Nuclear Factor of Activated T Cells Regulates the Expression of Interleukin-4 in Th2 Cells in an All-or-none Fashion

Background: Not every T helper type 2 (Th2) lymphocyte imprinted to express interleukin-4 (IL-4) does so when activated. Results: Preventing nuclear translocation of the nuclear factor of activated T cells (NFAT) reduces the number of Th2 lymphocytes reexpressing IL-4. Conclusion: NFAT is the limiting factor determining digital IL-4 expression in Th2 lymphocytes. Significance: This might help us to understand the regulation of immunopathology in allergy and asthma.

T helper (Th) 2 lymphocytes regulate immune responses by expression of cytokines instructing themselves and other cells to qualified reactions. Different cytokines are expressed by different lineages of Th cells, to adapt immune responses to the diversity of pathogens. Differentiation of activated Th cells into a particular lineage is induced by costimulatory signals and determined by lineage-determining transcription factors. T-bet, GATA-3, and ROR␥t determine the Th1, Th2, and Th17 lineages, respectively (reviewed in Ref. 1). Lineage-determining master transcription factors are both essential and sufficient for the differentiation of Th cells into a distinct lineage.
Expression of GATA-3 is under the control of a distal promoter responsive to T cell receptor stimulation, and of a proximal promoter responsive to GATA-3 itself and to STAT6, the signal transducer of the receptor for the cytokine interleukin-4 (IL-4) (2,3). Once induced, GATA-3 expression is stabilized by a positive feedback loop (4,5). IL-4 is the signature cytokine of Th2 lymphocytes. GATA-3 is critical for the epigenetic imprinting of IL-4 for reexpression in reactivated Th2 cells (6). GATA-3 binds to a conserved intronic regulatory element (CIRE) in the first intron of the Il4 gene and induces its demethylation, which correlates with its imprinting for reexpression (7). GATA-3 has been described to block methyl CpG binding domain protein-2, which links DNA methylation to silent chromatin (8). Other regulatory elements of the Il4 gene include the hypersensitivity site Va (9) important for lineagespecific binding of NFAT to the Il4 locus and a locus control region (LCR) within the Rad50 gene upstream of the Il4 gene (10). In addition to GATA-3, some other transcription factors participating in the transcriptional control of the Il4 gene such as STAT6 (11), Brahma-related gene 1 (Brg1) (12), and Crebbinding protein CBP/p300 (13) have the ability to recruit histone acetyltransferases and block DNA methyltransferases. Th2 reexpress their imprinted Il4 gene upon restimulation of the T cell receptor (14), however, not all of them.
A substantial fraction of Th2 cells will not reexpress Il4 in a given restimulation. This is not due to an insufficient imprinting of the gene because the very same cells can reexpress the Il4 gene in later restimulations, with similar efficacy as their sister cells in the original restimulation (15). The reason for the failure of a Th2 cell to reexpress Il4 in a given restimulation could be either a rate-limiting, stochastic availablility of transcription factors controlling Il4 expression in the nucleus, leading to monoallelic expression of the Il4 gene, with some cells not expressing it at all (16). Alternatively, one transcription factor could control the assembly of the Il4 transcriptional complex in an all-or-none fashion. This has been demonstrated for the control of reexpression of the cytokine IL-2, which is dependent on translocation of NFATc2 into the nucleus (17). This translocation is dependent on complete dephosphorylation of NFATc2 at 13 positions by calcineurin (18), a reaction of second order, resulting in an all-or-none translocation of NFATc2 into the nucleus in individual Th2 cells. In addition, dephosphorylation at serine residues in the N-terminal transactivation domain is required for transcriptional activation.
Here, we show that in restimulated Th2 cells NFATc2 controls the reexpression of Il4 in an all-or-none fashion. NFATc2 translocation into the nucleus is required for assembly of the transcription factor complex at the Il4 promoter, which occurs only in IL-4-expressing Th2 cells, upon restimulation of the T cell receptor (TCR). Modulation of TCR signaling strength by graded inhibition of NFAT results in decreasing frequencies of IL-4-expressing Th2 cells. The amount of IL-4 produced by expressing cells is not affected. Thus, in Th2 cells NFAT serves as a molecular switch that translates graded differences in TCR signal strength into a digital decision to express IL-4 or not.

