Nuclear Factor of Activated T3 Is a Negative Regulator of Ras-JNK1/2-AP-1–Induced Cell Transformation Free

The c-jun-NH₂-kinases (JNK) play a critical role in tumor promoter–induced cell transformation and apoptosis. Here, we showed that the nuclear factor of activated T3 (NFAT3) is phosphorylated by JNK1 or JNK2 at Ser²¹³ and Ser²¹⁷, which are located in the conserved SP motif. The transactivation domain of NFAT3 is found between amino acids (aa) 113 and 260 and includes the phosphorylation targets of JNK1 and JNK2. NFAT3 transactivation activity was suppressed in JNK1^−/− or JNK2^−/− mouse embryonic fibroblast (MEF) cells compared with wild-type MEF cells. Moreover, a 3xNFAT-luc reporter gene assay indicated that NFAT3 transcriptional activity was increased in a dose-dependent manner by JNK1 or JNK2. Double mutations at Ser²¹³ and Ser²¹⁷ suppressed NFAT3 transactivation activity; and SP600125, a JNK inhibitor, suppressed NFAT3-induced 3xNFAT-luciferase activity. Knockdown of JNK1 or JNK2 suppressed foci formation in NIH3T3 cells. Importantly, ectopic expression of NFAT3 inhibited AP-1 activity and suppressed foci formation. Furthermore, knockdown of NFAT3 enhanced Ras-JNK1 or JNK2-induced foci formation in NIH3T3 cells. Taken together, these results provided direct evidence for the anti-oncogenic potential of the NFAT3 transcription factor. [Cancer Res 2007;67(18):8725–35]

The JNKs signal transduction pathway has been shown to play an important role in coordinating various cellular responses such as apoptosis (1), proliferation (2), and neoplastic transformation (3). Mice that are deficient in both jnk1 and jnk2 exhibit embryonic death at E10.5 due to enhanced apoptosis in the hindbrain (4) and forebrain regions (4, 5), which clearly suggests that JNK1 and JNK2 are involved in cell survival during development. Furthermore, specific antisense oligonucleotides against JNKs inhibited tumor cell growth (6) and jnk2-deficient mice displayed significant suppression of skin papilloma development induced by 12-O-tetradecanoylphorbol-13-acetate (TPA; ref. 7). These types of observations may have been due to an enhanced apoptosis in jnk-deficient cells and mice (8).

When cells are stimulated by environmental stress, cytokines, or toxins (9), JNK phosphorylation is increased through MKK4/7 (10), and the activation signal is transmitted to downstream substrate(s) such as c-Jun (11). JNK1 and JNK2 are well known for the activation and phosphorylation of c-Jun at Ser⁶³ and Ser⁷³. However, other downstream target proteins include Elk-1 (12), c-Myc (13), p53 (14), and NFATc2 (15), as well as several members of the apoptosis-related family of proteins, including Bcl-2, Bcl-X_L, Bim, and Bad (16–18). These functions of JNKs have been primarily attributed to the fact that JNKs activate different substrates based on the specific stimulus or cell type.

Although the nuclear factor of activated T cell (NFAT) family of transcription factors has been primarily identified in immune cells, recent studies indicated that NFAT is functionally active in several other non-immune cell types, including vascular endothelial cells, embryonic exon cells, and 3T3-L1 fibroblasts (19–22). Four different isotypes of NFATs, including NFAT1 (1a, 1b, and 1c), NFAT2 (2a and 2b), NFAT3, and NFAT4 (4x, 4a, 4b, and 4c) were shown to have differential tissue distribution (23). These findings suggested that distinct NFAT isotypes play different roles in diverse tissues under various physiologic conditions (24). Calcineurin, a Ca²⁺/calmodulin-dependent protein phosphatase that is a downstream target of intracellular Ca²⁺ signaling, is a well-known effector of the NFAT family of transcription factors (NFAT1–4; ref. 23). Classically, calcineurin dephosphorylates NFAT1–4, allowing NFAT to translocate to the nucleus, bind to consensus DNA sites, and control gene transcription (24). Upon cessation of the Ca²⁺ signal, NFAT proteins are re-phosphorylated by kinases such as GSK-3 (25), resulting in the translocation of NFAT to the cytoplasm (24). However, recent studies indicated that the Ras signaling pathway positively regulates NFAT3 activity (26) by forming an activation complex to regulate PPARγ2 promoter activity, which leads to adipocyte differentiation (27). In addition, RSK2-mediated phosphorylation of NFAT3 regulates NFAT3 activity and induces muscle cell differentiation (28). Furthermore, the NFAT3 protein forms a complex with CBP to activate transcription machinery (29), and NFATc1 binds with AP-1 to enhance its transcriptional activity (30). Activation of NFATc1 was also reported to induce cell transformation (22). On the other hand, NFATc2 was shown to repress cyclin-dependent kinase 4 (CDK4), resulting in cell cycle arrest at G₀-G₁ (20, 31). In addition, when the lymphomagenic virus SL3-3 was infected in NFAT4-deficient mice, T-cell lymphoma developed faster and with higher frequency compared with wild-type mice (20). These reports strongly indicate that the oncogenic or anti-oncogenic activities of the NFAT proteins are dependent on the isotype and specific physiologic condition. However, the role of NFAT3 in the tumorigenesis is not yet understood.

