Constitutive dephosphorylation and activation of a member of the nuclear factor of activated T cells, NF-AT1, in Tax-expressing and type I human T-cell leukemia virus-infected human T cells.

The tax gene product of the type I human T-cell leukemia virus (HTLV-I) transactivates interleukin-2 (IL-2) gene through activation of an enhancer termed CD28 responsive element (CD28RE). Tax activation of the CD28RE is partially mediated by a member of the nuclear factor of activated T cells, NF-AT1. We have previously shown that NF-AT1 is constitutively active in Jurkat T cells stably transfected with the Tax cDNA, although the underlying molecular mechanism and physiological relevance of this finding remain unclear. In this report, we demonstrate that the active form of NF-AT1 is also present in the nuclei of HTLV-I-transformed T cells that express the Tax protein. Interestingly, the constitutive activation of NF-AT1 in these T cells is associated with its dephosphorylation. Furthermore, the dephosphorylated NF-AT1 can be rapidly rephosphorylated when the cells are incubated with cyclosporin A, an immunosuppressant inhibiting the serine/threonine phosphatase calcineurin. These results suggest that activation of NF-AT1 in Tax-expressing and HTLV-I-transformed T cells results from its dephosphorylation, which in turn may be due to deregulation of calcineurin.

leukin-2 (IL-2) and the ␣ subunit of its high affinity receptor complex (IL-2R␣) (reviewed in Ref. 6). The central role of these genes in normal T-cell activation and growth suggests that this specific action of Tax may be an important mechanism underlying HTLV-I-induced T-cell transformation (6,7).
Emerging lines of evidence suggest that Tax induces the target genes indirectly by modulating the activity or expression of specific host transcription factors (6,8). In this regard, we and others have recently shown that Tax induces the transcription of IL-2 gene through activation of cellular transcription factors that bind to an IL-2 gene enhancer termed CD28 responsive element (CD28RE) (9 -12). At least two families of transcription factors, NF-B/Rel and NF-AT, have been shown to bind to the CD28RE enhancer sequence (11)(12)(13)(14)(15).
The NF-B/Rel factors participate in the transcriptional activation of a large array of genes involved in cell growth and activation (16 -18). This family contains various dimeric complexes composed of a set of structurally related polypeptides, including p50 (NF-B1), p52 (NF-B2), RelA (previously named p65), RelB, and the proto-oncoprotein c-Rel (reviewed in Ref. 19). The biological activity of NF-B/Rel is regulated through cytoplasmic retention by association with specific inhibitory proteins termed IBs (20). Various immunological stimuli, as well as the HTLV-I Tax protein, induce the phosphorylation and subsequent proteolysis of IBs, leading to the nuclear translocation of the active forms of NF-B/Rel (for recent reviews, see Refs. [21][22][23]. NF-AT represents a family of enhancer binding proteins that participate in the regulation of a large number of cytokine genes (24 -26). The NF-AT family includes at least four structurally related proteins, NF-AT1 (previously named NF-ATp), NF-ATc, NF-AT3, and NF-AT4 (27)(28)(29)(30)(31). In normal T cells, the NF-AT proteins are sequestered in the cytoplasm as hyperphosphorylated inactive precursors (27,34). Treatment of T cells with stimuli of the T-cell receptor (TCR) complex or the calcium ionophore ionomycin results in the rapid dephosphorylation of NF-AT, which can be detected by its mobility change on SDS-polyacrylamide gels (32, 34 -36). Dephosphorylation of NF-AT is associated with its nuclear translocation and activation of its DNA binding activity (34,36). The serine/threonine phosphatase calcineurin appears to play an essential role in the activation of NF-AT, as the calcineurin inhibitors cyclosporin A (CSA) and FK506 block the signal-induced dephosphorylation and the subsequent nuclear translocation and DNA binding activity of NF-AT (34,36). In further support of this notion, NF-AT has been shown to physically associate with calcineurin (35,37,38) and serve as a substrate of calcineurin in vitro (39 -41).
The pattern of NF-AT regulation may differ between normal T cells and transformed T-cell lines. For example, in human Jurkat leukemic T cells, a portion of NF-AT1 is constitutively localized in the nucleus even when the cells are incubated with CSA, although this nuclear form of NF-AT1 is inactive in DNA binding (33). Interestingly, we have recently shown that the DNA binding activity of NF-AT1 is constitutively activated in Jurkat T cells stably transfected with the HTLV-I Tax cDNA expression vector. This NF-AT member is one of the major factors binding to the IL-2 CD28RE and appears to play an important role in the activation of IL-2 promoter in these Tax-expressing cells (11). In the present study, we demonstrate that NF-AT1 is also constitutively activated in HTLV-I-infected human T cells. Furthermore, activation of NF-AT1 in these cells results from its dephosphorylation through a CSAsensitive pathway.

