Dependence on nuclear factor of activated T-cells(NFAT) levels discriminates conventional T cellsfrom Foxp3+ regulatory T cellsMartin Vaetha, Ulrike Schliesserb, Gerd Müllerc, Sonja Reissigd, Kazuki Satohe, Andrea Tuettenberge, Helmut Jonuleite,Ari Waismand, Martin R. Müllerf,g,1, Edgar Serflinga, Birgit S. Sawitzkib, and Friederike Berberich-Siebelta,2

aDepartment of Molecular Pathology, Institute of Pathology, Julius Maximilians-University Würzburg, 97080 Würzburg, Germany; bInstitute ofMedical Immunology, Charité Campus Virchow-Klinikum, Humboldt-University of Berlin, 13353 Berlin, Germany; cDepartment of Tumor Genetics andImmunogenetics, Max-Delbruck-Center for Molecular Medicine, 13092 Berlin, Germany; dInstitute for Molecular Medicine and eDepartment of Dermatology,University Medical Center of the Johannes Gutenberg-University Mainz, 55131 Mainz, Germany; fProgram in Cellular and Molecular Medicine, ImmuneDisease Institute, Children’s Hospital Boston, Boston, MA 02115; and gDepartment of Pathology, Harvard Medical School, Boston, MA 02115

Edited by Anjana Rao, La Jolla Institute for Allergy and Immunology, La Jolla, CA, and approved August 13, 2012 (received for review March 6, 2012)

Several lines of evidence suggest nuclear factor of activated T-cells(NFAT) to control regulatory T cells: thymus-derived naturally oc-curring regulatory T cells (nTreg) depend on calcium signals, theFoxp3 gene harbors several NFAT binding sites, and the Foxp3(Fork head box P3) protein interacts with NFAT. Therefore, we in-vestigated the impact of NFAT on Foxp3 expression. Indeed, thegeneration of peripherally induced Treg (iTreg) by TGF-β washighly dependent on NFAT expression because the ability ofCD4+ T cells to differentiate into iTreg diminished markedly withthe number of NFAT family members missing. It can be concludedthat the expression of Foxp3 in TGF-β–induced iTreg depends onthe threshold value of NFAT rather than on an individual memberpresent. This is specific for iTreg development, because frequencyof nTreg remained unaltered in mice lacking NFAT1, NFAT2, orNFAT4 alone or in combination. Different from expectation, how-ever, the function of both nTreg and iTreg was independent onrobust NFAT levels, reflected by less nuclear NFAT in nTreg andiTreg. Accordingly, absence of one or two NFAT members did notalter suppressor activity in vitro or during colitis and transplanta-tion in vivo. This scenario emphasizes an inhibition of high NFATactivity as treatment for autoimmune diseases and in transplanta-tion, selectively targeting the proinflammatory conventional T cells,while keeping Treg functional.

gene regulation | tolerance | autoimmunity

Regulatory T cells (Treg) are T lymphocytes specialized forimmune suppression. They are necessary to maintain immune

homeostasis and to prevent autoimmune diseases. Treg are iden-tified by the expression of CD4, CD25, and the key transcriptionalregulator Foxp3 (Fork head box P3). Within the CD4+ com-partment Treg are represented as a heterogeneous population ofthymus-derived (naturally occurring or nTreg) and various pe-ripherally induced Treg (iTreg). Differentiation of nTreg requireshigh-affinity T cell receptor (TCR) signals as well as costimulatorysignals, both provided by thymic medullary epithelial cells. Incontrast, adaptive or iTreg are generated from conventional naïveCD4+ T cells (Tconv) in peripheral tissue. This can be mim-icked in culture by TCR (and coreceptor) engagement in thepresence of the cytokines TGF-β and IL-2 (1).Foxp3 is crucial for nTreg function (2). Mice and humans with

mutations in the Foxp3 gene suffer from a severe autoimmunedisorder known as scurfy or IPEX (immune dysregulation, poly-endocrinopathy, enteropathy, X-linked) syndrome, which mani-fests in lymphoproliferation, multiorgan lymphocytic infiltration,and systemic autoimmune inflammation. It can be prevented bythe adoptive transfer of CD4+CD25+ T cells. Foxp3 binds DNAthrough a winged helix-forkhead DNA binding domain and func-tions as a transcriptional activator/repressor by recruiting deace-tylases as well as histone acetyltransferases (3). In addition, severaltranscription factors, including nuclear factor of activated T-cells(NFAT), NF-κB (nuclear factor “kappa-light-chain-enhancer” of

activated B-cells), and Runx1/AML1 (runt-related transcriptionfactor1/acute myeloid leukemia1) have been identified as inter-action partners of Foxp3 (4–6). Interestingly, all three transcrip-tion factors have also been reported to regulate Foxp3 expression.Recently, several studies have demonstrated the importance of

the NF-κB family member c-Rel for thymic Foxp3 induction (7).c-Rel binds directly to the Foxp3 locus, thereby initiating chro-matin opening at a newly identified cis-regulatory element (CNS3)(8), concomitantly binding to further enhancer regions and theFoxp3 promoter (9).Accumulating evidence has pointed to a role of NFAT in Treg,

because the necessity of Ca2+ signals in nTreg development andfunction was emphasized (10, 11). TCR-initiated Ca2+ influx andsubsequent calmodulin/calcineurin activation is central for thetranslocation of NFAT transcription factors to the nucleus, wherethey bind to regulatory regions of numerous genes (12), includingat least one cis-regulatory element of Foxp3, namely CNS1 (13).The NFAT family comprises four calcium-regulated members:NFAT1/NFATc2, NFAT2/NFATc1, NFAT3/NFATc4, and NFAT4/NFATc3, with NFAT1, -2, and -4 being predominant in T cells.In contrast to the established role of Ca2+, previous data

revealed that nTreg in mice deficient for NFAT1 plus NFAT4were neither decreased in number nor impaired in their suppres-sive capacity (14). Therefore, it was concluded that NFAT2 mightbe the important family member for controlling nTreg de-velopment and/or function (10). Indeed, former analyses elicitedmRNA encoding the long isoforms of NFAT2 being present inperipheral CD4+ as well as CD4+CD25+ T cells. However, theactivation-induced NFAT2/αA was missing in CD4+CD25+ nTreg,in line with less overall NFAT2 protein in nTreg (15).In this study, we addressed the role of individual NFATs in

nTreg and iTreg development and function. We analyzed andcompared Treg from mice lacking only NFAT1, NFAT2, orNFAT4, as well as NFAT1 plus NFAT2 or NFAT1 plus NFAT4,in T cells. We found that decreased NFAT activity progressivelyimpaired Foxp3 induction in TGF-β–induced iTreg. Neverthe-less, in accordance with most Treg-associated markers beingstationary, NFAT-deficient iTreg as well as nTreg were fully

Author contributions: M.V., A.T., H.J., A.W., B.S.S., and F.B.-S. designed research; M.V.,U.S., G.M., S.R., K.S., E.S., and F.B.-S. performed research; G.M. and M.R.M. contributednew reagents/analytic tools; M.V., U.S., G.M., S.R., K.S., H.J., A.W., B.S.S., and F.B.-S. ana-lyzed data; and M.V. and F.B.-S. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: The microarray data reported in this paper have been deposited in theArrayExpress database, www.ebi.ac.uk/arrayexpress (accession no. E-MEXP-3687).1Present address: Department of Hematology, Oncology, and Immunology, UniversityClinic, 72076 Tübingen, Germany.

2To whom correspondence should be addressed. E-mail: path230@mail.uni-wuerzburg.de.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1203870109/-/DCSupplemental.

16258–16263 | PNAS | October 2, 2012 | vol. 109 | no. 40 www.pnas.org/cgi/doi/10.1073/pnas.1203870109

suppressive, suggesting that high levels of NFAT activity arenot required for their regulatory function. In accordance, humanand murine Treg have lower levels of both nuclear and cyto-plasmic NFAT than conventional CD4+ T cells. This designatesspecific NFAT-directed drugs as potent therapy in autoimmunedisease and transplantation.

ResultsNFAT Is Required for Foxp3 Expression in iTreg. NFAT2 wasreported to bind CNS1 of Foxp3 (13) [i.e., to an element that iscrucial for iTreg generation in gut-associated lymphoid tissues(8)]. Here, we analyzed the dependence of Foxp3 expression onNFAT2 in comparison with NFAT1 and -4. The offspring ofNfat2fl/fl mice were crossed with Cd4-cre mice (16) (Fig. S1 A–C).Together with anti-CD3/28 and IL-2, TGF-β induced robustFoxp3 expression in WT CD4+CD25− T cells, whereas inductionwas moderately diminished in the absence of NFAT2 and espe-cially when both NFAT1 and NFAT2 were missing (Fig. 1 A and Band Fig. S1D). Analysis of NFAT1 single- and NFAT1NFAT4double-deficient CD4+CD25− T cells yielded a similar result.Whereas lack of one family member led to some reduction ofFoxp3-expressing cells, loss of two members almost abrogated

iTreg induction (Fig. 1 C and D). Pharmacological inhibition ofall NFAT members by cyclosporine A (CsA) totally blockedFoxp3 induction in naïve CD4+ T cells during stimulation withanti-CD3/28 plus TGF-β/IL-2 (Fig. S2A). The difference be-tween WT and NFAT-deficient T cells was more pronouncedunder suboptimal doses of TGF-β (Fig. 1D and Fig. S2B).Skewing the same naïve CD4+ T cells toward T helper cellsubtypes demonstrated a similar dependence on the level ofNFAT for IFN-γ and IL-17, as well as a dominant influence ofNFAT2 on IL-17 expression (Fig. S2C).CD4 is first expressed at the CD4+CD8+ double-positive stage

of thymocytes, presumably ahead of Treg development. There-fore, Nfat2fl/fl × Cd4-cre created a thymocyte/T cell-specific Nfat2knockout (Fig. S1D) and enabled us to analyze nTreg developmentin the absence of NFAT2. However, the frequency of Foxp3+CD25+ nTreg among the CD4+ cell population in thymus, spleen,and lymph nodes (LN) were not affected by deficiency of any—alone or in combination—NFAT member (Fig. S3 A–F). That isnot due to lack of NFAT expression, because all three NFATmembers are similarly expressed in Tconv and Treg, even thoughup-regulation of NFAT2 does not occur in nTreg and only mar-ginally in iTreg (Fig. S3G). In summary, whereas nTreg developirrespective of NFAT expression, iTreg crucially rely on highNFAT levels with permissiveness for individual family members.

NFAT Influences Foxp3 Directly During iTreg Differentiation. Toelucidate whether NFAT2 was capable to bind to the regulatoryelements of Foxp3, we stimulated naïve CD4+CD25− T cells inthe presence or absence of TGF-β for 20 h, and performed ChIPassays. Specific binding of NFAT2 was observed at CNS1 in cellsstimulated in the presence of TGF-β, whereas Smad3 bound toCNS1 and -3 in cells just beginning to express Foxp3 (Fig. 1E).Electromobility shift assays (EMSA) with extracts from human ormurine T cells demonstrated some binding to their respectiveFoxp3 promoters. However, mobility of those complexes wasatypical, and unlabeled Foxp3 promoter probe could notcompete for NFAT binding to CNS1, whereas anti-NFAT1 oranti-NFAT2 supershifts were only found at CNS1 (Fig. S3 H–J).This indicates a specific, CNS1-directed, and cell-intrinsic in-fluence of NFAT on the Foxp3 locus during TGF-β–stimulatediTreg differentiation. To further investigate this point, we mixedcongenic WT CD4+CD90.1+ T cells with CD4+CD90.2+ T cellsfromWT, Nfat2fl/f× Cd4-cre, or Nfat1−/−× Nfat2fl/fl× Cd4-cremiceand induced Foxp3. Whereas NFAT-deficient CD4+CD90.2+

T cells showed reduced Foxp3 expression, congenic WT CD4+CD90.1+ T cells from the same TGF-β cultures remained un-affected (Fig. 1F). Therefore, iTreg differentiation by TGF-β ishighly dependent on the level of NFAT, from which at leastNFAT2 binds to CNS1 in vivo and seems to transactivate theFoxp3 gene for iTreg induction.

