Role of Cyclooxygenase Isoforms and Nitric-Oxide Synthase in the Modulation of Tracheal Motor Responsiveness in Normal and Antigen-Sensitized Guinea Pigs

Both nitric oxide (NO) and prostaglandins (PGs) are relevant mediators in airway physiology and pathology, and their production is regulated by different enzyme isoforms. NO is formed from l-arginine by NO synthase (NOS), of which three isoforms have been identified: neuronal NOS, inducible NOS, and endothelial NOS (Ogden and Moore, 1995). PG synthesis depends on cyclooxygenase (COX), two main isozymes of which are currently recognized: COX-1, constitutively expressed in different tissues, and COX-2, regarded as an inducible isoform, since an increase in its expression has been associated with inflammatory processes (Simmons et al., 2004).

It is generally believed that the production of NO and PGs by constitutive enzyme isoforms is important for physiological functions, whereas their formation via inducible isoforms is involved mainly in inflammatory and other pathological responses (Di Rosa et al., 1996). Increasing evidence suggests a link between NOS and COX pathways (Di Rosa et al., 1996), but there is no consensus on the functional consequences resulting from their mutual interactions. NO has been reported to regulate the activity of COX enzymes both by enhancing (Salvemini et al., 1993) and inhibiting (Stadler et al., 1993) their expression or modulating their activity (Di Rosa et al., 1996); on the other hand, a possible influence of PGs on NOS activity has been proposed, and current evidence suggests inhibition (Marotta et al., 1992; Milano et al., 1995), stimulation (Milano et al., 1995), or no interaction (Curtis et al., 1996).

Both NO- and COX-derived products influence airway smooth muscle responsiveness to contractile mediators, among which histamine plays a key role in the pathophysiology of allergic responses (Braunstein et al., 1988; Nijkamp et al., 1993). Nevertheless, the possible therapeutic usefulness of NOS and COX inhibitors in asthmatic hyper-reactive airways has not been clearly established. Advances in the development of selective COX-1 and COX-2 inhibitors have provided useful tools to analyze the contribution of COX pathways to various pathological conditions. However, selective COX-1 or COX-2 inhibitors have shown unclear effects in airways; a therapeutic effect has been reported for COX-2 inhibitors in allergic inflammation (Oguma et al., 2002) or hyper-responsiveness of guinea pig airways after ozone exposure (Nakano et al., 2000), whereas an increase in airway inflammation and muscle hyper-responsiveness has been observed in sensitized mice when treated with COX-1 or COX-2 inhibitors (Peebles et al., 2002).

The aim of the present study was to determine the influence of COX-1 and COX-2-selective inhibitors, either alone or in combination with a nonselective NOS inhibitor, on in vitro tracheal motor responsiveness to histamine to examine possible interactions between endogenous NO and COX products relevant to the control of airway smooth muscle reactivity. The study was extended from normal to ovalbumin (OVA)-sensitized and OVA-challenged animals to identify possible changes in the response to pharmacological treatments resulting from the allergic condition.

Materials and Methods

Male Dunkin-Hartley guinea pigs (300–400 g b.wt.) were used. The experiments were carried out in accordance with the legislation of Italian authorities (D.L. 27/01/1992 no. 116), which complies with European Community guidelines (CEE Directive 86/609) for the care and use of experimental animals. The experimental protocol was approved by the Animal Care Committee of Pisa University.

Immunohistochemical Analysis of COX-1 and COX-2. The tracheobronchial tree was removed and placed in phosphate-buffered saline (PBS) maintained at 37°C and oxygenated with carbogen (95% O₂ plus 5% CO₂). Tracheal specimens were immediately fixed in ice-cold 4% paraformaldehyde diluted in PBS overnight. Samples were dehydrated with ethanol, treated with xylene, and embedded in paraffin at 56°C. Five-micron-thick serial sections were processed for either routine staining by hematoxylin-eosin or immunostaining.

