Transcriptional complexes formed by NFAT dimers regulate the induction of T cell tolerance

The RHR of NFAT1 had been previously shown to have the ability to form homodimers on specific DNA sites containing two NFAT core binding sites separated by 1 or 2 nt (16, 17). To test the cooperative nature of these complexes, we analyzed binding of the NFAT1 RHR to probes containing two minimal consensus NFAT sites (GGAA) separated by 1 or 6 nt. As expected, recombinant NFAT1 RHR homodimers formed on the probe with a 1-nt spacer. Incubation of this probe with increasing amounts of recombinant NFAT1 RHR yielded increasing amounts of dimers (Fig. 1 A). However, dimer formation was almost completely absent when a 6-nt spacer was inserted between both NFAT monomer sites (Fig. 1 B). These data suggested that the binding of two NFAT1 RHR molecules to κB-like sites might be cooperative, and therefore disruption of the protein–protein contacts that maintained both NFAT molecules together should prevent binding of dimers onto the DNA.

Analysis of the crystal structure of the NFAT1 RHR dimers (16, 17) allowed us to identify a series of residues located in the dimer interface that, when mutated, should impair protein–protein interactions (Fig. 1 C). Deletion of the CC′ loop (aa616-623) in the RHR-C generated a mutant with reduced capacity to form dimers, although these could still be detected with high concentrations of the CC′ loop-deleted NFAT1 RHR (Fig. 1 D). Additional mutations on Q671 and Q673 produced a protein with a complete impairment of its ability to bind κB-like sites as a dimer (Fig. 1 D).

All mutations were performed in residues located in the RHR-C and, in theory, should not affect NFAT1's ability to form complexes with Fos and Jun proteins on composite sites because these interactions occur predominantly through the RHR-N (Fig. S1, available). To test this hypothesis, recombinant RHR was incubated with increasing amounts of c-Fos and c-Jun, and a probe containing the ARRE2 composite site of the murine Il2 promoter (see Supplemental materials and methods). Binding reactions showed that the mutant lacking the CC′ loop and bearing Q to A substitutions on Q671 and Q673 (ΔCC′QQ–AA), had the same ability to form cooperative complexes with Fos and Jun as the WT RHR (Fig. 1 E).

Next, we tested whether a full-length NFAT1 protein bearing those mutations would also show dimer formation impairment. Nuclear extracts prepared from HEK293 cells transfected with plasmids encoding WT NFAT1 or a ΔCC′QQ-EE mutant were analyzed by EMSA using a probe containing a κB-like dimer site. Glutamines were substituted for glutamic acid in this mutant to introduce electrostatic repulsion between monomers. This mutant showed almost no capacity to form dimers, but retained its ability to bind DNA and to form complexes with AP-1 (Fig. 1, F–H). Direct targeting of the dimer interface can, thus, prevent dimer formation without affecting NFAT interactions with AP-1 proteins.

Next, we determined if mutations in the dimer interface would inhibit NFAT-induced transcription from a promoter containing κB-like elements. Jurkat cells were transfected with vectors encoding constitutively active forms of WT or the ΔCC′QQ-EE NFAT1 and a reporter vector containing two tandem κB-like sites. We have previously shown that these active NFAT proteins, which contain several serine-to-alanine substitutions in key residues located in the regulatory region of the NHR domain, are constitutively nuclear and insensitive to the inhibitory effect of cyclosporine A (CsA) (19). Therefore, to assess the effect of the plasmid-encoded NFAT proteins, transfected cells were treated with CsA to block endogenous NFAT activation. WT NFAT1 was transcriptionally active on κB-like sites. Its transcriptional activity was similar in cells treated with or without ionomycin, suggesting that NFAT1 did not require cooperation with other calcium-induced transcription factors to activate that promoter (Fig. 2 A). Cooperative dimer formation was clearly necessary to induce transcription from κB-like sites, as the ΔCC′QQ-EE mutant showed almost no transcriptional activity (Fig. 2 A). All NFAT1 proteins tested were, however, equally active when a reporter containing three tandem ARRE2 sites from the Il2 promoter was used (Fig. 2 B), confirming that disruption of the dimer interface did not affect NFAT interactions with Fos or Jun.

To exclude that NFAT proteins could bind as monomers and activate transcription, the activity of constitutively active NFAT1 proteins was measured using a reporter plasmid containing 6 tandem copies of the core “GGAA” NFAT binding site. Neither the WT nor the ΔCC′QQ-EE mutant proteins showed any activity on this vector (Fig. 2 C). A reporter containing 6 9-bp κB-like sites was used as control and, as expected, only the WT protein could activate its transcription (Fig. 2 C).

