Treatment of Human T Cells with Bisperoxovanadium Phosphotyrosyl Phosphatase Inhibitors Leads to Activation of Cyclooxygenase-2 Gene*

Protein-tyrosine phosphatase (PTP) inhibitors are potent activators of T lymphocytes, most likely by affecting the early steps of T cell receptor (TCR) signaling. We have analyzed the effect of the PTP inhibitor bisperoxovanadium (bpV) on expression of the human cyclooxygenase 2 (COX-2) gene, which is induced following TCR triggering. Here we show that COX-2 promoter activity is markedly up-regulated following exposure of Jurkat T cells to bpV(pic). Interestingly enough, treatment of Jurkat cells with cyclic AMP-elevating agents such as forskolin, in combination with bpV, resulted in a more important COX-2 transcriptional activation. Such activation is inhibited by the immunosuppressive drugs FK506 and cyclosporin A. The two nuclear factor of activated T cells (NFAT) binding sites located within the COX-2 promoter region are involved in bpV-mediated positive effect on COX-2 promoter. Electromobility shift assays showed that NFAT1 and activator protein-1 are both translocated to the nucleus following bpV treatment. The active participation of p56 lck , ZAP-70, p36 LAT , and calcium in the bpV-dependent signaling cascade leading to COX-2 transcriptional activation was demonstrated using deficient cell lines and specific inhibitors. Although several PTPs are most likely targeted by bpV, our data suggest that the bpV-mediated signaling cascade is initiated by inhibition of SHP-1, which leads to phosphorylation of p56 lck and ZAP-70 and, ultimately, to NFAT and activator protein-1 nuclear translocation. These results suggest that PTP inhibitors can activate COX-2 gene expression in a manner very similar to the stimulation induced by TCR triggering.

Prostaglandins are important mediators in many physiological processes including cell growth, ovulation, and immune functions. These potent lipid molecules play key roles in the inflammatory response, and inhibition of their synthesis is the target of most non-steroidal anti-inflammatory drugs. Prostaglandin biosynthesis is controlled by the cyclooxygenase (COX) 1 enzyme, which catalyzes the initial conversion of arachidonic acid to prostaglandin H 2 , a precursor common to all prostanoids. Two COX isoenzymes are expressed in mammalian tissues, COX-1 and COX-2. COX-1 is expressed constitutively in most tissues, whereas COX-2 is induced by various proinflammatory cytokines and mitogenic agents in different cell types and is thought to be responsible for the increased production of prostaglandins in inflammatory disorders (1). COX-2 synthesis is regulated primarily at the transcriptional level via distinct pathways in various cell types (including vascular endothelial cells, epithelial cells, pancreatic cells, mast cells, and monocytes) and involves transcription factors such as nuclear factor of chain in B cells (NF-B), activator protein-1 (AP-1), CCAAT/enhancer-binding protein (C/EBP), cAMP-responsive element-binding protein (CREB), and nuclear factor of activated T cells (NFAT) (2)(3)(4)(5)(6)(7)(8)(9)(10). COX-2 seems to act as an early T cell receptor (TCR)-responsive gene, because TCR engagement results in COX-2 gene expression (11,12) through two NFAT motifs present in the COX-2 promoter region (3). Moreover, inhibition of COX-1 or COX-2 activity by specific inhibitors blocks T cell activation, and these two genes have been identified as integral and sequential components of TCR signaling to p38 mitogen-activated protein kinase in T cell activation (12,13).
One of the earliest events following TCR-mediated T cell activation is an increase in tyrosine phosphorylation of specific proteins such as cell surface receptors, adapter proteins, and kinases (14,15). The intracellular phosphotyrosine content is tightly regulated by a complex balance between protein-tyrosine kinase (PTK) and protein-tyrosine phosphatase (PTP) activities. The importance of PTP in T cell response has been demonstrated by studies using specific PTP inhibitors that reported an increase of T cell activation by these compounds (16). A panoply of second messengers such as p56 lck , p59 fyn , ZAP-70, and the mitogen-activated protein kinase cascade are induced following treatment of lymphocytes with PTP inhibitors such as pervanadate (17,18). We have demonstrated recently that treatment with potent PTP inhibitors bisperoxovanadium (bpV) compounds (19) strongly activates both NF-B and NFAT transcription factors in T cell lines and peripheral blood mononuclear cells (20,21). These observations, along with the previously reported implication of NF-B and NFAT in the control of COX-2 expression and the pivotal function of COX-2 in T cell activation, led us to scrutinize the effect of bpV molecules on the transcriptional induction of COX-2 isoenzyme. In this study, we show that bpV compounds strongly activate COX-2 promoter transcription in human T cells. The bpV-mediated activating effect on COX-2 requires calcineurin and two transcription factors, i.e. NFAT and AP-1.
