Nuclear factor-κB does not mediate the inhibitory effects of dexamethasone on granulocyte–macrophage colony-stimulating factor expression

Abstract

Human granulocyte–macrophage colony-stimulating factor (GM-CSF) reporter constructs containing up to 3·3 kb of upstream promoter sequence were transiently transfected into both Jurkat and HUT78 human T-cell lines. In Jurkat cells, stimulation with phorbol 12-myristate 13-acetate (PMA) plus phytohemaglutinin (PHA) produced robust increases in reporter activity, whereas HUT78 cells showed low levels of reporter induction attributable to constitutive nuclear factor (NF)-κB activity. Following mutation of either the proximal NF-κB site (−85/−76) or the activator protein1 (AP-1) motif within the conserved lymphokine element 0 (CLE0) site (−54/−31), reporter activity was markedly reduced in both cell lines. Despite this dependence on NF-κB and CLE0/AP-1, GM-CSF reporter activity was unaffected by dexamethasone in either cell line. Similarly, an NF-κB-dependent reporter was also not repressed by dexamethasone, yet GM-CSF release from HUT78 T cells was inhibited. These data therefore confirm a critical role for both NF-κB and CLE0 sites in GM-CSF promoter activation and indicate that NF-κB may not mediate glucocorticoid-dependent repression of GM-CSF in these cells.

Introduction

Granulocyte–macrophage-colony stimulating factor (GM-CSF) can be released from many of the cell types associated with inflammatory diseases such as asthma.¹ In particular, the activation of T cells following stimulation with phorbol 12-myristate 13-acetate (PMA), Ca²⁺ ionophore, phytohemaglutinin (PHA) or concanavalin A (ConA) results experimentally in the production of GM-CSF via mechanisms that involve both transcriptional and post-transcriptional regulation.^2,3 Thus, activation of the proximal GM-CSF promoter (−620 to +34) following PMA plus Ca²⁺ ionophore stimulation of Jurkat T cells was prevented by mutations in the conserved lymphokine element 0 (CLE0) (−54 to −31) or the nuclear factor (NF)-κB binding regions (−85 to −76) and binding of the activator protein1 (AP-1) to the regions in the CLE0 site was shown.^4,5 Similarly, PMA plus either Ca²⁺ ionophore or ConA treatment resulted in inducible occupation of the proximal NF-κB site by a complex containing p50/NFκB1 and p65/RelA,^6,7 whilst a distal enhancer region (∼ 3 kb upstream) was shown to contain functional NF-AT and AP-1 sites and was implicated in promoter activation and tissue specificity.^8–10

Glucocorticoids, such as dexamethasone, have potent immunosuppressive activity and are the drugs of choice for treatment of all but the most mild of asthmatics.¹ This class of drug has been previously shown to inhibit GM-CSF release from bronchial epithelial cells and peripheral blood mononuclear cells.^11,12 Furthermore the anti-inflammatory effects of glucocorticoids have been shown to act via several mechanisms at both the transcriptional and the post-transcriptional levels. Transcriptional mechanisms include the blockade of transcriptional activation via interaction with factors such as NF-κB and AP-1, whereas post-transcriptional mechanisms include both decreasing mRNA stability and preventing mRNA translation to protein (see reference¹³). In this paper we examine the mechanisms by which dexamethasone inhibits GM-CSF expression in the T-cell lines Jurkat and HUT78.

Materials and methods

Reagents and cell culture conditions

PMA, PHA and dexamethasone were obtained from Sigma (Poole, UK). HUT78 and Jurkat T cells (ECACC) were cultured at 5–10 × 10⁵ cells/ml in RPMI-1640 media (Sigma), supplemented with 10% (v/v) fetal calf serum (FCS), 2 mm l-glutamine, 100 U/ml penicillin, 100 µg/ml streptomycin and 2·5 mg/ml amphotericin B (Sigma) in a humidified atmosphere of 95% air, 5% CO₂ at 37°. Cells were washed and cultured in fresh medium at 3 × 10⁶ cells/ml for 24 hr prior to treatment.

Electroporation of T-cell lines

Electroporation of Jurkat and HUT78 T cells was as described previously,¹⁴ except that a volume of 500 µl and a cell concentration of 1 × 10⁷ cells/ml were used and resulted in a time constant (τ) of 22 ± 2·3 ms. After washing, cells were resuspended in 2 ml of serum-free medium and experiments were carried out with aliquots of 500 µl.

