Restoration of Corticosteroid Sensitivity by p38 Mitogen Activated Protein Kinase Inhibition in Peripheral Blood Mononuclear Cells from Severe Asthma

Background Severe asthma accounts for a small number of asthmatics but represents a disproportionate cost to health care systems. The underlying mechanism in severe asthma remains unknown but several mechanisms are likely to be involved because of a very heterogeneous profile. We investigated the effects of a p38MAPK inhibitor in corticosteroid sensitivity in peripheral blood mononuclear cells (PBMCs) from severe asthmatics and the profile of its responders. Methodology/Principal Findings Corticosteroid sensitivity was determined by measuring dexamethasone inhibition of CD3/28 and TNF-α induced IL-8 production in PBMCs by using ELISA. PBMCs from severe asthmatics were relatively less sensitive to dexamethasone (Dex) as compared to those of non-severe asthmatics and healthy volunteers. The IC50 values of Dex negatively correlated with decreased glucocorticoid receptor (GR) nuclear translocation assessed using immunocytochemistry (r = −0.65; p<0.0005) and with decreased FEV1 (% predicted) (r = 0.6; p<0.0005). A p38α/β inhibitor (SB203580) restored Dex-sensitivity in a subpopulation of severe asthma that was characterized by a defective GR nuclear translocation, clinically by lower FEV1 and higher use of oral prednisolone. We also found that SB203580 partially inhibited GR phosphorylation at serine 226, resulting in increased GR nuclear translocation in IL-2/IL-4 treated corticosteroid insensitive U937s. Conclusions/Significance p38MAPKα/β is involved in defective GR nuclear translocation due to phosphorylation at Ser226 and this will be a useful biomarker to identify responders to p38MAPKα/β inhibitor in the future.


Introduction
Most patients with asthma have mild to moderate forms of the disease and are well controlled by corticosteroids or a combination of corticosteroids and long-acting b 2 -adrenoreceptors agonists (LABA). However, between 5-10% of patients remain symptomatic despite treatment with high doses of corticosteroids [1,2]. This group of patients account for about 50% of total health care cost in asthma [3]. It remains unclear as to why these patients respond less to inhaled and oral corticosteroids. Therefore, it is important to investigate both clinical and molecular features of corticosteroid resistance in severe asthma in order to better understand the complexity of the disease and identify any specific treatment.
It is widely acknowledged that heterogeneous mechanisms are involved in corticosteroid insensitivity. Lymphocytes and monocytes have been shown to be less corticosteroid sensitive in severe asthmatics as compared to non-severe form [4,5]. Increased IL-2, IL-4 in bronchoalveolar lavage (BAL) cells and IL-13 in sputum and lung biopsies have been observed in severe asthmatics and these cytokines are known to cause in vitro loss of corticosteroid responsiveness [5][6][7]. An increase in the inactive GRb isoform and a decrease in nuclear translocation have also been identified as causes of corticosteroid insensitivity in severe asthma [8,9]. The phosphorylation status of GR is reported to play a crucial role in its function and localization [5,10]. Additionally, activated proinflammatory transcription factors NF-kB and AP-1 can sequester GR or compete for transcription co-factors [1,11]. Smoking asthmatics have also shown reduced systemic corticosteroid responsiveness and oxidative stress could affect histone deacetylase (HDAC) 2 level which is critical for the mechanism of GR transrepression [12,13].
Standard treatment for corticosteroid insensitive severe asthma includes high doses of inhaled corticosteroid combined with LABA [14] as well as the use of leukotriene receptor antagonists, anticholinergics or theophylline [15]. Systemic corticosteroids are needed in patients with severe unremitting disease although the risk of side effects is significantly increased [15]. Other therapies have been investigated with mixed results including immunosuppressants anti-IgE and anti-TNFa [16]. The search for new therapies is particularly directed towards add-ons treatment to corticosteroid that can overcome the decrease in sensitivity observed in severe asthmatics. Inhibitors to kinases (p38, JNK, ERK, PI3K) and pro-inflammatory transcription factors (AP-1, NF-kB) are of particular interest. The p38MAPK pathway regulates various pro-inflammatory transcription factors such as AP-1 and NF-kB [17,18]. p38MAPK activation can also stabilize pro-inflammatory cytokines and chemokines transcripts [19] but also lead to the phosphorylation and inactivation of the glucocorticoid receptor (GR) and subsequent corticosteroid insensitivity [5]. This is supported by recent evidence that confirmed an increase in p38MAPK activation in alveolar macrophages from severe asthmatics [20]. In addition, the p38MAPK inhibitor, SB681323, also inhibited cytokine production in blood stimulated ex vivo in COPD patients [21].
In the present study we showed that a p38MAPK inhibitor (SB203580) preferentially restored corticosteroid sensitivity in PBMCs from a subpopulation of severe asthma that were characterized by increased ex-vivo corticosteroid insensitivity, decreased GR nuclear translocation and clinically by a tendency for reduced lung function and higher use of oral corticosteroids.

