Outpatients attending for upper GI endoscopies were recruited for nasal brush sampling following approval of study protocol and consent forms by the Beaumont Hospital Ethics Committee. Subjects were excluded on the basis of pre-existing immunosuppression, pulmonary or nasal pathology, including current or recent (within 6 weeks) upper or lower respiratory tract infection and reported normal functional status.

Following informed consent, nasal brushing was performed under direct vision using a Cervibrush + (CellPath plc)using a modification of the technique of Bridges et al 8. Tracheobronchial epithelial cells were harvested as in the method of Kelsen et al 9. Samples were accepted for analysis if they contained at least 80% epithelial cells.

Human airway epithelial cells (A549, European Collection of Cell Cultures, Porton Down, UK) were cultured at 37°C in 5% CO₂ in Ham's F12 (Gibco-BRL), 10% FCS, 1% penicillin/streptomycin. Prior to agonist treatment, cells were washed with serum-free F12 and placed under serum-free conditions or in serum containing 1% FCS for LPS stimulations.

Fluticasone propionate and salmeterol were obtained from Glaxo SmithKline, Glaxo Wellcome UK Ltd, Stanley Park West, Uxbridge, Middlesex UB11 1BT, and reconstituted Ham's F12/0.01% DMA and Ham's F12/0.01% Methanol respectively to stock concentrations of 10^-6 M. Dexamethasone was purchased from Sigma-Aldrich, Tallaght, Dublin, Ireland and reconstituted in 10% Ethanol in PBS to a stock concentration of 1 mM, and serial dilutions prepared in PBS.

Cigarette smoke extract (CSE) was freshly prepared for each experiment by a modification of a previously published method 10. Briefly, 2 filtered Marlboro Red cigarettes, each containing 0.8 mg of nicotine and 10 mg of tar according to the manufacturer's report, were bubbled through 20 ml serum free F-12 medium, pre-warmed to 37°C, by a mechanical vacuum pump. The extract was filtered through a 0.45 μm pore filter (Millipore, Bedford, MA) to remove bacteria and particles, and serial dilutions 1:10 were made.

RNA isolation, cDNA synthesis and RTPCR were performed as previously described 3 using gene-specific primers (Table 1). Products were analyzed by densitometry and compared in a semi quantitative manner relative to GAPDH using ImageMaster^® TotalLab Software (Amersham Pharmacia, Amersham, UK).

TLR4 mRNA was quantified using commercially available SYBR Green assays as previously described 11 with primers listed in Table 1. The results are expressed as the ratio of the mean of triplicate target gene cDNA measurements to the triplicate housekeeping gene (β-actin) measurement.

IL-8 protein concentrations in cell supernatants were determined by sandwich ELISA (R & D Systems, U.K.). TLR4 protein was analysed in membrane and cytosolic fractions by Western Blot as previously described 12 and surface expression by Laser Scanning Cytometry as previously described 3.

Viability of A549 cells under stated treatment conditions was quantified using the Promega CellTiter 96 Aqueous One Solution Cell Proliferation Assay as recommended by the manufacturer.

Data were analyzed with GraphPad Prism 3.0 software package (GraphPad Software, San Diego, CA). Results are expressed as mean ± S.E. and were compared by Mann-Whitney test. Differences were considered significant when the P value was ≤ 0.05.

The demographics of the study population are shown in Figure 1A. There was no significant difference in the characteristics of the COPD subgroups or control subgroups in terms of age, gender or medication use. No patients or control subjects reported a clinical history suggestive of atopy. All COPD subjects were using inhaled LABA and corticosteroids. COPD patients were on average a decade older than the control subjects. There was difficulty in recruiting a population of "normal" older smokers, that is, smokers who had no history of respiratory disease and normal FEV1. The main objective of the study was to observe differences between COPD patients of different degrees of severity. Observed differences with control groups represent a "real world" differences between typical subjects with this condition and healthy control subjects. As all COPD patients were using both inhaled LABA and corticosteroids, differences between subsets of patients may be attributable to the disease process, while differences with control subjects may be the result of disease, smoking or medication.