To perform immunohistochemistry, sections were microwaved in citrate buffer and treated with 1% hydrogen peroxide in methanol for 30 min and normal swine serum (1:20) for 20 min at 37°C. The slides were incubated overnight at 4°C with the following polyclonal primary antibodies: rabbit anti-COX-1 (1:80, code 160109; Cayman Chemical, Ann Arbor, MI) and rabbit anti-COX-2 (1:2000, code no. ALX 210711; Alexis Biochemicals, Lausen, Switzerland). Primary immunoglobulins were diluted in PBS with 0.1% bovine serum albumin and 0.1% sodium azide. The immunoprecipitates were detected by an indirect streptavidin-peroxidase method (Dakopatts, Glostrup, Denmark) as previously reported (Bernardini et al., 1999; Virdis et al., 2005). In brief, sections were sequentially exposed to biotinylated secondary immunoglobulins and peroxidase-labeled streptavidin complex. The slides were then treated with 3,3′-diaminobenzidine tetrahydrochloride (DAB) and 1 mg/ml Tris buffer (Dakopatts) containing 0.02% hydrogen peroxide and counterstained with hematoxylin (1:30). All reactions were carried out at room temperature in a humidified chamber, and PBS was used for washes between each step. Negative controls were obtained by omitting primary antibodies or substituting the primary antibody with rabbit preimmune serum. Specificity controls included preadsorption of anti-COX-1 and -COX-2 primary antibodies with the corresponding blocking peptides, COX-1 (code no. CAY 360109; Cayman Chemical) and COX-2 (code no. ALX 153-063; Alexis Biochemicals), respectively. To test endogenous peroxidases and avidin-binding activity, slides were incubated with DAB alone or streptavidin-peroxidase complex plus DAB, respectively.

Specimens of six animals from different experimental group were immunostained for COX-1 and COX-2 and then blindly evaluated, using light microscopes, by two investigators (N. Bernardini and C. Ippolito), who were unaware of other results. The intensity of immunostaining was ranked as: –, absent; +, weak; ++, medium; and +++, strong, as previously reported (Fux et al., 2005).

Antigen Sensitization and Challenge. OVA sensitization was induced as described previously (Nieri et al., 1992). In brief, two consecutive administrations of OVA saline solution (100 mg/kg i.p. + 100 mg/kg s.c.) were performed 14 to 19 days before in vitro experiments. A group of sensitized guinea pigs were exposed to OVA aerosol (1% in saline) in a box connected to an ultrasonic nebulizer (Model Project 38080; Artsana, Como, Italy), on days 7, 14, 15, and 16 from sensitization. The aerosol was generated for 10 min or less if respiratory distress occurred. Respiratory distress was defined as the onset of sneezing, cough, and/or in-drawing of thoracic wall. Sham-challenged guinea pigs were exposed to aerosols with saline solution without OVA. Animals were sacrificed for tracheal removal 6 h after the last challenge. The multiple antigen-challenged guinea pig is a model that reproduces several of the characteristic features of asthma, including bronchial hyper-reactivity and inflammation (Samb et al., 2001). The presence of airway hyper-reactivity to histamine was confirmed 6 h after last antigen challenge by preliminary experiments in anesthetized guinea pigs, recording pulmonary inflation pressure as an index of airway resistance (data not shown).

