Materials and Methods

Cell Cultures.

Human mononuclear cells were obtained from heparinized venous blood of healthy volunteers through Ficoll Hypaque (Pharmacia Fine Chemicals, Uppsala, Sweden) centrifugation. The layer containing mononuclear cells was taken and the cells washed thoroughly by centrifugation in DMEM and finally resuspended in DMEM 2% fetal calf serum (FCS). Monocytes were separated by adherence to plastic disks for 2 h at 37°C. T cells were further purified by passing the nonadherent population through a nylon fiber wool column as described (Pimentel-Muiños et al., 1994). The purity of this population (detected by flow cytometry) was always greater than 95% CD3⁺ cells.

Purified T cells (10⁶/ml in DMEM medium containing 10% FCS) were seeded in 96-well U-bottom microtiter plates (10⁵/100 μl/well) and stimulated with immobilized (1 μg/ml-coated wells) anti-CD3 antibody (SPV3Tb kindly provided by Dr. J. E. de Vries; DNAX, Palo Alto, CA), in the presence of different concentrations of RP (Schering, Madrid, Spain), dBcAMP (Sigma, St. Louis, MO), and/or the protein kinase A-specific inhibitor KT 5720 (Kamiya Biomedical, Thousand Oaks, CA). The cultures were incubated for 72 h at 37°C and cell proliferation was evaluated by incorporation of [³H]thymidine (PerkinElmer Life Science Products, Boston, MA) into DNA during the last 16 h of culture. For cytokine assays, supernatants were harvested after 3 days of culture and cytokines quantified using commercially available specific enzyme-linked immunosorbent assays: (IL-2; R & D Systems, Minneapolis, MN; IL-10, IL-12, and TNF-α, Bender MedSystems, Vienna, Austria; and IFN-γ and IL-5, Endogen Corporation, Woburn, MA). The effect of PDE inhibitors on cell viability was assessed by propidium iodide staining. Briefly, cells were lysed with Triton X-100 (1.5%) and propidium iodide (5 μg/ml) and incubated for 20 min at room temperature in the dark and immediately analyzed in a cytoflourometer.

Transcription Assays.

Transcriptional activity was measured using reporter gene assay in transiently transfected normal resting T cells. The plasmid TNF-α-luc contains a region 850-base pairs upstream from the transcriptional initiation site of human TNF-α promoter. The NFAT-luc, containing three tandem copies of the NFAT binding site of the IL-2 promoter, and IL-2-luc, containing the −326 to +45 region of the human IL-2 promoter plasmids, were a generous gift from Dr. G. Crabtree (Durand et al., 1988). The AP-1-Luc plasmid includes the 73/+63-base pair region of the human collagenase promoter fused to the luciferase gene (Deng and Karin, 1993). The pNF-κB-luc contains three tandem copies of the NF-κB site of the conalbumin promoter driving the luciferase reporter gene (Navarro et al., 1998). The CRE-luc plasmid contains four copies of the CRE site of the human choriogonadotropin α gene promoter (−147 to −129) (Schwaninger et al., 1993). CMV-luc contains the luciferase gene under control of the CMV promoter.

For transfection assays, resting purified T cells were resuspended in RPMI supplemented with 10% FCS and electroporated at 320 V, 1500 μF by using a Bio-Rad Gene Pulser II (Bio-Rad, Hercules, CA) with 1 μg/10⁶ cells of purified plasmid (Navarro et al., 1998). After transfection the cells were cultured at 37°C for 14 h before being activated with phorbol-12-myristate-13-acetate (PMA) (10 ng/ml) plus ionomycin (Io) in the presence of RP, PTX (Sigma Chemical, St. Louis, MO), or dbcAMP (1 μM). Cells were incubated for an additional 12-h period, harvested, and lysed. The efficiency of transfection determined by cotransfection with a CMV-β-galactosidase expression plasmid varied between 5 and 10% of the cells. Luciferase activity was measured in a luminometer and expressed as relative luciferase units (RLU), calculated as light emission from experimental sample-light emission from untransfected cells/10⁶ cells. Data are represented as fold induction (observed experimental RLU/basal RLU in absence of any stimulus).