EXPERIMENTAL PROCEDURES
Mice-BALB/c, C57BL/6, OVA-TCRtg/tg DO11.10 (kind gift of Dennis Y. Loh and Kenneth Murphy, Washington University School of Medicine, St. Louis, MO), and OT-II mice were bred under specific pathogen-free conditions in our animal facility. The mice were sacrificed by cervical dislocation. All animal experiments were performed in accordance with institutional, state, and federal guidelines.
Isolation of Viable IL-4 Secreting Cells-Viable IL-4 secreting cells were isolated as described previously (14). The secreted IL-4 was detected with an anti-IL4 phycoerythrin-conjugated antibody (Miltenyi Biotec). The IL-4 producing cells and the IL-4 non-producing cells were separated by MACS using antiphycoerythrin microbeads (Miltenyi Biotec). After sorting, the purity of the sort was confirmed with a FACSCalibur (BD Biosciences).
Chromatin Immunoprecipitation-Cells were harvested at the indicated time points and fixed with 1% formaldehyde for 10 min at room temperature. The chromatin immunoprecipitation assay was performed as described previously (19). Intracellular Cytokine Staining-For intracellular cytokine staining, the cultured Th cells are harvested and restimulated with 10 ng/ml phorbol 12-myristate 13-acetate (PMA) and 1 g/ml ionomycin (Sigma) for 2 h followed by additional 2 h in the presence of 5 g/ml brefeldin A (Sigma Chemicals). The cells were washed with PBS and fixed with 2% formaldehyde (Merck-Millipore). The cell membrane was permeabilized with 0.5% sapinon in PBS/BSA for intracellular staining. The staining was measured with a FACSCalibur (BD Biosciences) or with a MACSQuant Analyzer (Miltenyi Biotec), and the data were analyzed using FlowJo (Treestar). For calcineurin/NFAT inhibition, cyclosporin A, BTP1 (3,5-bistrifluoromethyl pyrazole), or 11R-VIVIT (Calbiochem) was given to the cells at the indicated concentrations 15 min prior to addition of PMA and ionomycin. For siRNA-mediated inhibition of NFATc2, Accell siRNA (A-054724-16; Dharmacon, Lafayette, CO) specific for NFATc2 and control siRNA (D-001910-03-05; Dharmacon) were used as described in Ref. 20. Knockdown efficiency was determined by RT-PCR using the following primers: NFATc2 up, 5Ј-GGTTGCTCCTCTGCCCGCAG-3Ј and NFATc2 down, 5Ј-TTGGAGGGGATCCCGCAGGG-3Ј.
Image Cytometry-Th2 cells were harvested and restimulated with PMA/ionomycin for 3 h in the presence of 7 nM cyclosporin A (CsA). The cells were fixed with 1% formaldehyde. Cytokine staining was performed in 0.5% saponin. NFATc2 staining was done in Foxp3 staining buffer (eBioscience). Staining was analyzed using the Imagestream MKII (Amnis Merck-Millipore). For nuclear staining, DAPI was added before analysis. Data analysis was performed using the IDEAS software (Amnis). Nuclear localization of NFATc2 was determined by "similarity" of NFATc2 and DAPI on a per cell basis. The similarity score is a log-transformed Pearson's correlation coefficient of the pixel values of the DAPI and NFATc2 staining.

Assembly of the Activating Transcription Factor Complex at the Il4 Locus in Th2 Cells Is Dependent on T Cell Receptor
Stimulation-We generated Th2 cells by stimulating naive CD4 ϩ CD62L ϩ T cells from ovalbumin-specific T cell receptor transgenic DO11.10 mice with OVA 323-339 for 12 days in the presence of recombinant IL-4 and antibodies blocking IFN-␥ and IL-12. On day 6, fresh antigen-presenting cells, antigen, IL-4, and antibodies were added. On day 12, the resting Th2 cells were either fixed directly or restimulated with PMA and ionomycin for 3 h and used for chromatin immunoprecipitation (ChIP) to assay for binding of RNA polymerase II and the transcription factors NFATc2, NFATc1, NF-B p65, c-Maf, p300, Brg1, STAT6, and GATA-3 to the Il4 promoter and the Il4 HSS Va. Binding of GATA-3 to CIRE was also assayed. Binding to the Ifn␥ promoter was used as negative control. In resting Th2 cells, in which no IL-4 is detectable by intracellular cytokine staining (Fig. 1A), none of the transcription factors with the exception of STAT6 bound to any of the regions tested (Fig.  1B). STAT6 bound to both the Il4 promoter and Il4 HSS Va to the same degree in unstimulated and restimulated cells. In restimulated Th2 cells, all transcription factors analyzed bound to the Il4 promoter, except for GATA-3, which does not have a binding site there. NFATc2, NFATc1, p300, Brg1, and GATA-3 also bound to the Il4 HSS Va. GATA-3 also bound to the CIRE. Thus, the assembly of TCR dependent and independent transcription factors at the Il4 gene of Th2 cells is dependent on TCR stimulation, i.e. the activation of one or more TCR-dependent transcription factors.