In Vitro Recording of Tracheal Smooth Muscle Contraction. Guinea pigs were sacrificed by cervical dislocation and bleeding under light ether anesthesia. The tracheobronchial tree was immediately removed and placed in Krebs-Henseleit solution (118 mM NaCl, 4.7 mM KCl, 2.5 mM CaCl₂, 1.2 mM KH₂PO₄, 1.2 mM MgSO₄, 25.0 mM NaHCO₃, and 11.5 mM glucose). A set of experiments was carried out in epithelium-deprived tracheae after removing the epithelial layer by rubbing the internal surface of tracheal tube with the edge of a small spatula. Each specimen, examined by histology, revealed at least 70% removal of epithelium. Four multiple-ring preparations were obtained from a single tracheal tube. Each tracheal preparation was suspended, under a resting tension of 0.5 g, in a 20-ml organ bath containing the above saline solution maintained at 37°C and oxygenated with carbogen (95% O₂ plus 5% CO₂). Isometric contractile responses were recorded via a force-displacement transducer (model FT03D; Grass Instruments, Quincy, MA) connected to a polygraph (model WR 3101; Graphtec Corp., Yokohama, Japan). After an equilibration period of 10 min, a reference contraction to 40 mM KCl was obtained. After tone recovery by repeated washings with fresh Krebs-Henseleit solution, SC-560 (COX-1 inhibitor), DFU (COX-2 inhibitor), SC-560 + DFU or l-NAME (nonselective NOS inhibitor), or a combination of one COX inhibitor with the NOS inhibitor were added to the organ bath and incubated for 30 min. Thereafter, a single cumulative concentration-response curve to histamine (10^–7–10^–4 M) was constructed. Both SC-560 and DFU were tested at 10^–6 M, in agreement with selectivity studies (Riendeau et al., 1997; Smith et al., 1998). The effects of test drugs were compared with those obtained with their vehicle (controls). In each tracheal preparation retrieved from sensitized guinea pigs, 10^–5 M OVA was added at the end of experiment and the respective muscle contraction was recorded.

Assay of PGE₂ and Prostacyclin Release. PGE₂ and prostacyclin (PGI₂) release from tracheal preparations in response to histamine was assessed, both in normal and OVA-sensitized guinea pigs, to investigate their role in the modulation of tracheal contractile activity by COX-1 and COX-2 isoenzymes. Isolated tracheal rings were incubated using a 24-well plate at 37°C. Each well contained three tracheal rings in 0.5 ml of carbogenated Krebs-Henseleit buffer. Fifteen rings were isolated from each trachea; therefore, five different conditions per preparation could be tested. PGE₂ or PGI₂ levels were measured after histamine incubation, in the absence or in the presence of SC-560 (10^–6 M) or DFU (10^–6 M), either alone or in combination. The experimental design was the same as that employed for contraction recording, but histamine was applied only at the highest concentration (10^–4 M). The supernatant was collected after 20-min incubation with histamine and stored at –70°C until assay. PGE₂ was assayed as such, whereas PGI₂ was measured as its stable hydrolysis product 6-keto-PGF_1α. In both cases, an enzyme immunoassay was used, in accordance to the manufacturer's instructions (Cayman Europe, Tallinn, Estonia). The results are reported as picograms of PG per milligram of tissue weight.

Drugs and Chemicals. Histamine dihydrochloride, l-N^G-nitro-arginine methylester, ovalbumin (grade III), and SC-560 were obtained from Sigma-Aldrich (St. Louis, MO), and DFU was kindly provided by Merck Frosst Canada Inc. (Dorval, QC, Canada). l-NAME, histamine, and KCl were dissolved in Krebs-Henseleit solution; DFU and SC-560 were dissolved in dimethyl sulfoxide (10^–2 M). Dilutions of stock solutions were freshly prepared in Krebs-Henseleit solution for all drugs on the day of experiment. Concentration of dimethyl sulfoxide in the bath was always below 0.01% (v/v).

Data Analysis. In vitro responses to histamine, obtained in the presence or in the absence of test drugs, were expressed as percentage of active tension induced by 40 mM KCl in the same preparation. All data were reported as mean ± S.E. of n experiments. Comparisons between means were performed by Student's t test for unpaired data (two groups) or analysis of variance (ANOVA) (more than two groups) followed by Bonferroni post test. P < 0.05 was considered significant.

Results

Immunohistochemical Evaluation of COX-1 and COX-2 Expression

Both COX-1 and COX-2 are constitutively expressed in normal guinea pig trachea. In particular, COX-1 was found in the columnar cells of pseudostratified ciliated epithelium, which showed intense apical immunostaining at the luminal surface, corresponding to the ciliar layer, but scanty cytoplasmic staining (Fig. 1A). No specific immunoreaction was observed in subepithelial connective and muscle tissues (Figs. 1A and 2A). COX-2 was widely expressed in cytoplasm of columnar epithelial cells and in subepithelial resident inflammatory cells localized in lamina propria (Fig. 1C); low amounts of specific anti-COX-2 immunoreaction were detected in tracheal chondrocytes and smooth muscle cells of normal animals (Figs. 1C and 2C).