Enzyme-linked immunosorbent assay (ELISA)

GM-CSF in supernatants was determined by ELISA (Pharmingen, Cambridge, UK) as previously described.¹⁵

RNA extraction and reverse transcriptase–polymerase chain reaction (RT-PCR)

RNA was extracted and semiquantitative reverse transcription-polymerase chain reaction (RT-PCR) for GM-CSF and GAPDH performed using primers and conditions as previously described.¹⁵ For each experiment, the exponential phase of the PCR, where starting material is proportional to product formation, was determined empirically by creating ‘cycle profiles’ as previously described.^15,16 Appropriate cycle numbers ranged between 38 and 42 for GM-CSF and between 24 and 28 for GAPDH. Reaction products were analysed by agarose gel electrophoresis and gel images subject to densitometric analysis using GelWorks 1D Version 4·01 (UVP Ltd, Cambridge, UK). Southern blotting was used to confirm identity of products. In all cases a short cDNA dilution series was analysed in parallel with the experimental samples to provide confirmation of linearity.

Transcriptional reporters

The human GM-CSF promoter fragments GM-FL (−3298/+35) and GM-P (−624/+35) (GenBank accession number AJ22414) in pGL3basic (Promega, Southampton, UK) were previously described (see Fig. 1).¹⁵ Mutations in the proximal NF-κB (−85/−76) and CLE0 (−54/−31) sites as shown in Fig. 1 were introduced using the Quick-Change kit (Stratagene, Cambridge, UK) and were confirmed by sequencing. The NF-κB-dependent reporter 6κB.BG contains six copies of the consensus NF-κB binding site (5′-GGG ACT TTC C-3′) upstream of a minimal β-globin gene promoter as previously described.¹⁷ The reporter 6κB.BG.mut is identical except that the NF-κB sites are mutated to the non-DNA binding site (5′-GCC ACT TTC C-3′) (mutated bases underlined).¹⁷

Figure 1.

Schematic representation of the GM-CSF promoter. The promoter fragments cloned into pGL3basic (basic) are depicted. GM-FL represents the full-length GM-CSF promoter (−3286/+35). GM-P represents the proximal promoter region (−624/+35). The sequence of the promoter region containing the NF-κB (−85/−76) and CLE0 (−54/−31) sites is shown (below). The positions and sequence of the mutated sequences (sense strand only) are shown.

Statistical analysis

Statistical analysis was performed by anova with a Bonferroni post-test (with levels of significance P < 0·001, P < 0·01, and P < 0·05).

Results

GM-CSF promoter activation in Jurkat T cells

Jurkat T cells are widely used a model for analysis of T-cell activation and, in particular, have previously proved useful in transcriptional studies examining the function of promoters such as GM-CSF.^3–5,7,8 In the present study, we therefore used transfection analysis of the −624/+35 (GM-P) and the −3286/+35 (GM-FL) GM-CSF promoter constructs into Jurkat T cells to demonstrate inducibility in response to PMA + PHA treatment (Figs 2a and 2b). In both cases, the double mutant with both the proximal NF-κB and CLE0 sites mutated showed only low levels of inducibility. Analysis of the single mutants in the −624/+35 construct indicated a greater role for the −85/−76 NF-κB site than for the −54/−31 CLE0 site (Fig. 2a). Dexamethasone showed no effect on promoter activation of either the −624/+35 or the −3286/+35 constructs (Fig. 2c). Similarly, the NF-κB-dependent reporter 6 κBBG was highly responsive to PMA + PHA and was unaffected by concurrent dexamethasone treatment (Fig. 2d). Despite induction of mRNA as detected by RT-PCR (data not shown) or Northern blotting,² GM-CSF protein expression was not detected from two separate isolates of Jurkat T cells and consequently this cell line was only used for promoter analysis.

Figure 2.

Promoter activation in Jurkat T cells. (a, b) Jurkat T cells were prepared for transfection with wild-type or mutated (mut) GM-CSF promoter constructs, as indicated, and then either not stimulated or stimulated with PMA (50 nm) plus PHA (5 µg/ml). After 12 hr, cells were harvested for luciferase assay. Data (n = 3–5, except the FL and double mutant where n = 2) were expressed as the fold induction of PMA + PHA compared to unstimulated cells for each reporter construct. Values are plotted as means ± SEM. (c, d) Cells were prepared and transfected with various plasmids prior to incubation in the presence of PMA (50 nm) and PHA (5 µg/ml) and dexamethasone (1 µm), as indicated. After luciferase determination, data (n = 4 in each case) are expressed as fold induction for each reporter compared to unstimulated cells and are plotted as means ± SEM.

