen

MLN

REG

cell

(CD4

GFP

)

Thymus

REG

cell

(CD4

GFP

)

REG

cell

(CD4

GFP

)

CD4

CD3/28

IL-2

CD3/28

IL-2 + TGF

CONV

REG

Fth

1,15%

-10

79,4%

CD3/28
+IL-2

CD3/28

+IL-2/TGF

0.2

0.4

0.6

0.8

Fth/H3

Protein

CONV

REG

0.2

0.3

0.1

Fth/Arbp0

mRNA

CONV

REG

KDa

CONV

REG

Foxp3

GFP

Foxp3

GFP-Fth∆/∆

Foxp3

GFP

Foxp3

GFP-Fth∆/∆

Foxp3

GFP

MLN

Foxp3

-GFP

CD4

Foxp3

GFP-Fth∆/∆

Foxp3

GFP

Foxp3

-GFP

Nrp1

Nbr.

Foxp3

-GFP

Cells (x10

)

Nrp1

Foxp3

-GFP

(%)

2,15%

1,14%

***

Foxp3

-GFP

(%)

Foxp3

-GFP

(%)

Nbr.

Foxp3

-GFP

Cells (x10

)

Nbr.

Foxp3

-GFP

Cells (x10

)

Foxp3

-GFP

(%)

Nbr.

Foxp3

-GFP

Cells (x10

)

Nrp1

(tT

REG

)

Nrp1

(pT

REG

)

CD4

GFP

cells

100

***

82%

72,6%

Thymus

Spleen & MLN

Events (%)

Foxp3

GFP-Fth∆/∆

Foxp3

GFP

*** *** ***

Foxp3

-GFP

Thymus

Spleen

MLN

Foxp3

-GFP

MFI (10

)

Thymus

Spleen

MLN

REG

cell

CD4

GFP

Foxp3

-GFP

Foxp3

-GFP

Foxp3

-GFP

***

0.5

1.5

Qian Wu et al

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

1447

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

the accumulation of intracellular labile Fe

in T

REG

cells, isolated

from the mesenteric lymph nodes (MLN) of

Foxp3

GFP-Fth

vs.

control

Foxp3

GFP

mice (Fig. EV1B).

The frequency of thymic CD4

GFP

REG

cells was lower in

Foxp3

GFP-Fth

vs. control

Foxp3

GFP

mice (Fig.

H). This was not

associated, however, with a concomitant reduction in the number of
CD4

GFP

REG

cells (Fig.

H). In contrast, there was a marked

reduction of both the frequency and numbers of T

REG

cells in the spleen

(Fig.

I) and in the MLN (Fig.

J) 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 observa-

tions 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,

; Zhou et al,

), was reduced in thymic, splenic and MLN

REG

cells from

Foxp3

GFP-Fth

vs. control

Foxp3

GFP

mice (Fig.

K).

These observations suggest that FTH regulates

Foxp3

transcription in

the thymus as well as in circulating T

REG

cells, which is not suf

ﬁ

cient

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,

; Yadav et al,

). The frequency and number of

Nrp1

REG

cells and Nrp1

REG

cells was reduced, to the same

extent, as assessed in the MLN (Fig.

L) 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

ﬂ

ammatory ex-T

REG

cells

The reduction in T

REG

cells imposed by

Fth

deletion was associated

with an accumulation of activated CD4

CD44

high

CD62L

low

CONV

cells and CD8

CD44

high

CD62L

low

cytotoxic T (T

) cells, in the

spleen (Fig. EV1D) and in MLN of

Foxp3

GFP-Fth

vs. control

Fth

ﬂ

mice (Fig. EV1E). Concomitantly, there was an increase in the
frequency of interferon-

(IFN

)-expressing activated CD4

cells and CD8

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 home-

ostasis, preventing the activation and accumulation of in

ﬂ

amma-

tory CD4

cells and CD8

T cells.

Secreted ferritin complexes can restrain human T-cell prolifera-

tion 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

mice (Fig. EV2B).

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 pro

ﬁ

les of

CD4

CD25

GFP

REG

cells sorted from the lymph nodes of

Foxp3

GFP-Fth

vs. control

Foxp3

GFP

mice (Fig.

A). 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.

B).

Fth

deletion was associated with dysregulation of Foxp3-

dependent and -independent

“

REG

transcriptional signature

”

(Hill

et al,

), affecting at least 194 genes involved in T

REG

cell

function and lineage maintenance (Fig.

C). Pathway enrichment

analysis (Fig.

D,E) showed that T

REG

cells from

Foxp3

GFP-Fth

mice

presented T

1 and type 2 (T

2) transcriptional signatures (Fig.

D),

as illustrated by the induction of the transcriptional master
regulators of T

1 and T

2 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).

Consistently,

the

percentage

activated

CD4

GFP

CD44

high

CD62L

low

REG

cells was higher in the spleen and MLN

Figure 1.

REG

cell ferritin expression is a stable property of human and mouse T

REG

cells.

(

) FTH and

-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

) cells after two weeks of expansion with anti-CD3, anti-CD28 mAb and IL-2. (

) Relative quanti

ﬁ

cation of FTH protein expression,

normalized to

-Actin, detected by western blot as in (

3 independent experiments. (

) FTH and histone H3 protein expression detected by western blot in whole-

cell extracts from sorted mouse naive T cells (CD4

Foxp3

CD44

low

CD62L

high

; T

), (CD4

Foxp3

GFP

) T

REG

cells and memory T cells (CD4

Foxp3

−

CD44

high

CD62L

low

; T

by western blot. (

) Relative quanti

ﬁ

cation of FTH, normalized to histone H3, protein expression, detected by western blot as in (

). Data were normalized to FTH

expression in T

cells, pooled from four independent experiments. (

) Schematic representation of the protocol used for the generation of iT

REG

and representative

ﬂ

cytometry dot plots of mouse iT

REG

generated from sorted naive T cells (T

), stimulated with anti-CD3 and anti-CD28 mAb plus IL-2 and TGF

for 5 days. Control T

CONV

cells were subjected to the same experimental conditions, without TGF

. (

) FTH and histone H3 protein expression, detected by western blot in whole-cell extracts from

REG

and T

CONV

generated as depicted in (

). Data pooled from three independent experiments with similar trend. (

) The relative level of

Fth

mRNA expression, quanti

ﬁ

by qRT-PCR, using

Arbp0

as housekeeping gene. Data pooled from three independent experiments, with similar trend. (

–

) Schematic representation of the protocol used

(top panels), representative

ﬂ

ow cytometry dot plots (middle panels) and corresponding quanti

ﬁ

cation of percentage (%) in CD4

cells and cell number (Nbr.) (bottom

panels) of live (TCR

CD4

Foxp3

) GFP

REG

cells in (

) thymus, (

) spleen and (

) mesenteric LN (MLN). (

) Data from

8 mice per genotype, per organ, from two

independent experiments, with similar trend. (

) Data from

12 mice per genotype, per organ, from three independent experiments, with similar trend. (

) 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

ﬂ

ow cytometry histograms of GFP-hCre (bottom left panel). Relative quanti

ﬁ

cation of GFP-hCre expression (bottom right panel), represented as

mean

ﬂ

uorescence intensity (MFI). Data from

8 mice per genotype, pooled from two independent experiments with similar trend. (

) Schematic representation of the

experimental approach (left panel) used, representative

ﬂ

ow cytometry dot plots (top panel) and corresponding quanti

ﬁ

cation of percentage (%) (bottom left panel) and

cell number (Nbr.) (bottom right panel) of live (TCR

CD4

GFP

) Nrp1

and Nrp1

REG

cells in MLN. Data from

8 mice per genotype, pooled from two independent

experiments with similar trend. Data information: Data in (

) are presented as mean ± SD. Data in (

) are presented as mean ± SEM. Circles in (

–

) correspond to

individual mice.

values in (

–

) determined using unpaired

test with Welch

’

s correction, in (

–

) using ordinary one-way ANOVA, and in (

) using Two-way

ANOVA with Sidak

’

s multiple comparisons test. NS not signi

ﬁ

cant (

> 0.05); *

< 0.05; **

< 0.01; ***

< 0.001. Source data are available online for this

ﬁ

gure.

The EMBO Journal

Qian Wu et al

1448

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

eIF4/p70S6K

signaling

mTOR

signaling

EIF2

signaling

oit

pir

art

Fold Expression (log

)

-2 -1 0 1 2

AHR signaling

Interferon signaling

Acute phase response

ER stress pathway

BRCA1 in DNA damage response

DNA damage (14-3-3-σ signaling)

Inflammasome pathway

CHK in cell cycle checkpoint control

G1/S checkpoint regulation

cell differentiation

LXR/RXR activation

Cyclins and cell cycle regulation

GADD45 signaling

2 pathway

NRF2-oxidative stress response

Cell cycle control of chrom. replic.

1 pathway

Estrogen-mediated S-phase entry

G2/M DNA damage checkpoint

1 and T

2 activation

Akt1 Nqo1Dnajb11

Hmox1Sod2Mgst1Sqstm1

Nfe2l2MafkMap2k3

Txnrd1Dnajc15

Sod1Abcc1Fkbp5Bach1Ube2e3

Cat Pmf1GclmTxn1Vcp Prdx1Pi

k3c2a

Gsr Gclc A

bcc4

Gstt2Hacd3Enc1Maff Ft

MafgMaf Rras2Actg2Aox1Map3k5

Oxidative stress

Foxp3

GFP

Foxp3

GFP-Fth

−1

G-

CD25

-10

ait

Fth1

Mirt1

Top2a

Tigit

Hmox1

Anxa2

Myo1f

Myo6

Hist1h1b

Prune2

Mt1

Slc25a23

Stra6

Cdc42bpa

Hdac9

Nqo1

Ryr2

Cyp26b1

Tmem136

Cacna1i

Mt2

Gm13394

Mest

Clec2g

Pianp

Pls1

Pnlip

100

150

200

250

−8

−4

Adjusted p-value (-log

)

Base mean

expression

250000

500000

750000

Log

fold change

Foxp3

GFP-FthΔ/Δ

vs.