Transient Transfection and Stimulation of Cells-Cells were electroporated using a gene pulser I apparatus (Bio-Rad) (960 microfarad, 250 V) at room temperature. Cells were concentrated at 37.5 ϫ 10 6 per ml of complete culture medium. Cells (400-l aliquots) were electroporated with either 5 g of the reporter construct DNA alone or, in the case of reconstitution experiments, with 5 g of reporter construct DNA and 0, 5, or 15 g of the expression plasmid. The total DNA amount for the reconstitution experiments was maintained constant at 20 g using the empty vector.
To minimize variations in plasmid transfection efficiencies, cells were transfected in bulk and were separated at 36 h post-transfection into various treatments groups at a density of 10 5 cells/well (100 l) in 96-well flat-bottom plates. For studies using pharmacological inhibitors, cells were resuspended in fresh cell culture medium (1 ϫ 10 6 cells/ml) and were treated for 60 min with subcytotoxic and subcytostatic concentrations of cyclosporin A (1 to 100 ng/ml) (Fujisawa, Osaka, Japan), FK506 (1 to 10 ng/ml) (Sigma), TMB-8 (Sigma) (25 to 100 M), or piceatannol (Biomol) (0.1 to 50 M) before addition of bpV(pic). Cells were either left untreated or were treated with phorbol 12-myristate 13-acetate (PMA) at 20 ng/ml (Sigma) alone or in combination with ionomycin at 1 M (Sigma) or with various concentration of bpV(pic) (Alexis Biochemical), used either alone or in combination with 10 M forskolin (Cedarlane) in a final volume of 200 l. Next, cells were incubated at 37°C for 8 h unless otherwise specified. Luciferase activity was determined following a previously described protocol (20). -Fold induction was obtained by calculating the ratio between measured relative light units of treated over untreated samples. Results shown are the mean Ϯ S.D. of four determinations and are representative of three different experiments.
Nuclear Extract Preparation and Electrophoretic Mobility Shift Assay-Nuclear extracts were prepared according to a previously described protocol (20). Briefly, untreated or treated cells (5 ϫ 10 6 ) were first washed with phosphate-buffered saline. Cells were then resuspended in 400 l of hypotonic buffer (Buffer A) (10 mM HEPES, pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride) and kept for 15 min on ice before lysis with 25 l of Nonidet P-40 10%. After brief vortexing and centrifugation, the supernatant was discarded, and the pellet was resuspended with an hypertonic buffer (Buffer B) (20 mM HEPES, pH 7.9, 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride) followed by gentle agitation for 15 min. The solution was then centrifuged, and the supernatant was assayed for protein concentration by BCA assay (Pierce) and stored at Ϫ85°C until use.
Nuclear extracts (10 g) were incubated for 20 min at room temperature in 20 l of 1ϫ binding buffer (10 mM HEPES, pH 7.9, 4% glycerol, 1% Ficoll, 25 mM KCl, 1 mM dithiothreitol, 0.5 mM EDTA, 25 mM NaCl, 2 g of poly(dI-dC), and 10 g of nuclease-free bovine serum albumin fraction V) containing 0.8 ng of [␥-32 P]-labeled double-stranded DNA oligonucleotide. The following double-stranded DNA oligonucleotides were used as probes and/or competitors: the distal NFAT binding site of the human COX-2 promoter (nucleotides Ϫ117 to Ϫ191), 5Ј-tcgaCAA-GGGGAGAGGAGGGAAAAATTTGTGGC-3Ј; the NFAT/AP-1 proximal site of the COX-2 promoter (nucleotides Ϫ82 to Ϫ58), 5Ј-tcgaCAAAAG-GCGGAAAGAAACAGTCATTTC-3Ј; the distal NFAT binding site from the murine interleukin-2 promoter, 5Ј-TCGAGCCCAAAGAGGAAAAT-TTGTTTCATG-3Ј; the consensus NF-B binding site, 5Ј-ATGTGAGG-GGACTTTCCCAGGC-3Ј; and the consensus binding site for Oct-2A (used for nonspecific competition). DNA-protein complexes were resolved from free-labeled DNA by electrophoresis in native 4% (w/v) polyacrylamide gels. The gels were subsequently dried and autoradiographed on an Eastman Kodak Co. Biomax MR film at Ϫ85°C. Cold competition assays were conducted by adding a 100-fold molar excess of unlabeled double-stranded DNA oligonucleotide simultaneously with the labeled probe. Supershift assays were performed by preincubation of nuclear extracts with 1 l of the appropriate antibody in the presence of all of the components of the binding reaction for 30 min on ice before addition of the labeled probe.