Corticosteroid Sensitivity to IL-8 was Reduced in Severe Asthma
Ten healthy subjects, 20 patients with mid-to-moderate asthma and 20 patients with severe asthma have been recruited for this study (Table 1). Basal levels of IL-8 in PBMCs after overnight incubation were similar between patient groups ( Table 2). TNFa stimulation alone resulted in a 6 to 8 fold increase of IL-8 production in all patients with no significant differences between groups (Table 2). However, concentrationdependent inhibition curve of dexamethasone on TNFa-induced IL-8 shifted to the right in severe asthma compared with those of healthy volunteers and non-severe asthma patients, resulting in higher 50% inhibitory-concentration of dexamethasone (IC 50 dex) for severe asthmatics (median (range): 35.4 (19.8;48.4), n = 14; p,0.05) as compared to non-severe asthmatics (12.2 (9.2;34.3) nM, n = 14) and healthy volunteers (13.4 (8.4;20.3) nM, n = 8) ( Figures 1A and 1B).
When the co-stimulation of TNFa with anti-CD3/28 was used in order to stimulate lymphocytes as well as monocytes, the production of IL-8 increased by 3-5 fold compared to TNFa alone, and the levels were not different between groups (Table 2). This system was less sensitive to dexamethasone as IC 50 dex was higher, and PBMCs from some patients did not inhibit IL-8 production in the range of dexamethasone concentrations used. In that case, ''10 24 M'' was arbitrarily used as IC 50 dex as maximal quantifiable data. The IC 50 dex of anti-CD3/28 and TNFainduced IL-8 in PBMCs from healthy volunteers was 1.26 (median) (0.46;7.74) mM (n = 10), which was more than 100 times higher than IC 50 dex of TNFa-induced IL-8 production alone. IC 50 dex value of non-severe asthmatics was 0.09 (0.02;4.35) mM (n = 20) (Fig. S1A), significantly lower than theIC 50 dex value of severe asthmatics (2.87 (0.19;100) mM). The percentage of patients that did not response to dexamethasone up to 10 25 M was 33% in severe asthma, which was higher than in healthy volunteers and non-severe asthmatics (10% each) ( Table 2).

Impaired GR Nuclear Translocation is Associated with Loss of Corticosteroid Sensitivity and Disease Severity
GR nuclear translocation was determined as the ratio of mean fluorescence between the GR signals (cy3 channel; red) in a fixed area of the nuclei and the nuclear signal (cy5 channel; blue) of the same area ( Figure 1C). The fold induction of the signal ratio of 4 hours dexamethasone treatment over non-treatment was calculated as the index of GR nuclear translocation (GNI: GR nuclear translocation index). The antibody used for immunocytochemistry is specific only for GRa isoform. Ten cells were randomly selected in each slide and the average GNI was calculated (Fig. S1D). Figure 1C showed representative results of the PBMCs from four individuals treated with or without dexamethasone: one healthy volunteer (i; GNI = 4.2), one nonsevere asthmatic (ii; GNI = 2.4) and two severe asthmatics (iii; GNI = 1.2 and iv; GNI = 1.2). This semi-quantitative analysis demonstrated that the GNI in severe asthma (1.3 (1.1;2.6) ratio, n = 19) tended to be lower when compared with GNIs of healthy volunteers (2.2 (1.8;3.2), n = 9; p = 0.18) but significantly decreased when compared to non-severe asthmatics (2.4 (1.3;4.8), n = 14; p,0.05) (Fig. S1D). When comparing the IC 50 dex for IL-8 and the GNI in patients, there was a strong correlation (r = 20.65, p,0.0005, n = 25; Figure 1D), suggesting less GR nuclear translocation was associated with less inhibitory efficacy of dexamethasone on cytokine release. Furthermore, there was also a good correlation between FEV 1 (% predicted) and GNI in asthmatics (r = 0.62, p,0.0005, n = 32), suggesting that patients showing a defective GR nuclear translocation were more severe and more corticosteroid insensitive. No differences in GRa expression were observed between patient groups ( Table 2). GRb was not observed using this antibody which can detect both isoforms (data not shown).

