Sterile Krebs-Henseleit buffer (KHB) was obtained from Serag-Wiessner (Naila, Germany). The thromboxane-A₂ mimetic U46619 was supplied by Paesel-Lorei (Frankfurt, Germany) and iloprost by Schering (Berlin, Germany). All other chemicals were purchased from Merck (Darmstadt, Germany).

The perfused rabbit lung model has previously been described in detail 31. Briefly, rabbits of either sex weighing 2.6 to 2.9 kg were anticoagulated with heparin (1000 U/kg) and anaesthetized with intravenous ketamine/xylazine. Tracheostomy was performed and the animals were ventilated with room air. After mid-sternal thoracotomy, catheters were placed into the pulmonary artery and the left atrium, and perfusion with KHB was started. For washout of blood, perfusate was initially not recirculated and the lungs were removed without interruption of ventilation and perfusion. The lungs were placed in a temperature-equilibrated chamber at 37.5°C, freely suspended from a force transducer for monitoring of organ weight. In a recirculating system the flow was slowly increased to 120 ml/min (total volume 350 ml). Left atrial pressure was set at 2 mmHg in all experiments. In parallel with onset of artificial perfusion, an air mixture of 80.5% N₂, 15% O₂ and 4.5% CO₂ was used for ventilation. Tidal volume (11 ml/kg) and frequency (10 – 13 breaths/min) were adapted to maintain the pH of the recirculating buffer in the range between 7.35 and 7.37. A positive end-expiratory pressure of 1 mmHg was used throughout. The PO₂ and PCO₂ values in the post-lung buffer fluid ranged between 100 and 120 mmHg and 38 and 43 mmHg, respectively. Pressures in the pulmonary artery, the left atrium and the trachea were registered by means of small diameter tubing threaded into the perfusion catheters and the trachea and connected to pressure transducers (zero referenced at the hilum). The whole system was heated to 37.5°C.

Lungs included in the study had 1) a homogeneous white appearance with no signs of hemostasis, edema or atelectasis; 2) pulmonary artery and ventilation pressures in the normal range; and were 3) isogravimetric (lung weight gain < 0.2 g/h) during an initial steady state period of at least 45–60 min.

Rabbit pulmonary smooth muscle cells (PSMC) were prepared as described. Primary SMC were isolated from rabbit pulmonary artery by carefully preparing <1 mm³ pieces of media, devoid of adventitial tissue as assessed by microscopic control. The pieces of media were placed into 12-well cell culture plates with 500 μl culture medium. The isolated pulmonary smooth muscle cells (PSMC) identity was verified by characteristic appearance in phase-contrast microscopy, indirect immunofluorescent antibody staining for smooth muscle-specific isoforms of α-actin and myosin (at least 95% of cells stained positive), and lack of staining for von Willebrand factor and vimentin, indicating that the cultures did not contain significant numbers of endothelial cells or adventitial fibroblasts 37. Smooth muscle cells were grown in Dulbecco's modified Eagle's high glucose medium supplemented with 10% fetal bovine serum, 100 μg/ml streptomycin and 100 units/ml penicillin.

Smooth muscle cells were grown to 90% confluence in 24-well plates, as previously described. For desensitization assays, cells were treated with iloprost for the indicated times. Then, cells were washed three times with HBSS and preincubated in HBSS containing 1 mg/ml BSA, 10 mM HEPES (pH 7.3) and 1 mM IBMX for 10 min at 37°C. Afterwards cells were again stimulated with iloprost (100 nM) or forskolin (10 μM) for 10 min and cAMP measurements were performed as described below. Reactions were stopped by aspiration and addition of ice-cold 96% ethanol. Dried samples were overlaid with 300 μl RIA-buffer (150 mM NaCl, 8 mM Na₂HPO₄, 2 mM NaH₂PO₄, pH 7.4) and frozen overnight at -80°C. cAMP in the supernatant was determined by radioimmunoassay [Steiner et al., 1972]. Protein determination was performed according to the method of Bradford. Cells treated with vehicle alone (no ligand) or forskolin (50 μM) were served as negative and positive controls respectively for the cAMP measurement.

Values are means ± SEM. Student's t-test was performed for comparison of two values. For multiple comparisons, ANOVA was followed by post hoc test (Bonferroni/Dunn method). Statistical significance was considered at p < 0.05.

After termination of the steady state period, all lungs displayed P_PA values in the range between 5 and 7 mmHg.

