Activation of Nuclear Factor of Activated T Cells in a Cyclosporin A-resistant Pathway*

The mechanism of action of the immunosuppressive drug cyclosporin A (CsA) is the inactivation of the Ca 2 (cid:49) / calmodulin-dependent serine-threonine phosphatase calcineurin by the drug-immunophilin complex. Inac-tive calcineurin is unable to activate the nuclear factor of activated T cells (NFAT), a transcription factor required for expression of the interleukin 2 (IL-2) gene. IL-2 production by CsA-treated cells is therefore dra-matically reduced. We demonstrate here, however, that NFAT can be activated, and significant levels of IL-2 can be produced by the CsA-resistant CD28-signaling path- way. In transient transfection assays, both multicopy NFAT- and IL-2 promoter- (cid:98) -galactosidase reporter gene constructs could be activated by phorbol 12-myristate 13-acetate (PMA)/ (cid:97) CD28 stimulation, and this activation was resistant to CsA. Electrophoretic mobility shift as- say showed the induction of a CsA-resistant NFAT complex in the nuclear extracts of peripheral blood T cells stimulated with PMA plus (cid:97) CD28. Peripheral blood T cells stimulated with PMA/ (cid:97) CD28 produced IL-2 in conserved that PMA/ (cid:97) CD28

Nuclear factor of activated T cells (NFAT) 1 is required for IL-2 production after antigenic stimulation of T cells (1,2). Upon stimulation of T cells, an NFAT-containing protein complex appears in the nucleus and recognizes two or more sites in the IL-2 promoter (2,3). The complex is composed of a preexisting cytoplasmic component, which translocates to the nucleus, and an inducible nuclear component composed of members of the AP1 family (4 -6). Translocation of the cytoplasmic component involves Ca 2ϩ -induced activation of the calcium/ calmodulin-dependent serine/threonine phosphatase calcineurin (5,7). Inactivation of calcineurin by the immunosuppressive drugs CsA or FK506 results in the blockage of translocation and inhibition of IL-2 production (7,8).
CD28 is a T cell surface molecule that plays a major role as a coreceptor in the optimal activation of T cells (9). One of the major effects of CD28 stimulation is increased production of cytokines, such as IL-2, interferon ␥, granulocyte macrophagecolony stimulating factor, and others, through transcriptional and post-transcriptional mechanisms (10,11). One of the hallmarks of the CD28 costimulation is its insensitivity to the immunosuppressive drugs CsA and FK506 (10,12,13). We reasoned that if CD28 costimulation results in IL-2 production, and if NFAT is required for the expression of IL-2, there must be a CsA-resistant pathway for activation of NFAT. In this report, we demonstrate that NFAT can be activated in a CsAresistant manner.

MATERIALS AND METHODS
Cells and Tissue Culture-Freshly purified human peripheral blood T cells were greater than 95% CD3 ϩ cells. PMA was used at 10 ng/ml, ionomycin at 1 g/ml, and ␣CD28 monoclonal antibody (kindly provided by Dr. Carl H. June) at 100 ng/ml. Cyclosporin A (500 ng/ml) was added 30 min before addition of the other stimuli.
Plasmids-The thymidine kinase-␤-galactosidase (TK-␤gal) reporter plasmid was constructed by subcloning a HindIII/XhoI fragment containing the TK promoter from the pBLCAT2 vector (14) into the pEQ3 LacZ plasmid (15). The NFAT-TK-␤gal plasmid was constructed by synthesizing three copies of the IL-2 distal NFAT site (AAAGAAAG-GAGGAAAAACTGTTTCATACAG) 3 containing HindIII (at the 5Ј end) and BamHI (at the 3Ј end) restriction sites, and subcloning into the TK-␤gal plasmid. The NFB-TK-␤gal reporter plasmid was constructed as above, using the IgB sequence (ACAAGGGACTTTCCGCT). The expression vectors for NFAT c and for dominant negative NFAT c (16) were generous gifts from Dr. Gerald R. Crabtree. The pIL2-568 (15) construct (kindly provided by Dr. Christopher B. Wilson) was generated by subcloning the HindIII fragment (Ϫ568 to ϩ50) of the IL-2 promoter into the pEQ3 LacZ plasmid.
Transfection and ␤-Galactosidase Assay-Cells were transfected by a modified DEAE-dextran method (17). In brief, Jurkat cells were washed once with RPMI containing 50 mM Tris (pH 7.4). Cells were resuspended in the same medium at a density of 8 ϫ 10 6 /ml. An aliquot of a 0.5-ml cell suspension containing appropriate plasmids was added to 0.5 ml of Tris-RPMI containing 500 g/ml DEAE-dextran. The aliquot was incubated for 1 h with occasional stirring, then centrifuged, washed once with Tris-RPMI, and resuspended in 5 ml of RPMI (10% fetal calf serum) and plated in a six-well plate. The cells were stimulated after 24 h and harvested for ␤-galactosidase assay 18 h after stimulation. When multiple stimulations for a particular plasmid(s) were used, transfection was done in a single tube. After transfection, cells were divided into groups depending on the various stimulations. ␤-Galactosidase was assayed as reported (18). The results were expressed as activity, calculated as follows: (A 570 /g of protein) ϫ 10 3 .
Antisera-Antiserum ␣NFAT was raised against a peptide that is highly conserved in all NFAT proteins described to date. The peptide is found in the Rel homology domain, and its sequence is NH 2 -SDIEL-RKGETDIGRKNTR. Antiserum ␣NFAT p was raised against a peptide specific to NFAT p . Its sequence is VPAIKTEPSDEYEPSLI (sequence in murine NFAT p , but the antiserum also recognizes the human protein). Cytokine Testing-Human IL-2 was measured by ELISA (enzymelinked immunosorbent assay) test kits from Genzyme Corp. (Cambridge, MA). The assays were performed by the Clinical Immunology Services Program, SAIC, NCI-FCRDC, Frederick, MD.