COX-2 Promoter Activity Is Strongly Activated by the PTP
Inhibitor bpV(pic)-To assess whether PTP inhibitors can modulate the human COX-2 gene promoter activity, we performed reporter assays using various COX-2 promoter-based constructs. The P2-1900 construct, which comprises a region spanning from Ϫ1796 to ϩ104 bp of the human COX-2 gene placed in front of the luciferase reporter gene (3), was transiently transfected in Jurkat cells. As reported previously (3), treatment of Jurkat cells with the PMA/ionomycin combination led to an increase in COX-2 promoter-driven reporter gene activity (Fig. 1A). Interestingly, treatment with the PTP inhibitor bpV(pic) at 20 M induced an even stronger activation of COX-2 promoter. A greater bpV-mediated induction of COX-2 activity (i.e. 70-fold increase) was observed with P2-274 (Fig. 1B), a molecular construct that comprises the promoter region spanning from Ϫ170 to ϩ104. This suggests that the upstream region contains negative regulation elements. Dose-response and time course experiments indicated that the highest transcriptional induction was achieved following an 8-h incubation time period with 20 M bpV(pic) (Fig. 1, B and C). Altogether these data suggest that inhibition of PTP activity triggers COX-2 promoter activity in human T cells.
The Calcineurin-mediated Signaling Pathway Is Involved in COX-2 Transcriptional Activation by bpV(pic)-We have demonstrated previously (20,21) that bpV compounds can induce both calcineurin-dependent and -independent signaling pathways in T cells. To elucidate the signal transduction pathway(s) involved in bpV-mediated COX-2 transcriptional activation, we used the immunosuppressive drugs FK506 and CsA. Jurkat cells were first transiently transfected with P2-274 and next pretreated with increasing amounts of FK506 or CsA before activation by bpV(pic). As depicted in Fig. 2, treatment with both calcineurin inhibitors resulted in an almost complete inhibition of COX-2 transcriptional activation. Similar observations were made when using the P2-1900 COX-2 promoter construct (data not shown). No detectable cell toxicity was observed upon cell incubation with these inhibitors at the tested concentrations (data not shown). These results demonstrate that COX-2 transcriptional activation by bpV occurs through a pathway that relies on calcineurin activity.
The two NFAT Binding Sites in the COX-2 Promoter Are Involved in Transcriptional Activation by bpV(pic)-We next wanted to identify the precise cis-acting regions required for the inducibility of the COX-2 promoter region to the bpV(pic) compound. To this end, we first used several 5Ј deletion constructs that were created from the P2-274 vector (Fig. 3A). A deletion of 83 bp (i.e. P2-192 vector) resulted in a 60% decrease in the inducibility by bpV(pic) (Fig. 3B). A supplementary deletion of 42 bp (i.e. P2-150 vector) decreased transcriptional activity by an additional 50%, leaving only a 20% residual activation. These results suggest that regions spanning from Ϫ170 to Ϫ88 and Ϫ88 to Ϫ46 bp relative to the transcription start site of the human COX-2 gene include sequences responsible for the bpV-mediated positive effect on COX-2 gene. These regions contain a binding site for C/EBP, a CRE, and two NFAT binding sites, one of which contains an adjacent AP-1 motif (i.e. pNFAT) (Fig. 3A). Given that the two NFAT elements have been shown to be crucial for COX-2 transcriptional activation by PMA and ionomycin (3), and treatment of human T cells with bpV molecules resulted in induction of NFAT (21), we next tested the importance of NFAT binding sites in the noticed bpV-induced COX-2 activation. To this end, we used constructs bearing mutations destroying proximal (pNFAT-mut) or distal (dNFAT-mut) NFAT sites. Transfection of Jurkat cells with the