Results

The Effect of Corticosteroids on GATA-3 Nuclear Translocation and IL-4 mRNA
Corticosteroids are effective in inhibiting GATA-3-regulated IL-4 gene expression in vitro and in vivo [32]. We therefore investigated whether corticosteroids affect anti-CD3/CD28–stimulated nuclear import of GATA-3. Stimulation of cells with anti-CD3/CD28 resulted in a rapid cytoplasmic/nuclear GATA-3 translocation (Figure 1A), confirming our previous results [12]. We also confirmed a clear separation of nuclear and cytosolic fractions as indicated by histone H1 and MEK-1 markers (Figure 1B). The potent topical corticosteroid FP caused sustained loss of nuclear GATA-3 expression and cytoplasmic retention of GATA-3 at concentrations ranging from 10−12 to 10−8 M, which cover the therapeutic range [37]. This effect was concentration- and time-dependent, with a peak effect of 11.6-fold at 30 min at a concentration of 10−8 M (Figure 1C) and was associated with marked reductions in anti-CD3/CD28–stimulated IL-4 and IL-5 mRNA expression (Figure 1D) and a loss of GATA-3 binding to the native IL-5 promoter (Figure 1E).
10.1371/journal.pmed.1000076.g001 Figure 1  Fluticasone propionate down-regulates Th2 cytokine gene expression and inhibits GATA-3 nuclear import.
(A) Anti-CD3/CD28 treatment of HuT-78 cells results in translocation of GATA-3 from the cytoplasm to the nucleus within 30 min. (B) Histone H1 and MEK-1 were used to confirm distinct separation of cytoplasmic and nuclear extracts in three separate experiments. (C) Western blot analysis of FP-treated HuT-78 cells demonstrated impaired nuclear localization of GATA-3 induced by anti-CD3/CD28 co-stimulation in a time- (at 10−8 M FP) and concentration- (at 60 min after stimulation) dependent manner. Cells were pretreated with FP for 30 min prior to stimulation. MEK1 and histone H1 were used to demonstrate equal cytoplasmic and nuclear loading respectively. Results are presented graphically below as mean±SEM of at least three independent experiments. *** p<0.001 compared to t = 0. (D) RT-PCR showing that FP inhibits IL-4 and IL-5 mRNA expression in CD3/CD28-costimulated cells. GAPDH was used as a loading control. Lower panels show graphical analysis of results presented as mean±SEM of at least three independent experiments. ### p<0.001 compared to control, ***p<0.001 compared to anti-CD3/CD28–stimulated. (E) FP (10 nM) reduces the ability of anti-CD3/CD28-stimulated GATA-3 to associate with the native IL-5 promoter 60 min after stimulation. Data are also shown graphically as mean±SEM of three independent experiments. All data were analysed by ANOVA followed by Newman-Keuls post-test.

Ligand-Activated GR Competes with GATA-3 for Importin-α
We confirmed and extended previous data [20] to show that ligand-activated GR as well as GATA-3 uses importin-α for its nuclear import (Figure 2A and 2B). This interaction between GR and importin-α was significant at concentrations as low as 10−12 M and was maximal with 10−8 M FP. Subsequent GR nuclear translocation was rapid and sustained at significant levels for at least 14 h (Figure 2B). Using IP-Western blotting we showed that FP at 10−12–10−8 M decreased the association between GATA-3 and importin-α induced by anti-CD3/CD28 stimulation in a concentration-dependent manner (Figure 2C). In addition, using GFP-labelled GATA-3 and confocal microscopy we demonstrated that GATA-3 nuclear import following anti-CD3/CD28 stimulation for 30 min was attenuated by pretreatment with FP (10−8 M) (Figure 2D).
10.1371/journal.pmed.1000076.g002 Figure 2  Fluticasone propionate reduces GATA-3 association with importin-α and GATA-3 nuclear import.