Foxp3

GFP

Foxp3

GFP

Foxp3

GFP-Fth

-10

Foxp3

GFP

Foxp3

GFP-Fth

Sorting

Lymph nodes

CD25

Foxp3

(GFP

)

RNAseq

P-value (-log 10)

8,12%

89,9%

Fold Expression (log2)

Qian Wu et al

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

1449

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

from

Foxp3

GFP-Fth

vs. control

Foxp3

GFP

mice (Fig. EV2D). This was

associated with an increase in the frequency of CD4

GFP

REG

cells expressing IFN

in the spleen (Fig. EV2E). This suggests that

FTH is essential to prevent the transdifferentiation of T

REG

cells into

ﬂ

ammatory T

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.

D). This was associated with the

activation of other canonical stress responses, including the cell
cycle and DNA damage, unfolded protein, and hypoxic response
(Fig.

D) as well as an overall shutdown of eIF2, mTOR and eIF4/

p70s6K signaling transduction pathways (Fig.

E). 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.

F). These observations suggest that FTH is

essential to support a transcriptional pro

ﬁ

le that maintains T

REG

cell

redox homeostasis, similar to described in other cell types
(Blankenhaus et al,

; Vanoaica et al,

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 in

ﬂ

ammatory T

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.

A). The resulting

Foxp3

GFP-Fth

-tdT

mice allow distinguishing CD4

GFP

tdT

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.

A).

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.

B and EV3A). Concomitantly, there was an

increase in the frequency of circulating ex-T

REG

cells (Figs.

and EV3A), with 60% of circulating T

REG

cells becoming ex-T

REG

cells

Foxp3

GFP-Fth

-tdT

, 24 weeks after birth (Figs.

D 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.

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.

E). The percentage and

number of splenic ex-T

REG

cells remained relatively stable (Fig.

F), 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.

G), with over 80% of

REG

cells becoming ex-T

REG

cells (Fig.

G).

Fth

was deleted in ex-T

REG

cells from

Foxp3

GFP-Fth

-tdT

mice, as determined by qRT-PCR

(Fig. EV3B), con

ﬁ

rming 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-

ﬂ

ammatory cytokine, which is a feature of T

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 IFN

, as compared to the lack of

IFN

expression in T

REG

and ex-T

REG

cells from control

Foxp3

GFP-tdT

mice (Figs.

H and EV3C). Moreover, a signi

ﬁ

cant proportion of

REG

and ex-T

REG

cells in the LN (Fig.

I) and spleen (Fig. EV3D)

from

Foxp3

GFP-Fth

-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.

I 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 in

ﬂ

ammatory 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. Brie

ﬂ

y, sub lethally irradiated lymphogenic

Rag2-

ﬁ

cient (

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.

A). The

proportion of CD45.2

vs. CD45.1

REG

cells in LN (Fig.

–

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

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.

The proportion of CD45.2

vs. CD45.1

Foxp3

CD3

CD4

cells and CD3

CD8

cells was indistinguishable in the LN

(Fig.

–

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

high

CD62L

low

CONV

cells and CD45.2

CD8

CD44

high

CD62L

low

cells in the LN (Fig. EV4A

–

C) were also similar in these BM

chimeric mice. This con

ﬁ

rms that FTH sustains the number of

Figure 2.

FTH expression in T

REG

cells prevents transdifferentiation into in

ﬂ

ammatory ex-T

REG

cells.

(

) Schematic representation of experimental approach (left panel) and representative

ﬂ

ow cytometry dot plots of (CD4

CD25

GFP

) T

REG

cells from the lymph nodes

before and after sorting. (

) 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-Fth

and control

Foxp3

GFP

(

5 per genotype) mice.

values determined by Benjamini and Hochberg adjusted probabilities. (

) Heatmap

representation of T

REG

transcriptional signature genes differentially expressed in T

REG

cells sorted from

Foxp3

GFP-Fth

vs.

Foxp3

GFP

mice, as illustrated in (

). (

) Pathway

enrichment analysis of genetic programs overexpressed (

) or under-expressed (

) in T

REG

cells sorted from

Foxp3

GFP-Fth

vs.

Foxp3

GFP

mice, as illustrated in (

). Data

were analyzed using g:SCS multiple testing correction method with a signi

ﬁ

cance threshold of 0.05. (

) Heatmap representation of individual genes associated with

oxidative stress-responsive programs, differentially expressed in T

REG

cells sorted from

Foxp3

GFP-Fth

vs.

Foxp3

GFP

mice, as illustrated in (

). Source data are available

online for this

ﬁ

gure.

The EMBO Journal

Qian Wu et al

1450

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

2 4 7 9 11151924

Time [Weeks]

2 4 7 9 11 15 19 24

Time [Weeks]

2 4 7 9 11 15 19 24

Time [Weeks]

****

GFP

TdT

REG

Cells (%)

GFP

TdT

ex-T

REG

Cells (%)

GFP

TdT

Cells (%)

% (

GFP

TdT

)

REG

cells

% GFP

TdT

ex-T

REG

cells

Nbr

GFP

TdT

REG

cells

(x10

)

Nbr

GFP

TdT

ex-T

REG

cells

(x10

)

% GFP

TdT

cells

0.5

1 5

2.5

100

IFN

Events (%)

100

Spleen

10³

10⁴

10⁵

10³

10⁴

10⁵

CD71

Ki67

7 28%

14,3%

23,5%

16,1%

REG

ex-T

REG

(%)

Ki67

CD71

ex-

REG

***

REG

ex-T

REG

FMO

REG

cells

(

GFP

tdT

)

ex-T

REG

cells

(GFP

tdT

)

Gfp

hCre

Stop

LoxP

tdT

Rosa26

locus

Foxp3

Promoter

Fth

locus

LoxP

Fth

Promoter

BAC

LN, Spleen

IFN

REG

(

GFP

tdT

)

Ex-T

REG

(

GFP

tdT

)

Ki67
CD71

REG

(

GFP

tdT

)

Ex-T

REG

(

GFP

tdT

)

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

REG

cells

(

GFP

tdT

)

REG

cells

(

GFP

tdT

)

Blood

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

REG

cells

(

GFP

tdT

)

REG

cells

(

GFP

tdT

)

Spleen

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

****

**** ****

IFN

(%)

REG

ex-

REG

ex-

REG

Qian Wu et al

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

1451

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

circulating T

REG

cells, via a cell-autonomous mechanism that

acts irrespectively of the systemic in

ﬂ

ammatory response associated

with

Fth

deletion in T

REG

cells from

Foxp3

GFP-Fth

mice

(Fig. EV1D

–

G).

We then asked whether FTH restrains the transdifferentiation of

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

REG

cells (Fig.

D,E)

and CD45.2

CD3

CD4

GFP

tdT

ex-T

REG

cells (Fig.

D,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.

G). This suggests that FTH maintains peripheral T

REG

cell

lineage stability, via a cell-autonomous mechanism, irrespectively of
the systemic in

ﬂ

ammatory response associated with

Fth

deletion in

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 in

ﬂ

ammatory response associated with

Fth

deletion in T

REG

cells from

Foxp3

GFP-Fth

mice (Fig. EV1D

–

G).

To test this hypothesis, the gene expression pro

ﬁ

le of T

REG

cells

sorted from the LN of mixed BM chimeras (Fig.

H) was compared

to that of T

REG

cells sorted from the LN of non-chimeric mice

(Fig.

A). 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 down-

regulated 96 genes, compared to

Fth

-competent T

REG

cells from

control

Foxp3

GFP-tdT

mice (Fig.

I). 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 down-

regulated 36 genes, compared to

Fth

-competent ex-T

REG

cells from

control

Foxp3

GFP-tdT

mice (Fig.

J). 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.

I,J). This

suggests that FTH exerts cell-autonomous control of T

REG

cell redox

homeostasis, irrespective of the systemic in

ﬂ

ammatory response

associated with

Fth

deletion in T

REG

cells from

Foxp3

GFP-Fth

mice

(Fig. EV1D

–

G).

The in

ﬂ

ammatory transcriptional signature of T

REG

cells from

Foxp3

GFP-Fth

vs.

Foxp3

GFP

mice (Fig.

–

D) was not observed in T

REG

cells from BM chimeric mice, originating from

Foxp3

GFP-Fth

-tdT

vs.

Foxp3

GFP-tdT

(Fig.

I,J). Among the 245 differentially expressed genes in

Fth

-de

ﬁ

cient (CD4

GFP

) T

REG

cells from BM chimeric mice (Fig.

I,J),

134 matched those differentially expressed in

Fth

-de

ﬁ

cient T

REG

cells

from non-chimeric mice (Fig.

K,L). Gene ontology analyzes of the

overlapping genes, showed an enrichment for pathways related to
oxidative stress response, comprising several NRF2-regulated genes
(Fig.

–

M). In contrast,

Fth

-de

ﬁ

cient (CD4

GFP

) T

REG

cells from

BM chimeric mice did not show an enrichment for pathways
associated with T

cell activation (Fig.

–

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.

). In contrast, the in

ﬂ

ammatory pro

ﬁ

le associated with the

transition of

Fth

-deleted T

REG

cells towards in

ﬂ

ammatory ex-T

REG

cells,

observed in

Foxp3

GFP-Fth

mice (Fig.

) and

Foxp3

GFP-Fth

-tdT

mice

(Fig.

) requires, in addition, the development of systemic

ﬂ

ammation.

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,

), to compare the proliferative capacity of CD4

GFP

tdT

REG

cells from

Foxp3

GFP-Fth

-tdT

vs. control

Foxp3

GFP-tdT

mice. The

number of CD4

GFP

tdT

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.