Electrophoretic Mobility Shift Assays (EMSAs).

Nuclear extracts were obtained from activated T cells in the different conditions essentially by the previously described method (Pimentel-Muiños et al., 1994). The binding assays were performed as reported using as labeled probes: the double-stranded κB element of IL-2Rα promoter (5′ GCAGGGGAATCTCCCTCT 3′), the CRE consensus element (5′ AGAGATTGCCTGACGTCAGAGACCTAG 3′), the distal NFAT site from the IL-2 promoter (5′ GGAGGAAAAACTGTTTCATACAGAAGGCGT 3′), or an AP-1 consensus site (5′ CGCTTGATGAGTCAGCGGAA 3′). The binding complexes were separated in a 5% acrylamide gel and their specificity was determined by competition with 50× molar excess of the same unlabeled oligonucleotide (Pimentel-Muiños et al., 1994). Supershifting assay with anti-p65 NF-κB antibodies (a generous gift from Dr. Nancy Rice, Frederick Cancer Research and Development Center, National Cancer Institute, Frederick, MD) was carried out as described (Pimentel-Muiños et al., 1994).

Results

Effect of PDE4 Inhibition on T-Cell Activation.

To study the contribution of PDE4 to immune function, we have tested the effect of its blockade by RP in the activation of purified T cells. For this, purified T cells, depleted of the majority of monocytes, were activated through the T-cell receptor with immobilized anti-CD3 in the presence or absence of RP. In these cultures, no spontaneous secretion of cytokines was found. Upon activation, cell proliferation as well as IL-2, IL-5, IL-10, IFN-γ, and TNF-α secretion was induced (Fig.1A). The addition of RP inhibited all of these T-cell activities, although their sensitivity to inhibition was clearly different. Thus, RP inhibited with similar potency IL-10, IL-5, and TNF-α secretion by activated T cells (IC₅₀of ≈0.5–2 μM). IL-2 synthesis was somewhat less sensitive (IC₅₀ of ≈7 μM). In contrast, RP poorly affected IFN-γ secretion and T-cell proliferation (IC₅₀ of ≈100–200 μM). These effects were not due to nonspecific toxicity, because no significant decrease in viable cell number (tested by trypan blue exclusion) up to 1 mM was observed (data not shown). Besides, RP neither induced apoptosis in normal unstimulated T cells nor potentiated the small one induced by anti-CD3 stimulation (Fig. 1B). Similar effects, although requiring higher doses of the drug, were observed with a nonspecific PDE inhibitor PTX (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Effect of RP and dBcAMP on T-cell activation. A, T cells were stimulated with anti-CD3 antibody in presence of RP and dBcAMP at the indicated concentration, and cytokine production and proliferation were evaluated as indicated under Materials and Methods. Unstimulated T cells did not produce detectable amounts of cytokines or proliferated (data not shown). B, apoptosis in T-cell cultures. T cells unstimulated or stimulated with anti-CD3 as described above were checked for apoptosis by propidium iodide staining. Results shown are the mean ± S.D. from three experiments with different donors, each one carried out in duplicate.

So far, all the actions of RP have been attributed to increases in intracellular cAMP due to its ability to block its degradation by PDE4 (Teixeira et al., 1997). To test whether cAMP elevation was responsible for the above-mentioned effects, we studied the effect of a permeable analog, dBcAMP, in our system. dBcAMP added to the cultures strongly inhibited TNF-α secretion (IC₅₀ of ≈10 μM) and with lesser potency T-cell proliferation, and IL-2 and IFNγ secretion by activated T cells (IC₅₀ of ≈100 μM). In contrast, dBcAMP had a weak inhibitory effect on IL-10 and IL-5 secretion by activated T-cell cultures (Fig. 1).

Involvement of Protein Kinase A in RP Activities.

