Requirement for balanced Ca/NFAT signaling inhematopoietic and embryonic developmentMartin R. Muller, Yoshiteru Sasaki1, Irena Stevanovic, Edward D. Lamperti, Srimoyee Ghosh, Sonia Sharma,Curtis Gelinas, Derrick J. Rossi, Matthew E. Pipkin, Klaus Rajewsky, Patrick G. Hogan, and Anjana Rao2

Department of Pathology and Immune Disease Institute, Harvard Medical School, 200 Longwood Avenue, Boston, MA 02115

Contributed by Anjana Rao, January 6, 2009 (sent for review October 21, 2008)

NFAT transcription factors are highly phosphorylated proteins resid-ing in the cytoplasm of resting cells. Upon dephosphorylation by thephosphatase calcineurin, NFAT proteins translocate to the nucleus,where they orchestrate developmental and activation programs indiverse cell types. NFAT is rephosphorylated and inactivated throughthe concerted action of at least 3 different kinases: CK1, GSK-3, andDYRK. The major docking sites for calcineurin and CK1 are stronglyconserved throughout vertebrate evolution, and conversion of eitherthe calcineurin docking site to a high-affinity version or the CK1docking site to a low-affinity version results in generation of hyper-activable NFAT proteins that are still fully responsive to stimulation.In this study, we generated transgenic mice expressing hyperacti-vable versions of NFAT1 from the ROSA26 locus. We show thathyperactivable NFAT increases the expression of NFAT-dependentcytokines by differentiated T cells as expected, but exerts unexpectedsignal-dependent effects during T cell differentiation in the thymus,and is progressively deleterious for the development of B cells fromhematopoietic stem cells. Moreover, progressively hyperactivableversions of NFAT1 are increasingly deleterious for embryonic devel-opment, particularly when normal embryos are also present in utero.Forced expression of hyperactivable NFAT1 in the developing embryoleads to mosaic expression in many tissues, and the hyperactivableproteins are barely tolerated in organs such as brain, and cardiac andskeletal muscle. Our results highlight the need for balanced Ca/NFATsignaling in hematopoietic stem cells and progenitor cells of thedeveloping embryo, and emphasize the evolutionary importance ofkinase and phosphatase docking sites in preventing inappropriateactivation of NFAT.

The activities of many signaling proteins and transcription factorsare tightly regulated by phosphorylation and dephosphoryla-

tion. Protein kinases and phosphatases bind to specific docking siteson these intracellular proteins to allow their activation or inactiva-tion at the appropriate location and time. A well-studied exampleof a transcription factor regulated in this fashion is nuclear factorof activated T cells (NFAT) (1–3). In resting cells, NFAT proteinsare highly phosphorylated and reside in the cytoplasm; upon cellactivation, they are dephosphorylated by the calcium/calmodulin-dependent phosphatase calcineurin and translocate to the nucleus.NFAT transcription factors play a key role in orchestrating diversedevelopmental programs, including those of the immune, centralnervous, cardiovascular, and musculoskeletal systems (4–11).NFAT also is implicated in maintaining the quiescent state of stemcells in the skin (12).

NFAT activation is initiated by dephosphorylation of the NFATregulatory domain, a conserved 300-amino acid region locatedN-terminal to the DNA-binding domain (Fig. 1A) (2, 13). Thephosphorylated residues (serines) in this domain are distributedamong several classes of conserved serine-rich sequence motifs (14,15), and their phosphorylation status is maintained by the con-certed action of at least 3 families of kinases: CK1, GSK3, andDYRK (16–20). We have shown previously that enzyme–substratedocking interactions are required for efficient dephosphorylation ofthe NFAT1 regulatory domain by calcineurin (21, 22) and forefficient phosphorylation of the SRR-1 motif by CK1 (20). Themajor docking sites for calcineurin and CK1 are located near the N

terminus of NFAT (Fig. 1A), are conserved among NFAT proteins,and fit the consensus sequences PxIxIT and FxxxF, respectively(20–22) (Fig. 1B). Substitution of the calcineurin docking sequence,SPRIEIT, with its high-affinity variant, HPVIVIT, and substitutionof the CK1 docking sequence, FxxxF, with a low-affinity version,ASILA, both result in partial nuclear localization of NFAT1 in amanner that is still inhibited by CsA (20, 21). Thus, these mutantNFAT1 proteins are not constitutively (i.e., irreversibly) activated,but are hypereactive relative to wild-type NFAT1 in that theyremain responsive to stimulation.

Here, we have examined the effects of increased Ca/NFATsignaling by generating transgenic mice conditionally expressingdifferent hyperactivable mutants of NFAT1 from the ROSA26(R26) locus. We demonstrate that progressively hyperactivableNFAT1 proteins are increasingly deleterious during early embry-onic development and the development of hematopoietic stem cellsinto T and B cell lineages. We also show that low-level ectopicexpression of hyperactivable NFAT1 in the early embryo leads tomosaicism in many tissues and is barely tolerated in organs such asbrain, heart, and skeletal muscle, where NFAT function is knownto be essential. In contrast, expression of hyperactivable NFAT1proteins at a late stage of T cell differentiation is well tolerated andleads to a hyperresponsive phenotype in peripheral T cells. Takentogether, our data provide strong evidence for the necessity ofbalanced Ca/NFAT signaling in progenitor cells of the developingembryo as well as in lymphocyte development, and shed new lighton the importance of the evolutionary conservation of phosphataseand kinase docking sites in preventing inappropriate activation ofthe Ca/NFAT signaling pathway.