Having shown that NFAT1 dimers were transcriptionally active, we determined whether they would control the transcription of anergy-associated genes in T cells. For that purpose, we transduced primary Th1 cells with retroviruses expressing constitutively active forms of WT or ΔCC′QQ-EE NFAT1 proteins and analyzed the expression of four of the genes that had previously been suggested to be NFAT dependent (9, 30). Confirming previous data, quantitative real-time PCR (qPCR) analysis revealed that the expression of those genes was up-regulated in Th1 cells in response to increases in intracellular calcium caused by ionomycin, whereas CsA blocked their expression (Fig. 3 A).

Transcription of those genes was clearly NFAT dependent, as it was induced in cells expressing an active WT NFAT1 at similar levels to those detected in ionomycin-treated cells (Fig. 3 A). As in previous experiments, these cells were always treated with CsA to block endogenous NFAT activation. Whereas the expression of Grg4 and Dgka was also up-regulated by the ΔCC′QQ-EE mutant, this protein failed to induce the expression of Grail or Caspase3 (Fig. 3 A). These results indicated that the transcription of some anergy-inducing genes, such as Grail or Caspase3, was directly or indirectly regulated by transcriptional complexes containing NFAT dimers. Transcription of other anergy-associated genes, such as Grg4 or Dgka, although still dependent on NFAT proteins, would be activated by other NFAT-containing transcriptional complexes. To confirm that the ΔCC′QQ-EE mutant was still active on promoters containing nondimer sites, primary Nfat1^+/+ or Nfat1^−/− Th1 cells were cotransfected with plasmids expressing WT or ΔCC′QQ-EE NFAT1 proteins and a reporter vector in which the expression of luciferase was regulated by the murine proximal Il2 promoter. In these assays, the ΔCC′QQ-EE mutant showed similar transactivational activity as the WT NFAT1 protein (Fig. 3 B).

Given that we had previously shown that Nfat1^−/− T cells were less susceptible to anergizing stimuli and showed a markedly reduced ability to up-regulate the expression of most anergy-associated genes, we focused our study on the role that this NFAT protein had on the regulation of the expression of those genes. To determine if other NFAT proteins might also form dimers and be able to compensate for the lack of NFAT1, we analyzed the induction of the expression of Grail and Caspase3 in Nfat1^+/+ and Nfta1^−/− T cells. Our results showed that T cells that lacked NFAT1 were not able to up-regulate the expression of these dimer-dependent genes in response to calcium signaling (Fig. 3 C). To further study if, in any other circumstances, dimers containing other NFAT family members could activate the expression of those genes, we cotransfected NIH3T3 cells, which express very low levels of endogenous NFAT proteins, with plasmids expressing NFAT1, NFAT2, or NFAT4 (the three NFAT proteins that can be found in T cells) and a reporter vector in which the promoter of Grail directed the expression of luciferase. As expected, NFAT1 induced transcription from the Grail promoter (Fig. 3 D). Similar results were obtained for NFAT4, but NFAT2 did not show any transactivational activity on the Grail promoter, even though it was perfectly capable of inducing the expression of a reporter controlled by the Il2 promoter ARRE2 site, which contains composite NFAT–AP-1 sites (Fig. 3 D).

To determine whether the expression of Grail and Caspase3 could be induced by the NFAT dimer-mediated activation of their promoters, we measured the capacity of WT NFAT1 and the ΔCC′QQ-EE dimerization mutant to activate two reporter vectors where the expression of the luciferase gene was controlled by those promoters. We did not detect any NFAT-mediated activation of the Caspase3 proximal promoter (unpublished data). However, expression of an active form of NFAT1 was able to activate the Grail promoter (Fig. 4 A). Levels of activation were almost identical in cells treated with CsA and in cells treated with CsA and ionomycin, which suggested that, under the conditions used in the assay, NFAT did not require additional calcium-activated proteins to transactivate the Grail promoter (Fig. 4 A). Supporting the results obtained in primary Th1 cells, the ΔCC′QQ-EE dimerization mutant NFAT1 protein had a markedly reduced ability to activate transcription from the Grail promoter (Fig. 4 A), confirming that Grail expression was dependent on NFAT dimers.