In this study, we showed that NFAT3 is a strong binding partner of JNK1 and JNK2. The phosphorylation of NFAT3 at 213 and 217 by JNK1/2 induced NFAT3 transactivation activity. Importantly, overexpression of NFAT3 suppressed Ras^G12V-JNK1– or -JNK2–induced foci formation by inhibiting AP-1 activity. Taken together, these results indicated that NFAT3 inhibited neoplastic transformation through a negative feedback regulation of JNK1/2-AP-1 signaling.

Reagents and antibodies. DMEM and fetal bovine serum (FBS) were purchased from Life Technologies, Inc. Restriction enzymes and some modifying enzymes were obtained from New England BioLabs, Inc. The DNA ligation kit (version 2.0) was purchased from TAKATA Bio, Inc. The LipofectAMINE Plus transfection reagent for NIH3T3, JNK^WT, JNK1^−/−, and JNK2^−/− mouse embryonic fibroblast (MEF) cells were purchased from Invitrogen. The JetPEI transfection reagent for 293 cells was from Q-Biogene. The pcDNA3.1 plasmid was purchased from Life Technologies, Inc. The luciferase assay substrate was from Promega. Antibodies against the Flag or v5 epitope were purchased from Sigma-Aldrich or Invitrogen, respectively. Antibodies against JNK1 and NFAT3 were purchased from Santa Cruz Biotechnology, Inc., and the antibody against JNK2 was from Cell Signaling Technology, Inc.

Cell culture. The 293 (human embryonic kidney), JNK^WT, JNK1^−/−, and JNK2^−/− MEF cells were cultured in DMEM supplemented with 10% heat-inactivated FBS in a 37°C, 5% CO₂ incubator. NIH/3T3 cells were cultured in DMEM supplemented with 10% bovine calf serum in a 37°C, 5% CO₂ incubator. The cells were maintained by splitting at 80% to 90% confluence, and media were changed every 3 days.

Construction of expression and siRNA vectors. For the mammalian two-hybrid system assay, the pACT-transcription factors (TF), pBIND-JNK1 and -JNK2, and pcDNA3.1-v5-JNK1 and -JNK2 were constructed as previously described (28). The pcDNA3.1-FLAG-NFAT3 and various deletion mutants of NFAT3 from the COOH-terminal end (residues 1–853, 1–580, 1–450, 1–365, 1–260, 1–219, and 1–112) were described (28). The deletion GST-NFAT3 fusion vectors (1–112, 113–260, 261–365, 366–450, 451–580, 581–853, 853–902) were generated by PCR using wild-type NFAT3 full length (FL) subcloned into the BamHI/XbaI site of the pGEX-5X-C vector. The mutation of Ser²¹³ and Ser²¹⁷ to alanine was carried out using the QuickChange II Site–Directed Mutagenesis Kit (Strategene) according to recommended protocols. The deletion mutants of NFAT3-1–112, -113–260 and -261–902, NFAT3-113–260 and NFAT3-113-260S213,217A, and NFAT3S213A,217A were also subcloned into the pcDNA4 (Invitrogen) or pGEX-5X-C vector. The pU6pro vector (provided by David L. Turner, University of Michigan, Ann Arbor, MI) was used to construct small interfering RNA JNK1 (si-JNK1), siRNA JNK2 (si-JNK2), and siRNA NFAT3 (si-NFAT3) following the recommended protocols.³

All of the constructs were confirmed by restriction enzyme mapping and DNA sequencing.

Western blotting. The proteins were extracted with NP40 cell lysis buffer with freezing and thawing. The same amount of protein was resolved by SDS-PAGE, transferred onto polyvinylidene difluoride membranes, hybridized with appropriate antibodies, and then visualized using the enhanced chemiluminescence (ECL) detection kit (Amersham Biosciences).

NFAT3 activity assay. The 293 cells (2.0 × 10⁴) or JNK^WT, JNK1^−/−, and JNK2^−/− MEFs (4.0 × 10⁴) were seeded into 48- or 12-well plates and cultured in 10% FBS-DMEM for 18 h before transfection, respectively. The 3xNFAT-luc reporter plasmid, which was constructed by fusion of three NFAT binding consensus sequences on the 5′ upstream end of the minimal interleukin 2 promoter-luciferase reporter plasmid, was transfected with various recombinant plasmids as indicated in the respective figures. The cells were cultured for 36 h and were or were not stimulated with UVB or specific chemicals and then cultured for an additional 12 h. At each time point, cells were disrupted and analyzed for firefly luciferase activity. The 3xNFAT-luc luciferase activity was normalized against Renilla luciferase activity (pRL-SV40).