GM-CSF expression in HUT78 cells

Analysis of steady-state mRNA levels in HUT78 T cells treated with PMA + PHA showed increases in GM-CSF mRNA over 24 hr (Figs 3a and 3b). This response was reduced at each time point by approximately 50% by co-treatment with dexamethasone. Release of GM-CSF protein into the supernatant was similarly induced by PMA + PHA treatment and markedly repressed by dexamethasone, suggesting that additional, possibly translational mechanisms of dexamethasone action may exist (Fig. 3c).

Figure 3.

GM-CSF expression in HUT78 T cells. (a, b) HUT-78 cells were prepared and treated with PMA (50 nm) plus PHA (5 µg/ml) in the absence (solid bars) or presence (open bars) of dexamethasone (1 µm). Cells were harvested at the times indicated for RNA and RT-PCR analysis of GM-CSF and GAPDH. (a) Representative ethidium-stained agarose gels are shown. (b) After densitometric analysis, data (n = 4) as a ratio of GM-CSF/GAPDH are expressed as a percentage of values for stimulated cells at 24 hr and are plotted as means ± SEM. (c) Cells were treated as in (a), and supernatants were harvested at 24 hr for GM-CSF determination. Data (n = 4–8) are plotted as means ± SEM. Statistical analysis was performed by anova with a Bonferroni post-test (***P < 0·001).

Promoter activation in HUT78 cells

Previous studies have shown constitutive high-level NF-κB DNA binding in HUT78 cells.¹⁸ However, since the effect on transcriptional activation was not tested, we examined the regulation of NF-κB-dependent transcription in these cells. Transfection of the NF-κB-dependent reporter 6 κBBG, but not the promoterless vector pGL3basic, into HUT78 cells resulted in a high level of luciferase activity, which was unaffected by PMA + PHA stimulation or by dexamethasone treatment (Fig. 4a). Analysis of the reporter 6 κBBG.mut, in which the NF-κB sites are mutated, showed a low level of luciferase activity. These data are consistent with constitutive activation of NF-κB in these cells.

Figure 4.

GM-CSF promoter activation in HUT78 T cells. (a–c) Cells were transfected with various reporter plasmids as indicated and then treated with PMA (50 nm) plus PHA (5 µg/ml) in the absence or presence of dexamethasone (1 µm) as indicated. Cells were harvested after 12 hr and luciferase activity determined. Data (n = 3) are plotted as means ± SEM.

To test the role of promoter activation in the induction of GM-CSF, HUT78 cells were transfected with the wild-type −624/+35 promoter as well as with constructs with mutated NF-κB or CLE0 sites (Fig. 4b). Consistent with the role for NF-κB suggested by the above studies and the high level of activation in HUT78 cells, the proximal GM-CSF promoter was not substantially induced by PMA + PHA stimulation. Furthermore, this level of luciferase activity was markedly reduced by mutation of either the NF-κB or CLE0 site, again confirming the importance of these sites in promoter activation. As with the 6 κBBG construct, the proximal −624/+35 G-CSF construct was unaffected by dexamethasone in conjunction with the PMA + PHA (Fig. 4c). Taken together, these data confirm that the repression of GM-CSF mRNA in HUT78 by dexamethasone does not occur via interaction with NF-κB, but rather at some downstream post-transcriptional process.