MATERIALS AND METHODS
Cell Culture and Reagents-Human Jurkat leukemic T cells were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum, 2 mM L-glutamine, and antibiotics. A Jurkat cell line stably transfected with a Tax cDNA expression vector (Jurkat-Tax, see Ref. 42) was cultured in the same way as for Jurkat cells except that the medium was supplemented with 400 g/ml G418. The HTLV-I-infected T-cell lines C8166, HUT102, K3T, MT-2, and MT-4 cells, which had been characterized previously (43)(44)(45)(46), were cultured in RPMI supplemented with 20% fetal bovine serum. Except for MT-4, all these cells express the Tax  Immunoblotting and Electrophoresis Mobility Shift Assay (EMSA)-The HTLV-I-infected T cells and noninfected parental Jurkat or Jurkat-Tax cells were collected by centrifugation at 800 ϫ g for 5 min. Subcellular and whole-cell extracts were prepared as described previously (47). For immunoblotting analyses, protein samples (ϳ15 g) were fractionated by 7.5% reducing SDS-polyacrylamide gel electrophoresis, electrophoretically transferred to nitrocellulose membranes, and then analyzed for immunoreactivity with anti-NF-AT1 using an enhanced chemiluminescence detection system (ECL; DuPont NEN). EMSA was performed by incubating the nuclear extracts (ϳ5 g) with a 32 Pradiolabeled CD28RE probe derived from the human IL-2 gene (AAA-GAAATTCCAAAGAGT) followed by resolving the DNA-protein complexes on native 4% polyacrylamide gels (48). For antibody "supershift" assays, 1 l of diluted antisera (5-fold for anti-NF-AT1, 3-fold for the various anti-NF-B antisera) was added to the EMSA reaction 5 min before electrophoresis.

NF-AT1 Is Constitutively Active in HTLV-I-infected Human
T Cells-To examine whether NF-AT1 is active in HTLV-Iinfected T cells, EMSA was performed using a 32 P-labeled human IL-2 CD28RE, and nuclear extracts were isolated from various HTLV-I-infected T-cell lines as well as Jurkat cells or Jurkat cells stably transfected with the tax cDNA (Jurkat-Tax) ( Fig. 1). As shown previously (11), two major protein complexes, C1 and C2, were detected from the Jurkat-Tax cells (Fig. 1A, lane 2), while only low amounts of the C1 complex were found from the parental Jurkat cells (lane 1). Previous studies have shown that the intensity of the C1 complex varies in both the parental Jurkat cells and Jurkat-Tax cells; however, the C2 complex is detectable only in the Jurkat-Tax cells (11). Furthermore, the C2 complex, but not the C1 complex, has been shown to specifically immunoreact with an NF-AT1-specific antibody in antibody supershift assays, thus suggesting that NF-AT1 is involved in the formation of the inducible C2 complex (11). In the HTLV-I-infected cell lines (including C8166, HUT102, and K3T), two CD28RE-binding complexes were detected (lanes 3-5, C2 and C3), one of which (C2) comigrated with the NF-AT1 complex from the Jurkat-Tax cells (lane 2, C2), although all three cell lines lacked the C1 complex. In a separate experiment, the HTLV-I-infected MT-2 cells were also shown to express the C2 and C3 complexes (Fig. 1B, lane  2). However, another HTLV-I-infected cell line, MT-4, that was deficient in Tax expression (46,49), did not express either C2 or C3 although it expressed high levels of the C1 complex (lane 3).