–

C). While there were

no differences in the number of CD4

GFP

tdT

ex-T

REG

, the ratio of

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 signi

ﬁ

cance (Fig.

–

C).

Fth

expression in

CD4

GFP

tdT

REG

cells used in the adoptive transfer was

con

ﬁ

rmed by qRT-PCR (Fig. EV4D,E).

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 re

ﬂ

ecting, 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

cells activated in vitro with anti-CD3/

CD28 mAb plus IL-2 and TGF

, were indistinguishable regardless of

Figure 3.

FTH enforces T

REG

cell lineage stability.

(

) 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. (

–

) Percentage of circulating: (

) T

REG

and (

) ex-T

REG

cells in

Foxp3

GFP-Fth

-tdT

and control

Foxp3

GFP-tdT

mice,

and (

) 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 (

). Data from

–

6 mice

per genotype was pooled from three independent experiments with a similar trend. (

–

) Percentage and number of splenic (

) T

REG

cells, (

) ex-T

REG

cells and (

) relative

percentage of ex-T

REG

cells over total CD4

tdT

cells. Data from

4 mice per genotype, pooled from two independent experiments with similar trends. (

)

Representative

ﬂ

ow cytometry histograms of IFN

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 quanti

ﬁ

cation of the percentage of IFN

expressing (CD4

GFP

tdT

) T

REG

and

(CD4

GFP

tdT

) ex-T

REG

cells (right panels). Expression of IFN

was induced upon Phorbol-12-myristate-13-acetate (PMA) and Ionomycin re-activation in vitro. Data from

3 wells per genotype, in one experiment, representative of two independent experiments with similar trend. (

) Representative

ﬂ

ow cytometry dot plots (left panels)

and corresponding quanti

ﬁ

cation (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

6 mice per genotype, pooled from two independent experiments, with similar trend. Data information: Data in

(

–

) represented as mean ± SD. Circles correspond to mean values. Data in (

–

) circles correspond to individual mice and red bars to mean values. (

) represented as

mean ± SD.

values in (

–

) calculated using Two-way ANOVA analysis with Sidak

’

s multiple comparisons test and in (

–

) with unpaired

test with Welch

’

correction. NS not signi

ﬁ

cant (

> 0.05); *

< 0.05; **

< 0.01; ***

< 0.001; ****

< 0.0001. Source data are available online for this

ﬁ

gure.

The EMBO Journal

Qian Wu et al

1452

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

120

CD8

CD4

Foxp3

CD45.2

CD45.1

10³

-10³

10³

-10³

35.1%

64,8%

45%

54,2%

44%

54,4%

32,2%

67,7%

35,3%

63,8%

86,9%

12,7%

Chimerism

)

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

***

GFP

TdT

Nbr.

GFP

TdT

(x10

)

Nbr.

GFP

TdT

(x10

)

ex-T

REG

cells

GFP

TdT

10³

-10³

-10³ 0

10³

-10³ 0

TdT

GFP

2,77%

13,2%

79,7%

4,34%

2,82%

3,43%

91,1%

2,68%

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

REG

cells

Ratio

Lymph Nodes

GFP

TdT

CD45.1
CD45.2

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

CD8

CD4

Foxp3

***

Rag2

-/-

Lymph Nodes

CD45.1

vs.

CD45.2

C57BL/6

50%

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

50%

Fth1

Hmox1

Nqo1

Vmn1r80

Prune2

Fes

Mt1

Gm5441

Ly6c2

Mcoln2

Cxcr2

Atp8b4

Fth1−ps

Epdr1

Mt2

Padi4

Cnga1

Mageh1

Npdc1

Amz1

Susd4

Mrgprb1

Pde4c

Large2

Vmn1r68

−10 −5

10 15 20

Adjusted p-value (-log

)

Log

fold change

REG

cells (

GFP

tdT

)

REG

cells

(

GFP

tdT

)

ex-T

REG

cells

(

GFP

tdT

)

Foxp3

GFP-tdT

Foxp3

GFP-Fth

-tdT

vs.

Base mean

expression

Rag2

-/-

Lymph nodes

C57BL/6

50%

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

50%

CD45.2

RNAseq

Fth1

Hmox1

Nqo1

Nrn1

Ptpn5

Gm21909

Sema7a

Ly6d

Ccr6

Mt1

Gm16086

Hif3a

Gm29243

Gja10

Gm19951

Ift172

Fth1−ps

Sgcb

Cfap54

Mfap3l

Nr3c2

Cd83

Smim10l2a

Arhgap24

Plpp1

−10 −5

10 15 20

Log

fold change

Adjusted p-value (-log

)

ex-T

REG

cells

(

GFP

tdT

)

Foxp3

GFP-tdT

Foxp3

GFP-Fth

-tdT

vs.

Base mean

expression

Mixed BM chimeric/non-chimeric mice

REG

cells (

GFP

)

REG

cells (

GFP

)

REG

cells (

GFP

)

Chimeric

111

134

Non−chimeric

3711

Upregulated

Downregulated

300
100

100

150

-log

(p-value)

Mixed BM c

himeras

on-chimeric

Bio.

proc. KEGG Reac.

Cell redox

homeostasis

Cellular response to

chemical stimulus

Cell. response

to IFN-β

Detoxification

of ROS

BACH2

IRF1

IRF2

IRF

Ferroptosis

Oxid. stress

and redox

pathway

Oxid. stress

response

Response to

oxid. stress

Response

to ROS

Transcrip. activ. by

NRF2 in response

to phytochem.

−log

(p value)

Hmox1

Nqo1

Sqstm1

Sod1

Slc48a1

Txnrd1

Txn1

Prdx1 Mt1

Mafg

Fth1

Ftl1

Gclm

Gsr

CD45.2

CD45.1

Sorting

***

Qian Wu et al

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

1453

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

whether T

cells were sorted from

Foxp3

GFP-Fth

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,

; Pham et al,

FTH regulates T

REG

cell mitochondrial function

and bioenergetics

REG

cells rely on a core metabolic program whereby mitochondrial

oxidative phosphorylation is the major source of ATP (Angelin
et al,

; Weinberg et al,

). The observation that FTH

regulates mitochondrial integrity in parenchyma cells (Blankenhaus
et al,

) suggested that FTH also regulates the mitochondrial

integrity of T

REG

cells. However, the number of mitochondria per

CD4

GFP

REG

cells in the LN of

Foxp3

GFP-Fth

was similar to that

of T

REG

cells from

Foxp3

GFP

mice, corresponding to ~55 mitochon-

dria per T

REG

cell (Fig.

D). This suggests that FTH is not essential

to maintain the mitochondrial integrity of T

REG

cells.

The mitochondrial membrane potential of

Fth

-de

ﬁ

cient T

REG

cells from

Foxp3

GFP-Fth

mice was markedly reduced, compared to

control T

REG

cells from

Foxp3

GFP

mice (Fig.

E). 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 O

consumption rate (OCR), upon uncoupling of the

mitochondrial respiratory chain by carbonyl cyanide 4-(tri

ﬂ

uor-

omethoxy)phenylhydrazone

(FCCP).

Spare

mitochondrial

respiratory capacity was higher in (CD4

GFP

) T

REG

cells from

Foxp3

GFP-Fth

vs.

Foxp3

GFP

mice (Fig.

F,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.

F,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.

F,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 acidi

ﬁ

cation rate (ECAR), re

ﬂ

ecting the rate

of glycolysis, was higher in T

REG

cells from

Foxp3

GFP-Fth

vs.

Foxp3

GFP

mice (Fig.

H). This suggests that FTH plays an essential

role in supporting the metabolic and bioenergetic pro

ﬁ

le of T

REG

cells, favoring mitochondrial oxidative OXPHOS over glycolysis
(Angelin et al,

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

-de

ﬁ

cient 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

-ketoglutarate and isocitrate intracellular

concentrations was lower in splenic

Fth

-de

ﬁ

cient T

REG

cells from

Foxp3

GFP-Fth

mice vs. control T

REG

cells from

Foxp3

GFP

mice

(Fig.

I). In contrast, the ratio of intracellular

-ketoglutarate and

glutamate concentration was similar (Fig.

I). This suggests that

FTH modulates the rate of isocitrate to

-ketoglutarate conversion

(Fendt et al,

), catalyzed by isocitrate dehydrogenase (IDH) in

the mitochondrial TCA cycle. This occurs, most likely, without
interfering with glutaminolysis, whereby

-ketoglutarate is gener-

ated 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

ﬁ

cient 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 pro

ﬁ

le of T

REG

cells,

presumably favoring lactate production via glycolysis, similar to
observed in hepatocytes (Weis et al,

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,

; Zheng et al,

2010

). Cytosine demethylation is

catalyzed by the TET family of methylcytosine dioxygenases (Yue
et al,

; Yue et al,

), via redox-based reactions that use Fe

and

-ketoglutarate as an essential cofactor and obligate substrate,

respectively (Kohli and Zhang,

; Pastor et al,

). Having

Figure 4.

FTH enforces T

REG

cell lineage stability via a cell-autonomous mechanism.

(

) Schematic representation of bone marrow (BM) chimeric mice used for

ﬂ

ow 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

L) into

congenic C57BL/6-recipient

Rag2-

ﬁ

cient (

Rag2

−

) female mice, 2 h after irradiation (600 Gys). (

) Representative

ﬂ

ow cytometry dot plots and (

) Percentage of

CD45.1

vs. CD45.2

CD4

T cells, CD8

T cells and CD4

Foxp3

REG

cells in the lymph nodes of mixed BM chimeric mice, from (

). Data from

10 mice per genotype.