To corroborate the involvement of the cAMP-PKA pathway in RP activities, we tested the ability of KT 5720, a selective PKA inhibitor, to overcome the inhibitory effect of RP. Although KT 5720 had some effect by itself (enhancing TNF-α and IFN-γ but decreasing IL-10 secretion), this effect was not statistically significant. However, it prevented in a large percentage RP inhibition of TNF-α production and the small inhibitory effects on cell proliferation and IFN-γ production by activated T cells. However, it did not alter RP inhibition of IL-10 secretion in the same cell cultures (Fig.2). Another PKA inhibitor, H-8, had similar effects (data not shown). As expected, KT 5720 reverted dBcAMP inhibition of TNF-α production.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Effect of PKA inhibitors on RP activities. RP (100 μM), KT 5720 (200 nM) alone or in combination were added to the cultures of T cells stimulated with anti-CD3 and TNF-α, IL-10, and IFN-γ secretion were evaluated. Results shown (mean ± S.D. of two experiments) are standardized as percentage of stimulated control in the absence of drugs. Control values were TNF-α (350 ± 53 pg/ml), IFN-γ (2358 ± 220 pg/ml), and IL-10 (170 ± 35 pg/ml) in T cells. ■, RP; ▧, KT5720; ▪, KT5720 + RP.

Effect of PDE4 Inhibition on Cytokine Transcription of T Cells.

To determine whether the effect of RP on IL-2 and TNF-α was at the transcriptional level, we transfected reporter genes controlled by the IL-2 and TNF-α promoter regions in normal resting T cells. Stimulation with PMA plus Io, a treatment that mimics T-cell activation through the T-cell receptor, enhanced several times the activity of both TNF-α and IL-2 promoter reporter plasmids (Fig.3). Interestingly, RP and to a lesser extent PTX inhibited the transcriptional activity of both promoters in a dose-dependent manner. dBcAMP partially affected their transcription at 500 μM. Interestingly, KT 5720 did not prevent RP inhibition on IL-2 promoter and only partially reversed inhibition of TNF-α promoter. This reversion was more evident at low doses of RP. In contrast, KT 5720 completely prevented inhibition caused by dBcAMP.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Effect of RP on the transcriptional activity of TNF-α and IL-2 promoters. Human resting T cells were transfected with: pTNF-luc or pIL-2-luc and stimulated or not with PMA (10 ng/ml) plus Io (1 μM) in presence or absence of RP, dBcAMP, or PTX at the indicated millimolar concentration and/or KT 5720. Results shown are the mean ± S.D. of two experiments by using T cells from two different donors and normalized to fold induction values (observed experimental RLU/basal RLU in absence of any stimulus). ■, untreated; ▪, + KT5720.

The transcription of TNF-α and IL-2 is dependent on several nuclear factors induced by T-cell activation, such as NFAT, AP-1, and NF-κB (Fraser et al., 1993; Crabtree and Clipstone, 1994). So, we tested the effect of RP on primary T cells, transiently transfected with reporter genes under control of NF-κB, NFAT, and AP-1 elements. Again, a good activation of these reporter genes can be detected in transfected normal resting T cells upon activation with PMA plus Io (Fig.4A). RP (100 μM) was able to inhibit by 60 to 80%, depending of the experiment, the induction of NFAT activity. This inhibition of NFAT by RP was not prevented by KT 5720 (data not shown). It also inhibited the activation of NF-κB. In contrast, AP-1 activity was enhanced by RP over the low levels already induced by PMA plus Io. The activity of a reporter gene under the control of a CRE was enhanced by RP (Fig. 4A), indicating that RP was increasing intracellular cAMP levels. Interestingly, PTX had similar effects to RP on the activation of these transcription factors. In contrast to RP, dBcAMP minimally affected the induction of NFAT-dependent promoter activity, although it similarly inhibited the activation of the NF-κB-driven reporter and enhanced the AP-1 and CRE reporter genes. All the above-mentioned effects on promoter transcription observed with PTX and RP were specific, because these drugs as well as dBcAMP did not affect the transcription of a luciferase gene under control of CMV promoter (Fig. 4B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Effect of PDE4 inhibition on nuclear factor driven-transcription. T cells were transfected with reporter plasmids NF-κB-luc, NFAT-luc, AP-1-luc, or CRE-luc (A) or CMV-luc (B). Cells were stimulated or not with PMA (10 ng/ml) plus Io (1 μM) in presence or absence of RP (0.1 or 0.5 mM), dBcAMP (0.5 mM), or PTX (0.5 mM) as indicated. Results shown are the mean ± S.D. of three experiments by using T cells from three different donor and normalized to fold induction values.