We examined expression of TLR4 along with TLR2 and HBD2 in the nasal epithelium of healthy smokers and age matched controls (Figure 1B). Semi-quantitative analysis of mRNA expression revealed a very significant reduction in TLR4 expression in the nasal mucosa of smokers compared to controls (P < 0.005). There was no significant reduction in expression of TLR2 (P = 0.28) or HBD2 (P = 0.20).

There were no significant differences between mRNA expression of TLR4, TLR2 or HBD2 in nasal epithelium between smokers and non-smokers in either the severe (FEV1 < 1L) or less severe (FEV1 > 1L) COPD (data not shown). Smokers and non-smokers were therefore grouped together for further analysis. As demonstrated in Figure 1B, there was significant upregulation of TLR4 expression in mild to moderate COPD compared to smoking controls († P < 0.05), while severe disease was associated with a significant reduction in TLR4 expression compared to less severe disease (P < 0.05). There was no difference in TLR2 expression between the study groups. Changes in HBD2 expression mirrored those of TLR4, with significant upregulation in mild-moderate COPD compared to controls (P < 0.005), and reduced expression in severe COPD compared to mild-moderate disease (P < 0.05). HBD2 expression in severe COPD was statistically similar to normal controls. (P = 0.17).

In order to see if nasal expression of TLR4 could be extrapolated to expression in the lower respiratory tract, a subgroup of COPD patients underwent bronchoscopy and brush sampling of the tracheo-bronchial epithelium as well as nasal brushing. Data from each of nine subjects is presented in figure 1C, with linear regression analysis demonstrating good correlation between upper and lower respiratory tract expression of TLR4 mRNA (r² = 0.76, P = 0.001).

We next examined the effect of cigarette smoke on expression of TLR4 in respiratory epithelium in vitro. There was a dose dependant downregulation in TLR4 mRNA (Figure 2A) and protein (Figure 2B) following exposure to the cigarette smoke extracts. To ensure that the effect was not caused by direct toxicity of the cigarette smoke, a viability assay was performed which demonstrated 50% reduction in viability with undiluted CSE, but no toxic effect following dilution of the extracts (Figure 2C) which showed no significant difference in viability compared to untreated cells. We went on to examine functional effect by IL-8 ELISA. As expected, CSE has some direct inflammatory effect resulting in IL-8 production at dilute concentrations. However, concordant with the reduced expression of TLR4, the respiratory epithelial cells have dose dependent reduced secretion of IL-8 following treatment with higher concentrations of CSE, both with and without additional LPS (Figure 2D). Failure of the cells to produce any IL-8 following exposure to undiluted CSE may be a result of the direct toxic effects demonstrated in the viability assay.

To explore other potential modulators of TLR4 expression pertinent to COPD, we first examined the effect of Fluticasone on expression of, TLR4 mRNA by RT-PCR in the respiratory epithelial cell line A549 grown in culture (Figure 3A). A dose dependent downregulation of TLR4 compared to the housekeeping gene GAPDH was observed with an Inhibitory Concentration (IC) 50 between 10^-9 and 10^-8 M. Consistent with the data of Homma et, who found no upregulation of TLR2 in A549 cells treated with dexamethasone alone 13, we found no change in expression of TLR2 or of HBD2 (data not shown).

Fluticasone propionate is a synthetic trifluorinated glucocorticoid. Pharmacologic properties include high lipophilicity, low systemic absorption, rapid metabolism and clearance, and high affinity for the glucocorticoid receptor, resulting in a high therapeutic index as a topical anti-inflammatory agent 14. Its very low water solubility makes it unpredictable for use in cell culture, however. We therefore assessed whether the observed effect was a class effect of corticosteroids, using the more soluble glucocorticoid dexamethasone. A dose dependent downregulation of TLR4 mRNA (Figure 3B) and protein (Figure 3C) was observed. Consistent with the increased potency of fluticasone, which has approximately 8 times the binding affinity of dexamethasone, a higher dose of dexamethasone was required to achieve a comparable effect (IC₅₀ between 10^-8 and 10^-7 M), whilst acknowledging that these results are semi-quantitative.