Continuous infusion of U46619 provoked an increase in P_PA to 23.4 ± 0.9 mmHg within 25 min, with subsequent plateau of the pulmonary hypertension. This level of pulmonary hypertension was then maintained for at least 300 min with variations in P_PA of less than 2 mmHg. Total weight gain at the end of the experiments was 6.5 ± 2.7 g.

As in the preceding group, U46619 infusion for establishing stable pulmonary hypertension was performed. Bolus injection of 200 ng iloprost was followed by an infusion of 33 ng iloprost per hour. This procedure resulted in a significant vasodilatory response with P_PA values decreasing by a mean of 3.3 mmHg. Similarly, the continuous infusion of a more selective IP receptor ligand, PGI₂ also resulted in significant vasodilatory responses (Fig 1). Stable iloprost buffer concentrations in the range of 350–380 pg/ml were documented over the entire perfusion period. A slow increase in P_PA toward pre-infusion values was noted to commence after 100–120 min, and 200 min after beginning of the iloprost infusion, the vasodilatory effect was virtually fully lost. The total weight gain of the lungs was 5.3 ± 3.2 g, and thus did not differ from the groups with mono U46619 infusion.

As shown in Fig. 2A, the intravascular administered PDE3 inhibitor motapizone (dose range 0.01–100 μM) effected a dose-dependent reduction of the elevated P_PA values in lungs with U46619-elicited pulmonary hypertension. There was no shift in the dose response curve when applying identical doses of motapizone 240 min after onset of iloprost infusion. Similar experiments were performed with the PDE4 inhibitor rolipram, which dose-dependently reduced pulmonary artery pressure in U46619 preconstricted lungs, with identical dose-effect curves in the absence and presence of iloprost infusion (Fig. 2B). Total weight gain was 5.7 ± 2.9 g and 5.3 ± 1.9 g (iloprost infusion) in the motapizone treated lungs and 4.6 ± 2.2 g and 5.0 ± 2.6 g (iloprost infusion) in the rolipram treated lungs, respectively.

Inhalation of aerosolized iloprost (total dose 75 ng inhaled over 10 min) resulted in a significant reduction of the U46619-induced pulmonary hypertension, with PAP values decreasing by a mean of 6.3 mmHg (Fig. 3). In separate experiments, iloprost was infused in advance for 240 min, and then nebulization was started. There was virtually no response to inhaled iloprost after this preceding iloprost infusion period. The total weight gain of the lungs was 6.7 ± 2.9 g at the end of experiments.

As shown in Fig. 4, the adenylate cyclase activator forskolin (dose range 10–300 μM) reduced the U46619-elicited pulmonary hypertension in a dose dependant manner. After iloprost infusion for 240 min, an identical dose response curve was observed. Total weight gain was 5.6 ± 2.1 g at the end of the experiments.

As in the preceding groups, U46619 infusion for establishing stable pulmonary hypertension was performed. Subsequently, the PKC inhibitor BIM at a dose of 1 μM was applied alone or in combination with iloprost. In combination group, after 5 min of BIM application, bolus injection of 200 ng iloprost was followed by an infusion of 33 ng iloprost per hour. This procedure resulted in a significant vasodilatory response with P_PA values decreasing by a mean of 4.3 mmHg (Fig. 5A). However, the sole BIM application demonstrated no effects on U46619-elicited pulmonary hypertension (Fig. 5A). The calculated area under the curve (AUC) of the pressure curve was significantly increased as compared to sole iloprost infusion (Fig. 6). The loss of the vasodilatory response to iloprost was somewhat retarded, but not prevented, as compared to single iloprost application. The total weight gain of the lungs was 5.8 ± 2.5 g.

After adjusting stable pulmonary hypertension, the EP1 receptor antagonist AH 6809 at a dose of 3 μM was applied alone or in combination with iloprost, without effecting pulmonary artery pressure. In combination group, after 5 min of AH 6809 application, a bolus injection of 200 ng iloprost was followed by an infusion of 33 ng iloprost per hour. The maximum vasodilatory response to iloprost was 11.4 mmHg (Fig. 5B) and the area under the curve (AUC) of the pressure curve was 305 mmHg × min (p < 0.05 versus sole iloprost infusion) (Fig. 6). The total weight gain of the lungs was 4.8 ± 2.0 g. In contrary, the AH 6809 application alone had no effect on U46619-induced tension (Fig. 5B).