To corroborate the observed inhibition of NF-κB and NFAT activity by RP, we analyzed the presence of active transcription factors in the nucleus of activated T cells in the presence of RP or dBcAMP by EMSA. As expected, resting T cells have a low amount of specific complex(es) in the nucleus able to bind to specific oligonucleotide probe and anti-CD3 activation induced the appearance of active NF-κB. Careful examination of the gels indicates the presence of two complexes. The upper one was supershifted by anti-p65 NF-κB antibodies, and the lower band probably represents p450 homodimers as it has been described (Baeuerle and Henkel, 1994; Pimentel-Muiños et al., 1994). Addition of RP into the culture strongly inhibited (around 80%) NF-κB-active complexes. Interestingly, this inhibition was complete in the upper band, which is the transcriptionally active one. dBcAMP produced a partial inhibition of NF-κB binding (around 40% in agreement with the reporter data) (Fig.5). Activation by immobilized anti-CD3 also induced the appearance of a NFAT complex in the nucleus of T cells that can be outcompeted by the specific oligonucleotide. Its induced expression was completely inhibited by RP 100 μM (Fig. 5). In contrast, dBcAMP did not inhibit NFAT activation. The inhibition observed by RP on NFAT and NF-κB was observed at any time after anti-CD3 activation (data not shown). As expected, both RP and dBcAMP increased the nuclear proteins bound to both AP-1 and to a CRE DNA probe over the levels induced by anti-CD3 stimulation (Fig. 5, C and D). Similar results on transcription factors were found when T cells were stimulated with PMA plus Io (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Inhibition of NF-κB and NFAT activation by RP. T cells were unstimulated (uns) or stimulated with immobilized anti-CD3 antibody in the presence of RP (0.1 mM), dBcAMP (0.3 mM) as indicated. The binding activity of NFAT (A), NF-κB (B), AP-1 (C), and CREB (D) in the nucleus of T cells was assayed 14 h later, by using a κB-IL-2Rα, CRE, AP-1, or distal IL-2 NFAT site-labeled probes as described. Control specific binding was detected by using as competitor 50-fold excess of unlabeled corresponding oligonucleotides. In B, extracts from anti-CD3-stimulated control T cells were incubated with anti-p65 NF-κB antibody to identify the specific complexes, where indicated.

Discussion

PDE4-specific inhibitors are considered promising anti-inflammatory drugs for many diseases. However, their pharmacological actions have been restricted by their side effects. Therefore, there is challenge to identify the molecular mechanism by which PDE4 inhibitors exert their anti-inflammatory activities. Here, we have used primary T-cell cultures and efficient systems of transfection of normal resting peripheral blood T cells. These systems provide a sensitive and physiologically relevant model for study of the molecular mechanism resulting from PDE4 inhibition and to clarify its role in the regulation of cytokine secretion.

We have found that the specific inhibition of PDE4 by RP reduces the production of several cytokines such as IL-5, IL-10 (type 2 cytokines), TNF-α, and IL-2 but poorly affects IFN-γ (a type 1 cytokine) and T-cell proliferation in response to activation by anti-CD3. Similar poor sensitivity of T-cell proliferation (Essayan et al., 1994) and IFN-γ secretion (Sommer et al., 1995) to RP has been reported previously. Interestingly, our results show that PDE inhibitors exert an inhibitory effect on transcriptional activity of TNF-α and IL-2 promoters. Furthermore, our results indicate that RP suppresses not only NF-κB but also NFAT activation. Because both transcription factors are required for cytokine synthesis (Fraser et al., 1993;Crabtree and Clipstone, 1994), our results strongly suggest that the effect of PDE inhibitors on cytokine transcription may be attributed to their ability to inhibit NF-κB and NFAT activation.