NFAT Is Essential for iTreg Induction in Vivo. Induction of iTregoccurs primarily in gut-associated lymphoid tissues, where iTregbalance Th17-driven immune responses. To explore whether NFATdeficiency also impaired induction of iTreg in vivo, we first analyzedHelios expression in Foxp3+ T cells from mesenteric LNs(mLN). This allowed us to distinguish between nTreg, whichare Foxp3+Helios+, and iTreg, which are Foxp3+ but Helios−

(17), in untreated mice lacking NFAT2 and NFAT1 plus NFAT2in T cells. The data revealed that in vivo (i) the percentage ofFoxp3+Helios+ nTreg remained unaffected by the deficiency ofNFAT2 or NFAT1 plus NFAT2, but (ii) the NFAT1 plus NFAT2-deficient Foxp3+ Helios− iTreg were clearly reduced (Fig. 2A).Second, we addressed iTreg differentiation in a model of

murine colitis by transfer of naïve CD4+ T cells to lymphopenicrecipients (18). CD4+CD62L+ but CD25− T cells in a 1:1 mix-ture of WT CD90.1+ (to ensure disease onset) and CD90.2+ WTor Nfat2fl/fl × Cd4cre were delivered into Rag1−/− mice (Fig. S4A).Disease was monitored by colon miniendoscopy. In both casesthe transfer caused colitis (Fig. S4 B–D), whereas naïve CD90.2+

NFAT2-deficient CD4+ T cells were significantly disabled to

wild ty

pe

Nfat2fl/f

l x Cd4

-cre

IL-2

IL-2 & IL-10

IL-2 & TGFβ

A

Foxp3

CD

25

Cwild

type

Nfat1–/–

Nfat1–/–

x

Nfat

4–/–

Foxp3

CD

25

B

F

0

20

40

60

80

% C

D4+

CD

25+F

oxp3

+ c

ells

Foxp3

CD

90.2

wild type73%

71%

48%

63%

31%

68%

IL-2

IL-2 & IL-10

IL-2 & TGFβ

1.7 1.9 1.6

1.5 2.0 1.7

58.1 47.6 6.9

1.8 1.4 4.8

2.2 1.8 2.2

31.3 21.5 5.8

p < 0.0001

p = 0.0002

p < 0.0001

D

CNS3–

CNS1–

input isotype NFAT2 Smad3

20h anti-CD3/CD28

TGFβIL-2

– + – + – + – ++ + + + + + + +

wild ty

pe

Nfat2fl/f

l x Cd4

-cre

Nfat1–/–

x

Nfatc2

fl/fl Cd4

-cre

E

wild type

Nfat4–/–

IL-2 only2.5 ng/ml

TGFβ1.25 ng/ml

TGFβ0.63 ng/ml

TGFβ

1.6

2.5

86.9

74.9

67.7

35.9

27.6

5.6

Foxp3

CD

25

Fig. 1. Impaired induction of Foxp3 by TGF-β in NFAT-deficientCD4+CD25−Tconv. (A and B) CD4+CD25− T cells from WT, Nfat2fl/fl × Cd4-cre and Nfat1−/− × Nfat2fl/fl × Cd4-cre mice were stimulated for 3 d withplate-bound anti-CD3/28 in the presence of IL-2 only, IL-2 plus IL10, or IL-2plus TGF-β followed by rest of 4 d. (A) Representative FACS analysis ofCD4+CD25+Foxp3+ T cells and (B) summary of at least eight independent IL-2/TGF-β cultures. (C) CD4+CD25− T cells from WT, Nfat1−/−, and Nfat1−/− ×Nfat4−/− mice as in A. (D) CD4+CD25– T cells from WT and Nfat4−/−mice as inA, but with different concentrations of TGF-β. (E) ChIP assay (one out ofthree) of NFAT2 and Smad3 binding to the Foxp3-enhancer elements CNS1and CNS3 from twenty-hour stimulated CD4+CD25− T cells. (F) CD90.2+

CD4+CD25− T cells from WT, Nfat2fl/fl × Cd4-cre and Nfat2fl/fl × Cd4-cre ×Nfat1−/−mice were mixed 1:1 with congenic (CD90.1+) T cells and processed asin A. Frequency of CD4+Foxp3+ T cells among genotypes (CD90.1 vs.CD90.2) is denoted.

Vaeth et al. PNAS | October 2, 2012 | vol. 109 | no. 40 | 16259

IMMUNOLO

GY

develop into Foxp3+ iTreg both in spleen and mLN (Fig. 2 B–D).When cells from Nfat1−/− × Nfat2fl/fl × Cd4cre were transferred, weobserved even less iTreg in spleen and mLN, associated with anenhanced severity of colitis (Fig. 2 E–G). However, effectorfunction of conventional cells (e.g., cytokine production derivedfrom the identical naïve CD4+ T-cell pool) was additionallyimpaired (Fig. S2D). This probably masked the significance ofdiminished iTreg differentiation as seen in CNS deficiency (19).

NFAT Is Dispensable for iTreg Function in Vivo and in Vitro. An al-tered repressor capacity of iTreg cannot be monitored, when theirnumbers are changed during colitis and Tconv function is addi-tionally affected. An in vivo readout for the suppressive capacityof Treg is allograft acceptance after adoptive transfer of Tconvtogether with defined numbers of Treg. iTreg specific for alloan-tigens can be induced in vivo—and even in thymus-ectomizedmice, excluding nTreg involvement—by injection of the non-depleting anti-CD4 antibody YTS177.9, together with a donor-specific blood transfusion (DST) of allogeneic blood as antigen(Fig. 3A) (20, 21). DST pretreatment in WT vs. Nfat2fl/fl × Cd4-cre mice resulted in a different total yield of CD4+CD25+ Tregper spleen to less than half of Treg in NFAT-deficient comparedwith WT mice, once more indicating the dependence of iTreggeneration on NFAT expression in vivo (Fig. 3B). To evaluate

the suppressive function of this remaining iTreg population,Rag2−/− mice were injected with 2 × 105 WT or NFAT2-deficientCD4+CD25+ T cells from DST-pretreated mice together withCD45RBhi Tconv. One day after cell transfer, the mice receiveda BALB/c skin transplant. Both WT and NFAT2-deficient iTregguaranteed a clear increase in graft survival (Fig. 3C), dem-onstrating no obvious impairment in suppressor function ofNFAT2-deficient iTreg.To ensure purity of iTreg and avoiding contaminating nTreg

for functional tests we generated TGF-β–induced iTreg fromnaïve CD4+ T cells in vitro. Classic suppression assays dem-onstrated equal capability of WT and NFAT2-deficient iTreg(Fig. S5A). When allo-specific iTreg were generated in vitro—hindering differentiation upon NFAT2 deficiency—and trans-ferred in adjusted numbers, NFAT2−/− iTreg were at least asfunctional as WT iTreg in the model of skin transplantation (Fig. S5B–E). This verified the functional equivalence of in vivo generatedWT and Nfat2−/− iTreg (Fig. 3).

nTreg Function Independently of Individual NFAT Members. If iTregoperate with markedly reduced NFAT levels, the same mightapply to nTreg. NFAT1−/− plus NFAT4−/− nTreg are functionalin vitro (14). We tested whether NFAT2 is necessary for therepressor function of nTreg. Nonfunctional nTreg would leadto spontaneous autoimmunity, but peripheral lymphoid organsin 6-mo-old Nfat2fl/fl × Cd4-cre and Nfat1−/− × Nfat2fl/fl × Cd4cremice were without any pathological findings (Fig. S6 A–E). Toexclude that frequency and number of nTreg had recovered withaging (22, 23), we analyzed Treg development in neonates. Theratio of both thymic and splenic CD4 vs. CD8 exhibited somepeculiarities, but frequency of Foxp3+ T cells was unalteredupon deficiency of NFAT1 plus NFAT2 (Fig. S7). Furthermore,in vitro suppression assays revealed no functional dependence

A

B

5.3 8.8 4.4 9.2 2.8 8.4

Helios

Fox

p3

Cp = 0.0472

% C

D4+

CD

90.2

+F

oxp3

+

D

Foxp3

CD

90.1

Cd90.1+ WT

Cd90.2+ WT

Cd90.1+ WT

Nfat2fl/fl x Cd4-cre

3.7

6.6

3.2

1.2

7.9

10.6

7.5

4.9

0

2

4

6

8

0

2

4

6

8

0

5

10

15

0

2

4

6 to

tal C

D90

.2+ iT

regs

(x1

04)

% C

D4+

CD

90.2

+F

oxp3

+ to

tal C

D90

.2+ iT

regs

(x1

04)

p = 0.0045

p = 0.0171

Spleen mLN

Cd90.1+ WT

Cd90.2+ WT

Cd90.1+ WT

Nfat1–/– x Nfat2fl/fl x Cd4-cre

mLN

Foxp3

CD

90.1

11.2

8.3

9.2

1.4

E FG

wild type

Nfat2fl/fl x Cd4-cre

% C

D4+

CD

90.2

+F

oxp3

+ to

tal C

D90

.2+ iT

regs

(x1

05)

0

2

4

6

8

10

0

2

4

6

wild type

Nfat1–/– x Nfat2fl/fl x Cd4-cre

p = 0.0178

p = 0.0088

0

2

4

6

8

10

12

14

wild ty

pe

Nfat1

–/– x

Nfat2

fl/fl x

Cd4-c

re

mea

n cl

inic

al c

oliti

s sc

ore

Spleen mLN

p = 0.1335

Fig. 2. Impaired iTreg generation in NFAT-deficient T cells in vivo. (A) Rep-resentative FACS analysis of Foxp3+Helios− iTreg and Foxp3+Helios+ nTreg inmesenteric LN of untreated WT, Nfat2fl/fl × Cd4-cre, and Nfat1−/− × Nfat2fl/fl

× Cd4-cre mice. (B–D) 2.5 × 105 Cd90.1+ (WT) and 2.5 × 105 Cd90.2+ (WT orNfat2fl/fl × Cd4cre) CD4+CD62L+CD25− T cells were injected into Rag1−/− mice.The frequency of Foxp3+ iTreg was impaired in spleen (B, Left and C) andmesenteric LN (B, Right and D) after 6 wk, analyzed by FACS (B) and total cellnumber of CD90.2+Foxp3+ iTreg per organ compiled from five different mice(C and D). (E and F) As in B and D, but WT compared with Nfat1−/− × Nfat2fl/fl

× Cd4-cre T cells. (G) Mean clinical score of colitis by colon endoscopy frommice in E; each dot represents one individual mouse. The difference is notstatistically significant.

C57/BL6 Rag2–/– C57/BL6recipient

skin graftsurvival

CD45RBhigh T cells

BALB/c skin allograft

CD25+ iTreg

wild typeor

Nfat2fl/fl x Cd4-cre

day –28 day 0

BALB/c DST(donor specific

blood transfusion)

anti-CD4 blocking Ab

(YTS177)

A

B

0.0

0.4

0.8

1.2

Yie

ld o

f CD

4+C

D25

+ iT

reg

per

mou

se (

x106

)

p = 0.0119

0 20 40 60 80 1000

20

40

60

80

100

Tconv only

wild type iTreg

Nfat2fl/fl x Cd4-cre iTreg

C

graf

t sur

viva

l in

%

days

wild type Nfat2fl/fl x Cd4-cre

Fig. 3. Reduced iTreg generation but normal suppressive capacity in NFAT-deficient cells in vivo. (A) Schematic overview of adoptive transfer of B6WT or Nfat2fl/fl × Cd4-cre in vivo-induced iTreg along with CD4+CD45RBhi

cells into a BL6 Rag2−/− recipient mouse, receiving a BALB/c skin allograft.Induction of CD4+CD25+ iTreg by BALB/c DST along with anti-CD4 YTS177.9in vivo, 28 d before adoptive transfer. (B) Yield of CD4+CD25+ Treg fromDST/YTS177.9-treated WT or Nfat2fl/fl × Cd4-cre mice by FACS sorting (n > 5).(C) 2 × 105 Treg from WT or Nfat2fl/fl × Cd4-cre mice along with 2 × 105

CD4+CD45RBhi T cells were injected i.v., and BALB/c skin allograft survival inthe Rag2−/− recipient mouse was measured by log–rank test. All mice re-ceiving only CD4+CD45RBhi cells acutely rejected their skin transplants [n = 5,mean survival time (MST) = 12.8 d]. Addition of WT CD4+CD25+ Treg orNfat2fl/fl × Cd4-cre Treg could prolong skin graft survival to the same extend(WT: n = 5, MST = 64.0 d; Nfat2fl/fl × Cd4-cre: n = 5, MST = 64.8 d).