P2-274 vector bearing either COX-2 dNFAT or pNFAT mutated sites resulted in a 40 to 50% diminution in the responsiveness to bpV(pic) (Fig. 3C). Moreover, data obtained with dNFAT-mut were comparable with results seen when using the P2-192 deletion mutant, which lacks the entire dNFAT binding site. The inducibility by bpV(pic) was even further diminished when cells were transiently transfected with a P2-192-based construct carrying mutations in the remaining pNFAT site (i.e. P2-192 pNFAT-mut vector). These data support the idea that NFAT plays a crucial role in bpV- bpV Induces NFAT and AP-1 Binding Activities onto COX-2 Promoter-The COX-2 pNFAT element contains an adjacent AP-1-like binding site that is destroyed in pNFAT mutant. Thus, to establish the possible implication of the AP-1 transcription factor and to further incriminate the participation of NFAT in the observed phenomenon, we performed electromobility shift assays with nuclear extracts of Jurkat cells using DNA probes specific for the distal or the proximal NFAT elements. Two specific retarded complexes were induced with different kinetics when using the proximal NFAT probe (Fig.  4A). The intensity of complex I was maximal at 270 min whereas complex II reached a peak after only 30 min of bpV treatment. A single retarded complex, i.e. complex II, was seen when using the distal NFAT probe. Both complexes were efficiently competed away with 100-fold molar excess of the unlabeled specific probe but not by a nonspecific oligonucleotide (i.e. Oct-2A). The intense, faster migrating band was considered as nonspecific, because it was not consistently competed away by an excess of specific oligonucleotide. Migrating complexes were further identified using specific competitions and supershift assays. Complex I was specifically competed away by an NFATspecific oligonucleotide and also by an AP-1-specific sequence, but not by a control NF-B-specific oligonucleotide (Fig. 4B,  lanes 1-4), suggesting that this complex contains both NFAT and AP-1 bound to their specific sites. Complex II was also specifically competed away by a NFAT-specific oligonucleotide, but not by either NF-B-or AP-1-specific sequences, suggesting that NFAT is responsible for this complex (Fig. 4B, lanes 1-4). However, because this band was not completely competed away either by the consensus NFAT or by the distal NFAT sequences (lanes 2 and 5) but only by the pNFAT oligonucleotide, it is likely that the complex II represents a superimposition of a major NFAT complex and a minor AP-1 complex. Supershift assays indicated that NFAT1, but not NFAT2, was present in both the NFAT and NFAT⅐AP-1 complexes (lanes 8 and 9). Incubation with a pan-NFAT antiserum prevented the formation of both the NFAT and the NFAT⅐AP-1 complexes (lane 10). Similar observations were made when the distal NFAT se- quence was used as a probe (lanes [11][12][13][14][15][16][17][18][19][20], thus confirming that treatment of Jurkat cells with bpV molecules induces the binding of NFAT1 on both NFAT elements of the COX-2 promoter. Transcriptional Activation of COX-2 by bpV(pic) Is Independent of C/EBP Transcription Factors-Given that the COX-2 promoter contains a C/EBP binding site, we also assessed the importance of C/EBP transcription factors in bpV-dependent induction of the COX-2 promoter. The implication of the C/EBP family of transcription factors was assessed by using pCMV-LIP, a construct that encodes a truncated C/EBP protein that has only the DNA binding and leucine zipper domains and possesses a dominant negative function (28). Co-transfection of this expression vector, along with P2-274, had no effect on bpV-induced reporter gene expression (Fig. 5), thus ruling out any implication of C/EBP in COX-2 transcriptional regulation by the PTP inhibitor bpV(pic).