Mammalian two-hybrid assay. For the mammalian two-hybrid (M2H) assay, we followed the Promega Checkmate Mammalian Two-Hybrid System protocols. In brief, 293 cells were maintained in 10% FBS-DMEM, seeded into 48-well plates (2.0 × 10⁴), and incubated with 10% FBS-DMEM for 18 h before transfection. The various DNAs, pACT-TFs, pBIND-JNK1 or pBIND-JNK2, and pG5-luciferease were combined in the same molar ratio (1:1:1), and the total amount of DNA was not more than 100 ng per well. The transfection was done using jetPEI following the manufacturer's recommended protocols. For the luciferase assay, the cells were disrupted by the addition of lysis buffer and incubated for 30 min at room temperature with gentle shaking. The luciferase activity was measured automatically by the addition of 100 μL substrate buffer, and data were collected using the Luminoskan Ascent (MTX Lab, Inc.). The relative luciferase activity was calculated using the pG5-luciferase basal control and normalized against Renilla luciferase activity, which included the pBIND vector as an internal control.

In vitro kination assay. Wild-type GST-NFAT3 and deletion or point mutant proteins were used for the in vitro kination assay using an active JNK1 or JNK2 (Upstate Biotechnology, Inc.). Reactions were carried out at 30°C for 30 min in a mixture containing 50 μmol/L unlabeled ATP and 10 μCi of [γ-³²P] ATP and then stopped by adding 6× SDS sample buffer. Samples were boiled and then separated by 12% SDS-PAGE and visualized by autoradiography or Coomassie blue staining.

NFAT3 protein domain analysis. Data for analyzing the NFAT3 protein domains were downloaded from the ExPASy Proteomics Server (NiceProt view of Swiss-Prot: Q14934). Putative phosphorylation sites were predicted by the Netphospho 2.0 server.⁴

Construction and expression of GST-NFAT3 fusion proteins. The NFAT3 wild-type and deletion mutants were constructed as described (28). To express and purify the glutathione S-transferase (GST)-fusion NFAT3 fusion proteins, the BL21 bacterial strain was used. A single colony for each vector was cultured, and the GST-fusion protein was induced by isopropyl-l-thio-β-d-galactopyranoside (final concentration, 0.5 mmol/L) treatment at 25°C for 5 h. The protein was then partially purified with 1× IB wash buffer (Novagen) for disruption, binding to glutathione Sepharose 4B, and elution with 10 mmol/L reduced glutathione, followed by dialysis. Purified proteins were stored −70°C until used.

LTQ Orbitrap hybrid mass spectrometer analysis. Protein phosphorylation site assignment was carried out using ingel trypsin digest and nanoLC-ESI-tandem hybrid mass spectrometry with a ThermoFinnigan LTQ Orbitrap Hybrid Mass Spectrometer. Briefly, the Coomassie blue–stained protein bands were excised from the gel and cut into pieces no larger than 2 mm². Before digestion, the gel pieces were destained with 50% acetonitrile/50 mmol/L Tris-HCl (pH, 8.1) until clear. Gel pieces were then reduced with 20 mmol/L DTT/50 mmol/L Tris-HCl (pH, 8.1) at 55°C for 30 min and alkylated with 40 mmol/L iodoacetamide at room temperature for 30 min in the dark. Proteins were digested in situ with 30 μL (0.004 μg/μL) trypsin (Promega) in 20 mmol/L Tris-HCl (pH, 8.1) at 37°C overnight, followed by peptide extraction with 60 μL of 2% trifluoroacetic acid followed by 60 μL of acetonitrile. The pooled extracts were concentrated to <5 μL on a SpeedVac spinning concentrator (Savant Instruments) and then brought up in 0.1% formic acid for protein identification by nanoflow liquid chromatography electrospray tandem mass spectrometry (nanoLC-ESI-MS/MS) using a ThermoFinnigan LTQ Orbitrap Hybrid Mass Spectrometer (ThermoElectron Bremen) coupled to an Eksigent nanoLC-2D HPLC system (Eksigent). The peptide mixture is loaded onto a 250-nL OPTI-PAK trap (Optimize Technologies) packed with Michrom Magic C8 solid phase (Michrom Bioresources) and eluted with a 0.1% formic acid/acetonitrile gradient through a Michrom packed tip capillary Magic C18 column (75 μm × 150 mm). The LTQ Orbitrap mass spectrometer experiment was set to perform an Fourier transform (FT) full scan from 380 to 1,600 m/z with resolving power set at 60,000 (400 m/z), followed by linear ion trap MS/MS scans on the top three ions. The ambient air polycyclodimethylsiloxane 391 m/z ion was used as an internal lock mass for the FT full scans giving 2 ppm or better mass accuracy. To detect phosphorylated peptides, a MS scan is triggered if one of the top three ions in the MS/MS scan corresponds to the neutral loss of one or two phosphoric acid moieties with and without water for [M + 2H]²⁺ (49, 58, 98, and 107 m/z) and [M+3H]³⁺ (32.7 and 65.4 m/z) precursor ions. Dynamic exclusion was set to 2, and selected ions were placed on an exclusion list for 90 s. The MS/MS raw data were converted to DTA files using ThermoElectron Bioworks 3.2 and correlated to theoretical fragmentation patterns of tryptic peptide sequences from the Swissprot databases using both SEQUEST (ThermoElectron) and Mascot (Matrix Sciences) search algorithms running on 10 node clusters. All searches were conducted with fixed cysteine modifications of +57 for carboxamidomethyl-cysteines and variable modifications allowing +16 with methionines for methionine sulfoxide, +42 for protein NH₂-terminal acetylation, +80 at serine, threonine, and tyrosine for phosphorylation and −18 for serine and threonine H₃PO₄ loss. The search was restricted to trypsin-generated peptides allowing for two missed cleavages and was left open to all species. An additional search was done using only the expected sequence with no enzyme. Peptide mass search tolerances are set to 20 ppm, and fragment mass tolerance was set to ±0.8 Da. Matches to phosphorylated peptides were further interpreted manually to assign the site of phosphorylation using both +80 and −18 diagnostic ions. Preferences were given to matches of the more abundant ions.