Induction of NFAT Activity in a CsA-resistant Pathway-To
investigate whether NFAT can be activated by CD28 costimulation, three copies of the distal NFAT site from the IL-2 promoter were subcloned into a TK-␤gal reporter gene construct, designated as NFAT-TK-␤gal. This construct was transfected into Jurkat T cells, and NFAT activity was assayed after various treatments of the cells. Since we were looking for a CsA-resistant pathway, we could not use ␣CD3 as a stimulus, because one of the pathways it activates is Ca 2ϩ -dependent. Therefore, we chose to mimic the Ca 2ϩ -independent aspect of ␣CD3 signaling by PMA, as described previously (20). PMA alone had little effect on reporter activity (Fig. 1A) and ␣CD28 alone had no effect (data not shown), but the combination of PMA and ␣CD28 resulted in significant activity (about 4-fold induction relative to the vector alone). Strikingly, nearly all of this (PMA ϩ ␣CD28)-induced activity was resistant to CsA. In contrast, NFAT activity induced by PMA ϩ ionomycin was essentially completely CsA-sensitive (Fig. 1B). These results suggest that, in addition to the well-known Ca 2ϩ -dependent pathway, there is a second Ca 2ϩ -independent way to activate NFAT.
In the experiment described above, it was possible that the assay measured some other unknown factor able to bind to the NFAT sites in the reporter plasmid rather than bona fide NFAT activity. To test this, the assays were repeated in the presence of cotransfected NFAT c . Consistent with the results of others (16,21), NFAT c had little effect on the multicopy NFAT reporter in untreated cells (Fig. 1C), but had a marked effect in cells stimulated with PMA and ␣CD28 (15-fold induction of NFAT activity relative to the vector alone). As before, this induction was insensitive to CsA. In the same assay, promoter activity induced by PMA ϩ ionomycin was abolished in the presence of CsA (data not shown). These results indicate that overexpressed NFAT c protein can be activated through a CsAresistant pathway.
In order to determine whether the IL-2 promoter can also be activated by a CsA-resistant pathway, the 568-base pair IL-2 promoter-␤-galactosidase reporter construct (designated as pIL2-568) was tested in the transient transfection assay (Fig.  2). Like the multicopy NFAT-TK-␤gal construct, pIL2-568 showed much higher promoter activity after PMA/␣CD28 stimulation compared to either the unstimulated control or to PMA alone (panel A). This activity was further increased in the presence of overexpressed NFAT c protein (panel B). The vector alone did not have any activity (data not shown). The inducible promoter activity was completely resistant to CsA (panels A and B), whereas CsA suppressed PMA/ionomycin-inducible promoter activity (panel C). To test whether the promoter activity induced by PMA ϩ ␣CD28 was mediated by NFAT, a dominant-negative NFAT c protein (16) was tested in transient transfection assay. As shown in Fig. 2, A and B, overexpression indicating the involvement of AP1 in the inducible complex. To confirm that the inducible complex contained NFAT protein, we performed supershift analysis. As shown in Fig. 3B, antiserum raised against a peptide that is common to all NFAT proteins was able to block the inducible complex (lane 2). Another antiserum raised against an NFAT p -specific epitope showed a partial supershift (lane 4), whereas an irrelevant antiserum failed to react with the complex (lane 3). As shown in Fig. 3C, a PMA/␣CD28-inducible complex was also present in the nuclear extracts of peripheral blood T cells (compare lane 2 with 1). As in Jurkat cells, this inducible complex was insensitive to CsA, whereas the PMA/ionomycin-inducible complex was sensitive to CsA (compare lanes 10 and 11). Competition assays and supershift analysis revealed that the PMA/␣CD28 complex contained NFAT protein as well as the members of the AP1 family. These data show that in Jurkat cells and in peripheral blood T cells the NFAT DNA-binding complex can be induced in a CsA-resistant manner by PMA/␣CD28 stimulation.
Production of IL-2 by Peripheral Blood T Cells after Stimulation with PMA and ␣CD28 in the Presence of CsA-It has been shown previously that treatment of peripheral blood T cells with PMA ϩ ␣CD28 leads to IL2 production, and that this induction is largely resistant to CsA (10,12,13). We expected this to be true under our conditions as well, since reporter assays showed activation of the IL-2 promoter and EMSA showed induction of the NFAT complex. To test for IL-2 production, peripheral blood T cells were stimulated with PMA and ␣CD28 for 18 h with or without CsA. As shown in Fig. 4, stimulation in the absence of CsA resulted in the production of over 200 units/ml of IL-2; in the presence of CsA, IL-2 was produced at 70 -85% of this level (the slight drop may be due to toxic effects from the high concentration of drug used (500 ng/ml)). In contrast, CsA drastically reduced IL-2 production (to 7% of the level produced without CsA) in cells stimulated with PMA ϩ ionomycin. In absolute terms as well, CsA-resistant IL-2 levels are substantially higher in PMA-CD28 stimulated cells (140 units/ml) than in PMA/ionomycin-treated cells (24 units/ml). These data demonstrate the physiological significance of CsA-resistant NFAT activation.