p38MAPK Inhibition Restores Corticosteroid Sensitivity in Severe Asthma
A p38MAPKa/b inhibitor (SB203580), was incubated for 30 minutes prior to dexamethasone treatment and the individual changes in corticosteroid sensitivity (IC 50 dex without treatment/ IC 50 dex with treatment) were determined in each individual with severe asthma (Fig. S2A). SB203580 increased corticosteroid sensitivity more or less in all severe asthma patients (Figure 2A), and particularly the patients showed more than 6 improvement index (ratio of IC 50 dex with and without treatment) by SB203580 in 12 out of 20 severe asthmatics ( Figure 2C). In these samples formoterol (1 mM) did not improve corticosteroid sensitivity (Fig. S2A). SB203580 alone inhibited TNFa and CD3/CD28 induced IL-8 production in severe asthma (SB: 59786506 pg/ml vs. NT: 91026606 pg/ml; p,0.0001) ( Figure 2B Table 2. Molecular profile of healthy volunteers, non-severe and severe asthmatics.    Figure 3B (fold induction of nuclear translocation over NT: NT + dex = 3.1 vs. IL-2/4+ dex = 1.6; p,0.001) and SB203580 significantly increased dexamethasone induced GR translocation (fold induction of nuclear translocation: IL-2/ 4+ dex + SB = 4.9 vs. IL-2/4+ dex = 1.6; p,0.05). In addition, GR phosphorylation at Ser226 was clearly phosphorylated with IL-2/IL-4 treatment, and it was partially but significantly inhibited by SB203580 ( Figure 3C).