Tracheal tissues from OVA-sensitized animals showed different patterns of COX isoform expression compared with normal ones. The epithelium, which appeared thickened, with mucous metaplastic changes and extensive alteration at luminal surface, showed weak cytoplasmic COX-1 expression with scattered epithelial cells more intensely immunostained (Fig. 1B). Subepithelial tissues, including lamina propria and submucosa as well as tracheal cartilage (Figs. 1B and 2B) and smooth muscle (Fig. 2B), did not express detectable COX-1 staining. In contrast, COX-2 expression was markedly increased in tracheal specimens excised from OVA-sensitized animals. Large amounts of COX-2 were indeed detected in the epithelial layer, mainly in the cytoplasm of basal cells and columnar cells in their apical portion (Fig. 1D). COX-2 was also found to be intensely expressed in subepithelial inflammatory cells and vessels of lamina propria (Fig. 1D) as well as in tracheal smooth muscle cells and chondrocytes (Fig. 2D). Preadsorption of primary antibodies with the respective blocking peptides completely abolished the staining either in normal or sensitized trachea (data not shown). The degree of COX-1 and COX-2 immunoreactivity in tracheal samples from normal and sensitized animals is summarized in Table 1.

In Vitro Recording of Smooth Muscle Contraction

The responses to 40 mM KCl, used as an internal standard contracturant, were not significantly different among the experimental groups (298.6 ± 32.13 mg tension, n = 75 for normal guinea pigs; 297.2 ± 24.60 mg tension, n = 70 for OVA-sensitized guinea pigs; 275.1 ± 27.27 mg tension, n = 34 for OVA-challenged guinea pigs). SC-560, DFU, or l-NAME did not significantly modify the basal tone of tracheal preparations both in normal and sensitized guinea pigs, although in a few cases, a slight loss of basal tone was observed with COX inhibitors.

Normal Guinea Pigs. Histamine (10^–7–10^–4 M) induced concentration-dependent contractions, which were significantly enhanced by COX-1 and COX-2 inhibitors (Fig. 3, A and B). The increment of maximal contraction was 40.0% for SC-560 (10^–6 M) and 32.3% for DFU (10^–6 M). A similar enhancing effect was obtained upon concomitant application of SC-560 and DFU (increment of maximal contraction, 33.0%) (Fig. 3C). Both COX inhibitors, although enhancing the contractile response to histamine, did not modify its potency, with pD₂ values ranging from a minimum of 5.18 ± 0.23 in the DFU control group (n = 7) to a maximum of 5.54 ± 0.37 in the SC-560-treated group (n = 8). The NOS inhibitor l-NAME (10^–4 M) did not influence the tracheal responsiveness to histamine but suppressed the enhancing effect of SC-560 (Fig. 4A). In addition, the enhancing action of DFU remained largely unaffected by l-NAME (Fig. 4B), and the moderate decrease in the effect of DFU, occurring at histamine concentrations below 10^–6 M (Figs. 3B and 4B), was likely to reflect the biological variability of experimental settings.

OVA-Sensitized Guinea Pigs. Tracheal preparations from guinea pigs sensitized to OVA were used to evaluate whether antigen sensitization affects the influence of COX isoforms on airway muscle response to histamine. Under these conditions, the responsiveness to histamine was not significantly affected by SC-560 (10^–6 M), whereas a significant enhancement was observed with DFU (10^–6 M) (increment of maximal response, 23.6%) (Fig. 5, A and B). l-NAME (10^–4 M) did not significantly influence the control response to histamine, the enhancement induced by DFU, and the contractions evoked by histamine in the presence of SC-560 (Fig. 5, C and D).