(

–

) Representative

ﬂ

ow cytometry dot plots (

) and corresponding quanti

ﬁ

cation (

) of the percentage and number of (CD45.2

CD4

GFP

tdT

) T

REG

cells (

(CD45.2

CD4

GFP

−

tdT

) ex-T

REG

cells (

) and relative proportion of ex-T

REG

cells over total CD45.2

CD4

tdT

cells in lymph nodes (LN) (

) of mixed BM chimeric mice

(as in

). Data from

–

10 mice per genotype. (

) 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. (

) Volcano plot representations of RNA sequencing data of genes

overexpressed (red) or under-expressed (blue) in T

REG

cells (

) and ex-T

REG

cells (

) from lymph nodes of BM chimeric mice (as in

). Data from

–

4 mice per genotype.

values determined by Benjamini and Hochberg adjusted probabilities. (

) Euler plot of differentially regulated genes in (CD45.2

CD4

GFP

) T

REG

cells sorted from the

Foxp3

GFP-Fth

or control

Foxp3

GFP

REG

cells (non-chimeric; blue; from analysis described in Fig.

A,B) or from the mixed BM chimeric mice (chimeric; red; from analysis

illustrated in

–

). (

) Dumbbell plot, showing the adjusted

value of the 134 overlapping differentially regulated genes (from analysis described in

). Gray bars

connecting dots represent the difference in

values for the differentially regulated genes (from analysis described in

). (

) Functional enrichment analysis of the

overlapping genes (

134) (from analysis described in

), considering

ﬁ

ve 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 signi

ﬁ

cance threshold of 0.05. Data information: Data in (

) are represented as mean ± SD. Data in (

–

) are represented as mean, circles correspond to individual

mice and red bars are mean values.

value in (

) was determined by two-way ANOVA with Sidak

’

s multiple comparisons test and in (

–

) by unpaired

test with Welch

’

correction. *

< 0.05; **

< 0.01; ***

< 0.001. Source data are available online for this

ﬁ

gure.

The EMBO Journal

Qian Wu et al

1454

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

GFP

tdT

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

)

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

Lymph
Nodes

REG

ex-T

REG

GFP

TdT

REG

cells

(

GFP

tdT

)

ex-T

REG

cells

(

GFP

tdT

)

Rag2

-/-

Lymph nodes & spleen

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

REG

cells

(

GFP

tdT

)

Sorting

Transfer

Flow

Cytometry

TMRE

old

Change

)

0.4

0.8

1.2

100

200

300

400

Basal

Resp.

ATP

Prod.

Spare

Resp.

Capacity

0 20 40 60 80 100

FCCP

Oligo.

A/A

Rot.

Time (minutes)

***

0 20 40 60 80100

100

200

300

ECAR

(mmpH

cells

)

Time (minutes)

OCR

(pmol/min/10

cells)

hndri

(Number

)

200

400

600

800

-150

***

Lymph nodes

REG

cells

(

GFP

)

Spleen

TMRE

Sorting

OCR

(pmol/min/10

cells)

FCCP

Oligo.

A/A

Rot.

Spleen

Foxp3

GFP

Foxp3

GFP-Fth

Foxp3

GFP

Foxp3

GFP-Fth

Sorting

Foxp3

GFP

Foxp3

GFP-Fth

Foxp3

GFP

Foxp3

GFP-Fth

Foxp3

GFP

Foxp3

GFP-Fth

Genomic PCR

Events (%)

Seahorse

-ketaglutarate

/Isocitrate

0.01

0.02

0.03

-ketaglutarate

/Glutamate

Targeted

Metabolomics

Foxp3

GFP

Foxp3

GFP-Fth

Spleen

Sorting

67%

33%

42%

58%

REG

cells

(

GFP

)

REG

cells

(

GFP

)

REG

cells

(

GFP

)

Qian Wu et al

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

1455

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

established that FTH regulates intracellular catalytic Fe

, cellular

redox homeostasis and possibly the rate of

-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)

pro

ﬁ

ling

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.

A). 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.

B), revealing that FTH regulates the Methylome of

REG

and ex-T

REG

cells.

Fth

-de

ﬁ

cient CD45.2

CD4

tdT

cells, originating from the BM

Foxp3

GFP-Fth

-tdT

mice, presented a discrete number of hyper-

methylated and hypomethylated CpG-rich sequences, compared to
CD45.2

CD4

tdT

cells originating from the BM of control

Foxp3

GFP-tdT

mice (Fig.

C). 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.

D),

from different chromosomes (Fig.

E). 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.

F), controlling T

REG

cell lineage stability (Yue et al,

; Yue et al,

). 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.

G). 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.

H). 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.

I).

Taken together, these observations establish, as a proof of principle,
that FTH ferroxidase activity regulates TET methylcytosine dioxy-
genase 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 ques-
tioned 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.

A,B). This suggests

that FTH acts in a cell-intrinsic manner to regulate the expression
of FOXP3 in human T

REG

cells.

Having established that regulation of Fe metabolism by FTH

regulates

Foxp3

transcription in the thymus as well as in circulating

REG

cells (Fig.

K) we compared the relative level of

Foxp3-

GFP

expression in Nrp1

REG

and Nrp1

REG

cells. GFP expression

Figure 5.

FTH is a T

REG

cell-autonomous cytoprotectant.

(

) Schematic representation of

ﬂ

ow 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. (

) Representative

ﬂ

ow cytometry dot plots of (CD4

GFP

tdT

) T

REG

cells and

(CD4

GFP

tdT

) ex-T

REG

cells, analyzed, as illustrated in (

). (

) 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 (

). Data from

4 per genotype, in

one out of two independent experiments with similar trend. (

) 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), quanti

ﬁ

ed by genomic quantitative PCR. Data from

3 mice

per genotype in one experiment. (

) Schematic representation of the experimental approach used to quantify mitochondrial membrane potential in splenic mouse

(CD4

GFP

) T

REG

cells (left panel). Representative

ﬂ

ow cytometry histograms of tetramethylrhodamine ethyl ester (TMRE) staining (middle panels). Mean

ﬂ

uorescence

intensity (MFI) of TMRE (right panel). Data from

–

5 mice per genotype, pooled from two independent experiments, with similar trend. (

) 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

3 mice

per genotype, represented as mean ± SD (

–

5 technical replicates) in one out of three independent experiments, with similar trend. Oligomycin (Oligo.),

carbonilcyanide p-tri

ﬂ

ouromethoxyphenylhydrazone (FCCP), Antimycin A/Rotenone (A/A

Rot.). (

) Quanti

ﬁ

cation of basal respiration, ATP production and spare

respiratory capacity, from data represented in (

). (

) Extracellular acidi

ﬁ

cation rate (ECAR) in live splenic (CD4

GFP

) T

REG

cells represented as mean ± SD (

–

technical replicates) in one out of three independent experiments, with similar trend. (

) Schematic representation of the experimental approach (left panel), used to

quantify the ratio of

-ketoglutarate to isocitrate and

-ketoglutarate to glutamate (right panels) in live splenic (CD4

GFP

) T

REG

cells. Data from

4 mice per genotype

in one out of two independent experiments with similar trend. Data information: Data in (

–

) circles correspond to individual value and red bars to mean values. Data are

presented as mean ± SD. Data in (

) are presented as mean ± SD, circles correspond to technical replicates. Data in (

) are presented as mean ± SD of technical

replicates. Data in (

) are presented as mean ± SD, circles correspond to individual values.

values in (

(left panel),

) were calculated using two-way ANOVA with

Sidak

’

s multiple comparison test,

values in (

) were calculated using two-way ANOVA with Bonferroni

’

s multiple comparisons test.

values in (

, right panel),

)

were calculated using unpaired

test with Welch

’

s correction. NS not signi

ﬁ

cant (

> 0.05), *

< 0.05; **

< 0.01; ***

< 0.001. Source data are available online for this

ﬁ

gure.

The EMBO Journal

Qian Wu et al

1456

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

Promoter

Exon

Intron
Intergenic

0.4

0.2

0.6

0.8

Methylation rate

Genome wide

EM-Seq

(CD4

tdT

)

Hypermethylated

Hypomethylated

100

Percentage/chromosome

Chromosome

CD45.2

C57BL/6

50%

Foxp3

GFP-Fth

Foxp3

GFP-tdT

50%

Sorting

CD45.2

CD45.1

P<0.05

Rag2

-/-

Lymph nodes

-400 -200 0 200

-400

-200

200

PC1 (22.7%)

PC2 (15.1%)

0.2

0.4

0.6

0.8

Methylation

ate

CNS2

CpG sites (CNS1)

5
CpG sites (CNS2)

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

Foxp3

10 kb

CNS1

ctivit

Normalized to Control

TET3

VEC FTH FTH

mut

TET3-Flag

250

Kda

100

LaminA/C

TET3

VEC FTH FTH

mut

Nuclear

Cytosolic

0.5

1.5

FTH-Flag

GAPDH

Bone

Marrow

***

0.2

0.4

0.6

0.8

***

VEC FTH FTH

mut

VEC FTH FTH

mut

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

TET3

VEC FTH FTH

TET3-Flag

/LaminA/C

FTH-Flag

/GAPDH

Kda

mut

Qian Wu et al

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

1457

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

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.

C). Considering

that GFP is expressed under the control of a bacterial arti

ﬁ

cial

chromosome (BAC) transgene carrying the intact

Foxp3

promoter

(Chen et al,

; Zhou et al,

), these observations suggest that

FTH is essential to sustain

Foxp3

transcription in tT

REG

and

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.

D). This was also

observed in BM chimeric mice, that is, the relative levels of Foxp3
protein expression were lower in

Fth

-de

ﬁ

cient GFP

tdT

cells (T

REG

cells) originating from

Foxp3

GFP-Fth

-tdT

vs. control

Foxp3

GFP-tdT

mice

(Fig.

E). Moreover, GFP expression was lower in

Fth

-de

ﬁ

cient

tdT

cells (ex-T

REG

cells) originating from

Foxp3

GFP-Fth

-tdT

vs.

control

Foxp3

GFP-tdT

mice (Fig.

F). This suggests that FTH acts in a

cell-intrinsic manner to regulate Foxp3 transcription and expres-
sion in mouse T

REG

cells.