Focus forming assay. Transformation of NIH3T3 cells was conducted according to standard protocols (32). Cells were transiently transfected with various combinations of H-Ras^G12V (50 ng), JNK1 or JNK2 (450 ng), siRNA vector DNA (450 ng), and pcDNA3-mock (compensation for equal amount of DNA) as indicated in figures and then cultured in 5% FBS-DMEM for 2 weeks. Foci were fixed with methanol, stained with 0.5% crystal violet, and then counted with a microscope and the Image-Pro PLUS software program.

NFAT3 is a binding partner of JNK1 and JNK2 in vivo. To identify novel binding or substrate protein(s) for JNKs, we introduced JNK1 or JNK2 cDNA, including the open reading frame (ORF), into the pBIND mammalian two-hybrid system vector (pBIND-JNK1 or pBIND-JNK2) as bait. The ORF of each TF was amplified by PCR and then introduced into the pACT mammalian two-hybrid system vector (pACT-TFs). Each individual pACT-TF plus the pG5-luciferase reporter plasmid was cotransfected into 293 cells with pBIND-JNK1 or -JNK2. Each interaction activity was compared against the pG5-luc/pBIND-JNK1 or -JNK2 as the basal level (Fig. 1A,, lane 1 or 7, respectively). Elk-1, c-Jun, and activating transcription factor 2 (ATF2) were used as positive controls (Fig. 1A,, lanes 4–6 and 10–12). NFAT3 and NFAT4 showed strong interacting activity with pBIND-JNK1 or -JNK2 compared with the pG5-luc/pBIND-JNK1 or -JNK2 control (Fig. 1A,, lanes 2 and 3 versus 1 or 8 and 9 versus 7, respectively). NFAT4 has been identified previously as a binding partner for JNKs (33), and therefore, we selected NFAT3 for further study. To verify the interaction of NFAT3 with JNK1 and JNK2, we introduced the pcDNA3-v5-JNK1 or -JNK2 construct and the pcDNA3-Flag-NFAT3 construct into 293 cells. Results of immunoprecipitation (IP) experiments showed that NFAT3 could bind with JNK1 or JNK2 in vivo (Fig. 1B).

To identify the NFAT3 domain that is involved in the interaction of NFAT3 with JNK1 or JNK2, we constructed pcDNA3-Flag-NFAT3 deletion mutants (Supplementary Fig. S1A) that were then individually transfected into 293 cells with pcDNA3-v5-JNK1 or -JNK2, and expression was confirmed (Supplementary Fig. S1B). IP results using a v5 monoclonal antibody showed that the FL NFAT3 and deletion mutants from the COOH-terminal to aa 365 coprecipitated with JNK1 or JNK2 (Fig. 1C, and D, lanes 2–6). However, when aa 365–260 were deleted, the NFAT3 protein was not detected, and any mutants with deletions to aa 219 or 112 did not coprecipitate with the JNK1 or JNK2 protein (Fig. 1C, and D, lanes 7–9). Interestingly, when aa 450 to 580 were deleted, the binding affinity of NFAT3 with JNK1 or JNK2 was increased (Fig. 1C, and D, lane 5–6). In addition, the immunoprecipitated levels of JNK1 or JNK2 were almost the same (Fig. 1C  and D, bottom). These results suggested that aa 260 to 365 of NFAT3 are important for binding with JNK1 and JNK2.