16260 | www.pnas.org/cgi/doi/10.1073/pnas.1203870109 Vaeth et al.

on NFAT2 (Fig. S8A). nTreg can be activated and expandedby superagonistic CD28-specific mAb (CD28SA; D665) in mice(24, 25). Injecting CD28SA into mice with Treg-specific in-activation of NFAT2 (Nfat2fl/fl × FIC; Fig. S1E) evoked compa-rable expansion of nTreg to WT mice after 3 d (Fig. S8B). Toaddress the functional capacity of NFAT2-deficient nTreg in vivo,serum levels of IL-2 were measured after 3 h of CD28SA in-jection. If Treg were functionally impaired, an increase in IL-2production should be evident. Indeed, when Treg are completelydepleted, 10 times more IL-2 is secreted (25, 26). The similar lowamount of IL-2 produced from suppressed effector T cells inboth genotypes of mice revealed functional equivalence betweenWT and NFAT2-deficient nTreg (Fig. S8C).For therapeutic interest, acceptance after adoptive transfer of

Tconv together with nTreg was evaluated. We transferred CD25hinTreg from WT or Nfat2fl/fl × Cd4-cre together with CD45RBhi

Tconv from WT animals into Rag2−/− mice (cells and mice C57/BL6) that received a skin graft from allogeneic BALB/c mice(Fig. S8D). nTreg from NFAT2-deficient mice were as efficientas WT nTreg to ensure survival of skin allografts (Fig. S8E).In summary, nTreg lacking NFAT2 (or NFAT1 plus NFAT2and NFAT1 plus NFAT4) do not show any impairment in theirsuppressor function.To address the question of overall functional equivalence of

NFAT-deficient nTreg or iTreg, we analyzed surface markersassociated with suppressor function. Subtle, but no major dif-ferences could be observed ex vivo or after 3 d of stimulation,whereas CD25 was reduced to levels of activated Tconv exclu-sively on NFAT1 plus NFAT2-deficient iTreg after 4 d (Fig. S9A–C). Unbiased microarray experiments were carried out withRNA from nTreg and Tconv +/− TGF-β (Fig. S9D). Microarray

data are available under the accession number E-MEXP-12345on the ArrayExpress database, www.ebi.ac.uk/arrayexpress. Toevaluate direct target genes cells had only been stimulated for24 h. Evaluation focused on genes potentially regulated by NFAT:Foxp3-complexes (6, 27) and revealed a rather mild influence ofNFAT1 plus -2 double-deficiency, although IL-2 expression wasstrikingly diminished (Fig. S9E).

Impaired Nuclear Translocation of NFAT1 and NFAT2 in Treg. Givenour finding that the function of Foxp3+ Treg does not requirehigh levels of NFAT activity, we compared NFAT expressionand activation in Tconv vs. Treg. Our previous data from murinenTreg (15) (Fig. S3G) were extended to human CD4+ Tconv andCD4+CD25+ nTreg, which were analyzed for nuclear and cy-toplasmic proteins. Two isoforms of human FOXP3 (28) weredetected in the nucleus of isolated nTreg, but also to a limitedextent in Tconv (Fig. 4A). The strong staining for Galactin-10verified the identity of human nTreg (29). In Tconv NFAT pro-teins were both cytoplasmic and nuclear, where stimulation pro-moted expression and nuclear transportation. However, the overalllevels of both nuclear and cytoplasmic NFAT1 and especiallyNFAT2 were strongly diminished in human nTreg comparedwith Tconv. Analyzing murine Foxp3+ Treg, the overall levelsof both nuclear and cytoplasmic NFAT1 and NFAT2 proteinsappeared less in both types of Treg compared with Tconv (Fig. 4B and C). Quantitation demonstrated reduced nucleo-cytoplasmicratios of NFAT2 in CD4+CD25+ compared with CD4+ T cells,as well as less than 20% Foxp3+ nTreg and only 25–30% Foxp3+iTreg positive with robust nuclear NFAT2 (Fig. S10 A–D). Per-forming a time course revealed that indeed nuclear translocationof NFAT2 is less in nTreg (Fig. S10 E and F), reflected by slightly

w/o

TGFβ

CD4+CD25–

Tconv

CD4+CD25+

nTreg

CD4+CD25–

Tconv

CD4+CD25–

iTreg

NFAT1 Foxp3 DAPI

BA

CD4+CD25–

Tconv

CD4+CD25–

Tconv

CD4+CD25+

nTreg

CD4+CD25+

nTreg

NFAT2

NFAT1

Foxp3

Gal-10

loading

anti-CD3/28 0 24 48h 0 24 48h 0 24 48h 0 24 48h

nuclear fraction cytoplasmic fraction

ABC

FLΔ2

C

CD4+CD25–

Tconv

CD4+CD25–

iTreg

CD4+CD25+

nTreg

anti-CD3/28

anti-CD3/28+ CsA

anti-CD3/28

anti-CD3/28+ CsA

anti-CD3/28

anti-CD3/28+ CsA

NFAT2 Foxp3 CD3ε DAPINFAT2CD3ε

NFAT2DAPI

Fig. 4. Impaired nuclear translocation of NFAT inFoxp3+ iTreg and nTreg. (A) Immunoblot analysisof NFAT2 and NFAT1 in nuclear (Left) and cytosolic(Right) fractions of human CD4+CD25− Tconv andCD4+CD25+ nTreg. Cells were left unstimulated (0 h)or stimulated for 24 h or 48 h with anti-CD3/28before lysis. Arrows indicate isoforms of NFAT2 and(human) Foxp3, as loading control laminin (Left)and actin (Right) is given. (B) Three-color staining ofNFAT1 (red), Foxp3 (yellow), and chromatin (cyan)of freshly isolated CD4+ Tconv and CD4+CD25+ nTregstimulated for 6 h with anti-CD3/28 (Upper). Foxp3−

Tconv and TGF-β–induced Foxp3+ iTreg generatedfrom CD4+CD25− Tconv after 24 h stimulation withanti-CD3/28 (Lower). (C) Four-color staining of NFAT2(red), Foxp3 (yellow), CD3ε (green), and chromatin(cyan) of CD4+CD25− Tconv, CD4+CD25− TGF-β–induced iTreg, and CD4+CD25+ nTreg. Cells werestimulated with anti-CD3/28 in absence (Tconv) orpresence (iTreg) of TGF-β and IL-2 for 3 d, followedby 4 d resting. Restimulation for 6 h with anti-CD3/28.As a negative control, to avoid nuclear translocationof NFAT2, 10 nM CsA was used.

Vaeth et al. PNAS | October 2, 2012 | vol. 109 | no. 40 | 16261

IMMUNOLO

GY

less calcium flux (Fig. S10G) (30). Taken together, the low amountof NFAT present in nTreg and iTreg resembles their functionalintegrity upon loss of one or two NFAT family members.

DiscussionThe data presented here document dependency of iTreg onNFAT transcription factors to develop and especially to expressFoxp3. However, once differentiated, in vitro or in vivo, they canexert their suppressor function with severely reduced levels ofNFAT. Accordingly, Foxp3+ iTreg express and activate less NFATcompared with peripheral CD4+ Tconv. The reduced level andactivity of NFAT is common to both nTreg and iTreg from miceand men and is consistent with our finding that also nTreg do notprimarily depend on NFAT expression for suppressor function.TCR engagement is mandatory for all differentiation events

of CD4+ T cells toward lineages or subsets. It starts numeroussignaling cascades, including calcium mobilization and subsequentnuclear translocation of NFAT. Together with cytokine-inducedand other transcription factors, NFAT forms complexes at sev-eral loci of lineage-determining target genes encoding cytokines,cytokine receptors, and key regulatory transcription factors. Ina murine thymoma cell line, cooperative binding of NFAT andSmad3 to the enhancer1/CNS1 upon anti-CD3/28 plus TGF-βtreatment was demonstrated (13). This is in line with histoneacetylation (13) and demethylation of CpG residues (31) at thissite upon activation of CD4+ T cells in the presence of TGF-β.To address the in vivo situation, we tested different NFAT-

deficient mouse lines for their ability to develop Foxp3+ Treg.Our data prove that peripheral CD4+ T cells are dependent onNFAT for Foxp3 induction in response to TGF-β. It is mediatedby NFAT binding to enhancer1/CNS1 but not to CNS3 or thepromoter of Foxp3. This supports the identification of CNS1 asthe cis-regulatory element responsible for Foxp3 induction inperipheral T cells (8). Therefore, promoter occupancy of NFATis a distinct feature of human effector T cells (32), which—unlike murine Tconv—transiently express some FOXP3 afterTCR stimulation.CNS3 has recently been recognized as being initially re-

sponsible for Foxp3 expression in nTreg via the recruitment of ac-Rel enhanceosome (8, 33). This might explain why the severereduction in NFAT level does not hamper the development ofnTreg, but not why their development is crucially disturbed, whenCa2+ flux or calcineurin are blocked, which are involved in NFATactivation (10). Unexpectedly, Nfat1−/− and Nfat1−/− × Nfat4−/−

mice exhibited normal nTreg development (14). Likewise, usingnewly created conditional NFAT2-deficient mice, we demonstratehere that NFAT2 deficiency and even NFAT1 plus NFAT2double-deficiency leaves thymic development of nTreg untouched.We conclude that no individual NFAT member is necessary forFoxp3 induction in nTreg, and even deletion of two out of thethree NFAT proteins has no discernible effect. Of note, the datagathered with EL-4 cells (13) do not resemble the thymic situa-tion. In line with a high dependency on NFAT for the generationof iTreg, but not nTreg, a mouse strain with hyperactivatableNFAT1 gives rise to more iTreg—and even less nTreg (34). Theapparent contradiction between impaired nTreg developmentin Ca2+ signaling-deficient mice vs. the normal nTreg in NFAT-deficient mice can be attributed to two, not mutually exclusive,hypotheses. First, in thymocytes the threshold level of NFATactivity necessary for Foxp3 expression may be very low: inNfat1−/− × Nfat4−/− the constitutively expressed long isoforms ofNFAT2 (35) or in NFAT1 plus NFAT2 doubly-deficient T cellsthe amount of NFAT4 could be sufficient to support Foxp3 ex-pression. This is at least in sharp contrast to the periphery, wherethe loss of two NFAT members abolishes Foxp3 induction. Sec-ond, additional pathways such as for AP1 and NF-κB inductionare Ca2+/calcineurin-dependent (36). For example, calcineurincontrols the formation of the CARMA1/BCL10/MALT1 complexduring TCR-induced NF-κB activation (37, 38), whereas math-ematical calculations provide evidence that NF-κB is activatedat lower Ca2+ oscillation frequencies than NFAT (39). This

is consistent with the observation that severely reduced calcium fluxas observed in Stim1fl/fl × Cd4-cre allows regular developmentof nTreg, albeit Stim1/2 double-deficiency conducting analmost complete block of calcium flux abrogates thymic Tregdevelopment (40). Furthermore, development of nTreg—incontrast to iTreg—requires high-affinity/avidity TCR interactionsin conjunction with costimulatory signals through CD28. In com-parison, iTreg differentiate under suboptimal TCR stimulationwithout the necessity of CD28 costimulation (41). Most likely,this leads to activation of NFAT, but less NF-κB, reminiscent ofanergy induction in peripheral CD4+ T cells (42) and in agreementwith a crucial impact of NFAT on peripheral iTreg differentiation.In thymus, however, Ca2+/calcineurin might result in the essentialc-Rel/NF-κB activation.Our data reveal reduced NFAT activity in established iTreg

and nTreg, which has been suggested for human IL-4 inducedTreg (43) and because CsA treatment does not abolish sup-pressive activity of nTreg (44). This is in agreement with thefunctional intactness of nTreg and iTreg being single or double-deficient for NFAT members. Likewise, TCR-proximal signalinglike Ca2+ influx is impaired in nTreg (30, 45). It unravels anunexpected view of Foxp3–NFAT interactions analyzed in de-tail for NFAT1 (6): if Foxp3 does not depend on NFAT to build arepresseosome, it might interact to inhibit NFAT actions, whichwould be unfavorable for Treg. In line, recent data demonstratethat only a subset of genes is dependent on the interaction ofFoxp3 with NFAT, whereas Foxp3–Foxp3 homodimers pre-dominate repression (27). On the other hand, Foxp3 might acton additional levels, like repressing translocation and NFAT2expression (45, 46), again entailing the implication that highNFAT activity has to be avoided for Treg function.In conclusion, combined deletion of two of the three NFAT

family members expressed in T cells barely impairs Treg suppres-sive activity, indicating that either minimal levels of NFAT activitysuffice for regulatory function or that suppressive capacity is evenindependent of NFAT. This is of exciting importance for transplanttherapy and treatment of autoimmune diseases. Instead of thecalcineurin inhibitors CsA and FK506, new therapeutics like R11-VIVIT (47) and MCV1 (48) should be clinically improved. Thosewould reduce NFAT activation specifically, thereby functionallyinhibiting proinflammatory Tconv but not Treg suppression.