To establish whether FTH regulates endogenous

Foxp3

tran-

scription, we asked whether

Fth

deletion was associated with a

concomitant reduction of

Foxp3

and

Gfp

mRNA expression.

The relative level of

Foxp3

(Fig.

G) and

Gfp

(Fig.

mRNA expression were lower in

Fth

-de

ﬁ

cient 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

con

ﬁ

rmed by qRT-PCR (Fig.

H). While

tdT

mRNA expression

was not altered (Fig.

H). These observations suggest that FTH is

essential to sustain

FOXP3

transcription.

FTH expression in T

REG

cells limits the pathologic

outcome of autoimmune neuroin

ﬂ

ammation

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.

A).

Foxp3

GFP-Fth

mice had an increase in EAE incidence (Fig.

B) and

severity (Fig.

C), in response to immunization with myelin

oligodendrocyte

glycoprotein

(MOG)-derived

peptide

35-55

(MOG

35-55

) emulsi

ﬁ

ed in complete Freund

’

s adjuvant compared to

control immunized

Fth

ﬂ

mice. Of note, the immunization protocol

used was

“

sub-optimal

”

, as suggested by the relatively low disease

scores in control immunized

Fth

ﬂ

mice, likely favoring the increase in

EAE severity observed in

Foxp3

GFP-Fth

mice.

The frequency of T

cells expressing IFN

, IL-17A and

IL17

IFN

cells was higher in the spinal cord of MOG

35-55

immunized

Foxp3

GFP-Fth

vs. control

Fth

ﬂ

mice (Fig.

D). This

suggests that FTH expression in T

REG

cells limits the activation,

proliferation and/or in

ﬁ

ltration of self-reactive T

type 1 (T

1) and

type 17 (T

17) cells as well as pathogenic IL17

IFN

cells

(Duhen et al,

) into the central nervous system, in response to

MOG

35-55

immunization.

FTH expression in T

REG

cells limits malaria severity

REG

cells constraint the extent of immunopathology associated with

infectious diseases (Arpaia et al,

), supporting the hypothesis

that regulation of T

REG

cell lineage stability by FTH impacts on the

severity of infectious diseases (Soares et al,

). 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

Foxp3

GFP

mice that survived and cleared the parasite (Fig.

E).

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.

E). 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.

F), as

well as with an increase in the number (but not the frequency) of
IFN

-expressing T

cells in the spleen (Fig.

G). 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

ﬁ

ndings illustrate how dysregulation of Fe

metabolism promotes the pathogenesis of severe malaria (Ferreira
et al,

; Gozzelino et al,

; Ramos et al,

; Seixas et al,

; Wu et al,

FTH expression in T

REG

cells favors tumor progression

REG

cells are pathogenic, for example, when restraining anti-tumor

immunity (Curiel et al,

; Liu et al,

), suggesting that FTH

Figure 6.

FTH regulates mitochondrial function and cytosine methylation in T

REG

cells.

(

) Schematic representation of mixed bone marrow (BM) chimeric mice from which lymph node CD45.2

CD4

tdT

cells (i.e., GFP

REG

cells and GFP

−

ex-T

REG

cells) were

sorted for genome-wide EM-seq analyses. (

) 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 (

). Data from

4 mice per genotype is represented as individual circles in one experiment. (

) 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 (

). Statistical analysis for multiple testing correction was

performed with Sliding Linear Model (SLIM). (

) The relative percentage of total methylated regions (i.e., hyper- and hypomethylated regions) (

< 0.01 and methylation

difference >

10%) according to different genome regions (promoter, exon, intron and intergenic). (

) The number of hyper- and hypomethylation events (10% change in

methylation and a

value of 1%) per chromosome, shown as a percent of the differential sites. (

) Schematic representation of the

Foxp3

in enhancer regions CNS1 and

CNS2. (

) Relative quanti

ﬁ

cation 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 (

). Data from

4 mice per genotype in one experiment. (

) TET activity in nuclear extracts from

HEK293T cells transiently transfected with human TET3-

ﬂ

ag, FTH-

ﬂ

ag, or FTH

mut

ﬂ

ag cDNAs. Data shows technical replicates, pooled from four independent experiments.

(

) FTH-

ﬂ

ag, FTH

mut

ﬂ

ag, TET3-

ﬂ

ag, Lamin A/C and GAPDH protein expression, detected by western blot in nuclear and cytosol extracts from HEK293T cells transfected

as described in (

). Relative quanti

ﬁ

cation of TET3-

ﬂ

ag, normalized to Lamin A/C (bottom left panel), and FTH-

ﬂ

ag, normalized to GAPDH (bottom right panel). Data

from one experiment, representative of three independent experiments with similar trend. Data information: Data in (

) are presented as mean ± SD, circles correspond to

individual mice and red bars to mean values. Data in (

) are presented as mean ± SD, circles correspond to technical replicates.

values in (

) were calculated using

Two-way ANOVA with Sidak

’

s multiple comparison test, and in (

) using one-way ANOVA using with Sidak

’

s multiple comparison test. NS not signi

ﬁ

cant (

> 0.05),

< 0.05; **

< 0.01, ***

< 0.001. Source data are available online for this

ﬁ

gure.

The EMBO Journal

Qian Wu et al

1458

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

0.1

0.2

0.3

0.4

Spleen LN

Foxp3/Arbp0

(mRNA)

0.01

0.02

0.03

0.04

Spleen LN

***

Gfp/Arbp0

(mRNA)

0.05

0.1

0.15

0.2

0.25

Fth/Arbp0

(mRNA)

Spleen LN

*** ***

Foxp3

FOXP3 (MFIx10

)

FTH

-actin

Spleen

MLN

Foxp3 (MFI (x10

)

Spleen MLN

Ctrl

shRNA

***

Ctrl

FTH

429

FTH

432

shRNA

FTH

429

FTH

432

Events

(%)

Foxp3

GFP-tdT

Foxp3

GFP-Fth∆/∆-tdT

Unstained

Foxp3

Foxp3 MFI (x10

)

Spleen

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

Spleen LN

0.2

0.4

0.6

tdT/Arbp0

(mRNA)

Spleen

NS NS

CD 4

Foxp3

(

GFP

tdT

)

Rag2

-/-

C57BL/6

50%

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

50%

CD45.2

Events

(%)

REG

& ex-T

REG

cells

(

tdT

)

Spleen & LN

Sorting

****

GFP MFI (x10

)

Spleen LN

Spleen

Events

(%)

Unstained

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

REG

cells

(

tdT

)

Rag2

-/-

C57BL/6

50%

Foxp3

GFP-Fth

-tdT

Foxp3

GFP-tdT

50%

CD45.2

ex-T

REG

Foxp3

GFP-tdT

Foxp3

GFP-Fth∆/∆-tdT

REG

& ex-T

REG

cells

(

tdT

)

Sorting

Foxp3

GFP-Fth∆/∆

Foxp3

GFP

Spleen & LN

REG

cells

Spleen & LN

Foxp3

-GFP

Spleen & LN

Spleen &

MLN

Nrp1

(

REG

)

Nrp1

(

REG

)

CD4

GFP

cells

Foxp3

qRT-PCR

Gfp; Fth, TdT

qRT-PCR

Events (%)

Foxp3

GFP-Fth∆/∆

Foxp3

GFP

Foxp3

-GFP

Nrp1

Foxp3

-GFP

MFI

(10

)

***

Nrp1

Qian Wu et al

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

1459

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

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

mice (Fig.

H). This was

associated with lower frequency of tumor-in

ﬁ

ltrating T

REG

cells (Fig.

and higher frequency of activated Foxp3

CD4

CD25

effector T

cells

(Fig.

J). The frequency and number of activated CD4

IFN

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

IFN

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

cells,

albeit

not statistically signi

ﬁ

cant

(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

REG

cells respond to variations in the relative levels of nutrients,

vitamins and metabolites in their environment (Chapman et al,

; Shi and Chi,

), via dedicated transporter-receptor

coupled sensors that modulate T

REG

cell function and lineage

stability (Kempkes et al,

; Shi and Chi,

). In keeping with

Fe-regulatory genes being a core property of T

REG

cells (Cuadrado

et al,

), we found that the Fe-regulatory protein FTH is

essential to support T

REG

cell lineage maintenance in vivo (Figs.

–

)

and support immune homeostasis (Figs.

, EV1D

–

G, and EV2C).

FTH regulates T

REG

cell lineage stability (Figs.

and

–

without interfering with the antiproliferative function of T

REG

cells

(Fig. EV2A). Instead, FTH targets the intracellular pool of redox-
active Fe

(Fig. EV1B) to regulate T

REG

cell redox homeostasis

(Figs.

F and

–

M) in a manner that controls T

REG

cell: (i) energy

metabolism (Figs.

–

I and EV5A

–

E), (ii) TET activity (Fig.

H),

(iii) cytosine demethylation at CpG-rich sequences at CNS1 and
CNS2 in the

FOXP3

locus (Fig.

G), and (iv) FOXP3 transcription/

expression (Figs.

K,L and

–

H). As the latter is essential to

maintain the transcriptional program supporting T

REG

cell lineage

stability (Williams and Rudensky,

) (Nakatsukasa et al,

Ohkura et al,

; Yue et al,

)

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.

H,L). These

observations are consistent with FTH acting upstream of TET
methylcytosine dioxygenases, to control redox-based cytosine
demethylation in T

REG

cells (Fig.

) (Huang and Rao,

; Pastor

et al,

That FTH targets the intracellular pool of redox-active Fe

control TET enzymatic activity is suggested by the observation that
overexpression of a ferroxidase de

ﬁ

cient FTH

compromised TET

methylcytosine dioxygenase activity (Fig.

H, I). Several non-

mutually exclusive mechanisms might explain this observation.
First, FTH could act

“

directly

”

via sequestration of catalytic Fe

(Fig. EV1B), controlling the availability of this essential cofactor of
TET enzymatic activity (Kohli and Zhang,

; Pastor et al,

Second, FTH could act

“

indirectly

”

via the regulation of cellular

redox homeostasis (Figs.