ResultsMutation of Conserved Docking Sites for Calcineurin and CK1 MakesNFAT Hyperactivable. To assess the biological consequences ofincreased NFAT signaling in different tissues, we generated hyper-activable mutants of NFAT1. Previous studies have used NFATproteins bearing alanine substitutions in phosphorylated serines inthe regulatory domain, which are constitutively and irreversiblyactive (9, 15, 23). Instead, we chose to generate hyperactivable,stimulus-responsive versions of NFAT1 by mutating the CK1 andcalcineurin docking sites to lower and higher affinities, respectively.The conserved CK1 docking site (FSILF in NFAT1) was altered tothe low-affinity docking version ASILA (20), yielding ASILA-NFAT1 (abbreviated A-NFAT1); the conserved calcineurin dock-ing site (SPRIEITPS in NFAT1) was altered to the high-affinity

Author contributions: M.R.M., Y.S., K.R., P.G.H., and A.R. designed research; M.R.M., Y.S.,I.S., E.D.L., S.G., S.S., C.G., D.R., and M.E.P. performed research; D.R., M.E.P., and K.R.contributed new reagents/analytic tools; M.R.M., I.S., E.D.L., S.G., S.S., C.G., D.R., M.E.P., andP.G.H. analyzed data; and M.R.M., P.G.H., and A.R. wrote the paper.

The authors declare no conflict of interest.

1Present address: RIKEN Center for Developmental Biology, Laboratory for Stem CellBiology, 2-2-3 Minatojima-minamimachi, Kobe 650-0047, Japan.

2To whom correspondence should be addressed. E-mail: arao@idi.harvard.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0813296106/DCSupplemental.

7034–7039 � PNAS � April 28, 2009 � vol. 106 � no. 17 www.pnas.org�cgi�doi�10.1073�pnas.0813296106

Dow

nloa

ded

by g

uest

on

Oct

ober

21,

202

0

version HPVIVITGP (22), yielding VIVIT-NFAT1 (abbreviatedV-NFAT1); and both mutations were combined to yield a proteinexpected to be even more responsive to stimulation (ASILA-VIVIT-NFAT1, abbreviated AV-NFAT1). When tested by tran-sient transfection into HEK293 cells, the hyperactivable mutantswere increasingly dephosphorylated relative to wild-type NFAT1,in the expected order A-NFAT1 � V-NFAT1 � AV-NFAT1 (Fig.1C). We confirmed that the mutants were hyperactivable, notconstitutively active, by retrovirally expressing them in CD8 T cellsfrom NFAT1�/� mice (24) (Fig. 1D, Fig. S1). When stimulated withPMA plus increasing concentrations of the calcium ionophoreionomycin, T cells expressing the hyperactivable NFAT1 mutantsexhibited a clear shift, relative to cells expressing wild-type NFAT1,in their dose–response curve for expression of the cytokines IFN-�and TNF, with the order of responsiveness being AV � V � A �wild type. Cytokine expression was strictly dependent upon stim-ulation and was fully inhibited by the calcineurin inhibitor cyclo-sporine A (CsA).

Hyperactivable NFAT1 Proteins Are Deleterious During Early Embry-onic Development. To examine the effects of expressing the hyper-activable proteins in different cellular lineages in vivo, we generatedtransgenic mice conditionally expressing V-NFAT1 and AV-NFAT1 from the ROSA26 (R26) locus (25) (Fig. S2). In these mice,one or both alleles of the ROSA26 gene are replaced by a floxedcassette containing a neomycin-resistance (NeoR) gene and 3tandem transcriptional stop sites (abbreviated STOPflox), followedimmediately by the hyperactivable V-NFAT1 or AV-NFAT1 trans-gene. Expression of the hyperactivable proteins is controlled byCre-mediated excision of the STOPflox cassette, and it can bemonitored at a single-cell level by concomitant expression of EGFPfrom an internal ribosome entry site (IRES) (26). As a control, weused R26STOPflox-YFP reporter mice (27).

As a first step in the analysis, male mice of all 3 lines were bredto female CMV-Cre transgenic mice (deleter mice) (28). The Cre

transgene in this strain is under transcriptional control of a humancytomegalovirus minimal promoter and is expressed transientlyduring early embryogenesis (before implantation), leading to de-letion of loxP-flanked gene segments in all tissues, including germcells (Fig. S3). Because the Cre transgene in this strain is X-linked(28), female CMV-Cre transgenic mice were used for all crosses;this avoids the problem that the paternal X chromosome is inac-tivated before implantation, reactivated in blastocysts, and ran-domly inactivated in somatic tissues thereafter (29–33). In contrast,offspring of female CMV-Cre transgenic mice carry the Cretransgene on their maternal X chromosome, which is consistentlyactive before blastocyst implanation (29, 30).

Unexpectedly, we found that widespread expression of AV-NFAT1 in vivo was deleterious to embryonic development, withV-NFAT1 having a lesser effect. We bred heterozygous maleR26STOPflox-V-NFAT1/R26� or R26STOPflox-AV-NFAT1/R26� mice tohomozygous female CMV-CreCre/Cre (deleter) mice. Mendelian genet-ics predict that 50% of the offspring would have the genotypeR26V-NFAT1/R26�, CMV-Cre or R26AV-NFAT1/R26�, CMV-Cre[i.e., possess 1 copy of the wild-type ROSA26 allele (R26�), 1 copyof the expressible version of the hyperactivable NFAT1 in which theNeoR-STOP cassette has been deleted, and 1 copy of the Cre trans-gene], whereas the other 50% would have the genotype R26�/R26�,CMV-Cre (Fig. S4A). Instead, we observed an increasing competi-tive disadvantage in utero for pups capable of expressing hyperac-tivable NFAT1 (Fig. 2 A and B). Instead of the 50% expected,�43% of the offspring of R26STOPf lox-V-NFAT1/R26��CMV-CreCre/Cre breedings possessed an R26V-NFAT1 allele, andfewer than 10% of the offspring of R26STOPflox-AV-NFAT1/R26��CMV-CreCre/Cre breedings possessed an R26AV-NFAT1 allele capableof driving expression of hyperactivable NFAT1 (Fig. 2A). In linewith these observations, litter sizes decreased progressively in theoffspring of these crosses (R26YFP� R26V-NFAT1 � R26AV-NFAT1;Fig. 2B), emphasizing the dependence of the deleterious effect onthe degree of hyperactivability of the transgenic proteins.