NFAT3 is a substrate of JNK1 and JNK2 in vitro. To identify the target region of NFAT3 that harbors potential phosphorylation sites for JNK1 or JNK2, we constructed a series of GST-NFAT3 deletion fragments (Fig. 2A,, vector maps). These proteins were expressed partially purified as described in Materials and Methods and then subjected to an in vitro phosphorylation assay. The results indicated that JNK1 and JNK2 phosphorylated NFAT3 in the region spanning aa 113 to 260 (Fig. 2A,, top and middle), which includes a calcineurin binding site and two SP repeat regions. The p38 kinase is a well-known upstream kinase of NFAT3 that phosphorylates NFAT3 at Ser¹⁶⁸ and Ser¹⁷⁰ (34). To determine whether the phosphorylation sites for JNKs and p38 were identical, we constructed an NFAT3113-260 protein with mutations to replace Ser¹⁶⁸ and Ser¹⁷⁰ with alanine (NFAT3-113-260S168,170A) and conducted the in vitro phosphorylation assay (Fig. 2B). Results indicated that the phosphorylation of the mutant NFAT3-113-260S168,170A by JNK1 or JNK2 was the same as for wild-type NFAT3-113-260 (Fig. 2B , left and middle, lane 2 versus 3), suggesting that Ser¹⁶⁸ and Ser¹⁷⁰ of NFAT3 are not the phosphorylation targets of JNK1 or JNK2.

Identification of the NFAT3 sites phosphorylated by JNK1 and JNK2. To determine the specific site(s) of NFAT3-113-260 that are phosphorylated by JNK1 and JNK2, we used the human protein reference database⁵

to compare the amino acid similarity among various JNK substrates. Results showed that JNK1 and JNK2 recognized and phosphorylated various substrates at the SP/TP consensus sequences (Supplementary Fig. S2A). Moreover, we found that NFAT3-113-260 contains 13 SP/TP motifs (Supplementary Fig. S2B), including Ser¹⁶⁸ and Ser¹⁷⁰, which are phosphorylated by p38 (34), but not by JNK1 or JNK2 (Fig. 2B). Therefore, we next used the LTQ Orbitrap hybrid mass spectrometer to identify the NFAT3 sites that are phosphorylated by JNK1 and JNK2 as described in Materials and Methods. The results indicated that JNK1 and JNK2 phosphorylated two sites on NFAT3 at Ser²¹³ and Ser²¹⁷ (Supplementary Fig. S2B and C). To confirm these phosphorylation sites, we constructed mutants of GST-NFAT3-113-260, replacing Ser²¹³ and/or Ser²¹⁷ with alanine (GST-NFAT3-113-260S217A or GST-NFAT3-113-260S213,217A; Fig. 2C). These proteins were expressed and partially purified and then directly subjected to the in vitro phosphorylation assay using active JNK1 or JNK2. Results confirmed that mutation of NFAT3-113-260 at Ser²¹⁷ caused a marked decrease in phosphorylation by JNK1 (Fig. 2C,, left, lane 3) or JNK2 (Fig. 2C,, middle, lane 3). Furthermore, mutations at both Ser²¹³ and Ser²¹⁷ totally blocked phosphorylation of NFAT3-113-260 by either JNK1 (Fig. 2C,, left, lane 4) or JNK2 (Fig. 2C , middle, lane 4). These results showed that Ser²¹³ and Ser²¹⁷ are the amino acids targeted for phosphorylation by JNK1 and JNK2.

JNK1 and JNK2 regulate NFAT3 activity. To examine the role of JNK1 and JNK2 in the regulation of NFAT3, we first developed pGal4-NFAT3-1-902 and pGal4-NFAT3-261-902 constructs (Fig. 3A,, vector map) and analyzed the transactivation activity of NFAT3 using cotransfection of each of these constructs with the 5xGal4-luciferase reporter plasmid into JNK wild-type (JNK^WT) MEF cells. The transactivation activity of NFAT3 was significantly increased in cells transfected with Gal4-NFAT3-1-902 compared with cells transfected with the pGal4-mock control (Fig. 3A,, lane 2). On the other hand, the NH₂-terminal–deleted Gal4-NFAT3-261-902 truncated form markedly suppressed the transactivation activity of NFAT3 (Fig. 3A,, lane 3). These results indicated that the transactivation activity might be regulated in the NH₂-terminal area spanning aa 1 to 260 of the NFAT3 protein. Next, we transfected the pGal4-NFAT3-1-902 plasmid into JNK^WT, JNK1 knock-out (JNK1^−/−), or JNK2 knock-out (JNK2^−/−) MEF cells. The resulting 5xGal4-luciferase activity indicated that the increased transactivation activity of Gal4-NFAT3-1-902 observed in JNK^WT MEFs was significantly suppressed in JNK1^−/− or JNK2^−/− MEFs (Fig. 3B), suggesting that JNK1 and JNK2 might be positive regulators of the transactivation activity of NFAT3. To further delineate the role of JNK1 and JNK2 in the regulation of NFAT3 transactivation, we created Gal4-NFAT3-1-112 and Gal4-NFAT3-113-260 constructs and cotransfected each with p5xGal4-luc and analyzed NFAT3 transactivation activity in JNK^WT, JNK1^−/−, and JNK2^−/− MEFs (Fig. 3C). Results indicated that the transactivation activity of NFAT3-113-260 was strongly increased compared with the pGal4 mock control, but NFAT3-1-112 did not induce transactivation (Fig. 3C). Furthermore, the increased transactivation activity of Gal4-NFAT3-113-260 was significantly inhibited in JNK1^−/− or JNK2^−/− MEFs similar to that observed for NFAT3-1-902 in these same cells (Fig. 3B and C). To further verify a role for JNKs in regulating the transcriptional activity of NFAT3, we transfected increasing amounts of pcDNA3-Flag-NFAT3/pcDNA3.1-v5-JNK1 or pcDNA3-Flag-NFAT3/pcDNA3.1-v5-JNK2 with the 3xNFAT-luciferase reporter plasmid and analyzed luciferase activity (Fig. 3D). Results indicated that NFAT3-mediated 3xNFAT-luciferase activity was increased in a JNK1 or JNK2 dose-dependent manner (Fig. 3D). Taken together, these results indicated that JNK1 and JNK2 phosphorylate NFAT3 at Ser²¹³ and Ser²¹⁷ and positively regulate NFAT3 transcriptional activity.