Materials and MethodsMice and Cells. Nfat2fl/fl animals were generated in A. Rao’s laboratory (HarvardMedical School, Boston, MA). Nfat1−/− × Nfat4−/−, B6-Tg (Cd4-cre) 1Cwi/Cwilbcm(European mouse mutant archive, Rome, Italy) and Foxp3-IRES-cre havebeen described previously (16, 49, 50) (SI Materials and Methods).

Antibodies, Reagents, and Media. SI Materials and Methods gives clonenumbers, provider, and concentrations used.

T-cell Subsets. Human T-cell subsets (29, 51) and murine CD4+CD25+ nTregand CD4+CD25− Tconv (26) were isolated and stimulated as before. Detailscan be found in SI Materials and Methods.

Immunofluorescence and Immunoblot. See refs. 15 and 26. The followingprimary antibodies were used: anti-NFAT2 (7A6; BD Pharmingen), anti-NFAT1 (IG-209; immunoGlobe), anti-Smad3 (ab28379; Abcam), and anti-Foxp3 (FJK-16s; eBiosciences), anti-NFAT4 (F-1; Santa Cruz Biotechnology),and anti–β-actin (C4; Santa Cruz Biotechnology). Extended protocols aregiven in SI Materials and Methods.

ChIP Analysis. ChIP-IT Express kit (Active Motif) was used with enzymaticshearing followed by additional sonication. IP-Ab: anti-NFAT2 (7A6; BDPharmingen), anti-Foxp3 (FJK-16s, eBioscience), and anti-Smad3 (Abcam).Primers are given in SI Materials and Methods.

Adoptive Transfer Colitis and Endoscopy. Colitis was induced in Rag1−/−mice byinjecting i.p. 2.5 × 105 Cd90.1+ (WT) and 2.5 × 105 Cd90.2+ (WT or Nfat2fl/fl ×Cd4cre) CD4+CD62L+CD25− cells. Scoring (52) is described in SI Materialsand Methods.

16262 | www.pnas.org/cgi/doi/10.1073/pnas.1203870109 Vaeth et al.

Skin Transplant Model. To test alloantigen-induced or nTreg in vivo (20, 21),mice received 200 μg of anti-CD4 YTS177.9 mAb (Bio-Xcell) i.v. on day −28/−27.On day −27 the mice also received 250 μL DST from BALB/c mice. CD4+CD25+

T cells were FACS-sorted on day 0, and C57BL/6 Rag2−/− mice were recon-stituted i.v. with 2 × 105 C57BL/6 CD4+CD45RBhi cells along with 5 × 105

CD4+CD25+ nTreg cells isolated from naïve or 2 × 105 CD4+CD25+ Treg cellsfrom YTS177/DST-pretreated mice. Next day, BALB/c tail skin allografts weretransplanted onto flanks of reconstituted mice.

Statistical Analysis. Groups were compared with Prism software (GraphPad)using two-tailed Student’s t test.

ACKNOWLEDGMENTS. We thank Anjana Rao for providing Nfat2 floxed mice;Laurie H. Glimcher for Nfat1−/− and Nfat1−/− × Nfat4−/−; Kajsa Wing and ShimonSakaguchi for Foxp3-IRES-cremice; Niklas Beyersdorf for Cd90.1mice; FriederikeFrommerandKarin Elflein forRag1−/−mice;AndreaDietzel formaintainingmice;Niklas Beyersdorf and TeaGogishvili for providing CD28SA; Christina Kober,Mel-anie Schott, Ilona Pietrowski, Doris Michel, Miriam Eckstein, Anna Unverdorben,and Ralf Kielenbeck for excellent technical support; and Amiya Patra, ChristinaKober, Tobias Bopp, LenaDietz, Niklas Beyersdorf, and ThomasHünig for helpfuldiscussions. This work was supported by the German Research Foundation (DFG)with the following grants: Graduate College “Immunomodulation” (M.V.); GrantSPP1365 (to F.B.-S.); Grants TRR52/A2 (to H.J.), /A3 (to F.B.-S.), /C2 (to A.W.), and/C4 (to B.S.); and by the Federal Ministry for Education and Research: interdisci-plinary center of clinical research (IZKF) in Würzburg, Germany (F.B.-S.).

1. Walker MR, Carson BD, Nepom GT, Ziegler SF, Buckner JH (2005) De novo generationof antigen-specific CD4+CD25+ regulatory T cells from human CD4+CD25- cells. ProcNatl Acad Sci USA 102:4103–4108.

2. Fontenot JD, et al. (2005) Regulatory T cell lineage specification by the forkheadtranscription factor foxp3. Immunity 22:329–341.

3. Li B, et al. (2007) FOXP3 interactions with histone acetyltransferase and class II histonedeacetylases are required for repression. Proc Natl Acad Sci USA 104:4571–4576.

4. Bettelli E, Dastrange M, Oukka M (2005) Foxp3 interacts with nuclear factor of acti-vated T cells and NF-kappa B to repress cytokine gene expression and effector func-tions of T helper cells. Proc Natl Acad Sci USA 102:5138–5143.

5. Ono M, et al. (2007) Foxp3 controls regulatory T-cell function by interacting withAML1/Runx1. Nature 446:685–689.

6. Wu Y, et al. (2006) FOXP3 controls regulatory T cell function through cooperationwith NFAT. Cell 126:375–387.

7. Hori S (2010) c-Rel: A pioneer in directing regulatory T-cell lineage commitment? EurJ Immunol 40:664–667.

8. Zheng Y, et al. (2010) Role of conserved non-coding DNA elements in the Foxp3 genein regulatory T-cell fate. Nature 463:808–812.

9. Long M, Park SG, Strickland I, Hayden MS, Ghosh S (2009) Nuclear factor-kappaBmodulates regulatory T cell development by directly regulating expression of Foxp3transcription factor. Immunity 31:921–931.

10. Oh-hora M, Rao A (2009) The calcium/NFAT pathway: role in development andfunction of regulatory T cells. Microbes Infect 11:612–619.

11. Sakaguchi S, Sakaguchi N (1988) Thymus and autoimmunity. Transplantation of thethymus from cyclosporin A-treated mice causes organ-specific autoimmune disease inathymic nude mice. J Exp Med 167:1479–1485.

12. Serfling E, et al. (2000) The role of NF-AT transcription factors in T cell activation anddifferentiation. Biochim Biophys Acta 1498:1–18.

13. Tone Y, et al. (2008) Smad3 and NFAT cooperate to induce Foxp3 expression throughits enhancer. Nat Immunol 9:194–202.

14. Bopp T, et al. (2005) NFATc2 and NFATc3 transcription factors play a crucial role insuppression of CD4+ T lymphocytes by CD4+ CD25+ regulatory T cells. J Exp Med 201:181–187.

15. Nayak A, et al. (2009) Sumoylation of the transcription factor NFATc1 leads to itssubnuclear relocalization and interleukin-2 repression by histone deacetylase. J BiolChem 284:10935–10946.

16. Lee PP, et al. (2001) A critical role for Dnmt1 and DNA methylation in T cell de-velopment, function, and survival. Immunity 15:763–774.

17. Thornton AM, et al. (2010) Expression of Helios, an Ikaros transcription factor familymember, differentiates thymic-derived from peripherally induced Foxp3+ T regula-tory cells. J Immunol 184:3433–3441.

18. Powrie F, Leach MW, Mauze S, Caddle LB, Coffman RL (1993) Phenotypically distinctsubsets of CD4+ T cells induce or protect from chronic intestinal inflammation in C.B-17 scid mice. Int Immunol 5:1461–1471.

19. Josefowicz SZ, et al. (2012) Extrathymically generated regulatory T cells control mu-cosal TH2 inflammation. Nature 482:395–399.

20. Karim M, Kingsley CI, Bushell AR, Sawitzki BS, Wood KJ (2004) Alloantigen-inducedCD25+CD4+ regulatory T cells can develop in vivo from CD25-CD4+ precursors ina thymus-independent process. J Immunol 172:923–928.

21. Sawitzki B, et al. (2005) IFN-gamma production by alloantigen-reactive regulatoryT cells is important for their regulatory function in vivo. J Exp Med 201:1925–1935.

22. Marie JC, Letterio JJ, Gavin M, Rudensky AY (2005) TGF–beta1 maintains suppressorfunction and Foxp3 expression in CD4+CD25+ regulatory T cells. J Exp Med 201:1061–1067.

23. Zhang N, Bevan MJ (2012) TGF-β signaling to T cells inhibits autoimmunity duringlymphopenia-driven proliferation. Nat Immunol 13:667–673.

24. Beyersdorf N, et al. (2009) STIM1-independent T cell development and effectorfunction in vivo. J Immunol 182:3390–3397.

25. Gogishvili T, et al. (2009) Rapid regulatory T-cell response prevents cytokine stormin CD28 superagonist treated mice. PLoS ONE 4:e4643.

26. Vaeth M, et al. (2011) Regulatory T cells facilitate the nuclear accumulation of in-ducible cAMP early repressor (ICER) and suppress nuclear factor of activated T cell c1(NFATc1). Proc Natl Acad Sci USA 108:2480–2485.

27. Bandukwala HS, et al. (2011) Structure of a domain-swapped FOXP3 dimer on DNA

and its function in regulatory T cells. Immunity 34:479–491.28. Allan SE, et al. (2005) The role of 2 FOXP3 isoforms in the generation of human CD4+

Tregs. J Clin Invest 115:3276–3284.29. Kubach J, et al. (2007) Human CD4+CD25+ regulatory T cells: Proteome analysis

identifies galectin-10 as a novel marker essential for their anergy and suppressive

function. Blood 110:1550–1558.30. Gavin MA, Clarke SR, Negrou E, Gallegos A, Rudensky A (2002) Homeostasis and

anergy of CD4(+)CD25(+) suppressor T cells in vivo. Nat Immunol 3:33–41.31. Floess S, et al. (2007) Epigenetic control of the foxp3 locus in regulatory T cells. PLoS

Biol 5:e38.32. Mantel PY, et al. (2006) Molecular mechanisms underlying FOXP3 induction in human

T cells. J Immunol 176:3593–3602.33. Ruan Q, et al. (2009) Development of Foxp3(+) regulatory t cells is driven by the c-Rel

enhanceosome. Immunity 31:932–940.34. Ghosh S, et al. (2010) Hyperactivation of nuclear factor of activated T cells 1 (NFAT1)

in T cells attenuates severity of murine autoimmune encephalomyelitis. Proc Natl

Acad Sci USA 107:15169–15174.35. Chuvpilo S, et al. (1999) Alternative polyadenylation events contribute to the in-

duction of NF-ATc in effector T cells. Immunity 10:261–269.36. Kiani A, Rao A, Aramburu J (2000) Manipulating immune responses with immuno-

suppressive agents that target NFAT. Immunity 12:359–372.37. Frischbutter S, Gabriel C, Bendfeldt H, Radbruch A, Baumgrass R (2011) De-

phosphorylation of Bcl-10 by calcineurin is essential for canonical NF-κB activation

in Th cells. Eur J Immunol 41:2349–2357.38. Palkowitsch L, et al. (2011) The Ca2+-dependent phosphatase calcineurin controls

the formation of the Carma1/Bcl10/Malt1 complex during T cell receptor induced

NF-{kappa}B activation. J Biol Chem 286:7522–7534.39. Fisher WG, Yang PC, Medikonduri RK, Jafri MS (2006) NFAT and NFkappaB activation

in T lymphocytes: A model of differential activation of gene expression. Ann Biomed

Eng 34:1712–1728.40. Oh-Hora M, et al. (2008) Dual functions for the endoplasmic reticulum calcium sensors

STIM1 and STIM2 in T cell activation and tolerance. Nat Immunol 9:432–443.41. Curotto de Lafaille MA, Lafaille JJ (2009) Natural and adaptive foxp3+ regulatory

T cells: More of the same or a division of labor? Immunity 30:626–635.42. Borde M, Barrington RA, Heissmeyer V, Carroll MC, Rao A (2006) Transcriptional basis

of lymphocyte tolerance. Immunol Rev 210:105–119.43. Prots I, Skapenko A, Lipsky PE, Schulze-Koops H (2011) Analysis of the transcriptional

program of developing induced regulatory T cells. PLoS ONE 6:e16913.44. Li Q, et al. (2012) Constitutive nuclear localization of NFAT in Foxp3+ regulatory

T cells independent of calcineurin activity. J Immunol 188:4268–4277.45. Sumpter TL, Payne KK, Wilkes DS (2008) Regulation of the NFAT pathway discrim-

inates CD4+CD25+ regulatory T cells from CD4+CD25- helper T cells. J Leukoc Biol 83:

708–717.46. Torgerson TR, et al. (2009) FOXP3 inhibits activation-induced NFAT2 expression in

T cells thereby limiting effector cytokine expression. J Immunol 183:907–915.47. Noguchi H, et al. (2004) A new cell-permeable peptide allows successful allogeneic

islet transplantation in mice. Nat Med 10:305–309.48. Yu H, et al. (2012) Selective modulation of nuclear factor of activated T-cell function

in restenosis by a potent bipartite peptide inhibitor. Circ Res 110:200–210.49. Ranger AM, Oukka M, Rengarajan J, Glimcher LH (1998) Inhibitory function of two

NFAT family members in lymphoid homeostasis and Th2 development. Immunity 9:

627–635.50. Wing K, et al. (2008) CTLA-4 control over Foxp3+ regulatory T cell function. Science

322:271–275.51. Becker C, et al. (2009) Protection from graft-versus-host disease by HIV-1 envelope

protein gp120-mediated activation of human CD4+CD25+ regulatory T cells. Blood

114:1263–1269.52. Weigmann B, et al. (2008) The transcription factor NFATc2 controls IL-6-dependent

T cell activation in experimental colitis. J Exp Med 205:2099–2110.