NFAT1SP3SP2SRR-1SPRIE

IT

FSILF

NLS

CK1docking

Cndocking

400

TAD REGULATORY

KTS

kinase: CK1 GSK3 DYRK

700 928

DNA-BINDING C-TERM

1

Calcineurin docking site

NFAT1: SPRIEITPSNFAT2: SPRIEITSYNFAT3: CPSIRITSINFAT4: CPSIQITSIConsensus: PxIxIT

High affinity version HPVIVITGP

A

BCK1 docking site

NFAT1: FSILFNFAT2: FDFEFLFEFNFAT3: FKLVFNFAT4: FKLVFConsensus: FxxxF

Low affinity version ASILA

NFAT1

NFAT1- PNFAT1- de P

wt ASILA VIVIT AV

EGFP

control

TNFIFN-γ

Ionomycin [nM]

Pos

itive

cel

ls [%

of M

ax]

0

10

20

30

40

50

60

70

80

90

100

0 200 400 600 800 1000

0

10

20

30

40

50

60

70

80

90

100

0 200 400 600 800 10000

10

20

30

40

50

60

70

80

90

100

0 200 400 600 800 1000

Pos

itive

cel

ls [

%]

0

10

20

30

40

50

60

70

80

90

100

0 200 400 600 800 1000

wt NFAT1A-NFAT1V-NFAT1AV-NFAT1

C

D

Fig. 1. Conservation of kinase and phosphatase docking sites on NFAT proteins and analysis of hyperactivable NFAT1 mutants. (A) Schematic overview of NFAT1. TADindicates transactivation domain; Cn, calcineurin; NLS, nuclear localization signal. (B) Conservation of calcineurin and CK1 docking sites in NFAT proteins. Consensusmotifs and modified motifs with altered affinities (in NFAT1) are shown. (C) HEK293 cells were transduced with retroviral expression plasmids encoding EGFP orHA-tagged wild-type (wt) NFAT1, A-NFAT1, V-NFAT1 or AV-NFAT1, and NFAT1 phosphorylation status was assessed by immunoblotting with an anti-HA antibody. (D)CD8 T cells from NFAT1�/� mice were retrovirally transduced to express wt or hyperactivable NFAT1, then left unstimulated or stimulated with PMA and increasingconcentrationsof ionomycinfor4h.Resultsarerepresentedbothas thepercentageofpositivecells (topgraphs)andnormalizedtoset themaximumnumberofpositivecells in each series to 100% to show the shift in the dose-response curve more clearly (bottom graphs).

Muller et al. PNAS � April 28, 2009 � vol. 106 � no. 17 � 7035

DEV

ELO

PMEN

TAL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Oct

ober

21,

202

0

To further delineate the effects of hyperactivable NFAT1 inutero, we performed the crosses in the opposite direction and didtimed pregnancy experiments. We bred homozygous maleR26STOPflox-AV-NFAT1/R26STOPflox-AV-NFAT1 mice to heterozygousfemale CMV-CreCre/� mice and analyzed the offspring of thepregnant females at embryonic day 18.5. Again, 50% of theoffspring were expected to express Cre and delete the NeoR-STOPcassette, therefore becoming capable of expressing the hyperacti-vable NFAT1 (genotype R26AV-NFAT1/R26�). However, the aver-age fraction of viable embryonic day 18.5 embryos carrying theexpressible AV-NFAT1 transgene was again significantly lowerthan expected from Mendelian genetics (50% expected, 30%observed), and multiple dead, involuted embryos were seen (Fig.2C and S4B). These results indicate that mice globally expressinghyperactivable V-NFAT1 or AV-NFAT1 have a survival disadvan-tage and are not competitive with their littermate controls in utero.This is not due to deleterious effects of expression of the Cretransgene (34) (see SI Text). We monitored germ line transmissionof the hyperactivable NFAT1 alleles by breeding mosaic animalscarrying the recombined R26 locus (R26V-NFAT1/R26�, CMV-Creor R26AV-NFAT1/R26�, CMV-Cre) to wild-type C57BL/6 mice (Fig.S5). The expressed V-NFAT1 transgene was transmitted to off-spring at a significantly lower frequency than expected (50%expected, 26% observed), and the recombined AV-NFAT1 allelewas never transmitted through the germ line. Thus, there is efficientgerm-line selection against hyperactivable NFAT1 proteins in amanner that correlates with their degree of hyperresponsiveness.

Expression of Hyperactivable NFAT1 Leads to Mosaicism in ManyTissues and Is Not Tolerated in Brain, Heart, and Skeletal Muscle. Weattempted to force expression of hyperactivable NFAT1 bybreeding homozygous R26V-NFAT1 or R26AV-NFAT1 mice to ho-mozygous CMV-Cre mice. All of the offspring of these crossesare expected to possess the STOPf lox cassette-deletedR26V-NFAT1 or R26AV-NFAT1 allele, and so should be capable ofexpressing the hyperactivable NFAT1. The crosses yielded liveoffspring that appeared phenotypically normal. Consistent withthe deleterious effects documented above, however, litter sizesin these breedings were increasingly compromised comparedwith litters bearing the expressible R26YFP transgene (Fig. 2D).

