In this study the inflammatory conditions under which current asthma therapies down-regulate production of the mast cell chemoattractant CXCL10 by ASMC from people with and without asthma were established. The GC fluticasone and LABA salmeterol very effectively inhibited CXCL10 release induced by TNF-α or IL-1β, whereas they were completely ineffective, individually or together, against IFNγ- or cytomix-induced CXCL10 release. Surprisingly, salmeterol at sub-nanomolar concentrations increased CXCL10 release induced by all three cytokines together (cytomix). This effect of salmeterol was most likely mediated by increased NF-κB activity, but does not appear to be a class effect, as low formoterol concentrations did not affect release. In addition, fluticasone did not prevent the salmeterol-induced CXCL10 increase. Importantly, the thiazolidinediones ciglitazone and rosiglitazone were effective inhibitors of cytokine-induced CXCL10 release. However, their action appeared to be independent of PPARγ and NF-κB p65 activity. Nevertheless thiazolidinediones may provide an alternative therapeutic strategy for GC and LABA resistant asthma.
TNF-α- or IL-1β-induced CXCL10 release by ASMC from people with and without asthma was strongly inhibited by GC and LABA treatment in this study. This is in agreement with previous studies in which fluticasone or salmeterol inhibited release of other asthma-relevant chemokines induced by TNF-α [10–12]. In addition, Clarke et al.  showed fluticasone had an inhibitory effect on TNF-α-induced CXCL10 release in non-asthmatic ASMC. Similarly, IL-1β induced chemokine production from lung epithelial cells was strongly inhibited by fluticasone, salmeterol or both combined . Nie et al.  used ChIP analysis to demonstrate that the effects of fluticasone and salmeterol on ASMC TNF-α-induced eotaxin were mediated by inhibition of NF-κB p65 DNA binding to the eotaxin promoter. A similar mode of action could be assumed for TNF-α-induced CXCL10, as NF-κB also plays an important role in TNF-α-induced CXCL10 in ASMC .
In this study neither CXCL10 induced by IFNγ, nor CXCL10 induced by IFNγ combined with TNF-α and IL-1β (cytomix), were reduced when asthmatic and non-asthmatic ASMC were treated with GC or LABA individually, or in combination. In contrast, salmeterol has previously been reported to inhibit IFNγ–induced CXCL10 production by an airway epithelial cell line . However, similar to this study, others observed the inhibitory effects of fluticasone on TNF-α-induced CXCL10 disappeared when non-asthmatic ASMC were stimulated with a combination of TNF-α and IFNγ [13, 33]. Stimulation with TNF-α and IFNγ together inhibits ASMC glucocorticoid receptor (GR)α DNA binding and GC responsive element-dependent gene transcription [34, 35]. The findings reported here may have clinical relevance because CXCL10 production in the airway mucosa of people with asthma is increased by oral corticosteroids .
Subnanomolar concentrations of salmeterol further up-regulated cytomix-induced CXCL10 secretion in asthmatic ASMC, whereas formoterol did not. Formoterol was used at the same concentrations on the same ASMC in parallel with salmeterol in the experiments and did not cause any changes in CXCL10 levels. Thus this effect of salmeterol was not a drug class effect, or due to differences in lung donor age. A difference in effects of the two LABA is perhaps not unexpected as these agents differ markedly in their agonist properties. Salmeterol does not associate with the β2-adrenoceptor preferentially as formoterol does. Instead it partitions to the outer lipid bilayer of cells where it is retained. Two models have been proposed for salmeterol receptor engagement: 1) it is anchored to an exosite and engages and disengages from the receptor, or 2) it diffuses through the bilayer to engage the receptor through the side [37, 38]. In support of the latter, it has lower efficacy and onset of action than formoterol and increases membrane fluidity, which may affect receptor function and decrease its own efficacy reviewed in . There is extensive evidence from the development of salmeterol that β2-adrenoceptor antagonists inhibit all its effects . Given the reported efficacy of the β2-adrenoceptor antagonist butoxamine , its lack of effect on the salmeterol-induced increase in asthmatic ASMC CXCL10 release was unexpected. It is consistent with the effect of salmeterol not being mediated through classic β2-adrenoceptor engagement and signalling. However it does not eliminate the involvement of exosite binding or diffusion from the bilayer as salmeterol and butoxamine do not share all the same receptor binding sites [37, 41]. It may be possible for salmeterol to bind to the exosite/receptor sites not blocked by butoxamine and activate signalling leading to the increase in CXCL10 production. Salmeterol ≥ 1pM increased IL-6 and CXCL8 production by the lung epithelial cell line BEAS-2B stimulated with IL-1β and histamine. The β2-adrenoceptor antagonist ICI 118551 blocked the effects of 100nM salmeterol, confirming β2-adrenoceptor involvement at that high concentration, but it was not tested against subnanomolar concentrations  similar to those used in this study.