OVA-Challenged Guinea Pigs. Although in vivo airway hyper-responsiveness to histamine was observed 6 h after the last aerosol with OVA (data not shown), in vitro experiments did not show any tracheal hyper-reactivity to the autacoid, according to findings obtained by Mitchell et al. (2004) in pig bronchi. Figure 6A shows that the contractile activity evoked by histamine was enhanced by DFU (10^–6 M) (increment of maximal contraction, 37.1%), whereas it was not modified by SC-560 (10^–6 M). l-NAME did not affect the histamine-induced contraction when assayed alone or in the presence of either COX-1 or COX-2 inhibitor (Fig. 6B). As already observed for normal animals, in the presence of l-NAME, the enhancing action of DFU on the contractile responses to histamine in tracheal preparations from OVA-challenged guinea pigs was not apparent at histamine concentrations lower than 10^–6 M (Fig. 6B).

Epithelium-Denuded Trachea from Normal and OVA-Sensitized Guinea Pigs. In epithelium-denuded tracheal preparations from normal guinea pigs, DFU (10^–6 M) did not significantly affect the concentration-response curve to histamine (Fig. 7A). In contrast, in OVA-sensitized animals, histamine-induced contraction was significantly enhanced by DFU (increment of maximal contraction: 26.4%) (Fig. 7B). Epithelium removal also prevented the enhancing action of SC-560 (10^–6 M) on histamine-induced contraction in normal animals. In addition, as already observed in preparations with intact epithelium, COX-1 blockade did not modify the responsiveness to histamine following epithelial deprivation of tracheal tissues from OVA-sensitized guinea pigs (data not shown).

PGE₂ and PGI₂ Release

A similar basal release of PGE₂ was measured from normal and OVA-sensitized guinea pig trachea (Table 2). Following tracheal stimulation with histamine (10^–4 M) for 20 min, an increase in PGE₂ production was observed in both cases, with a greater increment in preparations from normal animals. SC-560 (10^–6 M) significantly inhibited histamine-induced PGE₂ release when applied to tracheal specimens isolated from normal and OVA-sensitized animals, whereas DFU (10^–6 M) was without effects (Table 2). At variance with PGE₂, basal PGI₂ release was significantly higher in tracheal tissues from OVA-sensitized guinea pigs (Table 2). A significant increment of PGI₂ content was detected in the medium after incubation of tracheal specimens with histamine, both in normal and OVA-sensitized animals. Both COX-1 and COX-2 blockade was associated with a decrease in histamine-induced PGI₂ release, the most prominent inhibition being observed upon exposure of tracheal preparations from OVA-sensitized guinea pigs to DFU (Table 2). Concomitant incubation with SC-560 and DFU lowered PGI₂ levels toward values detected in normal animals under basal conditions (Table 2).

Discussion

The role of COX isozymes in the modulation of airway smooth muscle in healthy subjects is currently debated. Although various authors reported the absence of COX-2 expression in the tracheobronchial system of normal guinea pigs (Amann et al., 2001; Oguma et al., 2002), the role of COX-2 has been proposed to be relevant in the regulation of tracheal intrinsic tone in this species (Charette et al., 1995), and a constitutive COX-2 expression has been reported in normal airways (Schlemper et al., 2005). Our results from normal guinea pigs, showing that DFU enhances histamineinduced contraction, indicate a functional role for COX-2 isoform that seems to mediate an inhibitory control on airway muscle. It can be excluded that COX-2 activity in normal trachea depends on inducing stimuli linked to surgical manipulation of isolated tissues since COX-2 expression was clearly revealed by immunohistochemistry in tracheal tissues immediately after excision. Therefore, our data suggest that COX-2 expression in airways does not occur exclusively in inflamed or manipulated tissues and that this isozyme regulates the contractile activity of tracheal smooth muscle under normal conditions.

In the present study, a functional activity, similar to that observed for COX-2, was shown for the COX-1 isoform in normal tissues, where the selective COX-1 inhibitor SC-560 enhanced histamine-induced tracheal contractions. An inhibition of the release of relaxant PGs from the epithelial compartment (Prié et al., 1990) is the most likely mechanism explaining tracheal muscle hyper-responsiveness to histamine after blockade of COX activities in normal animals. In support of this view, our experiments showed that immunostaining of both COX isoforms was prevalent in the epithelial layer and that the DFU-induced enhancement of histamine-induced contraction was abolished in epithelium-deprived tracheal preparations.