Vaeth et al. PNAS | October 2, 2012 | vol. 109 | no. 40 | 16263

IMMUNOLO

GY

Supporting InformationVaeth et al. 10.1073/pnas.1203870109SI Materials and MethodsMice and Cells.Nfat2fl/fl (nuclear factor of activated T-cells) animalswere generated in A. Rao’s laboratory (Harvard Medical School,Boston, MA). Nfat1−/− × Nfat4−/−, B6-Tg (Cd4-cre) 1Cwi/Cwilbcm (European mouse mutant archive, Rome, Italy) andFoxp3-IRES-cre have been described previously (1–3). Animalswere used at 6–16 wk and maintained in accordance with in-stitutional guidelines for animal welfare. Blood from healthydonors was obtained after informed consent, in accordancewith the Declaration of Helsinki, under a protocol that receivedapproval from the Institutional Review Board from the JohannesGutenberg university hospital in Mainz.

Antibodies, Reagents, and Media. mAbs against CD3ε (145-2C11),CD28 (37.51; both BD Pharmingen), and CD4 (YTS177.9; Bio-Xcell), as well as superagonistic CD28 mAb D665 (CD28SA;Serotec), were used as previously described (4). RecombinanthIL-2 (50 U/mL), hIL-6 (50 ng/mL), mIL-12 (10 ng/mL), hIL-21(65 ng/mL), mIFN-γ (50 ng/mL), hTGF-β1 (1–5 ng/mL), andmIL10 (20 ng/mL; all PeproTech), anti-IL-2 (5 μg/mL; eBio-science), anti-IL-4 and anti-IFNγ (both 2.5 μg/mL; R&D Systems),5 12-O-tetradecanoyl-phorbol-13-acetate (TPA, 10 ng/mL; Sigma),ionomycin (5 nM; Merck Biosciences), and cyclosporin A (CsA;Calbiochem) were used as indicated. CD4+ T cells were culti-vated in RPMI 1640 (5, 6).

Preparation of T-Cell subsets. Human T-cell subsets were isolatedand stimulated as shown previously (7, 8). Murine CD4+CD25+

naturally occurring regulatory T cells (nTreg) and CD4+CD25−

conventional naïve CD4+ T cells (Tconv) were isolated (6) usingDynal Mouse T Cell Negative Isolation Kit (Invitrogen) followedby staining with anti-CD25-PE mAb and anti-PE MACS beadsenrichment (Miltenyi Biotech).

Cell Culture and Stimulations. Priming and restimulation of primaryT cells was performed using plate-bound anti-CD3 mAb (145-2C11,5 μg/mL) plus anti-CD28 mAb (37.51, 1 μg/mL) (both BD Phar-mingen). After 72 h the cells were washed and cultured on freshplates for additional 96 h. For intracellular cytokine analysis cellswere stimulated for 6 h with TPA/Iono in presence of GolgiPlugand GolgiStop (both BD Pharmingen).

FACS Staining. FACS staining (6) was carried out with the fol-lowing Ab: fluorescein isothiocyanate (FITC)-conjugated CD4(GK1.5), CD8α (53-6.7), CD19 (1D3), GITR (DTA-1); phycoer-ythrin (PE)-conjugated CD3ε (145-2C11), CD4 (RM4-5), CD19(1D3), CD25 (PC61), CD39 (24DMS1), CD73 (eBioTY/11.8)CD103 (2E7), CD107a (eBio1D4B), OX-40 (OX-86), Lag3(eBioC9B7W), CTLA-4 (UC10-4B9); allophycocyanin (APC)-con-jugated CD8α (53-6.7); eFluor710-conjugated GARP (YGIC86)and LAP (TW7-16B4); biotin-conjugated CD3ε (145-2C11), CD4(GK1.5), and CD90.2 (53-2.1), and secondary streptavidin-HRP,streptavidin-APC or streptavidin-PE mAb (all BD Pharmingen).Intracellular Foxp3 (FJK-16s, FITC-, PE-, and APC-conjugated),Helios (22F6, FITC-conjugated, BioLegend), and cAMP (SPM486,Abcam), together with donkey-anti-mouse-AlexaFluor555 (In-vitrogen), staining was performed using the Foxp3 staining kit(eBiosciences). Cytokine staining for IL-2-APC (JES6-5H4),IFN-γ–APC (XMG1.2), IL-17-PE (eBio17B7), and TNFα-PE(MP6-XT22) was performed using the IC Fixation Buffer Kit(eBioscience). Samples were analyzed on a FACS Calibur (BD

Biosciences) with CellQuest (BD Biosciences) and FlowJo soft-ware (TreeStar).

ELISA. Cells (1 × 106/mL) were cultivated as indicated for 48 h.The supernatant was analyzed by IL-2 ELISA (eBiosciences).

Calcium Measurement. Splenic CD4+ T cells (1 × 107) were in-cubated in medium containing 5% (vol/vol) FCS, 1 μM Indo-1-AM (Invitrogen), and 0.015% Pluronic F127 (Invitrogen) at 30°C for 25 min. The cell suspension was then diluted with 700 μLmedium containing 10% (vol/vol) FCS and incubated at 37 °Cfor another 10 min. The cells were washed twice with PBS fol-lowed by surface staining with anti-CD4-PacificBlue, anti-CD25-PE, and biotinylated anti-CD3ε (all eBioscience). For calciummeasurement, cells were diluted 1:10 in PBS containing 0.5 mMEGTA, and a baseline was recorded for 60 s. Ca2+ movementwas assessed after streptavidin-HPR cross-linking (eBioscience),followed by the addition of 1 mM Ca2+ after 3.5 min and Ionoafter 9 min of recording. After 12 min, another baseline of un-stimulated cells was recorded as control. Increases in free in-tracellular Ca2+ were measured in real-time on an LSR II (BD),and data were analyzed as median in comparative overlays withFlowJo software (TreeStar).

Proliferation Assays. Proliferation assay (6) was measured usinga Mach 2 Harvester (Tomtec).

Immunofluorescence. For confocal microscopy (5, 6) the followingprimary antibodies were used: anti-NFAT2 (7A6; BD Pharmin-gen), anti-NFAT1 (IG-209; immunoGlobe), anti-Smad3 (ab28379;Abcam), and anti-Foxp3 (FJK-16s; eBiosciences). Secondarystaining was performed using Abs: anti-rabbit Alexa-Fluor 647,anti-mouse Alexa-Fluor 488, and anti-rat Alexa-Fluor 555 (allMolecular Probes). Slides were mounted with Fluoromount-G(Southern Biotechnology) containing DAPI. Images were takenwith a confocal microscope (Leica TCS SP2 equipment, objectivelense; HeX PL APO, 40×/1.25–0.75) and LCS software (Leica).For statistics, more than 100 cells from at least three independentexperiments were counted, and mean fluorescence intensity percell was calculated.

Immunoblot. Protein lysate was made with radioimmunoprecip-itation assay buffer (RIPA) buffer andmeasured using bicinchoninicacid (BCA) reagent (BioRad). Equal amount of protein was frac-tionated by 8–12% SDS/PAGE and electroblotted on membranes(5). For detection, anti-NFAT2 (7A6; BD Pharmingen), anti-NFAT1 (IG-209; immunoGlobe), anti-NFAT4 (F-1; Santa CruzBiotechnology), and anti-β-actin (C4; Santa Cruz Biotech-nology) with anti-mouse or anti-rabbit peroxidase-coupled sec-ondary antibodies were used.

PCR and Quantitative RT-PCR. Genomic DNA was prepared withDNA-lysis buffer (including 0.2% SDS). PCR was performedusing the following primers (5′→3′): Nfat2 CCTATTTAAAC-ACCTGGCTCCCTGCG plus CCATCTCTCTGACCAACAG-AAGCC AG, Δexon3 CTAGGCCTCAGGCGTTCCACC plusCCTGCCTCTCTCAGCCTTTGA, Cebp CGAGCCACCGCG-TCCTCCAGC plus CCGGTCGGTGCGCGTCATTGC. RNAwas extracted using the RNeasy Micro Kit (Qiagen) followed bycDNA synthesis which was performed with the iScript II Kit(BioRad). Real-time quantitative RT-PCR was carried out withan ABI Prism 7700 detection system using the following primers:Nfat1 TCATAGGAGCCCGACTGATTG plus CCATTCC-

Vaeth et al. www.pnas.org/cgi/content/short/1203870109 1 of 16

CATCTGCAGCAT, Nfat2 GATCCGAAGCTCGTATGGACplus AGTCTCTTTCCCCGACATCA, Nfat4 CACATCCCACA-GCCCAGTG plus CACATCCCACAGCCCAGTG normalizedto β-actin GACGGCCAGGTCATCACTATTG plus AGGAA-GGCTGGAAAAGAGCC.

ChIP Analysis. ChIP-IT Express kit (Active Motif) was usedaccording to the manufactures’ instructions, except for enzymaticshearing followed by additional sonication. The following IP-Abs were used: anti-NFAT2 (7A6; BD Pharmingen), anti-Foxp3(FJK-16s; eBioscience), and anti-Smad3 (Abcam). Primers (5′→3′)used were: Foxp3-Pr CTTCCCATTCACATGGCAGGC plusCTTTGCCCTTTACAAGTCATCTG; Foxp3-CNS1 GCACTTG-AAAATGAGATAACTGTTC plus CATCACAGTACATACGA-GGAAATG; and Foxp3-CNS3 CCAGATGGACGTCACCTACCplus GGCGTTCCTGTTT-GACTGTTTC.

Electromobility Shift Assay (EMSA). Nuclear proteins from humanand murine T-cell subsets were prepared and stored by ProteoJETKit (Thermo Scientific). hFoxp3-Prom-NFAT GTTCTTCTTCC-TTGTTTTTTTTT; mFoxp3-Prom-NFAT GACTTATTTTCCCT-CAGTTTTTTTTTT;mFoxp3-CNS1-NFATGCTTCATTTTTTCC-ATTTACTG; mIL2-Pubd CCCCAAAGAGGAAAATTTGTTT(boldface, NFAT consensus sites); anti-NFAT1 (D43B1, CellSignaling), and anti-NFAT2 (7A6, BD Pharmingen) were usedfor EMSA, which was performed as previously described (9).