(A) Western blot analysis demonstrates a time- (at 10−8 M FP) and concentration- (at 60 min after stimulation) dependent induction of FP-activated GR interaction with importin-α (Imp-α). A positive control for GR association with importin is shown. Quantification of the densitometry data is shown below. Each bar represents mean±SEM of at least three independent experiments. *** p<0.001 compared to control, ### p<0.001. (B) Western blot analysis demonstrated a time- (at 10−8 M FP) and concentration- (at 60 min after stimulation) dependent induction of FP-activated GR nuclear translocation measured by IP. Quantification of the densitometry data is shown below. Each bar represents mean±SEM of at least three independent experiments. ***p<0.001 compared to control. (C) Western blot analysis of HuT-78 cells treated with FP and anti-CD3/CD28 co-stimulation demonstrated a concentration-dependent decrease in GATA-3–importin-α association at 20 min. Quantification of the densitometry data is shown below. Each bar represents mean±SEM of at least three independent experiments. ### p<0.001 compared to control, ***p<0.001 compared to αCD3/CD28-stimulated cells. (D) GFP-tagged GATA-3 was overexpressed and cells stimulated (b, c) or not (a) for 30 min with anti-CD3/CD28. The effect of 30 min pretreatment of cells with FP (10−8 M, c) is also shown. All data were analysed by ANOVA followed by Newman-Keuls post-test.

Effect on MKP-1
Dexamethasone inhibits p38 MAPK function in a cell type–specific manner through the rapid induction of the dual kinase phosphatase MKP-1 (MAPK phosphatase-1), and this effect lasts for up to 24 h [28]. FP (10−8 M) treatment of HuT-78 cells activated by anti-CD3/CD28 in vitro significantly decreased p38 MAPK phosphorylation (Figure 3A) and activity measured by phosphorylation of the downstream target ATF-2 (Figure 3B). This effect was detected at 30 min and lasted for at least 14 h (Figure 3B). FP (10−8 M) also significantly reduced GATA-3 serine phosphorylation induced by anti-CD3/CD28 stimulation in both a time- and concentration-dependent manner (Figure 3C). This reduction in GATA-3 phosphorylation was also seen with lower concentrations of FP. We found that FP significantly induced MKP-1 mRNA in both a time- and concentration-dependent manner, reaching a plateau at 10−8 M after 10 min (Figure 3D and 3E). However, the effects of FP on GATA-3 nuclear import, importin-α association and IL-4 mRNA expression are seen at 10,000-fold lower concentrations (10−12 M, see Figure 2).
10.1371/journal.pmed.1000076.g003 Figure 3  Fluticasone propionate–mediated inhibition of p38 MAP kinase phosphorylation and activation is associated with a marked down-regulation of GATA-3 serine phosphorylation.
(A) Western blot analysis shows that FP (10−8 M, 30 min) treatment reduced dual phosphorylation (threonine-180 and tyrosine-182) of p38 MAPK in anti-CD3/CD28–co-stimulated HuT-78 cells. (B) Time course of the effect of FP (10−8 M) on phosphorylation of activated transcription factor 2 (ATF-2), a measure of p38 MAPK activity. (C) FP-induced inhibition of p38 MAPK activity is associated with the decrease of anti-CD3/CD28 co-stimulation–induced serine phosphorylation (P-Ser) of GATA-3. For (A–C), quantification of the densitometry data is also shown. Each bar represents mean±SEM of at least three independent experiments. ### p<0.001 compared to control, ***p<0.001 compared to αCD3/CD28-stimulated cells. (D) FP induced MKP-1 mRNA in a concentration-dependent manner. All results are representative of at least three independent experiments and where appropriate expressed as means±SEM, *p<0.05. (E) FP induces MKP-1 mRNA in a time-dependent manner. Results are representative of two independent experiments. All data except (E) were analysed by ANOVA followed by Newman-Keuls post-test.   Using an in vitro competition assay (Figure 4A) utilizing purified activated GATA-3, importin-α, and activated GR, we demonstrated that activated GR significantly increased GR-importin-α association in the presence and absence of activated GATA-3 (Figure 4B). This effect is not mutual, since activated GATA-3 did not block GR–importin-α association (Figure 4C). These data also suggest that both activated GR and phospho-GATA-3 can directly associate with importin-α (Figure 4D) and that activated GR attenuates the phospho-GATA-3/importin-α interaction in a concentration-dependent manner (Figure 4E). Together, this suggests that ligand-activated GR may compete with phospho-GATA-3 for importin-α and thereby limit GATA-3 nuclear import.
10.1371/journal.pmed.1000076.g004 Figure 4  Fluticasone propionate competes with phospho-GATA-3 for importin-α.