To examine the proteins involved in the formation of these CD28RE-binding complexes, EMSA was performed using the C8166 cell extract in the presence of specific antibodies for various known IL-2 regulating transcription factors, including members of the NF-B/Rel family and NF-AT1 (Fig. 1C). The antisera against the p50 (lane 3) and c-Rel (lane 4) subunits of NF-B/Rel selectively supershifted the C3 complex, thus suggesting that this protein complex was composed of both p50 and c-Rel. The antisera for the RelA (lane 5), p52 (lane 6), and RelB (lane 7) subunits of NF-B/Rel as well as a preimmune serum (lane 2) did not react with either C2 or C3. As expected, the C2 complex, which comigrated with the NF-AT1 complex in Jurkat-Tax cells (see Fig. 1A, C2), was supershifted by the anti-NF-AT1 antiserum (lane 8). The same result was obtained using the nuclear extracts isolated from HUT102, K3T, and MT2 cells (data not shown). Similar antibody supershift assays revealed that, as previously observed (11), no immunoreactivity was found between the C1 complex in Jurkat or MT-4 and the anti-NF-AT1 antibody. Together, these results suggest that activation of NF-AT1 DNA binding activity in both Jurkat and HTLV-I-infected T cells is associated with the expression of the Tax protein. Additionally, the CD28RE is also bound by the p50/c-Rel heterodimer in HTLV-I-infected T cells. In Jurkat-Tax cells, which express significantly lower levels of NF-B/Rel activity than in the HTLV-I-infected cells (50), the binding of NF-B/Rel to CD28RE was very weak. However, the p50/c-Rel complex could be detected in these cells, especially when the EMSA conditions were modified to favor the binding of this NF-B/Rel heterodimer (in the presence of about 60 mM NaCl, data not shown, also see Fig. 3C).
NF-AT1 Is Constitutively Dephosphorylated in Jurkat-Tax and HTLV-I-infected T Cells-Activation of NF-AT1 by immu-

FIG. 1. Activation of NF-AT1 in Jurkat-Tax and HTLV-I-infected T cells. A and B, EMSA analyses of CD28RE-binding NF-AT1
complex. Nuclear extracts were isolated from Jurkat, Jurkat-Tax, or the indicated HTLV-I-transformed T-cell lines and then subjected to EMSA using a 32 P-labeled CD28RE probe. The data presented in A and B were generated from two different experiments, in which the gels were run for different times (2 h for A and 3 h for B). The specific CD28RE-binding proteins were designated C1-C3. NS indicates a nonspecific protein that binds to different probes (data not shown). The C2 complex in Jurkat-Tax cells (lane 2) was previously shown to contain NF-AT1 (11). C, antibody supershift analysis of the CD28RE-binding proteins in HTLV-I-infected T cells. The nuclear extract isolated from C8166 cells was subjected to EMSA either in the absence (None) or presence of the indicated preimmune or immune sera. The supershifted bands are indicated with arrowheads. nological stimuli or the calcium ionophore ionomycin involves its dephosphorylation, and the dephosphorylated form of NF-AT1 migrates more rapidly than the fully phosphorylated form on SDS-polyacrylamide gel electrophoresis (32,34). To examine whether Tax-mediated activation of NF-AT1 also resulted from dephosphorylation of NF-AT1, immunoblotting studies were performed to analyze the phosphorylation state of NF-AT1 in Jurkat-Tax and the various HTLV-I-infected cells (Fig. 2). In the parental Jurkat cells, a single NF-AT1 protein band was detected, and, as previously reported (33), this phosphorylated form of NF-AT1 was located in both cytoplasm and nucleus ( Fig. 2A, lanes 1 and 2), although the phosphorylated form of nuclear NF-AT1 was inactive in CD28RE binding (see Fig. 1A, lane 1). Interestingly, in Jurkat-Tax cells, a large amount of NF-AT1 was present in its more rapidly migrating dephosphorylated form and localized in the nucleus ( Fig. 2A, lanes 3 and  4, NF-AT1*). The dephosphorylated form of NF-AT1 was also detected from the Tax-expressing HTLV-I-infected HUT102 (Fig. 2B, lane 1), K3T (lane 2), MT2 (lane 3), and C8166 (data not shown) cells. However, in the MT-4 cells, which are deficient in Tax expression, the entire NF-AT1 protein pool was in the phosphorylated form (Fig. 2B, lane 4, NF-AT1). Thus, the DNA binding activity of NF-AT1 was well correlated with its dephosphorylation (compare Figs. 1 and 2). The level of NF-AT1 dephosphorylation in the Tax-expressing T cells appeared to vary during different stages of cell culture, with fresh culture exhibiting higher levels (data not shown). However, no significant amount of the fast migrating dephosphorylated NF-AT1 was ever detected in the Tax-deficient Jurkat and MT4 cells ( Fig. 2 and data not shown). Taken together with the results shown in Fig. 1, it suggests that activation of NF-AT1 in Taxexpressing and HTLV-I-infected T cells likely results from its dephosphorylation.