Cyclic AMP-mediated Signal Transduction Pathways Potentiate bpV-induced COX-2 Activation-Because the CRE presents in the COX-2 promoter has been shown to be involved in COX-2 regulation in various cell types such as monocytes, macrophages, and endothelial and epithelial cells (6 -8, 33), we investigated the importance of cAMP-related effectors in induction of COX-2 upon treatment of T cells with bpV(pic). When Jurkat cells were transfected with COX-2 promoter constructs and were treated with the cAMP-elevating agent forskolin (FSK), a very modest increase in transcriptional activity was observed (never more than 2-fold) (Fig. 6, panels A and B). However, when FSK was used with bpV(pic), this combination resulted in an induction of COX-2 activity that was superior to the one seen when bpV(pic) was used alone. This up-regulating effect was observed with both COX-2 molecular constructs, i.e. P2-1900 and P2-274. Cyclic AMP is a ubiquitous intracellular second messenger that transmits information to several proteins including protein kinase A and a family of cAMP-responsive nuclear factors that bind to the CRE consensus sequence. Phosphorylation of these CREBs by the C subunit of protein kinase A modulates their overall activity. To assess the importance of CREB factors in COX-2 transcriptional activation, we used a dominant negative form of CREB named killer CREB (KCREB), which is unable to bind to CRE DNA sequence and blocks the binding of wild type CREB when present as a KCREB⅐CREB heterodimer. Jurkat cells were co-transfected with increasing amounts of the KCREB expression vector, along with P2-274, and were next treated with bpV(pic) and/or FSK. The dominant negative CREB was found to severely decrease COX-2 transcriptional activation that is seen following treatment with bpV(pic) alone (Fig. 6C). The enhancing effect of FSK was completely abolished in the presence of high concentrations of KCREB, therefore suggesting that the transactivating potential showed by FSK is mediated by CREB transcription factors. Surprisingly, mutation of the CRE binding site in the COX-2 promoter had no effect on induction by both bpV(pic) and the combination of bpV(pic) and FSK (Fig.  6D). These observations suggest that CREB transcription factors are not acting directly on COX-2 promoter but are activating the expression of other transcription factors that are able to regulate the COX-2 promoter. In an attempt to define whether bpV-dependent induction of intracellular cAMP could lead to an increased expression of AP-1 via an effect on CRE elements present in Fos and Jun promoters, we performed electrophoretic mobility shift assay using as a probe the COX-2 proximal NFAT sequence. Treatment of Jurkat cells with FSK did not induce any specific binding activities, but treatment with bpV(pic) and FSK resulted in a much stronger NFAT⅐AP-1 complex when compared with treatment with bpV(pic) alone (Fig. 6E).

COX-2 Transcriptional Activation by bpV(pic) Is Dependent on TCR-mediated Signaling Cascade Events-We next wanted
to characterize the bpV-induced signaling pathway(s) leading to COX-2 transcriptional activation. We reported previously (21) that induction of NFAT by bpV treatment in T cells is a process that relies on TCR-dependent biochemical events. Because NFAT is directly involved in the inducibility by bpV(pic) of the COX-2 promoter, we assessed the implication of different TCR-related signal transducers in this process. We used Jurkat-derived cell lines that are TCR-negative (i.e. J.RT3-T3.5) or deficient for either the PTK p56 lck (i.e. JCAM1.6) or the adapter protein p36 LAT (i.e. JCAM2). When such cells were transiently transfected with P2-274, the TCR-negative J.RT3-T3.5 cell line was found to be fully responsive to bpV(pic) stimulation, thus ruling out an involvement of the TCR in COX-2 activation by bpV compounds (Fig. 7A). In contrast, both JCAM1.6 and JCAM2 cell lines were unresponsive to bpV stimulation (Fig.  7A). To further establish the importance of p56 lck , a member of the Src family of non-receptor PTK, we cotransfected JCAM1.6 cells with increasing amounts of the p56 lck -encoding vector pEFneoLckWT, along with P2-274. The bpV-induced COX-2 transcriptional activation was partially restored (Fig. 7B), confirming the implication of p56 lck in the noticed up-regulation of COX-2 promoter activity. Similarly, co-transfection of a p36 LAT expression vector also partially restored the inducibility of the JCAM2 cell line, demonstrating that p36 LAT is also part of the bpV-induced signaling pathway (Fig. 7C). The incomplete restoration of inducibility to bpV treatment is most likely because of the low transfection efficiency of JCAM1.6 and JCAM2 cell lines. The implication of ZAP-70 was investigated using the Syk family kinase inhibitor piceatannol (34). Pretreatment of Jurkat cells transfected with P2-274 with increasing concentrations of piceatannol induced a dose-dependent inhibition of the COX-2 transcriptional activation by bpV(pic) (Fig. 7D).
More downstream effectors of the TCR signaling cascade include the release of intracellular calcium, an event that is known to be crucial for NFAT activation. To monitor the importance of calcium, we used the modulator of intracellular calcium homeostasis TMB-8 that blocks endogenous calcium release. Pretreatment of Jurkat cells transiently transfected with P2-274 with increasing concentrations of TMB-8 resulted in a marked diminution of COX-2 promoter-driven luciferase activity upon bpV(pic) stimulation (Fig. 7E). No loss of cell viability was observed with the tested doses of TMB-8 (data not shown).