Remarkably, expression of hyperactivable NFAT1 was mosaic inmany tissues of surviving R26V- or AV-NFAT1/R26�, CMV-Cre mice,and it seemed not to be tolerated in others. Circulating T and B cellsshowed increasingly mosaic expression of hyperactivable V-NFAT1or AV-NFAT1, as judged by expression of the linked EGFP (Fig.3A; representative primary data are shown in Fig. S6). Breeding ofhomozygous R26STOPflox-YFP/R26STOPflox-YFP control mice to ho-mozygous CMV-CreCre/Cre mice resulted in stop cassette excisionand transgene expression in close to 100% of T and B cells in the

Fig. 3. Mosaicism in mice expressing hyperactivable NFAT1 early in embry-onic development. (A) Mosaicism in B and T lymphocytes from 8–12 wk oldmice expressing different R26 transgenes. Peripheral blood from R26YFP/R26�,CMV-Cre (n � 19); R26V-NFAT1/R26�, CMV-Cre (n � 19) and R26AV-NFAT1/R26�,CMV-Cre (n � 21) transgenic mice was drawn from tail veins and analyzed byflow cytometry. The levels of IRES-EGFP expression in B220� B lymphocytes,CD4� and CD8� T lymphocytes are shown as box-and-whisker diagrams(lower whisker: first quartile, blue box: second quartile, green box: thirdquartile, upper whisker: fourth quartile, line separating blue and green box:median). (B–D) Immunohistochemistry for EGFP on different tissues fromR26YFP/R26�, CMV-Cre; R26V-NFAT1/R26�, CMV-Cre and R26AV-NFAT1/R26�, CMV-Cre transgenic mice. Size bars measure 100 �m as indicated. Tissue sectionswere counterstained with hematoxylin.

Litters 2 4 8

Animals 19 28 39

0

2

4

6

8

10

12

14

YFP V-NFAT1 AV-NFAT1

R26STOPflox/R26+ X CMV-CreCre/Cre

Lit

ter

size

[p

up

s]R26STOPflox-AV-NFAT1/ R26STOPflox-AV-NFAT1 X CMV-CreCre/+ (E18.5) (3 litters)

0

1

2

3

4

5

6

7

8

Ave

rag

e #

of

viab

le

em

bry

os/

litte

r

R26STOPflox-AV-NFAT1/R26+ (unrecombined)

R26AV-NFAT1/R26+, CMV-Cre (recombined)

0

5

10

15

20

25

expected actual

V-NFAT1 AV-NFAT1

n=28 n=39

R26STOPflox/R26+ X CMV-CreCre/Cre

Pu

ps

wit

h r

eco

mb

ined

R26

locu

s

n.s. p<1x10-8

p<0.052p<0.003

p<0.03p < 0.002

0

2

4

6

8

10

12

14

YFP V-NFAT1 AV-NFAT1

Lit

ter

size

[p

up

s]

Litters 2 3 5

Animals 19 19 21

R26STOPflox/R26STOPflox X CMV-CreCre/Cre

n.s.

A B

C D

Fig. 2. Expression of hyperactivable NFAT1 is deleterious during embryonicdevelopment. (A) Expected and actually detected frequencies of the recom-bined (NeoR-STOP cassette deleted) R26 transgenes in offspring of differentR26STOPflox/R26� � CMV-CreCre/Cre crosses. P values were determined using astandard �2 test. n.s., not significant. (B) Litter sizes of different R26STOPflox/R26� � CMV-CreCre/Cre crosses, P values were determined using a standardStudent’s t test. (C) Frequency of the unrecombined (R26STOPflox-AV-NFAT1/R26�)and recombined (R26AV-NFAT1/R26�, CMV-Cre) R26 allele at embryonic day 18.5(E18.5). Male R26STOPflox-AV-NFAT1/R26STOPflox-AV-NFAT1 mice were crossed withfemale CMV-CreCre/� mice. Pregnant females were sacrificed on E18.5 andembryos were assessed for viability and genotype. The average values forviable embryos from 3 litters with standard deviations are shown. P valueswere determined using a standard Student’s t test. (D) Litter sizes of differentR26STOPflox/R26STOPflox � CMV-CreCre/Cre crosses. P values were determined us-ing a standard Student’s t test.

7036 � www.pnas.org�cgi�doi�10.1073�pnas.0813296106 Muller et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

21,

202

0

vast majority of analyzed R26YFP/R26�, CMV-Cre animals (Fig.3A, left bars in each graph; median, 92.2–96.7%; n � 19). Incontrast, R26V-NFAT1/R26�, CMV-Cre mice were highly mosaic,exhibiting transgene expression in 0.15–89.6% of T and B cells(median, 25.2–39.5%; n � 19), and R26AV-NFAT1/R26�, CMV-Cremice showed transgene expression in only 0–30.8% of T and B cells(median, 5.3–9.4%; n � 21; Fig. 3A and Fig. S6). Again, therefore,the degree of mosaicism correlated with the degree of hyperacti-vability of the transgenic NFAT1 proteins.

We subsequently assessed the level of transgene expression indifferent tissues by immunohistochemistry. EGFP expression wasdetected in kidneys, lung, and spleens of R26V-NFAT1/R26�, CMV-Cre and R26AV-NFAT1/R26�, CMV-Cre mice, with staining gener-ally being more pronounced and involving a larger fraction of cellsin V-NFAT1-expressing relative to AV-NFAT1-expressing mice(Fig. 3 B and C). Even in R26V-NFAT1/R26�, CMV-Cre mice,however, there was a notable absence of EGFP staining in brain,heart, and skeletal muscle, suggesting that cells expressing even theless-hyperactivable V-NFAT1 protein were not competitive inpopulating these organs (Fig. 3D). We confirmed that EGFPexpression was tightly linked to expression of hyperactivableNFAT1 from the ROSA26 locus (see SI Text and Fig. S7).

Signal-Dependent Effects of Hyperactivable NFAT1 on T Cell Devel-opment in the Thymus. We further investigated the effect of AV-NFAT1 on hematopoietic stem cell differentiation down the T andB cell lineages by monitoring the extent of mosaicism in differentprecursor populations of R26AV-NFAT1/R26�, CMV-Cre mice. Wechose mice expressing different levels of EGFP in peripheral B andT cell populations, then analyzed bone marrow cells and thymocytesby flow cytometry for EGFP expression in hematopoietic stemcells, common lymphocyte precursors, and cells at different stagesof B and T cell differentiation. The range of EGFP expression inhematopoietic stem cells was 8.6–56.7% (n � 5). In all animals, thefraction of cells expressing AV-NFAT1 declined gradually fromhematopoietic stem cells to mature B lymphocytes in the B celllineage (Fig. 4A), with expression being �3 times higher in theearliest progenitor populations. In contrast, in the T cell lineage, asimilar gradual decline of AV-NFAT1 expression was overlaid bypeaks corresponding to the expression of the pre-TCR (DN1-to-

DN2 transition) and the TCR (DN4-to-DP transition; Fig. 4B) (35).These results strongly suggest that expression of hyperactivableNFAT1 in developing thymocytes modulates the strength of pre-TCR and TCR signaling so as to affect the proliferation and/orsurvival of T cells making these transitions.