Coordinated Assembly of Transcription Factors at the Il4 Locus-To determine the kinetics of transcription factor assembly to the Il4 locus, we fixed in vitro-generated Th2 cells before restimulation (0 h) and 1, 2, 3, 4, and 6 h following restimulation with PMA/ionomycin. Activated Th2 cells showed detectable levels of IL-4 mRNA already after 1 h and reached a maximum expression at 3 h, after which it declined again ( Fig. 2A). IL-4 protein expression follows a similar time course (14). Binding of RNA polymerase II to the Il4 promoter reached a maximum after 3h, with an abrupt drop to baseline levels after 4 h (Fig. 2B). The transcription factors NFATc2, NFATc1, c-MAF, p300, and Brg1 reached their maximal binding to the Il4 promoter after 3 h. The transcription factors analyzed bound with similar kinetics also to the Il4 HSS Va. No significant binding to the Ifn␥ promoter could be detected at any time point. STAT6 bound to the Il4 promoter and Il4 HSS Va at all time points tested. NFATc2 and Brg1 also bound to the locus control region, located in the Rad50 gene (LCR Rad50 ), reaching maximum binding after 3 h. GATA-3 binding to the CIRE increases after PMA/ ionomycin restimulation and continues to increase until 6 h after the onset of restimulation, the end of the period of observation. The kinetics of transcription factor assembly at the Il4 gene indicates the coordinated, interdependent assembly of all factors.

The Transcription Factor Complex Assembles at the Il4 Locus in IL-4 Expressing Th2 Cells but Not in IL-4-non-expressing Th2
Cells-Th2 cells were restimulated with PMA/ionomycin, and the IL-4 expression was determined (Fig. 3A). Of the Th2 cells, 55% expressed IL-4, whereas 45% did not express any detectable IL-4. IL-4-expressing and non-expressing Th2 cells were physically separated to Ͼ95% purity, using the IL-4 cytokine secretion assay, which we had developed earlier (14). IL-4 protein expression correlated with IL-4 mRNA expression as analyzed by quantitative PCR in the sorted populations (Fig. 3B). Both the IL-4 expressing and non-expressing Th cells expressed equal amounts of GATA-3, qualifying both fractions as bona fide Th2 cells (Fig. 3C).
Relative to the binding of the transcription factors to the Ifn␥ promoter, no significant binding to any of the Il4 gene regions analyzed from IL-4-non-expressing cells was observed for NF-B p65, c-Maf, and RNA polymerase II. NFATc2, NFATc1, p300, and Brg1 did not bind to the promoter and CIRE regions, whereas STAT6 and GATA-3 did (Fig. 3D). STAT6 and GATA-3 also bound to the CIRE, LCR Rad50 , and HSS Va regions of Th2 cells not expressing IL-4 (Fig. 3D). For all regions of Il4 analyzed, significantly more RNA polymerase II, NFATc2, NFATc1, NF-B p65, c-Maf, p300, and Brg1 was detected in IL-4 expressing versus non-expressing cells. Thus, the transcription factor complex, with the exception of GATA-3 and STAT6, efficiently assembles only at Il4 genes of IL-4-expressing Th2 cells.