Possible Involvement of SHP-1 and TC-PTP in COX-2 Transcriptional Activation by bpV(pic)-We next seeked to identify the tyrosine phosphatase(s) that, once inhibited by bpV(pic), is resulting in an increased phosphorylation of kinases such as p56 lck and ZAP-70 and, ultimately, in COX-2 transcriptional activation. We focused our efforts on CD45, SHP-1, He-PTP, and TC-PTP based on the notion that these PTPs are expressed at high levels in Jurkat cells (35). The implication of the transmembrane PTP CD45 was ruled out by the observation that bpV-mediated activation of COX-2 promoter in the CD45-negative cell line J45.01 is comparable with that seen in parental CD45-expressing Jurkat cells (data not shown). However, overexpression of SHP-1 and TC-PTP, but not He-PTP, caused a marked reduction in bpV-induced COX-2 transcriptional acti-

FIG. 8. bpV-mediated increase in COX-2 transcriptional induction is diminished by SHP-1 and TC-PTP overexpression.
Jurkat cells were transiently transfected with P2-1900 (5 g), along with 5 or 15 g of a vector encoding for wild type SHP-1, TC-PTP, or He-PTP. The total amount of DNA was kept constant in each transfection by the addition of an empty control vector. Cells were then either left untreated or were treated for 8 h with bpV(pic) before measuring luciferase activity. Results are expressed as percentage of the activation observed with P2-1900 alone, which was arbitrarily fixed at 100. vation (Fig. 8). The noticed inhibition of bpV-mediated increase in COX-2 activity with wild type SHP-1 and TC-PTP could not be compared, because the two proteins are expressed from different promoters. DISCUSSION Although it has been shown that COX-2 gene expression in T cells is regulated by various stimuli, the exact intracellular signaling events involved in this regulation have not yet been elucidated. In this study we analyzed the effect of the PTP inhibitor bpV(pic) on regulation of COX-2 transcriptional activity. We demonstrate here that PTP inhibition by means of a treatment with bpV(pic) results in a significant increase in COX-2 promoter activity through induction of both NFAT and AP-1 transcription factors. These results confirm previous observations demonstrating the importance of NFAT and AP-1 binding sites in the transcriptional induction of COX-2 isoenzyme that is seen upon T cell activation (3). Both the COX-2 distal NFAT and proximal NFAT/AP-1 binding sites were found to be involved in bpV-dependent enhancing effect on COX-2 promoter activity, because selective mutation of each site resulted in a diminution of the promoter inducibility (Fig.  3). Other cis-acting elements were described previously to be important for COX-2 gene regulation in non-T cells. For example, a C/EBP element appears to be crucial for COX-2 transcriptional regulation in macrophages, as well as in fibroblasts and osteoblasts, with the involvement of different members of the C/EBP family in each cell type (2,4,5,9,10). A cyclic AMPresponsive element is also involved in COX-2 regulation in various cell types including endothelial and epithelial cells (6 -8) and monocytes/macrophages (33). We did not observe any implication of these two motifs in COX-2 transcriptional activation induced by bpV(pic) (see Fig. 5 and Fig. 6D). Indeed, the COX-2 gene appears to be regulated by a particular set of transcription factors for each cell type and stimulus, a situation that could be related to the various physiological roles played by the COX-2 protein in various cell types, for example inflammation, tissue damage, tumorigenesis, or pain transmission.