Expression of Hyperactivable NFAT1 at a Late Stage of T Cell Differ-entiation Is Well Tolerated and Leads to a Hyperresponsive Phenotype.To evaluate the effects of hyperactivable NFAT1 at a late stage ofdifferentiation, we bred the R26-transgenic mice to CD4-Cre mice(36). These mice express the Cre recombinase under control of theCD4 promoter; thus, Cre is expressed beginning at the late ‘‘double-positive’’ stage of thymocyte differentiation, when both CD4 andCD8 are expressed, resulting in efficient excision of floxed DNAsegments in all peripheral T cells. More than 95% of peripheralCD4 and CD8 T cells from these mice were EGFP�, indicating thatthere was no selection against expression of the hyperactivableNFAT1 transgenes at the double-positive stage (Fig. 5A). Asexpected from earlier retroviral transduction experiments (Fig.1D), T cells from the mice showed significant hyperresponsivenessto stimulation, as assessed by accelerated nuclear translocation anddelayed nuclear export of NFAT (Fig. 5B), as well as substantiallyincreased cytokine expression upon stimulation with low concen-trations of stimulus (Fig. 5C). We took advantage of the uniformexpression to document that in these T cells, the expression levelsof hyperactivable NFAT1 from the ROSA26 locus were substan-tially lower than those of endogenous NFAT1 (Fig. S8). Thus, theobserved effects are due to hyperresponsiveness of the mutantproteins rather than mere overexpression.

DiscussionTo summarize, we have shown that low-level ectopic expression ofhyperactivable NFAT1 proteins has severe effects on progenitorcell function during embryonic and hematopoietic development inthe mouse. In the first generation, when R26V-NFAT1/R26� orR26AV-NFAT1/R26� mice are bred to CMV-CreCre/Cre mice, there isreduced representation of offspring that express the hyperactivabletransgene (Fig. 2). If, instead, global expression of hyperactivableNFAT1 is forced by breeding homozygous mice, the tissues ofsurviving progeny are variably mosaic for transgene expression,

Fig. 4. Effects of expression of hyperactivable NFAT1on B and T cell development. (A) (Upper) FACS analysis ofdifferent progenitor populations of B cell differentiationfor EGFP (AV-NFAT1) expression in 5 different mosaicAV-NFAT1 mice (R26AV-NFAT1/R26�, CMV-Cre). Hemato-poietic stem cells (HSCs), common lymphoid progenitors(CLPs), pro-B cells, pre-B cells, immature B cells, mature Bcells (bone marrow), and splenic B cells are shown.(Lower) Mean values and SDs for animals 1, 3, and 4,which exhibited comparable levels of mosaicism. (B) (Up-per) HSCs, CLPs, double-negative thymocytes 1 (DN1),DN2, DN3, DN4, early double-positive thymoctes (DPs),late DPs, and single-positives (SPs) are shown. (Lower)Mean values and SDs for animals 1, 3, and 4.

Muller et al. PNAS � April 28, 2009 � vol. 106 � no. 17 � 7037

DEV

ELO

PMEN

TAL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Oct

ober

21,

202

0

implying strong counterselection against developing transgene-positive cells (Fig. 3).

Because Cre expression in the CMV-Cre mouse occurs beforeimplantation (28), the mosaic expression of hyperactivable NFAT1in adult tissues of the surviving mice implies that Cre expression istransient, and that progenitor cells that escape the early wave ofCre-mediated deletion and therefore lack expression of hyperacti-vable NFAT1 have a competitive advantage over cells expressingthe hyperactivable protein (Fig. S9). Notably, the extent of thisadvantage correlates with the degree of hyperactivability of theNFAT1 protein. It is striking that the tissues that appear least

tolerant of forced expression of hyperactivable NFAT1—brain,heart, and skeletal muscle—are prominent among those in whichNFAT is known to be crucial for development and function (4–11).Our results also are consistent with a previous systematic analysisof physiological calcineurin substrates in yeast (37). The dockingaffinities of these substates for calcineurin ranged from 15 to 250�M; an engineered inappropriate increase in the affinity of onesuch substrate, Crz1, improved yeast growth under high-salt con-ditions but was deleterious under conditions of growth at high pH(37). Likewise, the hyperactivable NFAT1 that we have generatedappears to be deleterious during embryogenesis but advantageousin terms of increasing cytokine production by cultured T cells.

During lymphocyte development from hematopoietic stem cellsto terminally differentiated B and T cells, we observed a gradualdecline of the expression of hyperactivable NFAT1 through differ-ent progenitor cell populations. This is most likely due to a mildinability of the AV-NFAT1-expressing cell populations to competewith cells in which the hyperactivable protein is not expressed.These findings are consistent with a previous report showing thatas hematopoietic stem cells begin to differentiate, expression of allNFAT family members is downregulated (38, 39); in contrast,hyperactivable NFAT1 proteins expressed from the R26 locuswould not be downregulated, and progenitor cells expressing theseproteins, even at low levels, would display inappropriately sustainedNFAT activity, potentially conferring a competitive disadvantageon those cell populations. NFAT1 has been shown to repressexpression of the G0/G1 checkpoint kinase cyclin-dependent kinase4 (CDK4) (40). Sustained expression of AV-NFAT1 from the R26locus might thus lead to continuous repression of CDK4, whichcould at least in part explain the observed inability to compete ofthe AV-NFAT1-expressing cell populations. This considerationalso could account for the peaks observed during T cell develop-ment which, remarkably, correspond exactly with the time pointswhen either the pre-TCR or the TCR start to be expressed (35): Ifhyperactivable NFAT led to a signal-dependent decrease in the rateof cell cycle transit, cells expressing the hyperactivable protein couldpotentially increase in number relative to nonexpressing cells at thecheckpoint just before the pre-TCR/TCR signal was received.Alternatively, the hyperactivable NFAT might confer a signal-specific proliferation or survival advantage at these developmentalstages. Microarray and ChIP-chip analyses of enriched cell popu-lations will be needed to distinguish these possibilities.