D,F and

–

M), preventing catalytic

from catalyzing oxidative stress, which compromises TET

activity (Niu et al,

). Moreover, this should restrain NRF2

activation (Figs.

D,F and

–

M) from repressing

Foxp3

expression

and impair T

REG

cell function (Klemm et al,

Consistent with our

ﬁ

ndings, intracellular Fe mobilization via

lysosome-mediated ferritinophagy, was recently shown to regulate
TET-driven (de)methylation of the peroxisome proliferator-
activated receptor

(PPAR

) locus, the master regulators of

adipocyte development (Suzuki et al,

). This suggest that FTH

controls Fe-responsive epigenetic programs de

ﬁ

ning different

cellular developmental programs.

FTH regulates T

REG

cell mitochondrial TCA cycle and OXPHOS

(Figs.

–

I and EV5A

–

E), consistent with similar

ﬁ

ndings in other cell

types (Blankenhaus et al,

; Oexle et al,

1999

). While FTH does not

modulate the intracellular concentration of intermediate TCA cycle
metabolites (Fig. EV5E), including

-ketoglutarate (Fig.

I), it

does appear to modulate the rate of isocitrate conversion into

-ketoglutarate, likely acting irrespectively of

-ketoglutarate genera-

tion via glutaminolysis. It is possible therefore that FTH regulates
the production of this obligatory substrate of TET cytosine
dioxygenases (Kohli and Zhang,

; Pastor et al,

) via regulation

Figure 7.

FTH is required to sustain Foxp3 transcription and expression.

(

) FTH protein detected by western blot in whole-cell extracts from HEK293T cells infected with recombinant lentiviruses coding shRNAs targeting

FTH

(

FTH

429

and

FTH

432

) or control (Ctrl.) recombinant lentiviruses non-targeting shRNA. (

) Mean

ﬂ

uorescence intensity (MFI) of FOXP3 expression, detected by

ﬂ

ow cytometry in human

(CD4

CD45RA

CD25

) T

REG

cells infected with the same recombinant lentiviruses as in (

). Data from

6 samples per experimental group. (

) 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

ﬂ

ow cytometry histogram (middle panel) and quanti

ﬁ

cation of relative GFP expression (right panel), shown as mean

ﬂ

uorescence intensity (MFI). Data from

8 mice per genotype, pooled from two independent experiments with similar trend. (

) Schematic representation of the

experimental approach (left panel) used to monitor Foxp3 expression by

ﬂ

ow cytometry in mouse spleen and MLN (CD4

Foxp3

) T

REG

cells. Representative

ﬂ

ow cytometry

staining of Foxp3 (middle panel). Relative quanti

ﬁ

cation of Foxp3 expression (right panel), represented as mean of

ﬂ

uorescence intensity (MFI). Data from

9 mice per

genotype, pooled from two to three independent experiments with similar trend. (

) 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

ﬂ

ow cytometry of Foxp3 staining (middle panel).

Relative quanti

ﬁ

cation of Foxp3 expression (right panel), shown as mean

ﬂ

uorescence intensity (MFI). Data in (

) from

–

6 mice per genotype, representative of two

independent experiments with similar trend. (

) 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

ﬂ

ow cytometry of GFP (

) staining (middle panel). Relative quanti

ﬁ

cation of GFP expression

(right panel), represented as mean

ﬂ

uorescence intensity (MFI). Data from

10 mice per genotype, pooled from two independent experiments with similar trend. (

)

Schematic representation of the experimental approach (left panel), where (CD4

tdT

) cells were sorted from the spleen and LN for qRT-PCR (

, left panel). Relative

expression of

Foxp3

(right panel). (

)

Gfp, Fth

, and

tdT

mRNA expression normalized to

Arbp0

of cells sorted as in (

). Data in (

) from

–

4 mice per genotype

from one experiment. Data information: Data in (

) are presented as mean ± SD, circles correspond to individual wells and red bars to mean values. Data in (

–

) are

presented as mean ± SD, circles correspond to individual mice and red bars to mean values.

values in (

) were calculated using the Fiedman test with Dunn

’

s multiple

comparison test, in (

–

) using two-way ANOVA with Sidak

’

s multiple comparison test. NS not signi

ﬁ

cant (

> 0.05), *

< 0.05; **

< 0.01; ***

< 0.001;

****

< 0.0001. Source data are available online for this

ﬁ

gure.

The EMBO Journal

Qian Wu et al

1460

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

Fth

fl/fl

Foxp3

GFP-Fth

Fth

fl/fl

Foxp3

GFP-Fth

67,7%

32,6%

***

(%)

Foxp3

CD4

0 10

-2

-3

-2

1,32%

4,5%

CD25

(%)

CD25

Foxp3

***

Fth

fl/fl

Foxp3

GFP-Fth

Fth

fl/fl

Foxp3

GFP-Fth

100

Percent survival

***

Days

iRBCs/

L (x10

)

Foxp3

GFP

Foxp3

GFP-Fth

Days

9/9

1/9

CD4

Foxp3

-GFP

(x10

)

10%

4,86%

Foxp3

GFP

Foxp3

GFP-Fth

Tumor

Size

B16

Melanoma

Spleen

CD4

Foxp3

Pcc

CD4

IFN

10³

IFN

(

x10

)

IFN

(

)

12,6%

21,4%

CD4

IFN

Pcc

Spleen

Foxp3

GFP

Foxp3

GFP-Fth

Foxp3

GFP

Foxp3

GFP-Fth

Foxp3

GFP

Foxp3

GFP-Fth

Survival
Pathogen

0 3

Dis

0 5 10 15 20 25 30

0.5

1.5

2.5

Days

3.5

Disease Incidence

(%)

EAE

MOG

35-55

+CFA

0 5 1015 202530 35

100

Days

Foxp3

-GFP

Foxp3

-GFP (%)

***

13 15 17 19

200

400

600

800

1000

Days

Tumo

ize

(mm

)

IFN

(%)

IL-17A

(%)

Fth

fl/fl

Foxp3

GFP-Fth

(IFN

/IL-17)

CD4

Spinal

Cord

MOG

35-55

(%)

IFN

IL-17A

16%

22%

IFN

IL-17A

18,7%

33,9%

0 10

Fth

fl/fl

Foxp3

GFP-Fth

+CFA

1,4%

Fth

fl/fl

Foxp3

GFP-Fth

Qian Wu et al

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

1461

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

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

-speci

ﬁ

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,

). 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

ﬂ

ammatory ex-T

REG

cells (Figs.

and EV2C

–

E), via a non-cell-

autonomous mechanism that relies on systemic in

ﬂ

ammation

(Fig. EV1D

–

G), this is not associated with the accumulation of

ﬂ

ammatory ex-T

REG

cells in

Foxp3

GFP-Fth

-tdT

vs. control

Foxp3

GFP-tdT

mice (Fig.

–

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.

–

G). One possible interpretation is that FTH is essential not

only to restrain T

REG

cells from transdifferentiating into in

ﬂ

ammatory ex-

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-Fth

mice (Fig. EV2B). We note, however, that

Fth

deletion in

REG

cells does not compromise the T

REG

cell thymic output (Fig.

H) 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,

; Miyao et al,

; Zhou et al,

) together with the increase frequency of ex-

REG

cells among CD45.2

CD4

tdT

cells (Fig.

–

G), contributes

to the observed increase in the methylation of the

Foxp3

locus in

Fth

-deleted T

REG

cells (Fig.

–

G). This does not invalidate

however, that

Fth

deletion acts in a cell-autonomous manner to

decrease

Foxp3

transcription/expression (Figs.

K,L and

–

H),

therefore increasing the frequency at which T

REG

cells transdiffer-

entiate into ex-T

REG

cells (Fig.

–

G).

In contrast with

genetic

deletion of

Tet2

and

Tet3

in T

REG

cells,

which promotes the transdifferentiation of T

REG

cells into

ﬂ

ammatory ex-T

REG

cells and the development of autoimmunity

(Nakatsukasa et al,

; Ohkura et al,

; Yue et al,

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 in

ﬂ

ammatory

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.

–

C), induced by an immunization protocol

leading to low-grade disease severity in control mice (Fig.

C). This

was associated with a higher accumulation of activated T

1 and

17 cells as well as pathogenic IFN

IL-17A

cells (Duhen

et al,

) in the central nervous system (Fig.

D), likely

originating from auto-reactive T

cells and/from ex-T

REG

cells that

lost Foxp3 expression to become in

ﬂ

ammatory and presumably

pathogenic (Bailey-Bucktrout et al,

). 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

ﬁ

ndings, this

was linked to modulation of T

REG

(Gao et al,

2022a

), T

17 (Teh

et al,

), and FT

(Gao et al,

2022b

) cells and was associated

with regulation of DNA demethylation (Gao et al,

2022b

; Teh et al,

). 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.

E), consistent with dysregulation of Fe metabolism promoting

malaria lethality (Ferreira et al,

; Ramos et al,

; Ramos et al,

; Wu et al,

). This was associated with an increase in host-

parasite burden (Fig.

E), 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,

; Kurup et al,

; 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

Figure 8.

FTH expression in T

REG

cells controls the pathologic outcome of experimental immune-driven in

ﬂ

ammatory conditions.

(

) Schematic representation of the induction experimental autoimmune encephalomyelitis (EAE) in response to MOG

35-55

immunization. (

) EAE incidence (percentage)

and (

) EAE severity in MOG

35-55

immunized mice. Data from

–

20 mice per genotype, pooled from three independent experiments, with similar trend. (

)

Experimental approach (right panel), representative

ﬂ

ow cytometry dot plots (middle panel) and corresponding quanti

ﬁ

cation (left panel) of the relative percentage of

activated T

1 (CD3

CD4

IFN-

), T

17 (CD3

CD4

IL-17A

); and double positive IFN-

IL-17A

cells in the spinal cord, 22 days after MOG

35-55

immunization. Data

from

–

4 mice per genotype. (

) Survival (left panel) and number of circulating

Plasmodium chabaudi chabaudi

(

Pcc)

-infected red blood cells (iRBC) per

L of whole

blood (i.e., parasite burden) (right panel).