Calcineurin Digitalizes IL-4 Expression in Th2 Cells-Naive DO11.10 TCR transgenic CD4 ϩ Th cells were stimulated under Th2-polarizing conditions for 12 days and then restimulated with PMA/ionomycin. IL-4 expression was assessed by intracellular cytokine staining showing that 34% of the Th2 cells reexpressed IL-4. When the NFATc2 dephosphorylation by calcineurin was selectively blocked by 25 nM of the specific inhibitor BTP1, a 3,5-bistrifluoromethyl pyrazole derivative (21), IL-4 reexpression was completely blocked (Fig. 4A). In those cells, binding of RNA polymerase II, p300, NFATc2, c-Maf, Brg1, and NF-B p65 to the Il4 promoter, 3 h after restimulation (Fig. 4B), was decreased to levels observed for IL-4-non-expressing Th2 cells (Fig. 3D). This shows that the dephosphorylation of NFATc2 by calcineurin is critical for the assembly of a transcriptional activator complex at the Il4 gene.
Calcineurin, thus, translates graded differences in TCR signaling into an all-or-none expression of Il4 of restimulated Th2 cells. This became evident when calcineurin was inhibited by CsA in different concentrations. Increasing CsA concentrations resulted in dose-dependent, decreased frequencies of IL-4 expressing Th2 cells following restimulation with PMA/ionomycin (Fig. 4C) or anti-CD3/CD28 antibodies (data not shown). However, the amount of IL-4 expressed by individual IL-4-expressing cells remained the same. As CsA has been described to also affect NF-B activation (22), NFAT dephosphorylation was also blocked by the specific peptide inhibitor 11R-VIVIT (Fig. 4D) (23) and by specific siRNA targeting NFATc2 (Fig. 4E). Specific inhibition of either NFAT desphosphorylation by 11R-VIVIT or knockdown of NFATc2 itself by siRNA resulted in the reduction of the frequency of IL-4-expressing Th2 cells but not the amount of IL-4 expressed per cell.

IL-4 Expression Correlates with NFATc2 Nuclear Translocation-To visualize the nuclear translocation of NFATc2 in
Th2 cells on the single cell level, in vitro-generated Th2 cells were restimulated with PMA/ionomycin in the presence of 7 nM CsA and stained for NFATc2 and IL-4. The cells were analyzed by image cytometry. Among all IL-4-expressing Th2 cells (Fig. 5A), NFATc2 showed nuclear localization which was defined by a high similarity score representing the correlation coefficient between the NFATc2 fluorescent signal and the nuclear DAPI fluorescent signal (Fig. 5, B and C). Th2 cells that did not reexpress IL-4 showed a bimodal distribution having either only cytoplasmic or only nuclear NFATc2 (Fig. 4, B and  C). Taken together, our data indicate that in individual Th2 cells, calcineurin, by cooperative activating dephosphorylation of NFATc2, digitalizes graded differences in TCR signaling into all-or-none decisions to express IL-4.

DISCUSSION
Here, we show that in the TCR signaling cascade, calcineurin, by cooperative dephosphorylation of NFATc2, translates differences of signaling strength in individual, restimulated Th2 lymphocytes into an all-or-none decision to express or not the signature cytokine IL-4. NFAT is required to assemble at the Il4 gene a transcription factor complex containing GATA-3, RNA polymerase II, NFATc2, NFATc1, NF-B p65, c-Maf, CBP/ p300, Brg1, and STAT6. Of these, only STAT6 and GATA-3 can bind in the absence of NFAT in Th2 cells not expressing IL-4.
Th2 lymphocytes are imprinted epigenetically by DNA demethylation and histone modification of the Il4 gene (7,24) and transcriptionally by expression of the lineage-determining transcription factor GATA-3 (5,25), to express the signature cytokine IL-4 when restimulated by antigen. Surprisingly, and as noted early on, not all Th2 cells express IL-4 in a given restimulation (15). This is not a matter of lack of competence, as Th2 cells not expressing IL-4 in a given restimulation can express IL-4 in a subsequent restimulation at frequencies equal to cells that had expressed IL-4 (15). It has been speculated that the infidelity of IL-4 reexpression by individual Th2 cells might be due to stochastic, monoallelic expression of the Il4 gene, with some cells not expressing it at all. It remained unclear, whether accessibility of the Il4 gene (26 -28) or availability of the transcription factors necessary for expression were the ratelimiting determinants. Moreover, in these studies, monoallelic expression of Il4 was analyzed for genetically modified T lymphocytes that had one Il4 allele marked by knock-in of a reporter gene, either green fluorescent protein (Gfp) (15) or CD2 (29). For the Gfp knock-in Th lymphocytes, we have shown previously that the genetic insertion had replaced the GATA-3 binding site CIRE, disabling the epigenetic imprinting of the modified Il4 allele (7). For wild-type Th2 cells, stochastic monoallelic expression of Il4 would predict a subpopulation of cells expressing both alleles and consequently twice as much as the cells expressing only one allele. This was not observed.