Our experiments also provide insights into the evolutionaryconservation of the docking interactions of NFAT proteins with 2of their regulatory enzymes, calcineurin and CK1. The SPRI-EITPS � HPVIVITGP substitution of V-NFAT1 increases itsaffinity for calcineurin by 30- to 50-fold compared with the wild-type protein (from 25–30 �M for SPRIEITPS to 0.5–1 �M forHPVIVITGP) (41), indicating that a reasonably small change inbinding affinity for calcineurin (�2.5 kcal/mole) results in hyper-activability and heightened activation (and therefore dysregulation)of NFAT1. The decrease in CK1 affinity caused by the FSILF �ASILA substitution has not been accurately measured, but it islikely to be modest, because mutation of a similar sequence in apeptide from �-catenin did not have a significant effect on itsefficiency of phosphorylation by CK1 (42). Unexpectedly, even thisdegree of dysregulation of NFAT signaling led to a change in the‘‘fitness’’ of cells and embryos that was detectable in a singlegeneration. Expression of hyperactivable NFAT proteins later indevelopment appears less deleterious to cellular fitness, however,suggesting that expressing hyperactivable NFAT at late develop-mental stages, or acutely in tissues of the adult (by breeding the R26mice to Cre-ER mice, followed by tamoxifen administration, forinstance), will be a valuable strategy for understanding the biolog-ical role of NFAT and identifying NFAT target genes in diversetissues of interest.

t [min.]

Cells

with

nuc

lear

NFA

T1 [%

]

wt NFAT1

0102030405060708090

100

0 15 30 45 60 120 180 3000

102030405060708090

100

0 60 120 180 240 300 360

AV-NFAT1wt NFAT1

wt NFAT1 (CsA at t=30 min.)

CsA (30’)

CsA (-10’)

100

101

102

103

104

0

20

40

60

80

100

100

101

102

103

104

0

20

40

60

80

100%

of

Max

.

EGFP

CD4 CD8

95.2% 96.5%

R26STOPflox-AV-NFAT1/R26+ R26AV-NFAT1/R26+, CD4-CreCre/+

0

10000

20000

30000

40000

50000

60000

70000

80000

0 200 400 600 800 1000 1200

0102030405060708090

100

0 200 400 600 800 1000 1200

Ionomycin [nM]

Intracellular staining Cytokine Bead Assay

IFN

γ−po

sitiv

e ce

lls [%

]

IFN

γ [p

g/m

L]

CD4-CreCre/+ R26AV-NFAT1/R26+, CD4-CreCre/+ (G6;n=3)

R26AV-NFAT1/R26+, CD4-CreCre/+ (G7;n=3)

A

B

C

Fig. 5. Effects of expression of hyperactivable NFAT1 at a late stage of T celldifferentiation. (A) FACS analysis of CD4 and CD8 T cells from R26AV-NFAT1/R26�,CD4-Cre transgenic mice for EGFP expression (blue line). R26STOPflox-AV-NFAT1/R26�

mice were used as negative controls. (B) NFAT1 translocation assay for endoge-nous NFAT1 and AV-NFAT1. CD8 T cells were isolated from a C57BL/6 wild-typemouse and an R26AV-NFAT1/R26�, CD4-Cre mouse. Cells were expanded in IL-2,collected on day 5, and stimulated for various time intervals with 10 nM PMA and1 �M ionomycin (black and red curves; Right). To assess the effects of calcineurininhibition,1�MCsAwasaddedeither10minbeforestimulation(graycurve;Left)or 30 min after stimulation (green curve; Left). (C) IFN-� production in CD8 T cellsfrom R26AV-NFAT1/R26�, CD4-Cre transgenic mice. Cells from 6 R26AV-NFAT1/R26�,CD4-Cre transgenic mice (3 G6, 3 G7; G6 and G7 are 2 different clones of targetedES cells that have been used for blastocyst injection) were isolated and expandedin IL-2. On day 5, cells were stimulated with 10 nM PMA, and various concentra-tions of ionomycin and IFN-� production were assessed by using intracellularcytokine staining and flow cytometry (Left) or using a cytokine bead assay todetermine the accumulated IFN-� in the supernatant (Right) Identically treatedCD8 T cells from CD4-Cre mice were used as controls. Each value represents themean � SD.

7038 � www.pnas.org�cgi�doi�10.1073�pnas.0813296106 Muller et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

21,

202

0

Materials and MethodsTransfection of HEK293 Cells and Immunoblotting. HEK293 cells were transfectedwith10�goftheretroviral constructsKMV-wild-typeNFAT1,KMV-ASILA-NFAT1,KMV-VIVIT-NFAT1, and KMV-AV-NFAT1 by using calcium phosphate precipita-tion.EmptyKMV-EGFPwasusedasacontrol.Atotalof50�gofprotein lysatewasresolved by SDS/PAGE and analyzed by immunoblotting using the monoclonalmouse anti-HA antibody 12CA5 (1:1,000).

Mice. CD4-Cre mice were purchsed from Taconic. All mice were on the C57BL/6genetic background and were housed under specific pathogen-free conditions.All experiments were performed in concordance with protocols approved by theHarvard University Institutional Animal Care and Use Committee and by theImmune Disease Institute.

T Cell Isolation, Retroviral Transduction, and Differentiation. T cell isolation,retroviral transduction, and differentiation were performed as described previ-ously (43). A more detailed description is included in the SI Text.