Our results indicate that both NFAT and AP-1 binding activities are induced following bpV(pic) stimulation in Jurkat cells. NFAT1 is the most prevalent NFAT family member that is translocated to the nucleus (Fig. 4B). This is consistent with earlier reports demonstrating the predominance of NFAT1 in activated T cells (32), which is in line with the previous observation indicating that bpV stimulation of T cells leads to the binding of NFAT1 onto the regulatory elements of human immunodeficiency virus (21). Another study suggests that NFAT2 could also be implicated in COX-2 transcriptional regulation (3); however, this observation was made following overexpression of NFAT2 by transient transfection. We also identified by mobility shift assays an NFAT⅐AP-1 complex that is strongly induced upon bpV treatment. Such highly stable ternary complexes are known to act in synergy and to regulate expression of various genes (36 -38). NFAT alone is able to bind to the dual NFAT/AP-1 site (complex II in Fig. 4). However very weak, if any, binding of AP-1 alone was observed. This suggests that NFAT is the transcription factor that is crucial for COX-2 activation by bpV(pic). This is exemplified by the observation that bpV(pic)-mediated induction of COX-2 promoter activity is abolished by a pretreatment with the calcineurin inhibitors FK506 and CsA (Fig. 2). Activation of calcineurin is necessary to dephosphorylate NFAT proteins and to allow their translocation to the nucleus; however, no such role was described for AP-1. Although a calcineurin-dependent regulation of Jun proteins has been shown in activated T cells (39), induction of AP-1 was found to be inhibited by CsA in primary cells (40) but not in Jurkat cells (41). Consequently, the observed inhibitory ef-fect of FK506 and CsA is linked with a blockade of nuclear translocation of NFAT.
Differences in the kinetics of NFAT and AP-1 activation following bpV treatment were detected when performing mobility shift assays. Indeed, nuclear translocation of NFAT was already maximal at 30 min whereas the intensity of the NFAT⅐AP-1 complex reached a peak at 270 min after bpV treatment (Fig. 4A). Such a delayed translocation suggests that AP-1 activation by bpV compounds requires transcription and translation of Fos and/or Jun.
Additional experiments revealed that the PTKs p56 lck and ZAP-70, the adapter protein p36 LAT , and calcineurin are all implicated in bpV-dependent up-regulation of COX-2 promoter activity in human T cells. These observations suggest that the bpV(pic) PTP inhibitor affects the most proximal membrane events of TCR signaling. The TCR complex itself is not required for the triggering of COX-2 activity by bpV(pic) (Fig. 7A). Interestingly, ZAP-70 is able to interact with an immunoreceptor tyrosine-based activation motif distinct from the TCR/CD3 complex (42). The observed similarity between signaling pathways induced upon TCR engagement and bpV treatment is not surprising if one considers that addition of bpV to T cells treated with phytohemagglutinin or anti-TCR/CD28 antibodies results in activation of the interleukin-2 promoter through a concomitant induction of NF-B, NFAT, and AP-1 transcription factors (21). In this study, the working model to explain activation of NFAT by bpV molecules proposed that inhibition of PTPs led to p56 lck activation, which, via phosphorylation of ZAP-70, induced intracellular calcium increase and calmodulin activation (21).
Interestingly enough, CREB is also crucial for COX-2 gene activation by bpV(pic), with no implication of the CRE site present in the COX-2 promoter (Fig. 6, panels C and D). This observation suggests that PTP inhibition by bpV(pic) induces, via an increase in intracellular calcium, activation of calciumcalmodulin kinase IV, a kinase that is known to phosphorylate CREB serine 133 (43,44). This phosphorylated CREB does not directly transactivate the COX-2 promoter but rather acts indirectly by increasing c-Fos and/or c-Jun expression. Indeed, both fos and jun regulatory regions contain consensus sites for CREB (45). This indirect effect accounts for the delayed apparition of the NFAT⅐AP-1 complex (compared with the NFAT complex) observed in the electrophoretic mobility shift assay experiments. The residual CREB-independent transactivating potential could be explained by a signaling cascade initiated by ZAP-70 and p36 LAT . ZAP-70 could also activate the Ras/Raf pathway, leading to transcriptional activation of c-fos via the extracellular signal-regulated kinase pathway and phosphorylation of c-Jun by the c-Jun N-terminal kinase (JNK) (46,47). The implication of p36 LAT is also possible, because this adapter protein recruits phospholipase C␥ to the membrane, leading to the activation of protein kinase C via formation of diacylglycerol. Protein kinase C is involved in c-fos transcriptional activation by the Elk-1 and serum response factor (SRF) transcription factors via a p21 ras -dependent cascade (48), as well as in the direct activation of c-Fos by c-Fos kinase (49) and in c-Jun activation by JNK (50).
Another pathway leading to COX-2 activation following TCR triggering involves Cot kinase, a mitogen-activated kinase kinase kinase. This kinase was shown to activate NFAT-mediated transactivation of the human COX-2 promoter in a CsAindependent manner and to synergize with the calcineurindependant pathway (51). We showed that bpV-induced COX-2 transcriptional activation is inhibited by both CsA and FK506, which rules out any implication of Cot kinase in our findings.