Here, we show that reexpression of Il4 by Th2 lymphocytes is not only due to stochastic variations but is determined by activation of NFATc2 by calcineurin. NFAT has 23 phosphorylation sites, which are dephosphorylated by calcineurin in a coop- erative fashion, i.e. with strictly sigmoid kinetics, resulting in a "molecular switch" (30). NFAT has to be dephosphorylated at 13 of these sites to expose its nuclear translocation sequence. Dephosphorylation at serine residues at the N-terminal transactivation domain is required for NFAT to bind to its target DNA sequence (18). Translocation of NFATc2 into the nucleus of activated T lymphocytes, thus, is an all-or-none event (Fig.  5B) (17). For human Th lymphocytes, this results in all-or-none reexpression of IL-2, which is dependent on TCR signaling strength and mediated by calcineurin (17). Calcineurin and NFATc2, thus, qualify as molecular analog-to-digital converters, translating TCR signaling strength into different frequencies of cells expressing NFAT-dependent genes. In established Th effector/memory cells, it is the epigenetic imprint of a cell determining which genes are accessible, as we show here for murine Th2 lymphocytes.
Interestingly, independent of the frequency of IL-4-expressing cells in a given restimulation, the average amount of IL-4 expressed by the individual Th2 cell is the same, with stochastic cell-to-cell variability (31). This shows that under the conditions analyzed, none of the transcription factors required for Il4 expression is rate-limiting, except NFAT, as is evident from selective inhibition by BTP1 (32), 11R-VIVIT (22), or siRNA.
NFAT is required to assemble GATA-3, RNA polymerase II, NFATc2, NFATc1, NF-B p65, c-Maf, CBP/p300, Brg1, and STAT6 at the regulatory regions of the Il4 gene. The transcriptional activator complex may contain more proteins, which have not been analyzed here. STAT6 and GATA-3 did bind to the Il4 gene also in the absence of NFAT in restimulated Th2 cells not expressing IL-4, and GATA-3 remained bound to the Il4 gene in Th2 cells expressing IL-4 at late time points of restimulation. Apparently, on their own, they are not competent to assemble any of the other transcription factors analyzed to the Il4 gene, in particular not CBP/p300 and Brg1, which have been connected to epigenetic imprinting (12,33). Although GATA-3 itself has been shown to be critical for epigenetic imprinting of the Il4 gene (26) and is the lineage-determining transcription factor of Th2 cells (5,25), it is not required for the maintenance of the Th2 phenotype, with respect to IL-4 expression. Unlike inhibition of NFATc2 activity as shown here, conditional deletion of GATA-3 in already established Th2 cells did not change the frequency of Th2 cells reexpressing IL-4 but instead reduced the amount of IL-4 expressed per cell (6). In the Th2 cells analyzed here, GATA-3 expression obviously was not rate-limiting, as both Th2 cells, expressing IL-4 or not, expressed similar amounts of GATA-3.  The conversion of graded, analog differences in antigen receptor signal strength into expression of defined packages of cytokines in activated T lymphocytes by the calcineurin/NFAT switch, teaches us that in immune reactions, communication between individual cells via NFAT-dependent cytokines occurs in an all-or-none fashion, probably by direct contact and contact-directed secretion (34). This phenomenon is analogous to the signal transduction in neurons, where a stimulus leads to the opening of ion channels and the firing of an action potential. Increasing the strength of the stimulus does not increase the size of the action potential but rather increases the frequency of action potentials (35). This all-or-none principle ensures that neural signals are passed on in full strength once a certain threshold is passed.
Our data indicate that in adapting the magnitude of the immune response to different concentrations of antigens, it is the frequencies of responding cells among those able to respond, which is regulated by the calcineurin/NFAT switch. The advantage of this analog-to-digital conversion would be that the immune system is able to mount immune responses, even if by only a few cells, to antigens of low abundance, and it defines a threshold of reaction for the individual cell, minimizing background expression of potentially harmful genes, i.e. immunopathology.