PDE4 inhibition by RP increased intracellular cAMP in many systems (Teixeira et al., 1997) as well as in ours (J. L. Jiménez, R. A. Muñoz-Fernández, and M. Fresno, data not shown). Previous reports have assigned all activities of PDE4 inhibitors to elevations on cAMP. Thus, at the molecular level, the most obvious mechanism leading to the effects caused by RP may involve cAMP-dependent pathways. However, our results indicate that exogenous addition of permeable cAMP analogs cannot completely mimic RP activities. Thus, secretion of type 2 cytokines, IL-10 and IL-5, by activated T cells was poorly inhibited by dBcAMP, compared with TNF-α and IL-2, in agreement with other reports (Benbernou et al., 1997). In addition, dBcAMP showed no inhibition of NFAT activation in primary T cells, in contrast to RP. Besides we have found that other PDE inhibitors, although not specific of PDE4, such as PTX, behaved similarly to RP and not like cAMP. Because T lymphocytes contain PDE3 and PDE4 but PDE3 inhibition has no effect on cell function (Giembycz et al., 1996; Essayan et al., 1997) it is likely that the effects of PTX in T cells can be mostly attributed to PDE4 inhibition. Inhibition by RP of TNF-α but not of IL-10 production by activated T cells can be reverted (at least partially) by the PKA inhibitor KT 5720. Furthermore, the inhibition of TNF-α but not IL-2 promoter activity (or NFAT activity) observed in the presence of RP can be reverted by KT 5720. Taken together, our results clearly indicate that IL-10 inhibition by PDE inhibitors, such as RP, cannot be accounted for their ability to increase cAMP and are indicative that PDE inhibition may affect some activities independent of a cAMP-PKA pathway.

At the molecular level elevated cAMP has been shown to inhibit NF-κB activation in transformed T cells, measured both by EMSA and transient transfection of reporter genes (Chen and Rothenberg, 1994; Haraguchi et al., 1995). In contrast, dBcAMP did not inhibit NFAT activation (Chen and Rothenberg, 1994; Haraguchi et al., 1995), except when used at very high concentration (Tsuruta et al., 1995). On the contrary, cAMP stimulated AP-1 (Chen and Rothenberg, 1994; Haraguchi et al., 1995) as well as CREB (Haraguchi et al., 1995). Our results with dBcAMP in primary T cells are in agreement with those. However, we have found here that, in contrast to cAMP, PDE4 blockade with RP or PTX inhibits NFAT. RP and PTX also inhibit NF-κB and stimulate AP-1 and CRE-binding factors as cAMP does.