Ser²¹³ and Ser²¹⁷ of NFAT3 are involved in transactivation but not with the interaction of NFAT3 with JNK1 or JNK2. To examine the role of Ser²¹³ and Ser²¹⁷ in NFAT3 activity, we first analyzed the effect of SP600125, a JNK inhibitor. The p3xNFAT-luciferase plasmid was transfected into JNK^WT MEFs either with or without pcDNA3-Flag-NFAT3, and then cells were treated with 10 μmol/L SP600125. The results indicated that SP600125 suppressed NFAT3-induced 3xNFAT-luciferase activity (Fig. 4A). Next, we analyzed the binding affinity of JNK1 and wild-type NFAT3 or mutant NFAT3 (S213,217A) by the mammalian two-hybrid assay. The results indicated that NFAT3FL with mutations at Ser²¹³ and Ser²¹⁷ had no effect on the interaction with JNK1 compared with unmodified wild-type NFAT3 (Fig. 4B). In addition, we observed similar results for the binding affinity of JNK2 with the NFAT3 mutant (data not shown). To determine the effect of phosphorylation of Ser²¹³ and Ser²¹⁷ on the transactivation activity of NFAT3, we used JNK^WT and JNK1^−/− MEFs cotransfected with pGal4-NFAT3-113-260 or pGal4-NFAT3-113-260S213,217A and p5xGal4-luc. The results showed that mutation of NFAT3 at Ser²¹³ and Ser²¹⁷ inhibited NFAT3 transactivation activity by about 50% in either JNK^WT or JNK1^−/− MEFs (Fig. 4C). These results also revealed that Ser²¹³ and Ser²¹⁷ of NFAT3 are the target amino acids for JNK1 and JNK2 and play an important role in the positive regulation of NFAT3 activity.

Knockdown of JNK1 and JNK2 inhibits cell transformation. JNK1 and JNK2 are critical upstream kinases of c-Jun, which is a major component of the AP-1 complex. When c-Jun is phosphorylated by JNK1 or JNK2 at Ser⁶³ and Ser⁷³, c-Jun DNA binding affinity is increased, and the expression of many target genes is increased many times, resulting in increased cell proliferation and cell transformation (35–39). Therefore, we proposed that phosphorylation and activation of NFAT3 by JNK1 and JNK2 might play an important role in cell proliferation as well as cell transformation. To elucidate the physiologic significance of JNK1/2-NFAT3 signaling, we constructed siRNA vectors against JNK1 (si-JNK1) and JNK2 (si-JNK2; Fig. 5A). These si-constructs were transfected into NIH3T3 cells, and knockdown of endogenous JNK1 (∼75%) and JNK2 (∼95%) was confirmed by Western blotting (Fig. 5B). To examine the effect of JNK1 and/or JNK2 knockdown on cell transformation, we introduced various combinations of Ras^G12V, JNK1, JNK2, NFAT3, si-JNK1, and si-JNK2 expression vectors into NIH3T3 cells and conducted a focus-forming assay (Fig. 5C). As expected, constitutively active Ras (Ras^G12V) induced cell transformation in NIH3T3 cells (Fig. 5C, and graph lane 2). Moreover, overexpression of JNK1 or JNK2 combined with Ras^G12V resulted in an increased foci formation in NIH3T3 cells (Fig. 5C, and graph lanes 3 and 4). Notably, si-JNK1 or si-JNK2 almost completely blocked foci formation induced by Ras^G12V alone or Ras^G12V combined with JNK1 or JNK2 (Fig. 5C, and graph lanes 5–8). Interestingly, we found that Ras^G12V combined with NFAT3 did not increase, but actually suppressed foci formation in NIH3T3 cells compared with Ras^G12V alone (Fig. 5C, and graph lane 9 versus 2). Foci formation was even further suppressed by si-JNK1 or si-JNK2 (Fig. 5C, and graph lanes 10 and 11). The suppressive effect of si-JNK1 and si-JNK2 might be mediated by AP-1 and NFAT3 because AP-1 luciferase activity (Fig. 5D) and transactivation activity of NFAT3 (Fig. 3B) were suppressed in JNK1^−/− and JNK2^−/− MEF cells compared with JNK^WT MEF cells.