T Cell Stimulation and Measurement of Cytokines. On day 4 or 5 after isolation,T cells were stimulated with 10 nM PMA and various concentrations of ionomycin(0 nm to 1 �M) for 4 h. Brefeldin A (10 �g/mL; Sigma) was added for the last 2.5 hof stimulation. T cells were subsequently fixed in 2% paraformaldehyde, stainedintracellularly for TNF (phycoerythrin-conjugated anti-mouse TNF; eBioscience)and IFN-� (allophycoerythrin-conjugated anti-mouse IFN-�; eBioscience), and ana-lyzed by flow cytometry. TNF and IFN-� concentrations in the cell supernatantwere determined by using the BD Cytometric Bead Array (Mouse Th1/Th2 CytokineKit; BD Bioscences) according to the instructions provided by the manufacturer.

Conditional Gene Targeting and Genotyping. The cDNAs encoding for HA-tagged V-NFAT1 and AV-NFAT1 were cloned into a modified version ofpROSA26–1 (25), which also contains an frt-flanked IRES-EGFP cassette and abovine polyadenylation sequence (26) (Fig. S1). B6 ES cells (Artemis Pharmaceu-ticals) derived from the C57BL/6 mouse strain were transfected, cultured, andselected as described previously (44). Chimeric mice with targeted R26 alleleswere generated by blastocyst injection of heterozygous R26STOPflox-V-NFAT1 or

R26STOPflox-AV-NFAT1 embryonic stem cell clones. Two targeted clones were injectedfor each allele. Germ-line transmission of the targeted alleles was achieved bybreeding chimeric mice with C57BL/6 albino mice. Genotyping for the unrecom-binedR26allelewasperformedbyPCRusingtheprimerpair5-CTGCGTGTTCGAATT CGC CAA TGA-3 and 5-GGC AGC TTC TTT AGC AAC AAC CGT-3. Therecombined R26 allele (with the Neo-STOP cassette excised upon Cre recombi-nation) was detected by PCR using the primers 5-TTG AGG ACA AAC TCT TCGCGG TCT-3 and 5-CCC GCA TAG TCA GGA ACA TCG TAT-3 or by detecting EGFPexpression in lympocytes by flow cytometry.

Flow Cytometric Analysis. For analysis of blood samples, a volume of �50 �L ofperipheral blood was obtained from mouse tail veins by using heparinized glasscapillaries (Drummond). Blood cells were washed twice in FACS buffer, stainedwith fluorochrome-conjugated antibodies, and subsequently analyzed by flowcytometry. A detailed description of the analysis of specific subpopulations isincluded in the SI Text.

Immunohistochemistry. Tissues were fixed by immersion in 4% formaldehydeand then dehydrated and embedded in paraffin for sections at 5–6 �m thickness.Sectionsweredeparaffinatedandpretreatedwith1mMEDTA(pH8.0).Antibodyincubations were performed with reagents from a DAB/horseradish peroxidase-based staining kit (Dako), including peroxidase-block pretreatment. Primaryincubation was with a rabbit polyclonal antiserum to EGFP/YFP (ab290; Abcam).Following development of DAB staining, the sections were counterstained withhematoxylin.

NFAT1 Nuclear Translocation Assay. NFAT1 translocation was assessed as de-scribed previously (45). A detailed description is included in the SI Text.

ACKNOWLEDGMENTS. We thank D. Ghitza for help with blastocyst injectionsof ES cells and K. Ketman for help with cell sorting. This study was supportedby National Institutes of Health grants (to K.R. and A.R.); a T32 training grant(to S.G.), a Deutsche Krebshilfe postdoctoral fellowship (to M.R.M.), a CancerResearch Institute postdoctoral fellowship (to M.R.M.), a Canadian Institutesof Health Research postdoctoral fellowship (to S.S.), and a Leukemia andLymphoma Society postdoctoral fellowship (to S.S.).

1. Crabtree GR, Olson EN (2002) NFAT signaling: Choreographing the social lives of cells. Cell109(Suppl):S67–S79.

2. Hogan PG, Chen L, Nardone J, Rao A (2003) Transcriptional regulation by calcium, cal-cineurin, and NFAT. Genes Dev 17:2205–2232.

3. MacianF(2005)NFATproteins:KeyregulatorsofT-celldevelopmentandfunction.NatRevImmunol 5:472–484.

4. de la Pompa JL, et al. (1998) Role of the NF-ATc transcription factor in morphogenesis ofcardiac valves and septum. Nature 392:182–186.

5. Ranger AM et al. (1998) The transcription factor NF-ATc is essential for cardiac valveformation. Nature 392:186–190.

6. Chang CP et al. (2004) A field of myocardial-endocardial NFAT signaling underlies heartvalve morphogenesis. Cell 118:649–663.

7. Molkentin JD, et al. (1998) A calcineurin-dependent transcriptional pathway for cardiachypertrophy. Cell 93:215–228.

8. Chin ER, et al. (1998) A calcineurin-dependent transcriptional pathway controls skeletalmuscle fiber type. Genes Dev 12:2499–2509.

9. Winslow MM, et al. (2006) Calcineurin/NFAT signaling in osteoblasts regulates bone mass.Dev Cell 10:771–782.

10. Takayanagi H, et al. (2002) Induction and activation of the transcription factor NFATc1(NFAT2) integrate RANKL signaling in terminal differentiation of osteoclasts. Dev Cell3:889–901.

11. Graef IA, et al. (2003) Neurotrophins and netrins require calcineurin/NFAT signaling tostimulate outgrowth of embryonic axons. Cell 113:657–670.

12. Horsley V, Aliprantis AO, Polak L, Glimcher LH, Fuchs E (2008) NFATc1 balances quiescenceand proliferation of skin stem cells. Cell 132:299–310.