9 mice per genotype, pooled from two independent experiments, with similar trend. (

) Representative

ﬂ

ow cytometry

dot plot (left panels) and corresponding percentage and cell numbers (right panels) of splenic (CD4

Foxp3-GFP

) T

REG

cells (

) and IFN

CD4

activated T

cells (

7 days after

Pcc

infection. Data from

–

9 mice per genotype, pooled from two independent experiments, with similar trend. (

) Relative tumor (B16-F10-luc2) size,

–

19 days after inoculation (2 × 10

cells). Data from

–

11 mice per genotype, pooled from 3 independent experiments, with similar trend. (

) Representative

ﬂ

cytometry dot plots (left panels) and corresponding percentage (right panels) of live tumor-in

ﬁ

ltrating (CD4

Foxp3

) T

REG

cells (

) and (CD4

Foxp3

CD25

) effector T

cells (

), 3 weeks after tumor inoculation.

7 mice per genotype, pooled from three independent experiments, with similar trend). Data information: Circles in

(

) correspond to individual mice and red bars to mean values. Data in (

) are presented as mean ± SEM. Data in (

, right panel) are presented as mean ± SD.

values in (

), (

, right panel), and (

) were determined using Holm

–

Sidak method (multiple

tests), with alpha

0.05 under the assumption that both genotypes have

similar SEM, in (

) by log-rank (Mantel

–

Cox) test, and in (

) by unpaired

test with Welch

’

s correction. NS not signi

ﬁ

cant (

> 0.05); *

< 0.05; **

< 0.01;

***

< 0.001. Source data are available online for this

ﬁ

gure.

The EMBO Journal

Qian Wu et al

1462

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

host-pathogen burden, a defense strategy known as disease tolerance
(Medzhitov et

al,

;

Soares

al,

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.

H),

consistent with a relative reduction of tumor-in

ﬁ

ltrating T

REG

cells

(Fig.

I) and a more pronounced activation and/or in

ﬁ

ltration of

activated T effector cells (Fig.

J), including CD8

IFN

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,

; Consonni et al,

In conclusion, regulation of intracellular Fe metabolism by FTH

is essential to maintain T

REG

cell lineage and function in vivo,

ﬂ

ecting 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,

) that

control immunological tolerance to the microbiota (Xu et al,

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 in

ﬂ

ammatory 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

Identi

ﬁ

er or catalog

number

Experimental models

Human: HEK293T

ATCC

CRL-3216

™

Mouse: Tumor cells B16-F10-luc2
(B16)

CaliperLS

B16-F10-luc2

Mouse:

B6.C57BL/6 Fth

ﬂ

Lukas Kuhn, ETH,
Switzerland

(Darshan et al,

)

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,

N/A

Blood samples from anonymized
healthy male donors were
obtained in accordance with
guidelines established by the
Sanquin Medical Ethical
Committee.

This paper

Reagent/resource

Reference or
source

Identi

ﬁ

er or catalog

number

Recombinant DNA

psPAX

Addgene

Cat#12260

pMD2.G

Addgene

Cat#12259

pCMV-FTH-3tag3a

This paper

pCMV-FTH

mut

-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-

BV421

BioLegend

H57-597, 109229
(RRID:AB_10933263)

TCR-

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)

Qian Wu et al

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

1463

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

Reagent/resource

Reference or
source

Identi

ﬁ

er or catalog

number

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 Paci

ﬁ

c 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:

FTH

429

;

TRCN0000029429; target
sequence:
GCCTCGGGCTAATTTCCCATA

GPP Web Portal,
Broad Institute

RHS3979-9596837

Human FTH1 shRNA:

FTH

432

;

TRCN0000029432; target
sequence:
CCTGTCCATGTCTTACTACTT

GPP Web Portal,
Broad Institute

RHS3979-9596840

Primers for RT-qPCR, see Table

See Table

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
Scienti

ﬁ

L34957

eBioscience

™

Cell Stimulation

Cocktail

ThermoFisher
Scienti

ﬁ

00-4970-93

Protein Transport Inhibitor
Cocktail

ThermoFisher
Scienti

ﬁ

00-4980

CFSE

ThermoFisher
Scienti

ﬁ

C34554

Solid Phase Reversible
Immobilization (SPRI) beads

Beckman Coulter

Software

Include version where applicable

ImageJ

Schneider
et al,

https://
imagej.nih.gov/ij/

Flowjo

BD Sciences

Version 10.8.1

R Core Team
2014, Vienna,
Austria

Version 3.5.1

Reagent/resource

Reference or
source

Identi

ﬁ

er or catalog

number

FastQC method

Babraham
bioinform.

Version 0.11.5

Python (Linux/UNIX)

Python

Version 2.7.12

Trimmomatic

Bolger et al,

Version 0.36

HISAT2

Kim et al,

Version 2.1.0

HTseq

Anders
et al,

Version 0.6.1p1

DESeq2

Love et al,

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
Scienti

ﬁ

8804-6821-74

NEBNext

Enzymatic Methyl-seq

Kit

New England
Biolabs

E7120

Qubit HS dsDNA kit

ThermoFisher
Scienti

ﬁ

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 speci

ﬁ

c 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) (noti

ﬁ

cation 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,

; Zhou et al,

) with

Fth

ﬂ

mice

(Darshan et al,

). 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 arti

ﬁ

cial chromosome (BAC)

transgene (Chen et al,

; Zhou et al,

). 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

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

The EMBO Journal

Qian Wu et al

1464

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

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

–

peptide (s.c.;

200 µg), emulsi

ﬁ

ed in Complete Freund

’

s Adjuvant (CFA) contain-

ing

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,

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 × 10

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,

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 Tech-

nologies), 50 µg/mL Gentamicin (Life Technologies), and 50 µM
2-Mercaptoethanol (Life Technologies). After trypsin (Life Tech-
nologies) treatment, single-cell suspensions were resuspended in
ice cold calcium-free and magnesium-free HBSS (Life Technolo-
gies). Mice were injected subcutaneously in the right

ﬂ

ank with

2 × 10

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

sacri

ﬁ

ced when TD

≥

20 mm. By the end of each experiment,

tumor clearance (TD

≤

5 mm) was con

ﬁ

rmed 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. Brie

ﬂ

y, 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

ﬂ

ow 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

m) in PBS (3% FBS,

1 mM EDTA), pelleted (300 ×

, 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-

m cell

strainer, centrifuged (300 ×

, 4 °C, 10 min) and resuspended in PBS

containing 3% FBS and 1 mM EDTA.

Cell sorting

Mice were sacri

ﬁ

ced, and LN (i.e., inguinal, brachial, axillary,

mandibular, super

ﬁ

cial 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

cells

(CD11b/B220/

CD8

CD4

Foxp3

CD44

low

CD62L

high

memory/activated

cells

(CD11b/B220/CD8

CD4

Foxp3

CD44

high

CD62L

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

ﬂ

ow 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, TCR

, CD11b,

and CD304 (Nrp1). Intracellular Foxp3 staining was performed
using Foxp3/Transcription Factor Staining Buffer Set. Brie

ﬂ

y, upon

surface staining, cells were

ﬁ

xed, 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

ﬁ

xed and incubated with Fc block,

followed by incubation with antibodies directed against the
following surface markers: CD4, CD62L, CD44, CD3, TCR

. T

REG

cells and ex-T

REG

cells were distinguished based on endogenous

Foxp3

EGFP/icre

and Tomato expression. Follicular T cells were

ﬁ

xed

and incubated with Fc block, followed by incubation with
antibodies against surface markers: CD4, CD3, TCR

, 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

ﬁ

xation

with intracellular staining for the proliferation marker Ki67. Cell
acquisition was performed using a CYTEK Aurora (Cytek
Biosciences)

ﬂ

ow 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 Streptomy-
cin). 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 TCR

. Intracellular Foxp3, IFN

Qian Wu et al

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

1465

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

and IL-17 staining were performed using Foxp3/Transcription
Factor Staining Buffer Set. Brie

ﬂ

y, were

ﬁ

xed upon surface staining

cells, washed with permeabilization buffer, and incubated with
anti-Foxp3, anti-IFN

, and anti-IL-17 antibodies in permeabiliza-

tion buffer. Cell acquisition was performed using BD LSRFortessa
X-20 (BD Biosciences)

ﬂ

ow 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 TCR

. Intracellular

Foxp3, IFN

, IL-17 and IL10 staining were performed using

Foxp3/Transcription Factor Staining Buffer Set. Brie

ﬂ

y, upon

surface staining cells were

ﬁ

xed, washed with permeabilization

buffer, and incubated with anti-Foxp3, anti-IFN

, anti-IL-17 and

anti-IL-10 antibodies in permeabilization buffer. Cell acquisition
was performed using a CYTEK Aurora (Cytek Biosciences)

ﬂ

ow cytometer. Data were analyzed with FlowJo software Version

10.8.1.

Leukocyte staining with

ﬂ

uorescent 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-(tri

ﬂ

uoro-

methoxy) phenylhydrazone; 20

M, 10 min, 37 °C), to eliminate

mitochondrial membrane potential and positive TMRE staining. To
detect intracellular labile Fe

, cells were incubated with FerroFarRed

M; 1 h 37 °C) in RPMI without serum. After incubation with the

ﬂ

uorescent 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 TCR

Foxp3

cells were selected based on endogenous Foxp3

GFP

expression.

Cell acquisition was performed using BD LSRFortessa X-20 (BD
Biosciences)

ﬂ

ow cytometer and data were analyzed using FlowJo

software Version 10.8.1.