NFAT3 negatively regulates cell transformation mediated by JNK1 and JNK2. To more fully examine the JNK1- and JNK2-mediated physiologic function of NFAT3 in cell transformation, we constructed siRNA against NFAT3 (Supplementary Fig. S3A). This plasmid was transfected into NIH3T3 cells, and knockdown of endogenous NFAT3 protein level by about 90% was confirmed by Western blot (Supplementary Fig. S3B). We then introduced various combinations of Ras^G12V, JNK1, JNK2, NFAT3, and si-NFAT3 into NIH3T3 cells and conducted a focus-forming assay. As before, Ras^G12V alone induced foci formation, and JNK1 or JNK2 enhanced Ras^G12V-induced foci formation (Fig. 6A, and graph lanes 2–4). Interestingly, overexpression of NFAT3 suppressed Ras^G12V/JNK1- or Ras^G12V/JNK2-induced foci formation in NIH3T3 cells (Fig. 6A, and graph lanes 5 and 6). Very importantly, si-NFAT3 enhanced foci formation compared with Ras^G12V alone, Ras^G12V/JNK1, or Ras^G12V/JNK2 in NIH3T3 cells (Fig. 6A, and graph lanes 7 and 8). Moreover, the colony size in si-NFAT3–transfected cells seemed to be larger than Ras^G12V alone, Ras^G12V/JNK1, or Ras^G12V/JNK2 (Fig. 6A). We also found that UVB (4 kJ/m²) treatment increased AP-1–luciferase activity as well as 3xNFAT-luciferase activity compared with untreated control NIH3T3 cells (Fig. 6B). Moreover, mutation of NFAT3 at Ser²¹³ and Ser²¹⁷ suppressed 3xNFAT-luciferase activity under a variety of culture conditions as well as after UVB stimulation (Fig. 6C). Importantly, AP-1 luciferase activity was significantly inhibited by overexpression of NFAT3 in NIH3T3 cells (Fig. 6D).

Because NFAT transcription factors are found in a wide range of cell types and tissues and regulate genes involved in cell cycle progression, cell development, and differentiation and angiogenesis (19–22, 40, 41), NFAT3 is also now believed to play an important role in tumorigenesis or to possess antitumorigenic functions. Recent research reports have indicated that constitutively active NFATc1 induces a transformation phenotype in 3T3L1 fibroblasts (22). However, NFATc2 negatively regulates CDK4 gene expression, resulting in cells re-entering into the resting stage of the cell cycle (20). NFAT4 has also been reported to act as a tumor suppressor for the development of murine T-cell lymphomas induced by the retrovirus SL3-3 (31).

The mitogen-activated protein kinase p38 phosphorylates NFAT3 at Ser¹⁶⁸ and Ser¹⁷⁰ and induces cytoplasmic localization. When calcineurin is activated, Ser¹⁶⁸ and Ser¹⁷⁰ are dephosphorylated, and NFAT3 is localized to the nucleus. However, we found that JNK1/2 phosphorylated Ser²¹³ and Ser²¹⁷, and the mutation of Ser²¹³ and Ser²¹⁷ to alanine abolished NFAT3 transactivation and transcriptional activity (Figs. 4C and 6C). Furthermore, the NFAT3 transactivation activity was suppressed in JNK1^−/− and JNK2^−/− MEF cells (Fig. 3B). In addition, NFAT3-mediated 3xNFAT-luciferase activity was increased by cotransfection of JNK1 or JNK2 (Fig. 3D), indicating that phosphorylation of NFAT3 at Ser²¹³ and Ser²¹⁷ by JNK1/2 might induce nuclear localization. These results strongly indicated that distinct phosphorylation site(s) of NFAT3 might be involved in a different regulatory mechanism for NFAT3.

When NFAT3 was overexpressed with a combination of Ras^G12V, JNK1, or JNK2 in NIH3T3 cells, foci formation was suppressed compared with the expression of only Ras^G12V, Ras^G12V/JNK1, or Ras^G12V/JNK2 (Figs. 5C and 6A). In contrast, when siRNA-NFAT3 was cotransfected with various combinations of Ras^G12V, JNK1, and JNK2 into NIH3T3 cells, foci formation was increased compared with the expression of only Ras^G12V, Ras^G12V/JNK1, or Ras^G12V/JNK2 (Fig. 6A). Furthermore, we found that AP-1 activity was suppressed by ectopic expression of NFAT3 (Fig. 6D), indicating that NFAT3 suppressed cell transformation through the inhibition of AP-1 activity. Therefore, we hypothesized that NFAT3 is a negative regulator of Ras-JNK1/2-AP-1–induced cell transformation.