NF-AT1 Is Rapidly Rephosphorylated and Inactivated in Cells Treated with CSA-To explore the mechanism of NF-AT1 dephosphorylation and to further correlate the dephosphorylation and DNA binding activity of this transcription factor, the effect of CSA on the electrophoretic mobility and DNA binding activity of NF-AT1 was analyzed. For these studies, the HTLV-I-infected K3T cells (Fig. 3A) or Jurkat-Tax cells (Fig. 3B) were incubated with CSA and then subjected to immunoblot to analyze the expression of the different forms of NF-AT1 in both cytoplasm and nucleus. As shown in Fig. 3A, CSA treatment for as short as 5 min led to the disappearance of the dephosphorylated form of NF-AT1, which was associated with the increase of the phosphorylated form of NF-AT1 in both cytoplasm and nucleus (lanes 3 and 4). However, in contrast to that seen with immunologically stimulated murine T cells (34), the rephosphorylated NF-AT1 in these HTLV-I-transformed T cells did not relocate from the nucleus to the cytoplasm even after 30 min of CSA treatment (Fig. 3A, lanes 3-8). Similar results were obtained from Jurkat-Tax cells (Fig. 3B). Nevertheless, parallel EMSA revealed that the rephosphorylation of NF-AT1 was associated with the loss of its DNA binding activity in both the Jurkat-Tax (Fig. 3C) and K3T (data not shown) cells. Together, these results suggest that the constitutive DNA binding activity of NF-AT1 in Tax-expressing and HTLV-I-infected T cells apparently results from its dephosphorylation, which in turn is likely mediated by the CSA-sensitive phosphatase calcineurin. These studies also indicate that the nuclear translocation and DNA binding activity of NF-AT1 may be differentially regulated in these Tax-expressing cells. DISCUSSION Transcriptional activation of the IL-2 gene in normal T cells requires both the TCR-mediated primary and the CD28-mediated costimulatory signals. However, in T cells infected with HTLV-I or expressing the viral Tax protein, various pharmacological stimuli that mimic the TCR signaling are sufficient to induce the transcription of the IL-2 gene (12,42), thus suggesting that Tax may provide a costimulatory signal for T-cell activation. Interestingly, the Tax-responsive element within the IL-2 promoter has been identified to be the CD28RE (9 -12, 51), an enhancer element known to respond to the CD28 costimulatory signal (13,52). The transcription factors involved in Tax-mediated activation of CD28RE remain unclear. Although NF-B/Rel has been shown to bind to the CD28RE which may contribute to the activation of this enhancer (12), additional factors are clearly required since a Tax mutant (M47) that is capable of NF-B activation (53) fails to transactivate the CD28RE (10,18). In support of this notion, we have recently shown that NF-AT1 binds to the CD28RE in a Taxexpressing Jurkat cell line (Jurkat-Tax) and participates in the activation of CD28RE in a reporter gene assay.
In the present study, we have demonstrated that NF-AT1 is constitutively activated in both Jurkat-Tax and various HTLV-I-infected T-cell lines expressing the viral Tax protein (Fig. 1). This finding strongly suggests that activation of NF-AT may be part of the mechanism by which HTLV-I activates human T cells. As seen with immunological stimuli (32,34,36), activation of NF-AT1 in the HTLV-I-infected T cells is associated with the dephosphorylation of this normally heavily phosphorylated protein (Fig. 2). However, the dephosphorylated form of NF-AT1 is constitutively expressed in the nucleus of these virally infected T cells, which is in contrast to the transient activation of NF-AT1 in antigen-stimulated noninfected T cells (34). This apparent deregulation of NF-AT1 can be due to either the up-regulation of calcineurin activity or down-regulation of the opposing NF-AT kinase activity. We think the former possibility is more likely true since the dephosphorylated NF-AT1 can be rapidly rephosphorylated when the cells are incubated with the calcineurin inhibitor CSA (Fig. 3), which suggests that the NF-AT kinase is still active in these cells. It is important to note, however, that the rephosphorylated NF-AT1 in CSA-treated cells is largely retained in the nucleus even though it becomes inactive in DNA binding (Fig.  3). Although the precise underlying mechanism remains to be further investigated, it is possible that the nuclear form of NF-AT1 seen in the CSA-treated HTLV-I-transformed cells is not fully rephosphorylated. In support of this notion, the nuclear form of NF-AT1 appears to migrate slightly faster than the cytoplasmic form (see Fig. 3). Studies are in progress to directly analyze the phosphorylation state of the two different forms of NF-AT1.