Discussion
Severe asthma is characterized clinically by total or partial loss of corticosteroid sensitivity for the control of asthma symptoms [22]. Even at cellular level, macrophages [8], epithelial cells [23] and PBMCs [24] obtained from severe asthmatics have been reported to be corticosteroid insensitive in vitro. This study also confirmed corticosteroid insensitivity in severe asthma on TNFainduced IL-8 in PBMCs. IL-8 is reported to be increased in sputum, serum, monocyte, and robust read-out/marker [25][26][27]. We established this system to determine steroid sensitivity using PBMCs from severe asthma and COPD as well as culture cells (U937 cells) in many research [28][29][30]. Furthermore, the combination of TNFa and anti-CD3/CD28 were also used to Figure 2. p38MAPK inhibition restores corticosteroid sensitivity in severe asthma. A. PBMCs from severe asthmatics were incubated 30 minutes with SB203580 (5 mM) and followed by treatment with Dex (10 211 -10 26 M) for 1 hour and 24 hour with anti-CD3/28 plus TNFa. IL-8 release was measured by ELISA and IC 50 dexs were calculated with or without SB203580 (n = 20). B. PBMCs from severe asthmatics were treated with anti-CD3/28 plus TNFa and IL-8 concentrations calculated in pg/ml. Cells were pre-incubated with Dex (1 mM) alone or in combination with SB203580 (5 mM) prior anti-CD3/28 plus TNFa stimulation (n = 20). Data was plotted as median 6 SEM. p,0.05 is significant. C. The improvement on corticosteroid sensitivity was assessed for SB203580 by calculating the ratio (fold) change of IC 50 dex before and after treatment. Some patients did not show inhibition in the range of Dex concentrations used. In that case, ''10 24 M'' was arbitrarily used as IC 50 dex as maximal quantifiable data. When SB203580 incubation restored a response to Dex in Dex-insensitive patients, the improvement was arbitrarily designated as 1000 fold. Patients were divided into those who respond more to SB203580 (ratio .6, white dots, n = 12) than those who respond less (ratio ,6, black dots, n = 8). doi:10.1371/journal.pone.0041582.g002 stimulate lymphocytes and monocytes together, and it resulted in 3-5 times higher induction of IL-8 production and more than 100 times less corticosteroid sensitivity than with TNFa alone. In fact, anti-CD3 and anti-CD28 have been shown to synergistically stimulate TNFa and IL-8 in T cells [31], suggesting that TNFaproduced from T lymphocyte stimulated further monocyte production of IL-8 in our samples. Under this stimulation, PBMCs also tended to be dexamethasone insensitive, and 33% of severe asthma patients were totally dexamethasone resistant but only 10% of healthy volunteers and 10% of non-severe asthma group were dexamethasone resistant. Furthermore, it was found that the reduction in corticosteroid responsiveness correlated with a decrease in lung function, suggesting that patients with less corticosteroid sensitivity in PBMCs ex vivo displayed a more severe clinical phenotype. This reduction in lung function has been reported to be associated with a systemic increase in IL-8 and TNFa in blood serum and circulating [27] and infiltrated neutrophils in the airways [32,33] from severe asthmatics. Neutrophilic inflammation in the lung is thought to be corticosteroid-resistant [34]. Basal IL-8, a neutrophil chemoattractant, has been shown to be increased in BAL and sputum from severe asthmatics [32,33] although release from IL-8 from PBMCs was not. More importantly, we found that TNFa/anti-CD3/CD28induced IL-8 production was dexamethasone-insensitive particularly in severe asthma. Prednisolone was reported not to inhibit circulating neutrophils and IL-8 in the whole blood in patients with steroid-dependent asthma [35].
Several studies have been conducted in order to identify cell signalling relevant to pathogenesis of severe asthma. p38MAPK is one of the most studied signalling kinases and various compounds are currently being tested in a number of inflammatory diseases [36][37][38][39][40]. As shown in Figure 2B, the TNFa/anti-CD3/ CD28-stimulated IL-8 was inhibited by SB203580 in PBMCs, which was only inhibited at the highest concentration of dexamethasone. This shows the importance of the p38a/b pathway in the corticosteroid refractory pro-inflammatory cytokine regulation in severe asthma. The IL-8 promoter is regulated by transcription factors such as AP-1 and NF-kB [41] and p38MAPK is known to activate these transcription factors either by driving direct phosphorylation or indirectly by the phosphorylation of kinases responsible for the activation of transcription factors [17]. In contrast, dexamethasone only partially inhibited AP-1 and NF-kB inhibition [42]. In addition, p38a-dependent phosphorylation of MSK results in the phosphorylation of Histone-3 (H3) leading to the recruitment of NF-kB and the transcriptional regulation of IL-8 and other inflammatory cytokines [18]. Furthermore p38MAPK enhances the stability of pro-inflammatory cytokine mRNA by phosphorylation of an AUrich element in mRNA [19].
However, a more essential role of SB203580 was underlined in severe asthma by the restoration of corticosteroid sensitivity by this compound, particularly in the 17 most severe asthmatics (out of 20 patients we studied). Formoterol is reported to restore steroid sensitivity in PBMCs from severe asthma [24], the efficacy of which was limited in samples from patients recruited in this study, maybe due to routine medication of combination therapy of LABA and inhaled corticosteroids. Thus, our result highlights the importance of the p38a/b pathway in the restoration of corticosteroid sensitivity as previously published [43].
Earlier studies have demonstrated that treatment with IL-2/IL-4 can mimic the corticosteroid insensitivity seen in severe asthma Figure 4. Two populations in severe asthma based on response to SB203580. Severe asthmatics were arbitrarily separated into SB203580 ''responders'' or ''non-responders'' based on the ratio IC 50 dex without treatment/IC 50 dex with treatment .6 (SB resp, n = 12) or ratio ,6 (SB non-resp, n = 8) and compared using various parameters: A. IC 50 dex for IL-8. B. The % inhibition of anti-CD3/28 and TNFa-induced IL-8 after dex (10 27 M) treatment. C. Ratio of GR nuclear translocation dex (1 mM)/NT. D. Lung function by FEV 1 (% predicted). E. Use of oral prednisilone (mg). F. HDAC2 protein expression normalized with b-actin. Data was plotted as mean 6 SEM. p,0.05 is significant. doi:10.1371/journal.pone.0041582.g004 [5,44]. According to results obtained in a recent study, IL-2/IL-4 treatment induced p38MAPK activation [24]. Our study showed higher basal IL-2 and IL-4 production in PBMCs from severe asthma which could result in p38MAPK activation and subsequent loss of corticosteroid sensitivity. Other studies have also revealed both cytokines to be increased in blood serum of this patient group [6] and to be associated with increased corticosteroid resistance in PBMCs [44]. Goleva et al. confirmed these findings in T cells where IL-2/IL-4 induced corticosteroid resistance by activating p38MAPK resulting in reduced GR nuclear translocation [45]. Similarly, our in vitro studies confirmed that both cytokines could induce corticosteroid insensitivity in the monocytic cell line U937 with concomitant reduction of GR nuclear translocation.
In the present study, severe asthmatics showed a tendency for a reduction of GR nuclear translocation compared to healthy volunteers, mild and moderates asthmatics. However, two distinct patterns seem to present; one showing a defect in GR nuclear translocation and another with a normal GR nuclear shuttling in agreement with previous observations by Matthews et al. [9]. Impaired GR nuclear localization was associated with a decrease of corticosteroid sensitivity on IL-8 inhibition in severe asthmatics ( Figure 1D). In addition, severe asthma patients with lower GR nuclear translocation also showed lower lung function ( Figure 1E). Interestingly, ''higher responders'' to SB203580 in severe asthma patients had a significantly reduced GR nuclear translocation associated with reduced dexamethasone responsiveness as compared to ''low/non responders''. In fact, Irusen et al. demonstrated that p38MAPKa has the potential to phosphorylate GR leading to a defect of GR function [5]. Other studies have also demonstrated that another kinase, c-Jun N-terminal Kinase 1 (JNK1), and p38MAPK gamma could directly phosphorylate GR at a specific serine residue (S226) [24,46]. We confirmed that GR was phosphorylated at serine 226 in IL-2/4 corticosteroid insensitive model and that inhibition of p38MAPKa/b partially inhibited serine 226 phosphorylation. This implies that p38MAPKa phosphorylates GR in the cytoplasm and impairs GR nuclear translocation. Actually, restoration of corticosteroid sensitivity by SB203580 was shown to correlate with a defect in GR nuclear translocation.
Thus, IL-2 and IL-4 are likely to be responsible for reduced corticosteroid responsiveness as both cytokines were increased in severe asthma and induced corticosteroid insensitivity in vitro via reduced GR nuclear translocation due to excessive GR serine 226 phosphorylation. More importantly, p38MAPKa/b plays a major role on corticosteroid insensitive inflammation and defective steroid receptor function via increased GR phosphorylation at serine 226, resulting in reduced affinity for corticosteroid binding and decreased ability to translocate into the nuclei.
As several p38MAPK inhibitors are now been testing in clinical trials, this information or biomarkers will be useful in order to identify severe asthmatics that will respond to the treatment. This work also confirmed a heterogeneous phenotype of severe asthma based on signalling.