Three possibilities could explain the negative regulatory effect mediated by NFAT3. One is a possible competition between c-Jun and NFAT3 as a JNK1/2 substrate. When a cell is stimulated by, e.g., UV, JNK is activated and localized to the nucleus where it then phosphorylates the c-Jun protein (42). At the same time, JNK phosphorylates and activates NFAT3 because UV stimulation also induces 3xNFAT-luciferase activity (Fig. 6B). Therefore, NFAT3 and c-Jun might compete with JNK1/2 as substrates, resulting in the suppression of AP-1 activity. However, this is probably unlikely because the affinity of the interaction between JNK1/2 and c-Jun is very much higher than the potential interaction of JNK1/2 and NFAT3 as determined by the mammalian two-hybrid assay (Fig. 1A). The second possibility is that NFAT3 interferes with the nuclear localization of JNK1/2. We found that although JNK nuclear localization is induced by stimuli such as UV, only a relatively small amount of JNK is localized into the nucleus compared with the large amount of JNK still remaining in the cytoplasm (42). In cell culture, we found that NFAT3 is predominantly localized in the cytoplasm (28), and when cells are stimulated, JNK1/2 is phosphorylated and activated in the cytoplasm. After activation, JNK1/2 binds with NFAT3 and phosphorylates NFAT3 also in the cytoplasm. During this process, NFAT3 interfered with JNK1/2 localization, resulting in a large amount of JNK1/2 being localized in the cytoplasm. This could at least partially explain why the JNK1/2-NFAT3 interaction affinity is lower than the JNK1/2–c-Jun interaction, but could still inhibit AP-1 transcription activity. However, this idea does not explain the observation that UV induces a strong and rapid c-Jun phosphorylation at Ser^63/73. The third possibility involves JDP2 (c-Jun dimerization protein 2). JDP2 is a member of the bZIP family of transcription factors and binds with c-Jun and represses AP-1 transcriptional activity (43). JDP2 also binds with core transcription machinery such as ATF2 and CBP (44). The NFAT3 protein forms a complex with CBP (29), suggesting that a complex composed of NFAT3 and JDP2 is also possible, and that this complex might suppress AP-1 activity as well as induce cell differentiation. This idea is supported by findings showing that (a) JDP2 inhibited Ras-mediated cell transformation in NIH3T3 cells and suppressed tumor development in the xenograft severe combined immunodeficiency mouse model (45); (b) ectopic expression of JDP2 in myoblasts and rhabdomyosarcoma tumor cells strongly promoted muscle cell differentiation (46); (c) phosphorylation of NFAT3 (Ser²⁸¹, Ser²⁸⁵, Ser²⁸⁹, and Ser³⁴⁴) by RSK2 induced NFAT3 transcriptional activity, resulting in multinucleated myotube differentiation of C2C12 myoblasts (28); and (d) siRNA against NFAT3 enhanced Ras^G12V/JNK-mediated foci formation, and ectopic expression of NFAT3 suppressed Ras^G12V/JNK-mediated foci formation (Fig. 6A). Furthermore, although NFAT1 and AP-1 are cooperative for gene expression of many cytokine genes (30), no report has shown that NFAT3 and AP-1 bind to each other and cooperate directly. Therefore, the detailed molecular mechanism of AP-1 activity inhibition by NFAT3 requires further investigation.

Based on our observation that NFAT3 is a substrate of JNK1/2, resulting in increased NFAT3 activity, we hypothesized that JNK1/2-NFAT3 function might be involved in cell transformation as well as oncogenesis because JNK1/2 signaling plays a pivotal role for cell proliferation through the regulation of AP-1 activity. Surprisingly, we found that JNK1/2-mediated NFAT3 activation signaling suppressed Ras^G12V-mediated NIH3T3 cell transformation as well as AP-1 luciferase activity (Figs. 5C and 6D). In contrast, knockdown of NFAT3 induced Ras^G12V-mediated NIH3T3 cell transformation (Fig. 6A). Therefore, we concluded that NFAT3 might play a key modulator function to control proliferation and differentiation through the regulation of AP-1 activity. Our results are the first to show an anti-oncogenic function for NFAT3 that is mediated through the JNK signaling pathway.

Grant support: The Hormel Foundation and NIH grants CA77646, CA81064, CA27502, and CA111356.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Dr. R.J. Davis (Department of Biochemistry and Molecular Biology, University of Massachusetts Medical School, Worcester, MA) for the pcDNA3-Flag-JNK1, Dr. C.W. Chow (Department of Molecular Pharmacology, Albert Einstein College of Medicine, Bronx, NY) for the pcDNA3-Flag-NFAT3 deletion mutants, and Andria Hansen for secretarial assistance.