Microarray. CD4+CD25+ nTregs and CD4+CD25− Tconv fromWT or Nfat1−/− × Nfat2fl/fl × Cd4-cre (DKO) mice were isolatedand stimulated for 24 h with plate-bound anti-CD3/28 in absence(Tconv and nTreg) or presence (iTreg conditions) of 5 ng/mLTGF-β, followed by RNA extraction using the standard TRIzolmethod (Invitrogen). Biotin-labeled amplified aRNA was pre-pared using the GeneChip 3′ IVT Express Kit and hybridizedto GeneChip mouse genome 430 2.0 arrays (Affymetrix) ac-cording to the manufacturer’s protocols. The trimmed meansignals of the probe arrays were scaled to a target value of 500 andexpression values determined using Affymetrix GeneChip Oper-ating Software. Data were analyzed and visualized as a heat mapusing GeneSpring GX 12.0 software (Agilent Technologies).The original microarray data can be found in the ArrayExpress

database under the accession no. E-MEXP-12345 (www.ebi.ac.uk/arrayexpress).

CD28SA Treatment. Treatment of mice with superagonistic CD28mAb D665 was performed as previously described (4). Mice re-ceived a single i.p. injection of 250 μg CD28SA D665 (Serotec) orPBS as control. After 60 h serum samples were obtained from tailvein. Spleen and LN cells were harvested on day 3 post injection.

Adoptive Transfer Colitis and Endoscopy. Colitis was induced inRag1−/−mice by injecting i.p. 2.5 × 105 Cd90.1+ (WT) and 2.5 × 105

Cd90.2+ (WT or Nfat2fl/fl × Cd4cre) CD4+CD62L+CD25− cells.Mice were anesthetized [100 μL of a mixture of 100 mg/mLKetavest (Pfizer) and Rompun (Bayer Healthcare) i.p.] and clini-cal symptoms [murine endoscopic index of colitis severity (ME-ICS): total range 0–15 points; colon translucency (0–3 points),presence of fibrin (0–3 points), mucosa granularity (0–3 points),vascular pattern (0–3 points), and stool (0–3 points)] were ana-lyzed with a high-resolution video endoscopic system (KarlStorz) (10).

Skin TransplantModel.To test alloantigen-induced or nTreg in vivo(11, 12) mice received 200 μg of anti-CD4 YTS177.9 mAb (Bio-Xcell) i.v. on day −28/−27. On day −27 the mice also received250 μL donor-specific blood transfusions (DST) from BALB/cmice. CD4+CD25+ T cells were FACS-sorted on day 0, andC57BL/6 Rag2−/− mice were reconstituted i.v. with 2 × 105

C57BL/6 CD4+CD45RBhi cells along with 5 × 105 CD4+CD25+

nTreg cells isolated from naïve or 2 × 105 CD4+CD25+ Tregcells from YTS177/DST-pretreated mice. Next day, BALB/c tailskin allografts were transplanted onto flanks of reconstitutedmice. Graft survival between groups was monitored and com-pared using the log–rank test. For transfer of in vitro-generatediTregs, naïve CD4+ CD25− T cells were cocultivated with BALB/cCD19+ B cells in the presence of IL-2, TGF-β, anti-IL-12, anti–IFN-γ, and the anti-CD4 YTS177 for 2 wk, with restimulationafter 1 wk. FACS-sorted CD25hi T cells were verified for Foxp3expression, and equal numbers were used in transplantation.

Statistical Analysis. Groups were compared with Prism software(GraphPad) using two-tailed Student’s t test.

1. Lee PP, et al. (2001) A critical role for Dnmt1 and DNA methylation in T celldevelopment, function, and survival. Immunity 15:763–774.

2. Ranger AM, Oukka M, Rengarajan J, Glimcher LH (1998) Inhibitory function of twoNFAT family members in lymphoid homeostasis and Th2 development. Immunity 9:627–635.

3. Wing K, et al. (2008) CTLA-4 control over Foxp3+ regulatory T cell function. Science322:271–275.

4. Gogishvili T, et al. (2009) Rapid regulatory T-cell response prevents cytokine storm inCD28 superagonist treated mice. PLoS ONE 4:e4643.

5. Nayak A, et al. (2009) Sumoylation of the transcription factor NFATc1 leads to itssubnuclear relocalization and interleukin-2 repression by histone deacetylase. J BiolChem 284:10935–10946.

6. Vaeth M, et al. (2011) Regulatory T cells facilitate the nuclear accumulation ofinducible cAMP early repressor (ICER) and suppress nuclear factor of activated T cellc1 (NFATc1). Proc Natl Acad Sci USA 108:2480–2485.

7. Becker C, et al. (2009) Protection from graft-versus-host disease by HIV-1 envelopeprotein gp120-mediated activation of human CD4+CD25+ regulatory T cells. Blood114:1263–1269.

8. Kubach J, et al. (2007) Human CD4+CD25+ regulatory T cells: proteome analysisidentifies galectin-10 as a novel marker essential for their anergy and suppressivefunction. Blood 110:1550–1558.

9. Berberich-Siebelt F, et al. (2000) C/EBPbeta enhances IL-4 but impairs IL-2 andIFN-gamma induction in T cells. Eur J Immunol 30:2576–2585.

10. Weigmann B, et al. (2008) The transcription factor NFATc2 controls IL-6-dependentT cell activation in experimental colitis. J Exp Med 205:2099–2110.

11. Karim M, Kingsley CI, Bushell AR, Sawitzki BS, Wood KJ (2004) Alloantigen-inducedCD25+CD4+ regulatory T cells can develop in vivo from CD25-CD4+ precursors ina thymus-independent process. J Immunol 172:923–928.

12. Sawitzki B, et al. (2005) IFN-gamma production by alloantigen-reactive regulatoryT cells is important for their regulatory function in vivo. J Exp Med 201:1925–1935.

Vaeth et al. www.pnas.org/cgi/content/short/1203870109 2 of 16

Fig. S1. Conditional disruption of Nfat2 exon3 in mice. (A) Strategy for introduction of loxP sites into the mouse Nfat2 locus flanking exon 3. Genomic locus ofthe Nfat2 exon3 region (Upper) and the targeting vector containing a loxP sites flanked version of exon3 including a neomycin resistance cassette (Lower)bordered by frt sites. (B and C) Confirmation of conditional inactivation of Nfat2 in CD4+ T lymphocytes by CD4-mediated cre. (B) PCR analysis of genomicDNA of isolated CD4+ T cells with specific primers for the loss of Nfat2 exon3 (Δexon3) and total Nfat2 (wt Nfat2) of WT (Nfat2+/+ × Cd4-cre), heterozygous(Nfat2+/fl × Cd4-cre), and knockout (Nfat2fl/fl × Cd4-cre) mice. (C) Immunoblot of protein from CD4+ T cells of WT, heterozygous, and homozygous mice. Arrowsindicate isoforms of NFAT2. (D) Breeding of Nfat2fl/fl × Cd4-cre mice to Nfat1−/− mice resulted in NFAT1NFAT2 double-deficient T lymphocytes. Immunoblotanalysis of NFAT expression in total thymocytes from WT, Nfat2fl/fl × Cd4-cre, and Nfat1−/− × Nfat2fl/fl × Cd4-cre mice. (E) Foxp3-specific inactivation of Nfat2 inTreg cells by Foxp3-IRES-cre (FIC). Immunoblot with proteins isolated from CD4+CD25− and CD4+CD25− T cells of WT and Nfat2fl/fl × FIC mice.

Vaeth et al. www.pnas.org/cgi/content/short/1203870109 3 of 16

B

wild type

Nfat2fl/fl x Cd4-cre

A

5 ng/ml TGFβ 2.5 ng/ml TGFβ 1.25 ng/ml TGFβ 0.6 ng/ml TGFβ

45.0 43.7 37.6 19.2

36.1 22.5 17.0 6.9

Foxp3

IL-2 only

IL-2 & TGFβ

unstim. anti-CD3/28 anti-CD3/28+ 10 ng/ml CsA

3.3 2.4 7.4

5.3 71.5 9.3

wild type Nfat2fl/fl x Cd4-cre

Nfat1–/– x Nfat2fl/fl x Cd4-cre

C D

CD4

IL-2

TH0

TH1

TH17

iTreg

IL-17A

IFNγ

3.4

2.4

6.8

2.8

3.9

2.2

39.2

1.1

32.3

0.9

14.1

1.5

5.7

2.3

8.2

3.8

2.3

2.4

2.2

47.4

4.2

11.7

1.7

3.5

wild type Nfat2fl/fl x Cd4-cre

47.3 33.9

CD90.2+

27.6 20.3

CD4

IFNγ

32.6 22.1

CD4

TN

Fα

16.6 7.8

CD4

IL-1

7

Foxp3

CD

25

Fig. S2. Direct influence of NFAT on Foxp3 and cytokine induction. (A) TGF-β–mediated iTreg induction of WT CD4+CD25− T cells is abrogated in the presenceof 10 ng/mL CsA. (B) CD4+CD25− T cells from WT and Nfat2fl/fl × Cd4-cre mice were stimulated for 3 d with plate-bound anti-CD3/28 in the presence of IL-2 anddifferent concentrations of TGF-β, followed by rest of 4 d. Representative FACS analysis of CD4+CD25+Foxp3+ T cells in culture. (C and D) Impaired cytokineexpression in NFAT-deficient CD4+ T cells. (C) CD4+CD25− T cells from WT, Nfat2fl/fl × Cd4-cre, and Nfat1−/− × Nfat2fl/fl × Cd4-cre mice were cultivated for 3 dwith plate-bound anti-CD3/28 under TH0- (IL-2, anti-IL-4, anti-IFNγ), TH1- (IL-12, IFNγ, anti-IL-4), iTreg- (IL-2, TGF-β, anti-IL-4, anti-IFNγ), and TH17-polarizingconditions (IL-6, IL-21, TGF-β, anti-IL-4, anti-IFNγ, anti-IL-2). Cells were restimulated for 6 h with TPA/Iono followed by intracellular IFN-γ and IL-17 analysis usingFACS. (D) Measurement of intracellular IL-2, IFN-γ, TNF-α, and IL-17 of CD90.2+ WT or CD90.2+ Nfat2fl/fl × Cd4-cre T cells transferred into Rag2−/− (lymphopenic-induced colitis, parallel to Fig. 2 B–D; Fig. S4). Total cells from mLN were stimulated for 6 h with TPA/Iono followed by intracellular FACS-staining; gating onCD4 and CD90.2 identified WT or Nfat2fl/fl × Cd4-cre adoptively transferred T cells.

Vaeth et al. www.pnas.org/cgi/content/short/1203870109 4 of 16

H

mFoxp3-CNS1-NFATmIL2-Pubd mFoxp3-Prom. hFoxp3-Prom.

Tconv

TGFβnTreg

++–

*

anti-CD3/28+ –

+ ++ +– – –

+ – –

– – – –+++– + –

+ ++ +– – –

+ – –

– – – –+++– + –

+ ++ +– – –

+ – –

– – – –+++– + –

+ ++ +– – –

+ – –

– – – –+

Tconv

TGFβnTreg

anti-NFAT2

hFoxp3-Prom. mFoxp3-CNS1-NFAT

anti-NFAT1comp. hFoxp3-Prom.