To determine if NFAT-dimers would bind directly to the Grail promoter and regulate its expression, we followed two different approaches. First, we assessed the recruitment of NFAT1 to the Grail promoter by chromatin immunoprecipitation. We used an anti-NFAT1 antibody to pull down NFAT–DNA complexes in murine Th1 cells, and we quantified binding of NFAT1 by qPCR using specific primers for the proximal promoters of the Il2 and Grail genes (see Supplemental materials and methods). As expected, NFAT1 bound the Il2 promoter. Although some binding was already detected in cells treated with ionomycin, it was in fully activated T cells where NFAT1 binding to the Il2 promoter was clearly induced (Fig. 4 B). These results were consistent with the binding of NFAT–AP-1 complexes to composite sites on the Il2 promoter in activated T cells. NFAT1 was also able to bind to the Grail promoter; however, the pattern of binding was different from the one detected on the Il2 promoter. Consistent with the role of NFAT1 in the transcription of anergy-inducing genes, NFAT1 bound the Grail promoter preferentially in cells receiving an anergizing stimulus (ionomycin), whereas full activation of T cells with PMA and ionomycin reduced NFAT1 binding (Fig. 4 B).

To determine if two putative dimer sites we had identified in the Grail promoter (−355 gtaacgtttcc and −411 ggattcttcc) were binding sites for NFAT, and to simultaneously measure their contribution to the regulation of Grail expression, reporter vectors containing the Grail promoter with mutations that disrupted those sites were constructed (see Supplemental materials and methods). Analysis of Jurkat cells transiently transfected with these plasmids showed that disruption of any of those two sites resulted in an almost complete loss of NFAT-dependent transcriptional activity (Fig. 4 C).

Expression of an active form of NFAT1 is able to induce unresponsiveness in T cells (9). To determine the contribution of NFAT dimer complexes to the induction of the anergic phenotype, we transduced murine Th1 cells with retroviral vectors expressing a constitutively active WT NFAT1 or the ΔCC′QQ-EE mutant and analyzed the ability of these proteins to induce unresponsiveness. Western blot analysis demonstrated that equivalent amounts of both proteins were expressed in the sorted infected Th1 cells (Fig. 5 A). As previously reported (9), Th1 cells transduced with a constitutively active NFAT1 became resistant to stimulation, and activation-induced IL-2 expression decreased almost 12-fold when compared with control cells. However, the expression of the NFAT1 dimerization mutant, although still able to cause some inhibition of IL-2 expression, was much less efficient than the WT protein in blocking T cell responses to stimulation (3.5-fold compared to 11.8-fold; Fig. 5 A). Similar results were obtained when Nfat1^−/− T cells were transduced with retrovirus expressing nonconstitutively active forms of NFAT1 or the ΔCC′QQ-EE dimerization mutant. As previously shown (9), Nfat1^−/− cells were less susceptible to anergizing stimuli than Nfat1^+/+ T cells (Fig. 5 B). Restoring the expression of NFAT1 by retroviral infection in these cells also restored their ability to become anergic in response to ionomycin; however, no effect could be detected when cells were transduced with a retrovirus that expressed the ΔCC′QQ-EE mutant protein (Fig. 5 B).

To confirm these results in vivo, adoptive transfer experiments were performed. CD4⁺ T cells isolated from DO11.10 mice were adoptively transferred to BALB/c recipients and challenged with CFA or OVA_323-339 peptide in CFA. Upon analysis of the draining lymph nodes 7 d after challenge, proliferative responses could be observed only in mice challenged with the antigenic peptide (Fig. S2, available). When transduced T cells expressing GFP alone or GFP and constitutively active forms of WT NFAT1 or the ΔCC′QQ-EE mutant were transferred to BALB/c recipients, increased numbers of GFP⁺ T cells in response to OVA_323-339 challenge could be readily detected in control GFP-expressing cells, but they were almost completely absent in mice receiving cells expressing an active WT NFAT1 protein. When cells that expressed the ΔCC′QQ-EE dimerization mutant NFAT1 protein were transferred, it was evident that this protein had lost most of its ability to induce unresponsiveness in T cells and clear proliferative responses induced by OVA_323-339 challenge could be detected in those mice (Fig. 5 C).