At variance with COX inhibitors, l-NAME per se did not modify histamine-induced contraction. In this respect, controversial findings have been reported about the effect of NOS inhibition on airway responsiveness to spasmogens. In guinea pig trachea with intact epithelial layer, a weak potentiating effect by l-NAME in the proximal, but not distal, portion has been shown (Kloek et al., 2002). Moreover, the inability of l-NAME to modulate murine tracheal contractile responses to carbachol has been described by Kakuyama et al. (1999), who observed an l-NAME-induced small increase (approximately 15%) of maximal response to the cholinergic agonist only when NOS inhibition was tested in the presence of indomethacin. Other authors reported an NO-mediated relaxing effect counterbalancing in vitro tracheal contraction to spasmogens (Nijkamp et al., 1993), and the use of indomethacin in the organ bath throughout the whole experiment was hypothesized as the cause underlying the l-NAME-induced effect (Kakuyama et al., 1999). Other factors might be responsible for the conflicting results from different laboratories. For instance, open tracheal preparations (i.e., rings and strips) have been suggested to be insensitive to NO-mediated relaxation since epithelial-derived NO diffuses away in the medium, thereby decreasing its concentration in proximity of smooth muscle (Folkerts and Nijkamp, 1998). Moreover, circadian fluctuations of NOS activity have been described in many tissues, including lungs (Tunçtan et al., 2002), and a different metabolism of NO can give rise to formation of reactive molecular species modulating smooth muscle reactivity (Sadeghi-Hashjin et al., 1996) and even COX activity (Landino et al., 1996).

Of note, in the present study, there was a lack of additivity of the motor-enhancing effects exerted by SC-560 and DFU. Moreover, l-NAME, although being ineffective when assayed alone, prevented the stimulant action of SC-560 on histamine-induced contraction, without modifying the potentiating effect of DFU. These findings indicate that both COX isoforms and NOS operate within the same regulatory pathway at the level of the tracheal epithelial compartment and that a complex interplay among COX-1, NOS, and COX-2 is likely to exist. In particular, as illustrated schematically in Fig. 8A, our data suggest that epithelial COX-2 exerts an inhibitory influence on tracheal smooth muscle responsiveness, and NOS may exert an inhibitory control on COX-2, but such a modulating action is prevented by COX-1, which operates a tonic inhibition on NOS activity. In line with this functional model, the enhancing contractile effect observed with SC-560 may result from the removal of an inhibitory control of COX-1 on NOS activity, with subsequent NOS activation and onset of an NO-mediated inhibition on constitutive COX-2 function, leading to a reduced formation of relaxant COX-2-derived prostanoid(s). Consistently with this view, inhibitory actions of COX-1 products on NOS activity have been reported in airways (Kakuyama et al., 1999) and other tissues (Marotta et al., 1992; Illiano et al., 1996), and a COX-2 inhibition by NOS pathway has been demonstrated previously by various authors (Mollace et al., 2005).

In OVA-sensitized guinea pigs, the histamine contractile activity was differently modulated by COX inhibitors since only DFU, but not SC-560, enhanced the autacoid-induced contraction. These data suggest the occurrence of changes in the respective roles of COX isozymes in the control of airway smooth muscle reactivity as a consequence of allergic sensitization. The reduced significance of the COX-1 isoform in the modulation of histamine-induced contraction following OVA sensitization might depend on the specific localization of this isoform in the epithelial ciliar layer (instead of cytoplasmic compartment as observed for COX-2), which undergoes extensive alteration in sensitized animals, as shown by immunohistochemistry.

With regard to the role played by NOS pathway in sensitized animals, it does not seem to be significant in this setting since l-NAME did not modify the responsiveness of tracheal preparations to histamine both under control conditions and in the presence of COX-1 or COX-2 blockade. A direct relaxant effect by NO on airway smooth muscle has been excluded also by Larsson et al. (2005) in lung parenchymal strips from sensitized guinea pigs, where the authors revealed an indirect protective effect of NO on antigen-induced contraction via inhibition of leukotrienes release from mast cells.

