measuring conversion of either endogenous or exogenous arachidonic acid to three metabolites, PGE 2 , thromboxane B 2 or 6-ketoPGF 1a by radioimmunoassay. 3 Under control culture conditions HASM cells expressed COX-1, but not COX-2, protein. However, a mixture of cytokines (interleukin-1b (IL-1b), tumour necrosis factor a (TNFa) and interferon g (IFNg) each at 10 ng ml 71 ) induced COX-2 mRNA expression, which was maximal at 12 h and inhibited by dexamethasone (1 mM; added 30 min before the cytokines). Furthermore, COX-2 protein was detected 24 h after the cytokine treatment and the expression of this protein was also inhibited by dexamethasone (1 mM) and cyclohexamide (10 mg ml 71 ; added 30 min before the cytokines). 4 Untreated HASM cells released low or undetectable amounts of all COX metabolites measured over a 24 h period. Incubation of the cells with the cytokine mixture (IL-1b, TNFa, IFNg each at 10 ng ml 71 for 24 h) caused the accumulation of PGE 2 and 6-keto-PGF 1a .5 In experiments where COX-2 metabolized endogenous stores of arachidonic acid, treatment of HASM cells with IL-1b in combination with TNFa caused a similar release of PGE 2 to that when the three cytokines were given in combination. 6 In other experiments designed to measure COX-2 activity directly, cells were treated with cytokines for 24 h before fresh culture medium was added containing exogenous arachidonic acid (30 mM for 15 min) after which PGE 2 was measured. IL-1b and TNFa increased COX-2 activity and an additional small increase was produced by the three cytokines in combination. 7 These ®ndings suggest that the increased expression of COX-2 is intimately involved in the exaggerated release of prostanoids from HASM cells exposed to pro-in¯ammatory cytokines. These data indicate a role for airway smooth muscle cells, in addition to their contractile function, as in¯ammatory cells involved in the production of mediators which may contribute to the in¯ammatory response seen in diseases such as asthma.

Due to an error in manuscript preparation, an incorrect shRNA sequence for IRS2 was published. The correct hairpin sequence is a 19-nt stretch beginning from nt 703 of the published IRS2 cDNA sequence (XM_357863). The oligonucleotides cloned into the U6 construct for the IRS2U6 adenovirus are as follows: tcgagGTGACGCTGCAGCTTATGAttcaagagaTCATAAGCTGCAGCGTCACttttt (forward) and ctagAAAAAGTGACGCTGCAGCTTATGAtctcttgaaTCATAAGCTGCAGCGTCACc (reverse). In addition, the shRNA cassettes were cloned into the adenoviral cosmid pAxcwit, which lacks a promoter, and not the cosmid pAxCAwtit, as published.. The authors regret this error.

1 We investigated the eect of the p38 kinase inhibitor SB 203580 on airway in¯ammation induced by aerosolized lipopolysaccharide (LPS) in male Wistar rats. SB 203580 signi®cantly inhibited (ED 50 =15.8 mg kg 71) plasma levels of TNF-a in rats challenged with LPS (1.5 mg kg 71 , i.p.). 2 Aerosolized LPS induced a peak in TNF-a levels and the initiation of a neutrophilic response in bronchoalveolar lavage (BAL)¯uid at the 2 h time point. Furthermore, the 4 h time point was associated with the peak in IL-1b levels and the initial plateau of neutrophilia observed in the BAL uid. 3 SB 203580 (100 mg kg 71 ), had no eect on peak TNF-a levels or the associated neutrophilia in the BAL. Interestingly, the PDE 4 inhibitor RP 73401 (100 mg kg 71) signi®cantly reduced both TNF-a levels and neutrophilic in¯ammation. However, the BAL¯uid from rats pre-treated with either compound signi®cantly inhibited TNF-a release from cultured human monocytes 18 h after LPS treatment (83.6 and 44.5% inhibition, respectively). 4 Alternatively, SB 203580 (100 mg kg 71) produced dose-related inhibition of BAL IL-1b levels (67.5% inhibition, P50.01) and BAL neutrophilia (45.9% inhibition, P50.01) 4 h after LPS challenge. 5 P38 protein was present in lung tissue and the level of expression was not aected by LPS treatment.6 P38 kinase appears to be involved in the release of IL-1b and the sustained neutrophilic response in the BAL¯uid. This data may suggest a role for p38 inhibitors in the treatment of airway in¯ammatory diseases in which neutrophilia is a feature of the lung pathology.

1 Eotaxin is a novel C-C chemokine with selective chemoattractant activity for eosinophils. We determined whether eotaxin could be produced by human airway smooth muscle (HASM) cells in culture and examined its regulation by interleukin-10 (IL-10) and the corticosteroid, dexamethasone. 2 Stimulation of the cells with interleukin-1b (IL-1b) or tumour necrosis factor (TNFa) each at 10 ng ml 71 induced the release of eotaxin protein with maximal accumulation by 24 h. Interferon-g (IFNg) alone at 10 ng ml 71 had no e ect and there was no synergy between these cytokines on the release of eotaxin. 3 Reverse phase high performance liquid chromatographic (HPLC) analysis of supernatents from cells treated with TNFa (10 ng ml 71 for 96 h showed immunoreactivity to eotaxin which eluted with the expected retention time of 34.5 ± 35 min. 4 Both IL-1b and TNFa-induced release of eotaxin was not inhibited by dexamethasone (1 mM), however IL-10 (10 ng ml 71 ) had a signi®cant inhibitory e ect. Dexamethasone and IL-10 did not inhibit the induction of eotaxin mRNA induced by IL-1b or TNFa. 5 Thus, human airway smooth muscle cells can release eotaxin and could be an important source of chemokine production during airway in¯ammatory events.

1 In this study we have evaluated the pharmacological pro®le of the muscarinic antagonist glycopyrrolate in guinea-pig and human airways in comparison with the commonly used antagonist ipratropium bromide. 2 Glycopyrrolate and ipratropium bromide inhibited EFS-induced contraction of guinea-pig trachea and human airways in a concentration-dependent manner. Glycopyrrolate was more potent than ipratropium bromide. 3 The onset of action (time to attainment of 50% of maximum response) of glycopyrrolate was similar to that obtained with ipratropium bromide in both preparations. In guinea-pig trachea, the oset of action (time taken for response to return to 50% recovery after wash out of the test antagonist) for glycopyrrolate (t 1/2 [oset]=26.4+0.5 min) was less than that obtained with ipratropium bromide (81.2+3.7 min). In human airways, however, the duration of action of glycopyrrolate (t 1/2 [oset]496 min) was signi®cantly more prolonged compared to ipratropium bromide (t 1/2 [oset]=59.2+17.8 min). 4 In competition studies, glycopyrrolate and ipratropium bromide bind human peripheral lung and human airway smooth muscle (HASM) muscarinic receptors with anities in the nanomolar range (K i values 0.5 ± 3.6 nM). Similar to ipratropium bromide, glycopyrrolate showed no selectivity in its binding to the M 1 ± M 3 receptors. Kinetics studies, however, showed that glycopyrrolate dissociates slowly from HASM muscarinic receptors (60% protection against [ 3 H]-NMS binding at 30 nM) compared to ipratropium bromide. 5 These results suggest that glycopyrrolate bind human and guinea-pig airway muscarinic receptors with high anity. Furthermore, we suggest that the slow dissociation pro®le of glycopyrrolate might be the underlying mechanism by which this drug accomplishes its long duration of action.