In order to determine the functional relevance of this effect, we stimulated the cells with the TLR4 agonist LPS. Stimulation of the cells with LPS 10 μg/ml for 24 hours resulted in a significant induction of IL-8, as measured by ELISA of the cell culture supernatant (P < 0.05) (Figure 4). Pre-treatment of the cells with dexamethasone dose-dependently abrogated this effect, reaching statistical significant at a dose of 10^-7 M (P < 0.05). A similar effect was seen on the induced expression of HBD2 mRNA in response to LPS (data not shown).

We next examined the effect of the long acting beta agonist salmeterol on expression of TLR4 mRNA by RT-PCR over a dose range of 10^-9 and 10^-6 M. Cells were incubated with the drug for 6 hours. Semi-quantitative analysis suggested a small increase in TLR4 expression over control at the highest dose of 10^-6 M, however the lack of a dose response cast doubt on the functional relevance of this observation. We therefore went on to quantify this effect by Real Time RT-PCR and found no significant effect of Salmeterol 10^-6 M on TLR4 gene expression (Figure 5A). Analysis of protein expression in total cell lysates similarly showed no significant change in total TLR4 (TLR4t) expression (Figure 5B lower panel), however levels in cytosolic extracts (TLR4c) were decreased (Figure 5B upper panel). A concomitant increase in membrane expression was evident at doses of 10^-7 M and 10^-6 M. (Figure 5C), an effect which was confirmed by Laser Scanning Cytometry (Figure 5D). Pre-treatment of cells with the beta-blocker Butoxamine abrogated the effect of Salmeterol 10^-6 M on membrane expression of TLR4, indicating a specific beta-adrenoreceptor mediated effect (Figure 5C). Taken together these data indicate that salmeterol induces a post-translational transport effect on TLR4.

Because inhaled long acting beta agonists are most often prescribed in combination with inhaled corticosteroids, we examined the effect of these compounds used in combination. The lowest dose of dexamethasone at which a functionally significant downregulation of TLR4 was observed, namely 10^-7 M, was used in combination with the dose of salmeterol required to produce upregulation of the same receptor, namely 10^-6 M. TLR4 gene expression was determined by Real Time PCR (Figure 6A). Again, treatment with salmeterol alone caused no significant change in TLR4 expression above untreated cells, while dexamethasone down regulated TLR4 expression. At the mRNA level, the dexamethasone effect persists when the two compounds are used in combination, resulting in significant downregulation in TLR4 mRNA expression compared to untreated cells. Looking at membrane protein expression however, salmeterol reverses the effect of dexamethasone on TLR4 expression resulting in no net change in TLR4 membrane expression with the two drugs used in combination (Figure 6B). Cell viability was not affected by either drug (data not shown). A similar pattern was observed in LPS-induced IL-8 expression, where the addition of salmeterol partly reversed the impaired IL-8 response to LPS observed with steroid treatment alone (Figure 6C).

As previously demonstrated in figure 2D, IL-8 production in response to LPS was downregulated following exposure to CSE 10^-1. IL-8 production was further inhibited by pre-treatment with dexamethasone consistent with an additive effect of downregulation of TLR4 expression by both treatments in isolation. Salmeterol treatment was not able to enhance LPS-induced IL-8 expression in the presence of CSE however, and similarly the "protective" effect of salmeterol on dexamethasone-induced inhibition of TLR4 signalling was lost in the presence of CSE. In fact there was further downregulation of IL-8 production (Figure 7). These findings are in keeping with recent report that combination of fluticasone and salmeterol potentiates the suppression of cigarette smoke-induced IL-8 production by macrophages 15. Salmeterol was found to have no effect on CSE induced IL-8 production in airway smooth muscle cells 16, although the effect of LPS was not examined in these studies.