NF-κB plays an important role in cytokine induced CXCL10 production. In intestinal epithelial cells, IL-1β and IFNγ alone, and in combination, up-regulated CXCL10 in an NF-κB dependent manner . In ASMC from people with chronic obstructive pulmonary disease (COPD), TNF-α induced CXCL10 via NF-κB activation, whereas IFNγ only induced a weak activation of NF-κB  and activated STAT-1 as well, but not AP-1 . Stimulation with these cytokines together results in synergistic increases in CXCL10 production by ASMC [9, 13]. In this study, we showed in the asthmatic ASMC that cytomix stimulation strongly induced NF-κB p65 DNA binding activity and treatment with very low salmeterol concentrations further enhanced it. This increased p65 DNA binding activity may contribute to the increases in cytomix-induced CXCL10 and is consistent with the hypothesis that salmeterol increases asthmatic ASMC cytomix-induced CXCL10 production in an NF-κB dependent manner. However further studies are needed to establish whether or not the increase in p65 DNA binding activity is specific to asthmatic ASMC, occurs in the CXCL10 promoter and over what salmeterol concentration range this occurs.
The salmeterol enhancement of cytomix-induced CXCL10 release was specific to asthmatic ASMC. Whereas the salmeterol-induced increases in IL-6 and CXCL8 secretion by BEAS-2B cells were GC sensitive , the salmeterol enhancement of asthmatic ASMC CXCL10 production reported here was not. In ASMC from people with asthma there is increasing evidence of reduced sensitivity to steroids and intrinsic changes in intracellular signalling. These include altered calcium homeostasis and increased mitochondrial biogenesis [44, 45]. Cytomix-induced NFκB transcriptional activity may be increased in the presence of salmeterol as a result because ASMC NFκB transcriptional activity is sensitive to changes in calcium . As well, asthmatic ASMC β2-agonist–induced cAMP levels are lower due to increased PDE4 levels and activity  and cytomix-induced MAPK signalling is altered . Production of cAMP activates cAMP-dependent protein kinase (PKA) and exchange protein directly activated by cAMP (Epac). Whether PKA and/or Epac expression or activity is altered in asthmatic ASMC warrants investigation. Notably, β2-agonists regulate non-asthmatic ASMC functions via Epac [48, 49] and β2-agonist enhancement of bradykinin-induced CXCL8 release is mimicked by activation of either PKA or Epac and involves the MAPK ERK . We reported recently that the MAPK JNK, but not p38 or ERK, is involved in CXCL10 production following cytomix stimulation, but JNK activation is markedly reduced in asthmatic compared to non-asthmatic ASMC . Thus the salmeterol-induced increase in CXCL10 would seem less likely to be mediated via PKA or Epac with ERK. The alternate β-arrestin-2 pathway involved in receptor desensitization, leads to increased inflammation again through ERK/p38 MAPK activation reviewed in . It may still be of interest however, as β-arrestin-2 also interacts with PI3K and some differences in expression of PI3K isoforms in asthmatic and non-asthmatic ASMC have been reported [51, 52]. These scenarios are only likely if salmeterol does stimulate some receptor-mediated signalling in the asthmatic ASMC despite the presence of butoxamine. Clearly multiple signalling pathways are potentially involved and carefully controlled follow-up studies are needed to investigate which contribute to the salmeterol-induced increase in CXCL10 release by asthmatic ASMC and the functional consequences of this enhanced response.