We also demonstrate that cAMP-mediated signaling events cooperate with bpV(pic) to induce a more powerful activation of COX-2 (Fig. 6, panels A and B). This effect is mediated by the CREB family of transcription factors (Fig. 6C) and, most likely, via CREB phosphorylation on its serine 133 by protein kinase A. However, the CRE element located in the COX-2 promoter is not necessary for this transactivating potential (Fig. 6D), thus suggesting an indirect effect of CREB on COX-2 transcriptional activity, most likely via an increase in Fos and/or Jun expression. This hypothesis in strengthened by the observation that FSK treatment induced a marked increase of the NFAT⅐AP1 complex that is seen upon bpV(pic) treatment (Fig. 6E). The inability of cAMP-mediated signals to fully transactivate COX-2 gene expression can then be explained by the fact that AP-1 can activate transcription only when complexed with NFAT, which is not induced upon FSK treatment. This is paralleled by the very inefficient transcriptional activation induced by PMA or ionomycin alone as compared with the PMA/ ionomycin combination (Fig. 1A). Another possible effect of cAMP would involve serine phosphorylation of He-PTP by protein kinase A, which, in turn, would inhibit binding of He-PTP to MAP kinases, leading to their activation (52). This effect would be sufficient to induce transcription from the c-fos promoter and could then contribute to the increase in AP-1 binding to the COX-2 promoter that is observed upon FSK treatment. It can be proposed that bpV molecules act by inhibiting one or several PTP(s) that normally attenuate a constitutive kinase activity. Several PTP are most likely targeted by bpV(pic), leading to the activation of multiple kinases and to COX-2 transcriptional activation. However, our results show that the PTK p56 lck and ZAP-70 are both critical to the bpV-induced activation, suggesting that the signaling cascade is initiated by the inhibition of a PTP that normally dephosphorylates p56 lck and/or ZAP-70. Signaling events transduced through the TCR complex following treatment with the PTP inhibitor pervanadate was shown to be mediated by inhibition of CD45 (18). The fact that bpV-mediated activation of COX-2 promoter was still detected in the CD45-negative cell line J45.01 is an indication that this T cell membrane PTP is not participating to this process (data not shown). The Src homology-containing PTP SHP-1 represents an ideal candidate, because it has been proposed as a negative regulator of T cell activation by decreasing the activity of several molecules involved in TCR signaling machinery, including p56 lck , phospholipase C␥, ZAP-70, and Syk (25,30,(53)(54)(55). Our data showing an inhibition of bpVinduced transcriptional activation when SHP-1 is overexpressed further strengthen this hypothesis. Overexpression of SHP-1 may counteract the inactivation caused by bpV(pic) and restore the inhibition of kinase activity. Our results are in contradiction with a previous report stating that a PTP distinct from CD45, SHP-1, and SHP-2, present in T cell membranes, is responsible for the p56 lck -dependent tyrosine phosphorylation events that are induced by PTP inhibitors (56). It should be mentioned that PTP inhibitors used in this study were hydrogen peroxide and vanadate. It is thus possible that bpV molecules target PTPs that are different from the ones affected by hydrogen peroxide and vanadate.
Our data do not suggest an implication of He-PTP, although this cytosolic PTP was shown to inhibit the TCR-induced activation of NFAT/AP-1 (31). This discrepancy is probably related to some differences in the signaling pathway induced by bpV compared with TCR stimulation. On the other hand, we observed an inhibition of bpV-induced activation following overexpression of TC-PTP. Localization of TC-PTP in the nucleus or endoplasmic reticulum suggests that this PTP is not playing an active role in the most proximal TCR signaling events (35). The inhibitory effect that is seen following overexpression of wild type TC-PTP is then most likely related with dephosphorylation of some transcription factors in the nucleus, leading to a decrease in COX-2 transcriptional activity. Last, we cannot totally exclude the implication of the Pro-Glu-Ser-Thr (PEST)enriched phosphatase that acts as a potent negative regulator of T cell activation through an association with Csk and a possible direct dephosphorylation of Src family PTK (27).
In summary, we have demonstrated that COX-2 transcriptional activity is augmented following treatment by the PTP inhibitor bpV(pic), in a manner very similar to the stimulation induced by TCR triggering. These findings render bpV compounds very attractive for the restoration of immune functions in immunosuppressed individuals, especially knowing that COX-2 is an integral component of the TCR signaling cascade and is crucial for T cell activation.