Subjects
Ten healthy volunteers, 11 patients with mild asthma, 9 patients with moderate asthma and 20 patients with severe asthma were recruited. Asthma severities were characterized using the Global Initiative for Asthma [14] guidelines and patients characteristics is summarized in Table 1. This study was approved by the Ethics Committee of the Royal Brompton & Harefield Hospitals National Health Service Trust, and all subjects gave written informed consent.

Isolation of PBMCs
Blood was collected in acid citrate dextrose (ACD) syringes and PBMCs were separated using the ACCUSPIN TM System-HISTOPAQUEH (Sigma, Poole, UK) following manufacturer's instructions.

Cell Culture of U937s
U937 (human monocytic cell lines) cells were purchased from the American Tissue Culture Center (ATCC, Teddington, UK) and maintained in continuous cell culture at 37uC, 5% CO2 in RPMI-1640 medium containing 10% FCS and 15 mM glutamine. Cells (5610 6 ) were incubated with or without human IL-2 (20 ng/ ml) and IL-4 (10 ng/ml) for 48 hours in RPMI-1640 medium containing 1% FCS and 15 mM glutamine.

Whole Cell and Nuclear Extraction
PBMCs (8225610 6 cells) were stimulated in the presence/ absence of dexamethasone (1 mM) (Sigma) and incubated at 37uC, 5% CO 2 in RPMI-1640 medium (10% FCS and 15 mM glutamine) for 4 hours. Whole cell extractions were performed using the Active Motif Nuclear Extraction kit (Rixensard, Belgium) following manufacturer's instructions. U937s that were stimulated with IL-2 and IL-4 for 48 hours, then with SB203580 for 30 minutes followed by dexamethasone treatment for 1 or 4 hours. Cells were then collected for nuclear extraction using the Active Motif Nuclear Extraction kit following manufacturer's instructions.