The loss of influence of COX-1 and NOS pathways on COX-2-mediated control of airway muscle responsiveness, taken together with the observation that the enhancing effect of DFU on histamine-induced contractions persisted after epithelium deprivation of tracheal preparations, suggests that, in antigen-sensitized guinea pigs, the contractile responsiveness to histamine is modulated by COX-2 at extraepithelial sites and that the regulatory network of COX isoforms and NOS at epithelial level loses its functional significance in this setting (Fig. 8B). In support of this proposal, we obtained immunohistochemical evidence that, following OVA sensitization, COX-2 expression was markedly increased at subepithelial level, particularly in smooth muscle cells.

A number of different agents, including histamine, have been reported to induce the release of relaxant PGs, namely PGE₂ and PGI₂, in airways, and the inhibition of their release is regarded as a possible cause of tracheobronchial hyper-responsiveness elicited by COX inhibitors (Folkerts and Nijkamp, 1998; Carey et al., 2003). To investigate the role of these prostanoids in the present contractile effects elicited by COX-1 and COX-2 inhibitors, we assayed PGE₂ and PGI₂ release from isolated tracheal preparations. Both in normal and in OVA-sensitized animals, histamine-induced PGE₂ release was not affected by COX-2 blockade while being inhibited only by SC-560. However, this observation, indicating COX-1 as the source of PGE₂ release in airways, is consistent with previous findings by Gavett et al. (1999) in mice, where a constitutive expression of both COX-1 and COX-2 in control lung was present, but PGE₂ in bronchoalveolar lavage fluid decreased only in COX-1 knockout animals. When considering PGI₂, both COX isoforms seem to contribute to its release from tracheal specimens of normal animals, whereas in OVA-sensitized guinea pigs, COX-2 accounted mostly for histamine-induced PGI₂ release, these observations being consistent with immunohistochemical data. These findings, taken together with the results of our experiments on tracheal contractility, suggest that PGI₂, but not PGE₂, is responsible for COX-2-mediated modulation of tracheal responsiveness both in normal and OVA-allergic guinea pigs. In support of such a proposal, previous studies have shown that PGI₂ exerts relaxant effects on airway smooth muscle and can be released from tracheal tissues (Prié et al., 1990). In addition, the COX-2 pathway has been reported to mediate PGI2 production also in smooth muscle cells of extrarespiratory sites, including the vascular system (Schildknecht et al., 2004).

Tracheal preparations from antigen-challenged guinea pigs did not display substantially different response patterns with respect to animals subjected only to OVA sensitization, indicating that, in our experimental model, sensitization alone is able to determine alterations in tracheal responsiveness to COX and NOS inhibitors that persist in the asthma model of antigen-challenged guinea pig. The absence of an NO-mediated influence on the control of smooth muscle responsiveness in asthmatic airways has been already reported in guinea pigs as a result of arginase activity up-regulation (Meurs et al., 2002) or neuronal nitric-oxide synthase protein down-regulation (Samb et al., 2001). Therefore, our data extend the available evidence indicating a lack of influence by NO on the control of airway smooth muscle responsiveness in multiple antigen-challenged animals in the presence of COX isoform inhibition.

In conclusion, the main findings obtained in the present work show that both COX-1 and COX-2 isoforms are constitutively expressed in guinea pig trachea, and their inhibition induces tracheal hyper-responsiveness to histamine in normal guinea pigs, whereas in antigen-sensitized and -challenged animals, only COX-2 inhibition is able to evoke such enhancing action; in normal guinea pigs, the NOS pathway mediates a complex interplay between COX-1 and COX-2 at the epithelial level, where COX-1 inhibits NOS, thus preventing the inhibitory action of NOS on COX-2; after antigen sensitization, the epithelial COX-1/NOS/COX-2 network is impaired, and COX-2 isoform, up-regulated at the subepithelial level, particularly in smooth muscle cells, exerts its modulatory function through the release of PGI₂.