Bone marrow transplants

Bone marrow (BM) cells were harvested by

ﬂ

ushing 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 ×

, 10 min, 4 °C) and

ﬁ

ltering (0.22 µm). BM cell suspensions

(1 × 10

/mL in PBS) were incubated with an anti-Thy1.2 mAb

(0.5 µg/mL) and mixed gently (every 15 min) with rabbit comple-
ment (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

ﬁ

ltered (70-

mesh, cell strainer), washed (2x in PBS, 2% FBS and 1× in PBS
without serum), and cell numbers adjusted (10

/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-

ﬁ

cient (

Rag2

−

) female mice, 2 h after

irradiation (600 Gys). Hematopoietic chimerism was monitored by
immunophenotyping, 6 weeks after bone marrow reconstitution
and thereafter.

REG

cell in vivo homeostatic expansion

LN (i.e., super

ﬁ

cial, 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-

m mesh tissue to isolate leukocytes. Cell suspensions were washed

in cold PBS, red blood cells were lyzed (i.e., ammonium chloride),
passed through 40-

m cell strainer and CD4

GFP

TdT

REG

cells

were sorted upon surface marker staining, as described in

“

Cell

sorting

”

. To test T

REG

cell stability in vivo, CD4

GFP

TdT

REG

cells from

Foxp3

GFP-tdT

and

Foxp3

GFP-Fth

-tdT

mice were adoptively

transferred (i.v., 1 × 10

cells) to

Rag2

-/-

mice. LN and spleen were

collected six weeks later, disrupted, and RBC was lysed. The
number of CD4

GFP

TdT

REG

cells and CD4

GFP

TdT

ex-T

REG

cells was quanti

ﬁ

ed by

ﬂ

ow cytometry, as described in

“

Immuno-

phenotyping

”

. An additional aliquot of spleen cells was used to sort

CD4

GFP

TdT

REG

cells, as described in

“

Cell sorting

”

. These

were used to extract mRNA and monitor

Fth

mRNA expression in

the CD4

GFP

TdT

REG

cells used for the adoptive transfer.

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) (Thermo-

ﬁ

sher; 1/1000 in PBS without serum) at RT in the dark for 15 min.

Staining was stopped by adding

ﬁ

ve volumes of complete media

(containing 10% FBS). Sort-puri

ﬁ

ed T

REG

cells were plated and

serially twofold diluted, starting at 2.5 × 10

cells/well in round-

bottom 96-well plates with 1 µg/mL soluble anti-CD3 mAb and
5 × 10

irradiated splenocytes. CTV-labeled T

cells were plated

(2.5 × 10

cells/well), resulting in T

REG

NAIVE

ratio ranging from 1:1

to 1:64. Cultures were set in triplicates in a

ﬁ

nal volume of 200 µL.

On day 3 of culture, CTV intensity was measured in responder
T cells de

ﬁ

ned as Thy1.1

Thy1.2

TCRb

CD4

, live lymphocytes.

In vitro induction of T

REG

cells and

ﬂ

ow 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 TGF

(5 ng/mL) (iT

REG

cell differentiation medium). Alternatively,

naive T cells were cultured in the same media, without TGF

(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

ﬁ

xed 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)

ﬂ

ow 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 × 10

/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.

The EMBO Journal

Qian Wu et al

1466

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

Lentiviral transduction

Lentivirus was produced by transfecting con

ﬂ

uent 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 (Poly-
sciences, 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 puri

ﬁ

ed by ultracentrifugation at 50,000 ×

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 contain-
ing 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

) 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 Scienti

ﬁ

c) 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

ﬁ

lm (Fuji Medical). Sorted cells mouse T

REG

, T

, and

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
quanti

ﬁ

ed 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
Scienti

ﬁ

c). ECL signal was developed using Pierce ECL substrate

kit followed by autoradiographic detection on

ﬁ

lm (Fuji Medical).

Alternatively, western blots were developed using Amersham
Imager 680 (GEHealthcare), equipped with a Peltier-cooled
Fuji

ﬁ

lm 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

ﬁ

rst

strand cDNA synthesis kit (Roche) or Xpert cDNA Synthesis Kit

(GRiSP). Quantitative real-time PCR (qRT-PCR) was performed
using 1

g 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

were

designed using Primer Blast (Ye et al,

Serology

Mice were euthanized using CO

inhalation. Whole blood was

collected by cardiac puncture and transferred into an EDTA for
hemogram analyses or heparin tubes for serology (Iron, Transfer-
rin, and Transferrin saturation). Analysis was performed by
DNAtech (Lisbon).

Histology

Organs were harvested,

ﬁ

xed (10% formalin), embedded in

paraf

ﬁ

n, sectioned (3-

m-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 quanti

ﬁ

cation of

mitochondrial DNA (mtDNA) in comparison to nuclear DNA
(nDNA) using a qRT-PCR-based method (Quiros et al,

Brie

ﬂ

y, 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,

) are listed in Table

. 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 acidi

ﬁ

cation

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. Speci

ﬁ

cally sorted

TREG and TCONV cells were plated on poly-L lysine-coated
Seahorse XF96 plates (15 × 10

cells/well) XF medium (10 mM

glucose, 1 mM sodium pyruvate, 2 mM

-glutamine, pH 7.4), and

centrifuged (400 ×

, for 5 min) to promote cell adhesion. Cells

were incubated in a non-CO

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

M); FCCP (2

M); Antimycin A/Rotenone (1

M); and three

Qian Wu et al

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

1467

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

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.

) 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.

). RNA sequencing data from non-chimeric mice was

analyzed as follows: RNA sequencing was performed using Illumina

’

next-generation sequencing (Bentley et al,

). Quality check and

quanti

ﬁ

cation 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 quanti

ﬁ

cation 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 protocol

. Library preparation,

including cDNA

“

tagmentation

”

, PCR-mediated adaptor addition and

library ampli

ﬁ

cation 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 × 10

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)

. 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,

https://may2021.archive.ensembl.org

). 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

ﬁ

t a

binomial GLM

ﬁ

tting for

i coef

ﬁ

cient and Wald statistics (by

nbinomWaldTest

). Pairwise comparisons were performed with the

function

results

(alpha = 0.05), and the log

fold change for each

pairwise comparison was shrunken with the function

lfcShrink

using

the algorithm

ashr

(v.2.2-47)

. Differentially expressed genes were

considered as genes with an adjusted

value < 0.05 and an absolute

log

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

—

) 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 gpro

ﬁ

ler2 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

”

)

Table 1.

RT-qPCR primers.

Oligonucleotides

Sequences

Reference

Arbp0

Fwd

′

-CTTTGGGCATCACCACGAA-3

′

Blankenhaus et al,

Arbp0

Rev

′

-GCTGGCTCCCACCTTGTCT-3

′

Blankenhaus et al,

Fth

Fwd

′

-CCATCAACCGCCAGATCAAC-3

′

Blankenhaus et al,

Fth

Rev

′

-GCCACATCATCTCGGTCAAA-3

′

Blankenhaus et al,

Ftl

Fwd

’

-AAGATGGGCAACCATCTGAC-3

’

This work

Ftl

Rev

’

-GCCTCCTAGTCGTGCTTGAG-3

’

This work

Hk2

Fwd

′

-GCCAGCCTCTCCTGATTTTAGTGT-3

′

Quiros et al,

Hk2

Rev

′

-GGGAACACAAAAGACCTCTTCTGG-3

′

Quiros et al,

Nd1

Fwd

′

-CTAGCAGAAACAAACCGGGC-3

′

Quiros et al,

Nd1

Rev

′

-CCGGCTGCGTATTCTACGTT-3

′

Quiros et al,

GFP

Fwd

′

-CGACGTAAACGGCCACAAGTTCAG-3

′

Liu et al,

GFP

Rev

′

-CCGTAGGTCAGGGTGGTCACGAG-3

′

Liu et al,

Tdt Fwd

′

-GCCGACATCCCCGATTACAAGA-3

′

Wienert et al,

Tdt Rev

′

-CGATGGTGTAGTCCTCGTTGTGG-3

′

Wienert et al,

Foxp3

Fwd

′

-GGCCCTTCTCCAGGACAGA-3

′

Fontenot et al,

Foxp3

Rev

′

-GCTGATCATGGCTGGGTTGT-3

′

Fontenot et al,

http://www.bioinformatics.babraham.ac.uk/projects/fastqc/

The EMBO Journal

Qian Wu et al

1468

The EMBO Journal

Volume 43 | Issue 8 | April 2024 | 1445

–

1483

Downloaded from https://www.embopress.org on April 22, 2024 from IP 2607:f720:1902:10::389.

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

values. Only

statistically signi

ﬁ

cant (user_threshold=0.05) enriched functions are

returned (signi

ﬁ

cant=TRUE) after multiple testing corrections with

the default method g:SCS (correction_method =

“

analytical

”

Gpro

ﬁ

ler2 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). Gpro

ﬁ

ler2 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;

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
signi

ﬁ

cant differences were de

ﬁ

ned 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 signi

ﬁ

cance. Func-

tional enrichment analysis was performed with gPro

ﬁ

ler (Kolberg et al,

). Data were analyzed using g:SCS multiple testing correction

method with a signi

ﬁ

cance threshold of 0.05.

Targeted metabolomics

Targeted metabolomics analyses of cell extracts for intermediates of the
tricarboxylic acid cycle (TCA cycle; citrate, isocitrate,

-ketoglutarate,

succinate, fumarate, malate, cis-aconitate) and additional metabolites
closely linked to the TCA cycle (pyruvate, lactate, DL-2-hydroxyglu-
tarate, 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,

https://zenodo.org/record/2555454/export/xd

). For

metabolite quanti

ﬁ

cation, 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 quanti

ﬁ

ed 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

ﬁ

ciency as per the manufacturer

’

s protocol. Samples were fragmen-

ted 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 quanti

ﬁ

ed 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)
(

) nf-core (Ewels et al,

). 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,