+ + + –– – –– – – + + +

*

ss NFAT1–ss NFAT2–

+ + +

+ + ++ + ++ –– ––– –

–– ––– – –– – –

–– – –– – –– – – – –+– – + – – + – – + – – +–– + – – + – – + – – + ––

I

Tconv

TGFβnTreg

anti-NFAT2

EL-4

CsA

mFoxp3-CNS1-NFAT mFoxp3-Prom-NFAT

ss NFAT2–

+ +– –– –– –– –– +

– –– +

– –– +

– –– + – +

+ ++ + – –– – – –– – – –– –+ +

– – – –– – – –– – + + + + + +

+ +– –– –– –– –– +

– –– +

– –– +

– –– + – +

+ ++ + – –– – – –– – – –– –+ +

– – – –– – – –– – + + + + + +

J

wild ty

pe

Nfat2fl/f

l x Cd4

-cre

Nfat1–/– x

Nfat2fl/f

l x Cd4

-cre

Thymus

Spleen

LN

13.7 13.4 14.6

1.3 0.9 1.2

12.4 13.3 9.3

A B

wild ty

pe

Nfat2fl/f

l x Cd4

-cre

Nfat1–/–

x

Nfat2fl/f

l x Cd4

-cre

Foxp3

CD

25

Cwild

type

Nfat1–/–

Nfat1–/–

x

Nfat

4–/–

Thymus

Spleen

LN

Foxp3

CD

25

Thymus

Spleen

Lymph nodes%

CD

4+C

D25

+F

oxp3

+

0.6 0.7 1.4

14.0 10.5 11.3

9.1 8.6 10.3

E

Thymus

Spleen

mLN

Foxp3

CD

25

0.9 0.8

15.3 22.1

17.1 13.3

Spleen

% C

D4+

CD

25+

Fox

p3+

cel

ls

F

0

10

20

30

0

5

10

1520

25

0

0.5

1.0

1.5

2.0

2.5

0

5

10

15

20

25

0

5

10

15

20

25

wild type Nfat2fl/fl x FIC

wild type

Nfat2fl/fl x FIC

LN

Dwild type Nfat4–/–

0.8 0.6

15.414.7

13.7 15.9

Thymus

Spleen

mLN

Foxp3

CD

25

G

n.s.

rela

tive

expr

essi

on to

β-a

ctin

unstim. anti-CD3/28

Nfat2 Nfat1 Nfat4

nTreg Tconv Tconv+TGFβ

nTreg Tconv Tconv+TGFβ

nTreg Tconv Tconv+TGFβ

% C

D4+

CD

25+

Fox

p3+

%

CD

4+C

D25

+F

oxp3

+

0

1

2

3

0

1

2

3

0

1

2

3

Fig. S3. Normal numbers of nTreg in NFAT-deficient mice. (A and B) Nfat2fl/fl × Cd4-cre and Nfat1−/− x Nfat2fl/fl × Cd4-cre mice show normal frequencies ofCD4+CD25+Foxp3+ nTreg in thymus, spleen, and LN. (A) Representative FACS analysis of gated CD4+ cells and (B) synopsis of those from different lymphoid organs; eachsymbol represents one littermate. (C) AppearanceofnTreg is unchanged inNfat1−/−andNfat1−/−×Nfat4−/−mice comparedwithWT. FlowcytometryofgatedCD4+ cells inthymus, spleen,andLN. (D)Nfat4−/−miceshownormal frequenciesofCD4+CD25+Foxp3+nTreg in thymus, spleen,andLN, representativeFACSanalysisofgatedCD4+Tcells.(EandF) Treg-specificdeletionofNfat2viaFoxp3-IRES-cre (FIC) doesnotchangethenTregcompartment. (E) RepresentativeFACSdeterminationofCD4+CD25+Foxp3+cellsin thymus, spleen, and LN and (F) summary of four analyzed mice. (G) RNA of Nfat1, Nfat2, and Nfat4 is expressed in unstimulated and anti-CD3/28−stimulated Tconv,nTreg, and Tconv in presence of TGF-β (iTreg induction), detected after 24 h by real-time PCR and normalized to β-actin. Human (H and I) andmurine (J) nuclear proteinextracts exhibit binding activity of NFAT1 and NFAT2 to Foxp3-CNS1, but hardly to the promoter (cells harvested for EMSA after 48 h; ss, supershift; *unidentified band).

Vaeth et al. www.pnas.org/cgi/content/short/1203870109 5 of 16

Rag1–/–

2.5 x105 CD4+CD62L+CD25– cells:Cd90.2+ wild type or

Cd90.2+ NFAT-deficient

6 weeks

2.5 x105 CD4+CD62L+CD25– cells:Cd90.1+ congenic

wild type

A

Cd90.1+ wild type & Cd90.2+ wild type

Histology FACSCytokines

weight changeendoscopy

70

80

90

100

110

120

0

2

4

6

8

10

Cd90.1+ wild type & Cd90.2+ Nfat2fl/fl x Cd4-cre

B C

weeks after T cell transfer weeks after T cell transfer

0 1 2 3 4 5 6 1 2 3 4 5

D

100 x

400 x

wei

ght c

hang

e in

%

clin

ical

sco

re (

endo

scop

y)healthy control

Cd90.1+ wild type & Cd90.2+ wild type

Cd90.1+ wild type & Cd90.2+ Nfat2fl/fl x Cd4-cre

n.sn.s

Fig. S4. Normal course of lymphopenia-induced colitis by NFAT2-deficient T cells. (A) Schematic overview of adoptive T cells transfer to induce colitis. 2.5 × 105

CD90.2+CD4+CD62L+CD25− WT or NFAT-deficient T cells (Nfat2fl/fl × Cd4-cre or Nfat1−/− × Nfat2fl/fl × Cd4-cre) along with 2.5 × 105 CD90.1+CD4+CD62L+CD25−

congenic WT cells were injected i.p. into Rag1−/− recipient mice. (B–D) Normal course of colitis in Rag1−/− mice receiving NFAT2-deficient CD4+CD62L+CD25−

T cells. (B) Monitoring weight change and (C) the clinical symptoms by colon endoscopy over 6 wk revealed only marginal differences in the characteristicsof colitis in mice receiving either WT or Nfat2fl/fl × Cd4-cre T cells (five mice per group). (D) Representative H&E staining of paraffin-embedded colonsections showed no alteration of colitis in mice receiving either WT or Nfat2fl/fl × Cd4-cre T cells. As a reference a healthy colon (Left) in 100× and 400×magnification is given.

Vaeth et al. www.pnas.org/cgi/content/short/1203870109 6 of 16

A

prol

ifera

tion

in c

pm (

x104

)

wild typeiTreg

Nfat2fl/fl x Cd4-creiTreg

CD4+ gatedFoxp3

CD

4

0

5

10

15

unstim

1 : 0.8

1 : 01 : 0

.71 : 0

.41 : 0

.31 : 0

.2

unstim

1 : 0.8

1 : 01 : 0

.71 : 0

.41 : 0

.31 : 0

.2

wild type Nfat2fl/fl x Cd4-cre

1 : 0.8

1 : 0.7

1 : 0.4

1 : 0.3

1 : 0.2

47.1 45.7

37.5 40.0

29.4 32.6

19.1 19.7

16.3 15.8B

Rag2–/– C57/BL6recipient

skin graftsurvival

CD45RBhigh T cells

BALB/c skin allograft

CD25+Foxp3+ iTreg

wild type or Nfat2fl/fl x Cd4-creC57/BL6 CD4+CD25– Tconv

+BALB/c CD19+ B cells +IL-2/TGFβ

+blocking anti-IL-12/IFNγ+ anti-CD4 mAb (YTS177, 1µg/ml)

day 14

day 0

CD

4+C

D25

+ iT

regs

in %

Nfat2fl/fl x Cd4-cre

wild type

C

D

E

Nfat2fl/fl x Cd4-crewild type

Tconv only (n=5)

wild type iTreg (n=4)

Nfat2fl/fl x Cd4-cre iTreg (n=2)

graf

t sur

viva

l in

%

0.3

5.4

21.7

72.7

0.1

9.2

15.2

75.5

tota

l CD

4+C

D25

+ iT

regs

(x1

07)

0

1

2

3

4

5

0

20

40

60

Nfat2fl/fl x Cd4-cre

wild type

p=0.0009 p=0.0388

0 5 10 15 20 25 300

20

40

60

80

100

days after transplantation

Foxp3

Hel

ios

2.8 1.4

92.5 92.53.4

0.3 1.4

5.7

CD4+CD25+

iTreg

CD4+CD25–

non-iTreg

Fig. S5. Normal in vitro and in vivo suppressive capacity of NFAT2-deficient TGF-β–induced Treg. (A) CD4+CD25− Tconv from WT and Nfat2fl/fl × Cd4-cre micewere stimulated for 3 d with anti-CD3/28 in presence of 5 ng/mL TGF-β followed by 12 d rest in medium containing IL-2 and TGF-β. CD25hi cells were enrichedand Foxp3 expression was analyzed by FACS. Different dilutions of Foxp3+ iTreg (Right) were cocultured together with WT responder CD4+CD25− T cells andAPCs for 3 d. Ratio of responder vs. iTreg cells in culture is denoted (unstim, unstimulated 1:0.8 ratio as control). After 48 h proliferation was measured bythymidine incorporation (Left). Data are mean ± SD of triplicates. (B–E) NFAT2-deficient iTreg are fully suppressive in vivo. (B) Scheme of experimental setup toexamine suppressive capacity of NFAT2-deficient iTreg in a model of allogenic skin transplantation. (C) CD4+CD25− Tconv from WT and Nfat2fl/fl × Cd4-cre mice

Legend continued on following page

Vaeth et al. www.pnas.org/cgi/content/short/1203870109 7 of 16

were cocultivated for 2 wk with allogenic BALB/c splenocytes in the presence of IL-2, TGF-β, anti-IL-12, and anti-IFNγ. Frequency of CD4+CD25+ allogenic-induced iTreg was controlled by FACS (n > 5). (D) Foxp3 and Helios expression of WT and Nfat2fl/fl × Cd4-cre CD4+CD25+ iTreg from these cultures. (E) 2 × 105

WT or Nfat2fl/fl × Cd4-cre iTreg depicted in C and D were injected along with 2 × 105 CD4+CD45RBhi T cells i.v. in the Rag2−/− recipient mice and the BALB/c skinallograft survival was measured by log–rank test. All mice receiving only CD4+CD45RBhi cells acutely rejected their skin transplants [n = 5, mean survival time(MST) = 15 d]. Addition of Nfat2fl/fl × Cd4-cre iTreg prolonged skin graft survival as efficiently as WT CD4+CD25+ iTreg (WT: n = 4, MST = 22.0 d; Nfat2fl/fl × Cd4-cre:n = 2, MST = 27 d). Shown is one representative experiment out of two.

Vaeth et al. www.pnas.org/cgi/content/short/1203870109 8 of 16

CD3ε

B22

0

B

CD4

CD

8α

C

0

100

200

300

IL-2

in p

g/m

l

*

42.6 54.1 56.4

26.9 16.7 12.5

20.2 26.1 29.3

47.5 43.2 41.6

p < 0.0005

p = 0.036

21.8

34

13.2

20.5

2.31

7.89

5.87

14

9.35

10.9

19.4

25.4

4.87

7.6

12.5

19.1

18.2

23.9

Spleen

LN

Spleen

LN

Thymus

wild ty

pe

Nfat2fl/f

l x Cd4-cr

e

Nfat1–/– x

Nfat2fl/f

l x Cd4-cr

e

Nfat1–/– x Nfat2fl/fl x Cd4-cre

Nfat2fl/fl x Cd4-cre

wild type

Spleen Lymph nodes

wild

type

Nfa

t2fl/

fl x

Cd4

-cre

Nfa

t1–/

– x

Nfa

t2fl/

fl x

Cd4

-cre

Foxp3CD3ε

E

Nfat1–/– x Nfat2fl/fl x Cd4-cre

Nfat2fl/fl x Cd4-cre

wild type

0

20

40

60

80

0

50

100

150

Spleen Lymph nodesto

tal c

ell n

umbe

r in

106

A D

tota

l cel

l num

ber

in 1

06

Fig. S6. NFAT-deficient mice display no evidence of autoimmune disorders. (A) Total cell numbers in spleen and LN of WT, Nfat2fl/fl × Cd4-cre, and Nfat1−/− ×Nfat2fl/fl × Cd4-cre mice. Data are mean ± SD from at least six littermates. (B) Representative FACS analysis of B and T-cell compartments in spleen and LN ofWT, Nfat2fl/fl × Cd4-cre, and Nfat1−/− × Nfat2fl/fl × Cd4-cremice. (C) Frequency of CD4+ and CD8+ T cells in thymus, spleen, and LN of WT, Nfat2fl/fl × Cd4-cre, andNfat1−/− × Nfat2fl/fl × Cd4-cre mice examined by flow cytometry. (D) Impaired IL-2 production by NFAT-deficient CD4+ T cells in vitro. Isolated CD4+CD25− Tconvfrom WT, Nfat2fl/fl × Cd4-cre, and Nfat1−/− × Nfat2fl/fl × Cd4-cre mice were stimulated with plate-bound anti-CD3/28 for 48 h, and produced IL-2 was assayedfrom supernatants by ELISA. Data are mean ± SD of triplicates done in one experiment and representative of two; asterisk denotes nondetectable levels of IL-2in 100-fold diluted supernatants. (E) Paraffin-embedded section of spleen and LN from untreated WT, Nfat2fl/fl × Cd4-cre, and Nfat1−/− × Nfat2fl/fl × Cd4-cremice were used for two color staining of Foxp3 (red) and CD3ε (green). (Scale bar, 80 μm.)