Tolerizing stimuli induce the expression of a specific set of genes that are different from those expressed during a productive immune response (9, 27). In various models of immune tolerance, an elevation of intracellular Ca²⁺ can be consistently detected (8, 36, 37). In T cells, weakly agonistic antigens that induce T cell anergy also elicit sustained increases in the levels of cytoplasmic Ca²⁺ (8). Thus, tolerizing stimuli that cause suboptimal activation of T cells may only induce low levels of sustained increase in intracellular Ca²⁺ that have been shown to be sufficient to activate NFAT, but not AP-1 or NF-κB, proteins (38, 39). Several lines of evidence support the involvement of Ca²⁺ and NFAT in the expression of anergy-associated genes: cells treated with a Ca²⁺ ionophore become unresponsive and up-regulate the expression of anergy-associated genes, which can be blocked by the calcineurin inhibitor CsA; T cells from Nfat1^−/− mice are hyperactive and more resistant to becoming anergic; and, finally, expression in Th1 cells of a constitutively active NFAT1 induces a state of unresponsiveness and up-regulates many anergy-associated genes (9).

In activated T cells, NFAT cooperates with other transcription factors on multiple cytokine promoters (21, 40). However, during anergy induction, many of the NFAT transcription partners, such as AP-1, are not fully activated. Thus, other NFAT-containing transcriptional complexes must regulate the expression of anergy-associated genes. Recently, two groups characterized the dimer structure of the NFAT1 RHR bound to the κB-like sites of the human IL-8 promoter and the HIV-1 LTR (16, 17). Our results show that NFAT dimers are able to induce transcription from κB-like sites and regulate the expression of at least two genes responsible for blocking TCR signaling in anergic T cells. Whereas NFAT1 can induce the expression of Grail and Caspase3, a mutant NFAT1 protein unable to form dimers fails to transactivate those two genes. Transient reporter assays indicate that the NFAT-dependent regulation of Caspase3 expression does not occur through its proximal promoter. It is possible, though, that distal control elements containing NFAT sites are responsible for the regulation of the expression of this gene, as it has been described for several cytokines (41–43). Grail, however, is regulated by direct binding of NFAT dimers to at least two κB-like sites located in its promoter. Grail is also posttranslationally regulated by two isoforms of the ubiquitin-specific protease Otubain 1 (44). After T cell activation, Otubain1 binds Grail and inhibits its deubiquitination, leading to Grail degradation. During anergy induction, a different spliced form, Otubain1-ARF, is expressed and binds and stabilizes Grail by recruitment of USP8 (44). It is still unknown how the expression of these isoforms of Otubain is regulated, and it would be interesting to determine if NFAT might also be involved.

Our data indicates that NFAT homodimers are cooperative. Increasing the distance between the NFAT binding hemi-sites in a κB-like site abrogates binding of NFAT dimers. Similarly, mutations of the NFAT1 dimer interface that inhibit protein–protein interactions result in a markedly reduced ability of NFAT1 to bind and transactivate from κB-like sites. Most of the interactions between NFAT and DNA or AP-1 occur through the RHR-N (Fig. S1) (15), which explains why mutations in the RHR-C do not affect the formation of NFAT–AP-1 complexes. NFAT-containing transcriptional complexes activated during a productive immune response are different from those formed during the induction of T cell tolerance, and involve different protein surfaces on NFAT. Molecules could thus be designed to specifically manipulate these cellular responses. For instance, targeting the critical residues needed for NFAT–AP-1 interactions should interfere with the formation of these complexes without affecting dimer formation, thus skewing T cell responses toward tolerance. Conversely, targeting the dimer interface should not affect AP-1 interactions, and should selectively disrupt dimer transcriptional complexes, tilting T cell responses toward activation.

The NFAT1 residues involved in the dimer interface are conserved in NFAT2 and NFAT4 (16, 17). Thus, structurally, it is likely that other NFAT proteins can also homodimerize or even heterodimerize. It has been well characterized that redundancy within the NFAT family exists. In general, knocking out one NFAT protein results only in a mild phenotype, whereas knocking out two or more members leads to striking defects (45–53). The phenotypes of these mice seem to indicate that although NFAT1 and NFAT2 proteins regulate the expression of activation-induced genes, it is NFAT1 and NFAT4 that might be specifically involved in controlling cellular programs that lead to the negative regulation of T cell responses. As suggested by the phenotypes of those mice, NFAT1 and NFAT4 had a much higher activity on the promoter of Grail than NFAT2. NFAT4 dimers are unlikely to compensate for NFAT1, as T cells that lack NFAT1 showed an almost complete lack of up-regulation of Grail and Caspase3 when anergized. Although we cannot rule out a possible role for NFAT1–NFAT4 heterodimers, recent reports have indicated that, because of differences in the residues involved in dimer interaction between NFAT1 and NFAT4, these complexes are not likely to occur (16).