Discussion

Regulation of GM-CSF gene expression appears to be a dynamic process, involving transcriptional and post-transcriptional mechanisms. Consistent with an important role for transcriptional activation, we found that up to ∼3·3 kb of the human GM-CSF promoter was inducible in response to PMA, but not to PHA, in Jurkat cells. These observations are consistent with those of previous studies using run-on analysis and GM-CSF reporter constructs to demonstrate that GM-CSF was regulated, at least in part, by transcriptional mechanisms.^19,20 However, although previous studies have demonstrated the importance of both the distal enhancer and the proximal promoter regions in GM-CSF promoter activation,^4,6,8,9 using the Jurkat cell model, we found similar levels of inducibility for a −3286/+35 construct, which contains the enhancer region, and the proximal −624/+35 construct, suggesting that, at least for PMA + PHA stimulation, the upstream enhancer is not required. Whilst these current data are broadly consistent with the findings of previous studies, the lack of additional enhancer responsiveness could also be attributed to the non-physiological arrangement of the promoter.¹⁰ However, this element contains a number of functional NF-AT as well as AP-1 binding sites and was identified using stimulation by PMA + ionomycin, which gives rise to a strong and prolonged elevation in intracellular Ca²⁺.⁸ By contrast, co-stimulation with PHA produces a more physiological rise in intracellular Ca²⁺ and may consequentially result in lower levels of enhancer activation.²¹ Certainly this possibility is consistent with the view that the upstream enhancer primarily responds to PMA and Ca²⁺ ionophore and could therefore provide an explanation for the seemingly low level of enhancer activation observed here.³ Interestingly, a parallel analysis in primary human (CD3⁺) T cells gave a similar result, with both proximal (−624/+35) and full-length (−3286/+35) promoter constructs giving rise to similar levels of inducibility following PMA + PHA stimulation.²² We therefore believe the proximal promoter region to contain the dominant cis-regulatory regions controlling GM-CSF expression in T cells.

Activation of the proximal GM-CSF promoter is thought to involve a NF-κB-like site at −85/−76 and a CLE0 element at −54/−31.^6,9,23 Consistent with these previous studies, we have now shown that, in Jurkat cells, promoter activation by PMA + PHA was also substantially reduced by mutations in either the NF-κB or the AP-1 part of the CLE0 site or in both of these sites (Fig. 2). Similarly, HUT78 cells, despite showing constitutive activation of NF-κB and consequently showing little inducible activation of the GM-CSF promoter,¹⁸ also revealed a substantial reduction in GM-CSF promoter activity if either or both of these sites were mutated (Fig. 4). The necessity for both of these transcription factors and the fact that a similar result was obtained in primary human T cells highlight the importance of cross-talk or interaction between these two pathways in the transcriptional regulation of GM-CSF.²²

Because activation of the glucocorticoid receptor (GR) by ligand binding causes translocation of the GR–ligand complex to the nucleus, it is generally believed that glucocorticoids exert their principal anti-inflammatory effects by modulation of gene transcription.¹ Indeed, numerous studies have reported that glucocorticoids inhibit the transcription of pro-inflammatory cytokines, such as interleukin (IL)-6 and IL-8, by GR-mediated transrepression of NF-κB.^13,24 However, whilst, in the present study, dexamethasone could still inhibit GM-CSF release from transfected HUT78 cells, there was no clear effect of dexamethasone on NF-κB-dependent transcription or on the GM-CSF promoter constructs tested. Taken in conjunction with our studies in primary human T cells,²² the current data further confirm the role of both the NF-κB and CLE0 sites and the functionality of the reporter system. This analysis therefore suggests that the ability of dexamethasone to repress promoter activity via these sites may not be important in the dexamethasone-dependent repression of GM-CSF. One possible explanation for a lack of effect of dexamethasone in the studies reported here is a failure of the transfected reporters to adopt the correct chromatin structure. However, this charge can be levelled at all the promoter studies listed above and may even be extended to include stable transfection analysis. Furthermore, this issue fails to explain why stimulated transcription occurs as expected whereas dexamethasone-dependent repression does not. Whilst inappropriate chromatin structure is always an important caveat in any reporter analysis, we believe that our data point to the existence of post-transcriptional and/or translational mechanisms for the repression of GM-CSF expression by glucocorticoids. In addition, the fact that no promoter activation was observed in HUT78 cells, yet GM-CSF mRNA and protein were induced by PMA + PHA, also highlights the existence of post-transcriptional or translational mechanisms and is consistent with a major role for such mechanisms in the regulation of GM-CSF.² As such mechanisms have already been proposed in the regulation of other inflammatory genes by glucocorticoids,¹³ and GM-CSF itself is subject to post-transcriptional regulation,²⁵ we consider these observations to be highly pertinent.

In summary, the data presented here confirm the key role of the NF-κB (−85/−76) and CLE0 (−54/−31) elements in the proximal region of the human GM-CSF promoter in T-cell lines. Taken together with the findings of our parallel study in primary human T cells,²² these data collectively suggest that therapeutic strategies to target these pathways may be of benefit in the context of diseases such as asthma in which T cells play a role. Finally, these analyses indicate that the glucocorticoids do not target the signalling pathways that lead to GM-CSF promoter activation. Importantly, this suggests that other mechanisms, possibly post-transcriptional or translational, must account for the repressive effects of glucocorticoids, such as dexamethasone, on GM-CSF expression.