13. Graef IA, Gastier JM, Francke U, Crabtree GR (2001) Evolutionary relationships among Reldomains indicate functional diversification by recombination. Proc Natl Acad Sci USA98:5740–5745.

14. Beals CR, Clipstone NA, Ho SN, Crabtree GR (1997) Nuclear localization of NF-ATc by acalcineurin-dependent, cyclosporin-sensitive intramolecular interaction. Genes Dev11:824–834.

15. Okamura H, et al. (2000) Concerted dephosphorylation of the transcription factor NFAT1induces a conformational switch that regulates transcriptional activity. Mol Cell 6:539–550.

16. Beals CR, Sheridan CM, Turck CW, Gardner P, Crabtree GR (1997) Nuclear export of NF-ATcenhanced by glycogen synthase kinase-3. Science 275:1930–1934.

17. Zhu J, et al. (1998) Intramolecular masking of nuclear import signal on NF-AT4 by caseinkinase I and MEKK1. Cell 93:851–861.

18. Arron JR, et al. (2006) NFAT dysregulation by increased dosage of DSCR1 and DYRK1A onchromosome 21. Nature 441:595–600.

19. Gwack Y, et al. (2006) A genome-wide Drosophila RNAi screen identifies DYRK-familykinases as regulators of NFAT. Nature 441:646–650.

20. Okamura H, et al. (2004) A conserved docking motif for CK1 binding controls the nuclearlocalization of NFAT1. Mol Cell Biol 24:4184–4195.

21. Aramburu J, et al. (1998) Selective inhibition of NFAT activation by a peptide spanning thecalcineurin targeting site of NFAT. Mol Cell 1:627–637.

22. Aramburu J, et al. (1999) Affinity-driven peptide selection of an NFAT inhibitor moreselective than cyclosporin A. Science 285:2129–2133.

23. Macian F, et al. (2002) Transcriptional mechanisms underlying lymphocyte tolerance. Cell109:719–731.

24. Xanthoudakis S, et al. (1996) An enhanced immune response in mice lacking the tran-scription factor NFAT1. Science 272:892–895.

25. Zambrowicz BP, et al. (1997) Disruption of overlapping transcripts in the ROSA beta geo26 gene trap strain leads to widespread expression of beta-galactosidase in mouse em-bryos and hematopoietic cells. Proc Natl Acad Sci USA 94:3789–3794.

26. Sasaki Y, et al. (2006) Canonical NF-kappaB activity, dispensable for B cell development,replaces BAFF-receptor signals and promotes B cell proliferation upon activation. Immu-nity 24:729–739.

27. Srinivas S, et al. (2001) Cre reporter strains produced by targeted insertion of EYFP andECFP into the ROSA26 locus. BMC Dev Biol 1:4.

28. Schwenk F, Baron U, Rajewsky K (1995) A cre-transgenic mouse strain for the ubiquitousdeletion of loxP-flanked gene segments including deletion in germ cells. Nucleic Acids Res23:5080–5081.

29. Huynh KD, Lee JT (2003) Inheritance of a pre-inactivated paternal X chromosome in earlymouse embryos. Nature 426:857–862.

30. Huynh KD, Lee JT (2005) X-chromosome inactivation: A hypothesis linking ontogeny andphylogeny. Nat Rev Genet 6:410–418.

31. Cheng MK, Disteche CM (2004) Silence of the fathers: Early X inactivation. Bioessays26:821–824.

32. Okamoto I, Otte AP, Allis CD, Reinberg D, Heard E (2004) Epigenetic dynamics of imprintedX inactivation during early mouse development. Science 303:644–649.

33. Mak W, et al. (2004) Reactivation of the paternal X chromosome in early mouse embryos.Science 303:666–669.

34. Schmidt-Supprian M, Rajewsky K (2007) Vagaries of conditional gene targetting. NatImmunol 8:665–668.

35. Borowski C, et al. (2002) On the brink of becoming a T cell. Curr Opin Immunol 14:200–206.36. Lee PP, et al. (2001) A critical role for Dnmt1 and DNA methylation in T cell development,

function, and survival. Immunity 15:763–774.37. Roy J, Li H, Hogan PG, Cyert MS (2007) A conserved docking site modulates substrate

affinity for calcineurin, signaling output, and in vivo function. Mol Cell 25:889–901.38. KianiA,etal. (2004)ExpressionandregulationofNFAT(nuclear factorsofactivatedTcells)

in human CD34� cells: Down-regulation upon myeloid differentiation. J Leukoc Biol76:1057–1065.

39. Kiani A, et al. (2007) Expression analysis of nuclear factor of activated T cells (NFAT) duringmyeloid differentiation of CD34� cells: Regulation of Fas ligand gene expression inmegakaryocytes. Exp Hematol 35:757–770.

40. BakshS,etal. (2002)NFATc2-mediatedrepressionofcyclin-dependentkinase4expression.Mol Cell 10:1071–1081.

41. Li H, Zhang L, Rao A, Harrison SC, Hogan PG (2007) Structure of calcineurin in complex withPVIVIT peptide: Portrait of a low-affinity signalling interaction. J Mol Biol 369:1296–1306.

42. Bustos VH, et al. (2006) The first armadillo repeat is involved in the recognition andregulation of beta-catenin phosphorylation by protein kinase CK1. Proc Natl Acad Sci USA103:19725–19730.

43. AnselKM,etal. (2004)Deletionofaconserved Il4 silencer impairsThelpertype1-mediatedimmunity. Nat Immunol 5:1251–1259.

44. Schmidt-Supprian M, et al. (2000) NEMO/IKK gamma-deficient mice model incontinentiapigmenti. Mol Cell 5:981–992.

45. Oh-Hora M, et al. (2008) Dual functions for the endoplasmic reticulum calcium sensorsSTIM1 and STIM2 in T cell activation and tolerance. Nat Immunol 9:432–443.

Muller et al. PNAS � April 28, 2009 � vol. 106 � no. 17 � 7039

DEV

ELO

PMEN

TAL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Oct

ober

21,

202

0