We have also observed that the regulation of other anergy-inducing genes, such as Grg4 and Dgka, is NFAT dependent but dimer independent. How are these genes regulated? Is it possible that NFAT monomers may drive their expression? The crystal structure of the NFAT1 RHR bound to a single NFAT cognate DNA site shows the formation of four different DNA–NFAT binary complexes (18). In all four, the RHR-N recognizes the DNA and provides a relatively rigid domain to interact with other transcriptional factors. A flexible linker exists between the RHR-N and RHR-C that permits the RHR-C to acquire different conformational changes, providing a dynamic surface for protein–protein interactions. This flexibility should facilitate the formation of higher order NFAT transcriptional complexes. Our experimental data supports this conclusion and shows that NFAT1 is not able to induce any significant transcriptional activity in a reporter vector driven by NFAT monomer sites. Overall, the data suggests that NFAT1 must cooperate with other transcription factors to activate gene expression. It is also possible that NFAT may indirectly regulate the expression of some anergy-inducing genes. It has recently been shown that anergic T cells up-regulate transcription of Cblb by inducing the expression of the transcription factors Egr2 and Egr3 in a NFAT-dependent way (28).

Could NFAT dimers be involved in other cellular functions as well? T reg cells are also intrinsically anergic; they do not produce IL-2 and proliferate poorly upon stimulation in vitro. A hallmark of T reg cells is the expression of the Foxp3 transcription factor, which seems to be necessary for T reg cell development and suppressor activity (54–56). Recently, it has been reported that T reg cells express up to 10 times more Grail than naive CD4⁺ T cells. Grail may thus be necessary to maintain the anergic status of T reg cells and to acquire suppressor activity, as T cells overexpressing Grail gain suppressor capacity (57). The mechanism of action and the regulation of Grail in T reg cells are unknown. Recent evidence indicates that NFAT–Foxp3 complexes are crucial for the development of T reg cell suppressor function (24). However, we cannot rule out that NFAT dimers might also be involved in the regulation of the program of gene expression that leads to the establishment of a regulatory phenotype.

The identification of the transcriptional complexes that activate distinct programs of gene expression in T cells should allow us to understand the mechanisms that regulate specific cellular responses. At the same time, this information should provide valuable insight to design tools that may interfere with the formation or functioning of these complexes, and therefore may modulate T cell function.

WT or Nfat1^−/− C57BL6/J, BALB/c, and DO11.10 mice were maintained in pathogen-free conditions. All animal work was approved by the Animal Institute Committee of the Albert Einstein College of Medicine.

Primary CD4⁺ T cells were isolated from lymph nodes and spleen of 4–6-wk-old mice using anti-CD4–coupled magnetic beads, stimulated with 0.5 µg/ml plate-bound anti-CD3ε and anti-CD28, and differentiated for 7 d in the presence of 10 ng/ml IL-12, 10 µg/ml anti–IL-4, and 10 U/ml of recombinant hIL-2. Cells were cultured in DME supplemented with 10% fetal calf serum, 2 mM l-glutamine, nonessential amino acids, and essential vitamins, 550 nM l-arginine, 240 nM l-asparagine, 14 nM folic acid, and 50 µM β-mercaptoethanol. Jurkat, HEK293, and NIH-3T3 cells were cultured in DME supplemented with 10% fetal calf serum and 2 mM l-glutamine.

Jurkat cells were transfected by electroporation in serum-free medium, and murine T cells were transfected by electroporation using a Nucleofector electroporator (Amaxa) following the manufacturer's recommendations. 24 h after transfection, cells were stimulated with 1 µM ionomycin and/or 20 nM PMA in the presence or absence of 1 µM CsA. 8 h after stimulation, cells were lysed and assayed for luciferase activity using the Promega Dual-Luciferase Reporter Assay System. HEK293 and NIH-3T3 cells were transfected using PolyFect transfection reagent (QIAGEN) following the manufacturer's recommendations.

The retroviral vectors RV-IRES-GFP and RV-CA-HA-NFAT1-IRES-GFP have been previously described (9). Phoenix Ecotropic cells (a gift from G. Nolan, Stanford University, Palo Alto, CA) were transfected with retroviral vectors. Supernatants were collected 48 h after transfection, supplemented with 8 µg/ml polybrene, and used to infect CD4⁺ T cells 24 h after stimulation. Infected cells were sorted for GFP expression. NFAT expression was assessed by immunoblot using an anti-HA antibody.