(A) schematic representation of the in vitro binding competition assay. (B) GR isolated from FP (10−8 M) stimulated cells enhances GR–importin-α binding in the presence (•) and absence (▪) of activated GATA-3. * p<0.05 compared to no activated GR. (C) GATA-3 isolated from anti-CD3/CD28–stimulated cells does not attenuate GR–importin-α association. *p<0.05 compared to control. (D) Activated GR blocks the ability of purified phospho-GATA-3 isolated from anti-CD3/CD28–stimulated cells interacting with immobilised importin-α in an in vitro binding assay. *p<0.05 compared to GATA-3 isolated from unstimulated cells. # p<0.05 compared to stimulated GATA-3-importin binding. (E) The effect of activated (•) versus unstimulated (○) GR on attenuation of GATA-3–importin-α association was concentration-dependent. *p<0.05, **p<0.01 between groups. All results are expressed as mean±SEM of three independent experiments and analysed by ANOVA followed by Newman-Keuls post-test.   Other possible interpretations of our results could include an effect of FP on GATA-3 nuclear export and/or degradation. Leptomycin B, which inhibits nuclear export, did not affect the ability of FP to block GATA-3 nuclear localization (Figure 5A). Additionally, FP had no effect on whole cell GATA-3 expression during the time course of these experiments (Figure 5B). Nor did addition of FP subsequent to anti-CD3/CD28 nuclear translocation affect GATA-3 nuclear residency (Figure 5C), suggesting that activated GR does not enhance GATA-3 nuclear export. Finally, the effect of FP on GATA-3 nuclear import was not nonspecific, since FP (10−8 M) had no effect on p65 nuclear translocation measured at 60 min (Figure 5D).
10.1371/journal.pmed.1000076.g005 Figure 5  Fluticasone propionate does not affect GATA-3 nuclear export.
(A) Western blot analysis showing that the nuclear export inhibitor leptomycin B (2 nM) does not affect the ability of FP (10−8 M) to prevent anti-CD3/CD28–stimulated GATA-3 nuclear localization measured at 60 min. ***p<0.001 compared to unstimulated cells, ### p<0.001 compared to anti-CD3/CD28–stimulated cells. (B) Western blot analysis showing that FP (10−8 M) does not affect whole-cell GATA-3 degradation over 17 h. (C) GFP-tagged GATA-3 is overexpressed and cells stimulated (b–j) or not (a) with anti-CD3/CD28. The effect of treating cells with FP (10−8 M, f–j) after 30 min stimulation with anti-CD3/CD28 is also shown. (D) FP (10−8 M) does not prevent anti-CD3/CD28–stimulated p65 nuclear translocation at 60 min after stimulation. **p<0.01 compared to unstimulated cells. All results are representative of at least four independent experiments and are shown as mean±SEM. Results were analysed by ANOVA followed by Newman-Keuls test.

The Inhibitory Effect of Corticosteroids on GATA-3 Nuclear Localization in Primary T Lymphocytes Ex Vivo and In Vivo
Treatment with FP ex vivo demonstrated a concentration-dependent decrease in the direct interaction between phospho-GATA-3 and importin-α in PBMCs from patients with asthma (Figure 6A and 6B), which was significantly inhibited at 10−12 M FP (p<0.001, ANOVA and Newman-Keuls test) and completely attenuated by 10−8 M FP (p<0.001, ANOVA and Newman-Keuls test).
10.1371/journal.pmed.1000076.g006 Figure 6  Fluticasone propionate impairs GATA-3 interaction with importin-α and GATA-3 nuclear localization in vivo and ex vivo.