FIG. 2. Activation of an IL-2 promoter-␤-galactosidase reporter gene construct in the presence of CsA.
A, the pIL2-568 reporter gene (5 g) along with expression vector alone or in combination with the expression vector for a dominant negative form of NFAT c (5 g) was transiently transfected into Jurkat cells and was subsequently stimulated as described above. B, the pIL2-568 reporter gene along with the NFAT c expression vector (5 g each) in the presence or absence of the dominant negative NFAT c (5 g) was transfected into Jurkat cells that were subsequently stimulated as described in Fig. 1. The amount of transfected plasmids was normalized by adding the empty vector, pBJ5. C, similar experiment as B except the stimulation was with PMA and ionomycin. D, the NFB-TK-␤gal reporter gene was transfected into Jurkat cells along with the expression vector for the dominant negative NFAT c (5 g each). Cells were subsequently stimulated as described in Fig. 1.

DISCUSSION
Previous studies have shown that the immunosuppressive drug CsA drastically inhibits, but does not completely block, IL-2 production following stimulation by ␣CD3 and ␣CD28 (10). The drug-resistant activity can most likely be traced to the CD28 signaling pathway, since stimulation of cells with PMA (to mimic the Ca 2ϩ -independent aspects of ␣CD3 signaling) and ␣CD28 is entirely resistant to CsA (10). In this case, IL-2 is produced at significant levels in the continued presence of CsA or FK506. Since the transcription factor NFAT is required for expression of the IL-2 gene, these results imply that, in addition to the well-known Ca 2ϩ -dependent pathway, NFAT can be activated by a CsA-resistant pathway. Data presented here show that this is the case.
The involvement of NFAT in the activation of reporter gene constructs (both NFAT-TK-␤gal and pIL2-568) after PMA/ ␣CD28 treatment in the presence of CsA was demonstrated by the facts that (i) overexpressed wild type NFAT augmented reporter activity only after PMA/␣CD28 stimulation, and (ii) the inducible reporter activity was essentially eliminated in the presence of a dominant negative mutant of NFAT. The presence of NFAT in the PMA/␣CD28-inducible DNA-protein complexes was also demonstrated by the supershift assays. In agreement with the transient transfection assays, the peripheral blood T cells were also shown to produce IL-2 after PMA/ ␣CD28 treatment, and this production was resistant to CsA.
The established pathway for NFAT activation involves Ca 2ϩdependent activation of the serine-threonine phosphatase calcineurin. Subsequent dephosphorylation of NFAT leads to its nuclear translocation and, in conjunction with AP-1, DNA binding. The data presented here suggest an alternate pathway for NFAT activation that is calcineurin independent. Since dephosphorylation of NFAT appears to be necessary for its activity, a different phosphatase must be involved in this pathway. A candidate second messenger (instead of Ca 2ϩ ) is ceramide, which has been reported to be involved in CD28 signaling (22,23). Interestingly, a serine/threonine phosphatase has been reported recently that is involved in the sphingomyelin pathway (24). It is an attractive hypothesis that this or another ceramide-induced phosphatase can lead to NFAT activation in a CsA-resistant manner. This idea is currently being tested.
The present finding may explain the ineffectiveness of CsA in the treatment of graft versus host disease following allogenic bone marrow transplantation (25,26). The cytotoxic T cells thought to be responsible for graft versus host disease (27) express CD28 on their surface and are capable of producing cytokines responsible for immune reaction in the presence of CsA.