Western Blot
Proteins were separated using sodium dodecylsulphate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred into a nitrocellulose membrane using the i-Blot TM Dry Blotting System (Invitrogen, Carlsbad, CA, USA) following manufacturer's instructions. Primary antibodies against HDAC2 (Sigma), b-actin, TBP (Abcam, Cambridge, UK), Glucocorticoid Receptor (GR, E-20, Santa Cruz Biotechnology, California, USA), anti-S226 GR (Abcam) and Lamin A/C (Santa Cruz Biotechnology) were used for protein detection. Briefly, membranes were incubated with primary antibody overnight after blocking with 5% dry skimmed milk in TBS-Tween (0.05% v/v) and then with HRP conjugated secondary antibody. Bound antibodies were visualized by the ECL system (Amersham Biosciences, Buckinghamshire, UK).

Immunocytochemistry of GR
PBMCs, previously incubated with/without dexamethasone (1 mM), were cytospined into slides and fixed using IntraPrep TM Reagent 1 and permeabilized with IntraPrep TM Reagent 2 (Beckman Coulter, High Wycombe, UK). The method used was adapted from Li et al. (2007) [47]. The slides were analysed by confocal microscopy with imaging analysis Leica Confocal Software Lite TM (Leica, Heidelberg, Germany).

Detection of Phosphorylated GR
Human monocytic U937, maintained in continuous cell culture at 37uC, 5% CO 2 in RPMI-1640 medium containing 10% foetal calf serum (FCS) and 15 mM glutamine were stimulated with IL-2 and IL-4 for 48 hours in minimal media (1% FCS) to induce corticosteroid insensitivity. Cells were then treated with SB203580 (5 mM) for 30 minutes prior whole-cell extraction and SDS-PAGE/western-blotting analysis. Phosphorylation of Serine 226 of level was determined with anti-S226 GR antibody (Abcam) and normalized to GR expression.

Statistical Analysis
Clinical data are expressed as median and interquartile range. The effect of SB203580 on clinical data is expressed as average 6 SEM. Results were analysed using Kruskal Wallis and comparisons were made by Mann-Whitney using the Graph Pad Prism Software (Prism, San Diego, CA). Correlation was analysed by non-parametric Spearman method. Comparisons of two values were analysed with Wilcoxon matched pair test or paired t-test. The in vitro results were expressed as average 6 SEM. P,0.05 was considered statistically significant. Figure S1 A. PBMCs from healthy volunteers (HV) (n = 10), non-severe asthmatics (NSAA) (n = 20) and severe asthmatics (SA) (n = 20) were incubated 1 hour with Dex (10 211 210 -6 M) followed by 24 hours with anti-CD3/28 plus TNFa. IC 50 dex was measured for IL-8 in all patients. Some patients became completely resistant to Dex and their IC 50 dexs could not be calculated. They were given a nominal value of 10 24 M. Data was plotted as median 6 SEM. B. PBMCs from HV (n = 10), NSA (n = 20) and SA (n = 20) were seeded in 96-well plates and IL-2 cytokine release was measured using ELISA. C. PBMCs from HV (n = 10), NSA (n = 20) and SA (n = 20) were seeded in 96-well plates and IL-4 cytokine release was measured using ELISA. D. PBMCs were incubated with/ without Dex (1 mM) for 4 hours. GNI was measured by immunocytochemistry in HV (n = 9), NSA (n = 14) and SA (n = 19). E. HDAC2 protein expression was determined by SDS-PAGE/Western blotting and normalized using the expres-  Figure S2 A. Add-on treatments in severe asthma. A. PBMCs from severe asthmatics were treated with formoterol (1 nM) or SB203580 (5 mM) for 30 minutes followed by 1 hour stimulation with Dex (10 211 -10 26 M) and 24 hour with anti-CD3/28 plus TNFa. IL-8 release was measured by ELISA and IC 50 dexs calculated. The improvement on corticosteroid sensitivity was assessed for each add-on treatment by calculating the ratio (fold) change of IC 50 dex before and after treatment. A ''heat-map'' was constructed using the ratio for each svere asthmatic. B. PBMCs from severe asthmatics were treated with formoterol (1 nM) for 30 minutes followed by 24 hour with anti-CD3/28 plus TNFa. IL-8 release was assessed using ELISA. (TIF)