(A and B) Co-immunoprecipitation analysis of PBMCs from steroid-naïve asthma patients treated with FP in vitro demonstrated impaired interaction between GATA-3 and importin-α measured at 60 min. Each bar represents the mean±SEM of at least three independent experiments; *** p<0.001 compared with control as determined by ANOVA/Newman-Keuls analysis. (C and D) Co-immunoprecipitation analyses of PBMCs from steroid-naïve asthma patients treated with inhaled FP (500 µg via a spacer) in vivo demonstrated decreased association between GATA-3 and importin-α. The individual values for each treatment are presented graphically. (E) Representative Western blot showing that importin-α expression was unaffected by inhalation of FP. Blot is representative of gels from three participants.   Our previous T cell line studies indicated that 10−12 M FP suppresses IL-4 and -5 gene expression and attenuated the interaction of GATA-3 with importin-α (see Figures 1D and 2). This concentration is close to peak plasma levels obtained from asthmatic patients treated with inhaled FP (500 µg) [27]. Inhaled FP (500 µg) treatment of seven steroid-naive asthma patients significantly reduced GATA-3–importin-α interaction in vivo in a time-dependent manner. This produced a >90% decrease in GATA-3–importin-α association at 2 h (median [95% CI], 13,494 [6,828–17,829] versus 879 [597–1,165]; p<0.05 Friedman's analysis). However, this did not reach significance using Wilcoxon's post-test analysis (W = 6.00) probably due to low numbers of participants. Similar results were observed when GATA-3–importin-α association was measured (Figure 6C and 6D). The lower dose of FP (100 µg) was not effective. The attenuated interaction of GATA-3 did not result from the defective recycling of importin-α, as a significant decrease in the abundance of importin-α in the cytoplasmic pool was not detected (Figure 6E).
We further examined whether inhaled FP could affect cellular localization of GATA-3 in peripheral blood T cells. Treatment with inhaled FP (500 µg) for 2 h significantly increased GR nuclear translocation (Figure 7A) and concomitantly decreased the number of nuclear GATA-3 immunoreactive peripheral blood T cells (37%±4.2% versus 58.2%±4.95%, p = 0.016, W = 28.0, Wilcoxon's rank test) compared with placebo as measured by immunocytochemistry (Figure 7A and 7B). This was confirmed by Western blotting, which also indicated that this effect was both time- and dose-dependent (Figure 7C and 7D). Thus, inhaled FP (500 µg) induced significant loss in nuclear GATA-3 at 2 h (median [95% CI], 0.40 [0.27–0.53] versus 0.14 [0.11–0.19], p<0.05, W = 21.00, Wilcoxon's rank test) (Figure 7C) and cytoplasmic GATA-3 levels were enhanced by inhaled FP in a dose-dependent manner (median [95% CI], 0.0032 [0.0026–0.0039] versus 0.658 [0.592–0.720], p<0.05, W = −21.00, Wilcoxon's rank test) (Figure 7D). In addition, FP (500 µg) inhibited p38 MAPK phosphorylation in primary T cells in vivo at 2 h in samples from two patients (Figure 7E).
10.1371/journal.pmed.1000076.g007 Figure 7  Inhaled fluticasone propionate impairs GATA-3 nuclear localization in PBMCs.
(A) Representative immunocytochemistry of showing the effect of inhaled FP (500 µg) on GR and GATA-3 nuclear localisation. (B) Nuclear GATA-3 immunoreactivity in PBMCs from seven steroid-naïve asthma patients 2 h following inhaled FP treatment (100 or 500 µg via spacer). The median and interquartile ranges for each treatment are presented as a box-and-whiskers plot (n = 7); * p<0.05 Wilcoxon's rank test compared with placebo. (C) Immunoblotting analyses of PBMCs demonstrated a time-dependent decrease in nuclear expression of GATA-3, and increased cytoplasmic GATA-3 expression after inhalation of FP. (D) Immunoblotting analyses of PBMCs demonstrated a dose-dependent decrease in nuclear expression of GATA-3, and increased cytoplasmic GATA-3 expression 2 h after inhalation of FP. Histone H1 and MEK-1 immunoblotting confirmed equivalent total protein loading for the nuclear and cytoplasmic fractions respectively. Quantification of the densitometry data in (C) and (D) is shown as a box-and-whiskers plot of results from n = 6 participants for which data were available. *p<0.05 compared to control. (E) Western blot analyses of PBMCs demonstrated a time-dependent decrease in dual phosphorylation (threonine-180 and tyrosine-182) of p38 MAPK after inhalation of FP (500 µg). The results shown in (E) are representative of samples from two participants.   Taken together, our data suggest that inhaled FP reduces nuclear localization of GATA-3 in vivo by acutely inhibiting phospho-GATA-3–importin association. This effect may be direct, through competition for importin-α or associated molecules, or secondary to an effect on p38 MAPK-mediated GATA-3 phosphorylation via rapid induction of MKP-1. The combination of these two interacting effects can result in complete suppression of GATA-3 nuclear import and thus Th2 cytokine gene expression.