The observed insensitivity of IFNγ- or cytomix-induced CXCL10 to current asthma medications emphasizes the need for novel compounds that can inhibit these nonresponsive pathways. In this respect thiazolidinediones show some promise. They are anti-inflammatory in animal models of asthma [20, 21] and have bronchodilatory effects in smokers with asthma . In vitro they do have anti-inflammatory and anti-remodelling effects, but these can be cell- and stimulus-specific and involve diverse and often complex mechanisms that are still being elucidated. Rosiglitazone, by increasing PPARγ signalling, inhibits CXC3CL1 signalling  and hypoxia-induced increases in lung TGF-b signalling and collagen deposition , but both pioglitazone and rosiglitazone reduce orbital fibroblast hyaluronan synthesis and T-cell adhesion to the fibroblasts in a PPAR-independent manner . In ASMC they also have PPARγ-dependent and independent effects to inhibit ASMC proliferation and secretion of some key pro-inflammatory mediators including CCL11 (eotaxin), CCL2 (MCP-1), CCL5 (RANTES) and IL-6 [15, 18, 23].
In this study the thiazolidinedione ciglitazone strongly inhibited IFNγ- and cytomix-induced ASMC CXCL10 secretion. This appears to be a class effect as rosiglitazone, which is currently used to treat people with type II diabetes mellitus, also inhibited cytomix-induced CXCL10 release. Although Ward and colleagues  found the anti-proliferative effects of rosiglitazone in ASMC were sensitive to the PPARγ antagonist GW-9662, in this study GW-9662 did not prevent ciglitazone/rosiglitazone inhibition of asthmatic and non-asthmatic ASMC CXCL10 production. Further, in contrast to reports that PPARγ agonists can inhibit NF-κB activity  and subsequent CXCL10 production  in other cell types, there was no evidence that ASMC NF-κB activity was inhibited by ciglitazone or that it affected early CXCL10 gene transcription. In addition any reported thiazolidinedione effects on JAK-STAT signalling  are not applicable to inhibition of CXCL10 induced by IL-1β or TNF-α Thus we propose ciglitazone inhibits asthmatic and non-asthmatic ASMC CXCL10 release via a PPARγ-independent post-transcriptional mechanism. Nie and colleagues  have also provided evidence that PPARg agonists can have post-transcriptional inhibitory effects on ASMC chemokine production. They found the PPARγ agonist 15d-PGJ2 and thiazolidinedione troglitazone inhibited TNF-α-induced CCL2 post-transcriptionally, but prevented NFκB binding to the CCL11 gene promoter and inhibited CCL11 gene transcription. Zhu and colleagues  recently reported similar findings to our study: rosiglitazone and/or troglitazone inhibition of ASMC cytokine-induced IL-6 and CCL5 production was not prevented by GW-9662, or PPARγ knockdown, and was not accompanied by inhibition of NF-κB activation.
Although further studies elucidating the different mechanisms of action of the thiazolidinediones are needed, the strong inhibition of ASMC CXCL10 release observed, irrespective of cytokine stimulus, is evidence of another beneficial anti-inflammatory effect the thiazolidinediones might have in asthma and other obstructive respiratory diseases. However in contrast to the findings of Nie and colleagues for CCL5 and CCL11 , in this study there were no additional/synergistic inhibitory effects on CXCL10 release when ciglitazone was used in combination with fluticasone or salmeterol.