Vaeth et al. www.pnas.org/cgi/content/short/1203870109 9 of 16

A C

B

thym

us p

opul

atio

n in

%

wild type Nfat1–/– xx x Nfat2fl/fl x Cd4-cre

D

E

2.782.95

0.78 1.03

7.3 71.3

14.0 4.7

6.1 55.9

31.4 6.4

4.6 2.5

4.3 1.6

0.7 0.5

90.4 95.7

CD4+ gated

SS

C

Foxp3

SS

C

CD8α

CD

4

CD8α

CD

4

DP CD4 CD8 DN0

10

20

304050607080

CD4 CD80

1

2

3

4

5

F

spl

een

popu

latio

n in

%

wild type

Nfat1–/– xx x Nfat2fl/fl x Cd4-cre

day 1

Thymus Spleen0

1

2

3

4

5

6

Thymus Spleen0.00.51.01.52.0

68

101214

day 1

day 12

day 1

day 12

day 1

day 12

day 1

day 12

0.390.26

10.3 10.0

11.6 49.6

34.1 4.6

8.7 82.7

5.6 3.0

3.1 0.2

93.7 3.0

2.6 0.1

94.5 1.9

CD

4+F

oxp3

+ i

n %

Thymus Spleen0

20

40

60

80

CD4 CD80

1

2

3

4day 12

day 1 day 12

DP CD4 CD8 DN0

10

20

305060708090

Thymus Spleen 00.20.40.60.81.0

6

8

10

12

tota

l cel

l num

ber

(x10

6 )

day 1 day 12

day 1 day 12

**

*

*

*

*

**

**

Fig. S7. Regular frequency of NFAT1 plus -2 double-deficient nTreg in newborn and young mice. (A) Representative FACS analysis of CD4/CD8 cell distribution(Upper) and nTreg frequency among CD4+ cells (Lower) in the thymus of newborn and 12-d-old WT and Nfat1−/− × Nfat2fl/fl × Cd4-cre mice. (B) RepresentativeFACS analysis of CD4/CD8 cell distribution (Upper) and nTreg frequency among CD4+ cells (Lower) in the spleen of newborn and 12-d-old WT and Nfat1−/− ×Nfat2fl/fl × Cd4-cremice. (C–F) Quantification of at least three independent experiments shown as frequency in % (C–E) and total cell number (F) of thymus andspleen. *P ≤ 0.05; **P ≤ 0.005.

Vaeth et al. www.pnas.org/cgi/content/short/1203870109 10 of 16

0

10000

20000

30000

40000

0

10000

20000

30000

40000

50000

0

10000

20000

30000

40000

50000

A

B

unstim

0 : 11 : 0 1 : 11 : 0

.5

1 : 0.25

1 : 0.125

unstim

0 : 11 : 0 1 : 11 : 0

.5

1 : 0.25

1 : 0.125

unstim

0 : 11 : 0 1 : 11 : 0

.5

1 : 0.25

1 : 0.125

0

5000

10000

15000

unstim

1 : 0 1 : 11 : 0

.5

1 : 0.25

0

5000

10000

15000

20000

25000

unstim

1 : 0 1 : 11 : 0

.5

1 : 0.25

Nfat1–/– nTreg Nfat1–/– x Nfat4–/– nTreg

wild typecontrol

Nfat2fl/fl x FIC+CD28SA

Spleen Lymph nodes

wild type+ CD28SA

D

C

Ser

um IL

-2 in

pg/

ml

prol

ifera

tion

in c

pm

prol

ifera

tion

in c

pm

14.0 14.8

25.2 21.3

24.5 25.5

0

50

100

150

200

250

300

Foxp3

CD

25

E

C57/BL6donor

Rag2–/– C57/BL6recipient

skin graftsurvival

CD45RBhigh T cells

BALB/c skin allograft

CD25+ nTreg

wild typeor

Nfat2fl/fl x Cd4-cre

*0 20 40 60 80 100

0

20

40

60

80

100

graf

t sur

viva

l in

%

survival days

Tconv only

wild type nTregNfat2fl/fl x Cd4-cre nTreg

Fig. S8. NFAT-deficient nTreg exhibit normal suppressive capacities in vitro and in vivo. (A) CD4+CD25+ nTreg from WT, Nfat2fl/fl × Cd4-cre, and Nfat1−/− ×Nfat2fl/fl × Cd4-cre (Upper) or Nfat1−/− and Nfat1−/− × Nfat4−/− mice (Lower) were stimulated together with WT responder CD4+ T cells and APCs for 3 d. Ratioof responder vs. nTreg in culture is denoted (unstim, unstimulated 1:1 ratio as control). After 48 h proliferation was measured by thymidine uptake. Data aremean ± SD of triplicates done in one experiment and representative of four. (B) Frequency of Nfat2fl/fl × FIC nTreg is not impaired in spleen and LN of micetreated with 250 μg superagonistic CD28 mAb D665 (CD28SA) for 3 d. (C) Serum IL-2 was assayed by ELISA 2.5 h after CD28SA injection. Data are mean ± SD oftriplicates. (D and E) NFAT2-deficient nTreg fully prevent skin allograft rejection in an adoptive transfer mouse model. (D) Schematic overview of adoptive

Legend continued on following page

Vaeth et al. www.pnas.org/cgi/content/short/1203870109 11 of 16

transfer of B6 WT or Nfat2fl/fl × Cd4-cre CD4+CD25+ nTreg along with CD4+CD45hi T cells into a B6 Rag2−/− recipient mouse, receiving a BALB/c skin allograft. (E)5 × 105 nTreg along with 2 × 105 CD4+CD45RBhi cells were injected i.v. and BALB/c skin allograft survival in the Rag2−/− recipient mouse was measured by log–rank test. All mice receiving only CD4+CD45RBhi cells acutely rejected their skin transplants (n = 5, MST = 12.8 d). Addition of WT (n = 4, MST = 51.0) or Nfat2fl/fl

× Cd4-cre Treg (n = 5, MST = 60.4) prolonged skin graft survival.

Vaeth et al. www.pnas.org/cgi/content/short/1203870109 12 of 16

Fig. S9. Analysis of NFAT-deficient Treg on mRNA and protein levels. (A and B) Analysis of molecules involved in Treg-mediated suppression and homeostasis.(A) Freshly isolated splenic CD4+ T cells from WT or Nfat1−/− × Nfat2fl/fl × Cd4-cre (DKO) mice were stained for indicated molecules. Subsequent intracellularFoxp3 staining identified nTreg and Tconv using FACS; i.c., intracellular; s, surface staining. (B) CD4+CD25+ nTreg and CD4+CD25− Tconv from WT or Nfat1−/− ×Nfat2fl/fl × Cd4-cre (DKO) mice were isolated and stimulated for 3 d with plate-bound anti-CD3/28 in absence (Tconv and nTreg) or presence (iTreg conditions)of TGF-β. Denoted molecules on Foxp3− Tconv, Foxp3+ nTreg, or TGF-β–induced iTreg were analyzed by FACS. (C) Flow cytometry of CD25 expression in variousNFAT-deficient Treg populations. CD4+CD25+ nTreg and CD4+CD25− Tconv from WT, Nfat2fl/fl × Cd4-cre, Nfat1−/− × Nfat2fl/fl × Cd4-cre, Nfat1−/−, and Nfat1−/− ×

Legend continued on following page

Vaeth et al. www.pnas.org/cgi/content/short/1203870109 13 of 16

Nfat4−/− mice were isolated and stimulated for 3 d with anti-CD3/28 in the presence of TGF-β. After 4 d the CD25 expression on Foxp3+ nTreg and iTreg wasmeasured by FACS. (D and E) Analysis of genes potentially regulated by NFAT:Foxp3-complexes using microarray gene profiling. (D) Venn diagram illustrationsof genes regulated more than twofold between Nfat1−/− × Nfat2fl/fl × Cd4-cre (DKO) and WT in CD4+CD25+ nTreg and CD4+CD25− Tconv stimulated for 24 hwith anti-CD3/28 in presence (Tconv+TGF-β, iTreg polarizing conditions) or without TGF-β (nTreg and Tconv). Data represent two completely independentbiological experiments. (E) Representation of suggested NFAT:Foxp3-regulated genes in DKO and WT nTreg, Tconv, and Tconv+TGF-β stimulated for 24 h withanti-CD3/28 as heatmap (Left) or bar graphs (Right). Clustering analysis and heatmap of expression values show the log2 transformed expression intensity ofthe genes. Genes regulated ≥twofold between DKO and WT are tagged with an asterisk. Data from two independent biological experiments are given as anaverage of reliable entities representing denoted genes.

Vaeth et al. www.pnas.org/cgi/content/short/1203870109 14 of 16

% n

ucle

ar N

FAT

2

0

20

40

60

80

100

0

20

40

60

80

100

CD4+ T cells w/o TGF CD4+CD25– T cells + TGF

Foxp3+Foxp3–Foxp3+Foxp3–

p < 0.0001 p < 0.0001

CD4+CD25– Tconvw/o stimulation

CD4+CD25+ nTregw/o stimulation

CD4+CD25– Tconvanti-CD3/28

CD4+CD25+ nTreganti-CD3/28

NFAT2 Foxp3 DAPI

D

C

A

B

NFAT2 Foxp3 Smad3 DAPI

CD4+CD25–

Tconv

CD4+CD25+

nTreg

w/o

TGFβ

CD4+CD25–

Tconv

CD4+CD25–

iTreg

0.0

0.5

1.0

1.5

nTregTconv

w/o stimulation anti-CD3/28

ratio

nuc

leus

/cyt

osol

NFA

T2

fluor

esce

nce

sign

al p =0.002

% n

ucle

ar N

FAT

2

E F

0 0.5 1 1.5 2 4 6 8 16 24 480

20

40

60

80

100

time (h) on plate-bound anti-CD3/28

cells

with

nucl

ear

NFA

Tc1

( %

)

G

0 h

0.5 h

1 h

1.5 h

2 h

4 h

6 h

8 h

16 h

24 h

48 h

Indo1-A

M (

viole

t/blu

e)

CD4+CD25– TconvCD4+CD25+ nTreg

time (min)

CD3e crosslink

1 mMCaCl2

Iono baseline

CD4+ neg. isolated cells

CD

25

CD4

CD4+CD25– TconvCD4+CD25+ nTreg

NFAT2 Foxp3 DAPINFAT2DAPI

Fig. S10. Impaired nuclear translocation of NFAT2 in Foxp3+ cells. (A and B) Four-color staining of NFAT2 (red), Foxp3 (yellow), Smad3 (blue), and chromatin(cyan) of freshly isolated CD4+ conventional T cells (Tconv) and CD4+CD25+ nTregs stimulated for 6 h with anti-CD3 and anti-CD28 (Upper). Foxp3− Tconv andTGF-β-induced Foxp3+ iTregs generated from CD4+CD25− Tconv after 24 h stimulation with anti-CD3 plus anti-CD28 (Lower). (B) Quantification of nuclear andcytosolic fluorescence signal of NFAT2 in freshly isolated CD4+ Tconv and CD4+CD25+ nTregs by confocal microscopy, nuclei were demarcated by DAPI staining.The ratio of nuclear/cytosolic signal was calculated of unstimulated (w/o) and 6 h anti-CD3/CD28 stimulated cells, data are mean ± SEM from at least 100 cellsout of three individual experiments. (C) Histograms of one representative Foxp3− Tconv and nTreg cell, respectively, show different NFAT2 nuclear trans-location after 6 h stimulation by anti-CD3/28. Red line represents NFAT2 staining, yellow line Foxp3, and the nucleus is demarcated by cyan DAPI staining.(D) Quantification of cells with nuclear NFAT2 in CD4+ Tconv and CD4+CD25+ nTreg (Left) and 24 h TGF-β–induced Foxp3+ iTreg (Right), data are mean ± SEM

Legend continued on following page

Vaeth et al. www.pnas.org/cgi/content/short/1203870109 15 of 16

from more than 100 cells out of four individual experiments. (E and F) CD4+ T cells were stimulated with plate-bound anti-CD3/28 for the indicated time points,and NFAT2 nuclear translocation was recorded using confocal microscopy. Representative pictures (E) and the quantification of more than 30 individual cells(F) are shown. (G) Calcium-influx measurement in CD4+CD25− Tconv (red line) and CD4+CD25+ nTreg cells (blue line) by real-time flow cytometry. CD4+ T cellswere loaded with calcium-sensitive dye Indo1-AM and stained subsequently with anti-CD25-PE and biotinylatd anti-CD3ε. After 1 min the CD3ε Abs were cross-linked with streptavidin followed by the addition of 1 mM CaCl2 after 3.5 min. As a positive control Iono was added after 9 min of measurement, as a negativecontrol nonactivated cells were analyzed again as a baseline after 12 min.

Vaeth et al. www.pnas.org/cgi/content/short/1203870109 16 of 16