The QuickChange Site-Directed Mutagenesis kit (Stratagene) was used to remove the CC′ loop and substitute Q671 and Q673 for A or E in the NFAT dimer interface. The two NFAT dimer sites at position −355 and −411 on the Grail promoter were mutated from GTAACGTTTCC to GTCTCGTAACC and from GGATTCTTCC to TAATTCAACC, respectively.

Recombinant proteins were expressed fused to hexahistidine tags and purified using Ni-NTA agarose. Nuclear extracts were prepared from transfected HEK293 cells, and equal amounts of recombinant or HEK293-expressed NFAT proteins were incubated with [³²P]-labeled probes containing κB-like dimer sites with different 1-, 2-, or 6-bp spacers or a composite NFAT–AP-1 site in a buffer containing 10 mM Hepes, pH 7.0, 125 mM NaCl, 10% glycerol, 0.25 mM DTT, 100 ng/µl of poly dI-dC, 0.8 mg/µl of BSA for 30 min, and then resolved in 4% polyacrylamide gels.

Th1 cells were treated with 1 µM ionomycin for 16 h. In some experiments, CsA at 1 µM was also added 30 min before the ionomycin treatment. Cells were then washed and rested for 2–4 h in fresh medium before stimulation.

25–50 × 10³ Th1 cells were stimulated with 0.5 µg/ml plate-bound anti-CD3 and anti-CD28 in 96-well plates. Supernatants were collected 8 h after stimulation, and IL-2 levels were measured in a sandwich ELISA.

Total RNA samples were prepared from sorted GFP⁺ retrovirally transduced mouse Th1 cells. cDNA was synthesized using Superscript II reverse transcription. Gene expression was analyzed using SYBR Green in a Smart Cycler II thermocycler. A threshold was set in the linear range of the amplification curve (fluorescence = f[cycle]), and the number of cycles needed to reach it was calculated for every sample. Expression of each gene was normalized to levels of actin, and the fold induction was calculated relative to the resting values.

NFAT1 binding was assayed using a ChIP assay kit (Millipore). Nuclear lysates from 10⁷ Th1 cells untreated or treated with 1 µM ionomycin or 1 µM ionomycin/20 nM PMA for 2 h were subjected to immunoprecipitation overnight at 4°C with an anti-NFAT1 antibody (ABR). Taqman probes were used to analyze the ChIP products by qPCR. Results were expressed as fold induction over values in resting cells and were normalized by input values.

5 × 10⁵ retrovirally transduced, GFP⁺ DO11.10 T cells were transferred into BALB/c recipients by tail vein injection. 1 d after the transfer, mice received a subcutaneous injection of 200 µl of complete Freund's adjuvant in the presence or absence of 400 µg of OVA_323-339 peptide. On day 7, draining lymph nodes were collected. Cell suspensions were prepared and labeled with a PE-coupled anti-CD4 antibody, and the number of CD4⁺GFP⁺ T cells was measured by FACS.

Differences in the inhibition of IL-2 production and gene expression in anergic T cells were analyzed using a Student's t test.

Fig. S1 shows an amino acid sequences comparison of the of the RHRs of NFAT1, 2, and 4, highlighting the residues involved in contacts with Fos and Jun, those involved in NFAT-NFAT dimer interactions, and the amino acids mutated in this study. Fig. S2 shows the proliferative responses of adoptively transferred DO11.10 TCR transgenic T cells in BALB/c mice in response to stimulation with OVA_323-339 peptide. The Supplemental materials and methods describes the sequences of the DNA probes and primers used in this study.

We thank members of our laboratory for helpful discussions. We also thank Drs. Lin Chen, at the University of Southern California, Stephen Harrison, at Harvard Medical School, and Patrick Hogan, at the Immune Disease Institute, for helpful suggestions and discussion. Human IL-2 and antibodies against murine IL-4 were obtained from the Biological Resources Branch preclinical repository.

This work was supported by National Institutes of Health (NIH) grants AI059738 (F. Macian), CA42471 (A. Rao), and AI48213 (A. Rao); NIH training grants GM007288 (N. Soto-Nieves) and GM007288 (B.T. Abe and I. Baine); and the Irene Diamond Foundation (F. Macian).

The authors have no conflicting financial interests.