Evidence indicates that type 2 cytokines are controlled by a subset of transcription factors different from those that control proinflammatory TNF-α and IL-2 transcription. Thus, NF-κB does not seem to be required for IL-10 (Platzer et al., 1994) and IL-5 (Lee et al., 1995;Stranick et al., 1997) transcription, whereas it is required for IL-2 and TNF-α transcription (Baeuerle and Henkel, 1994). In contrast, NFAT is clearly required for IL-5 (Lee et al., 1995; Rao et al., 1997), IL-2, IL-10, and TNF-α transcription (Rao et al., 1997). Type 2 cytokines such as IL-10 and IL-5 (Platzer et al., 1994; Lee et al., 1995), but not IL-2 (Crabtree and Clipstone, 1994), also have CRE elements in their promoters. TNF-α promoter contains several AP-1/CRE-like binding sites that may bind to those factors and are stimulated by PTX treatment (Newell et al., 1994). For these reasons, cAMP will be effective against NF-κB-dependent cytokines (TNF-α and IL-2), whereas PDE4 inhibitors will also affect cytokines that require NFAT, as do type 2 cytokines. cAMP has been shown to induce IL-10 and IL-5 secretion in several cell types probably through binding to the CRE sites of their promoters. It is likely that the weak inhibition of type 2 cytokines secretion by dBcAMP may be secondary to the inhibition on cell proliferation, because all those parameters have similar sensitivity to cAMP and cytokine production in T cells is dependent on proliferation. Therefore, the unique NFAT inhibition by RP may explain why this drug and not dBcAMP inhibits IL-5 and IL-10 production. The smaller inhibitory effect of cAMP compared with RP on TNF-α transcription may be due to its exclusive inhibitory effect on NF-κB and not on NF-κB and NFAT as in the case of RP or partially compensated by enhanced AP-1/CRE binding (Newell et al., 1994). On the other hand, IL-2-dependent transcription in normal human T cells is more dependent on NFAT/AP-1 than on NF-κB (Tsuruta et al., 1995). This may explain why IL-2 transcription is poorly inhibited by cAMP and why PKA inhibitors did not prevent RP inhibition. Thus, the relative contribution of the different nuclear factors to transcription of the various cytokines and the sensitivity of those factors to inhibition by the different drugs will determine the outcome.

The strong inhibitory effect of RP on type 2 cytokines and not of IFN-γ is somewhat surprising, because RP has been proven effective to treat inflammatory diseases, such as multiple sclerosis, which are thought to be Th1-mediated and in which type 2 cytokines play a beneficial role (Muñoz-Fernández and Fresno, 1998). However, although RP ameliorates the disease, it had no effect on IFN-γ production by autoimmune Th1 cells ex vivo (Sommer et al., 1995), in agreement with our results. Neither NFAT nor NF-κB are required for IFN-γ transcription, whereas AP-1 is stimulatory (Barbulescu et al., 1997). This would explain why IFN-γ secretion is rather insensitive to RP.

It is generally thought that pharmacological agents that increase cAMP, such as cholera toxin, PGE2, and forskolin inhibited type 1 cytokine production (IL-2 and IFN-γ) but had no effect on type 2 cytokine (IL-4, IL-5, and IL-10) production (Muñoz et al., 1990; Katamura et al., 1995). However, in agreement with our results, an increasing number of recent reports indicate that PDE inhibitors, such as PTX and RP, also inhibit type 2 cytokine (IL-4, IL-5, and IL-10) secretion by T cells (Chan et al., 1993; Essayan et al., 1995; Crocker et al., 1996; Foissier et al., 1996). These apparent discrepancies could be now explained by our results on the inhibition of NFAT and NF-κB by PDE4 inhibitors and only NF-κB by cAMP.

Taken together, our results may provide a molecular explanation for the described apparently discrepant results of PDE inhibitors on the synthesis of several cytokines and the lack of correlation with cAMP. It is likely that the effect of PDE4 inhibitors in the in vitro transcription of a particular cytokine may depend on the relative contribution of AP-1 and CREB (enhanced by RP) versus NF-κB and NFAT (inhibited by RP) to the transcriptional activation of their respective promoters. Some activities resulting from PDE4 inhibition (i.e., TNF-α and NF-κB suppression) are likely due to stimulation of the cAMP-PKA pathway, whereas others are due to its ability to block other cellular functions such as NFAT activation, which does not seem to involve this pathway. Our experiments with KT 5720 indicate, indirectly in the case of the IL-2 promoter or directly with NFAT-reporter genes, that these effects are not mediated by PKA. Experiments are in progress to elucidate the molecular link between PDE4 inhibition and decreased NFAT activity. Reported cellular changes induced by the PDE inhibitor PTX include not only cAMP elevation but also alterations in Ca²⁺ intracellular content (Yang et al., 1995). This latter effect may contribute to explain its effect on NFAT, because ts activation is strongly dependent on Ca²⁺ (Rao et al., 1997) Besides, our results confirm the role of PDE4 in T-cell activation (Jiang et al., 1998) and indicate that PDE4 might be an additional therapeutic target for treatment of immune dysfunctions.