We assayed the desensitization kinetics of iloprost and forskolin in rabbit pulmonary smooth muscle cells (Fig. 7). Incubation of cells with iloprost led to a time-dependent reduction of iloprost induced cAMP formation to 88% (1 h), 71% (2 h), 76% (3 h), 66% (5 h), 54% (7 h) and 26% (24 h) of control. Incubation of cells with iloprost and stimulation of cells with forskolin led to a reduction of cAMP formation to 83% (1 h), 84% (2 h), 91% (3 h), 83% (5 h), 79% (7 h) and 75% (24 h) of control.

Phosphodiesterases (PDE) are enzymes that hydrolyze cyclic AMP and cyclic GMP, the second messengers of PGI₂, and NO 4243. The characterization of the various PDEs currently known has profited from the employment of selective PDE inhibitors. Concerning the lung vasculature, the presence of the PDE isoenzymes 1, 3, 4 and 5 in the cytosolic and particulate phases (homogenized human pulmonary artery tissue) has been demonstrated 44. There is strong evidence of PDE activation and upregulation in response to intracellular elevation of cAMP. It has been shown recently that cyclic AMP upregulates isoforms of PDE4 in airway smooth muscle cells 45, monocytes 46, endothelial cells 47 and Jurkat T-cells 48. In addition, short term activation (phosphorylation) of PDE4 via PKA was demonstrated by treatment of cells with forskolin, a stimulator of adenylate cyclase 49. Furthermore, treatment of rats with 7-oxo-prostacyclin resulted in a significant increase in PDE4 activity, which attenuated the elevation of isoprenaline-induced cAMP levels as well as the contractile force development in a Langedorff-heart preparation 50. Similar results were obtained for PDE3 which has been shown to be regulated on the level of activity 48 and expression 32 by increased cAMP levels. In the current study, we employed motapizone and rolipram, selective inhibitors of PDE3 and PDE4, respectively and performed dose response studies on vasodilation in normal and desensitized (iloprost-treated) isolated rabbit lungs. As reported earlier, these PDE inhibitors are potent pulmonary vasodilators 51 and we found no shift in the dose response curve in desensitized lungs. Thus, there is no evidence of increased PDE activity in iloprost-treated isolated rabbit lungs.

We employed forskolin as direct activator of adenylate cyclase and performed dose response studies on pulmonary vasodilation in isolated rabbit lungs. The response to forskolin in the desensitized (iloprost-treated) organs remained intact (and seemed to be slightly increased), indicating no regulation of adenylate cyclase. Similar results were obtained in rats chronically treated with albuterol in which acetylcholine (ACh)-induced bronchoconstriction was reversible by forskolin, but not by albuterol itself, suggesting that the intracellular site of desensitization is upstream of adenylate cyclase 27.

Moreover, the cAMP production of iloprost-pretreated rabbit pulmonary smooth muscle cells in response to forskolin was also not significantly reduced, except in 24 hours of iloprost-pretreated group (to 75% of control). This is quite contrary to iloprost responses, where a significantly reduced cAMP levels (to 25% of control) was observed, thereby excluding the down-regulation of adenylate cyclase as a possible reason for reduced cAMP formation after continuous stimulation of cells with iloprost. In this context, desensitization of IP receptors by iloprost may possibly explained by its direct influence on the receptor, a mechanism upstream of adenylate cyclases. This involves rapid receptor phosphorylation, which results in uncoupling of receptor-G protein interactions and subsequent receptor internalization. These events result in a diminished response to agonist 26. This may also be due to preferential reduction of domain-bound pool of the cognate Gproteins GsαL and GsαS 52.

On the other hand, iloprost induced vasorelaxation responses may also be via cAMP-independent manner. Consistent with the observation, PGI₂ analogues iloprost were shown to activate MaxiK and K_IR channel and subsequent vasorelaxation through a cAMP-independent, G(s)-protein mediated mechanism in vascular smooth muscle cells 5354. Concomitantly in guinea-pig aorta, cyclic AMP-dependent and -independent pathways were recently proposed to underlie the relaxation induced by the PGI₂ analogues, iloprost and Beraprost 555657. In contrast, in rat tail artery, it is thought that channel phosphorylation by cyclic AMP-dependent PKA is responsible for the MaxiK channel-mediated vascular relaxation due to IP receptor stimulation by iloprost 5859. This is well in line with investigations in human fibroblasts, where cells were challenged at any time point of iloprost pre-treatment with forskolin and fully maintained their cAMP response 30. In contrast, Sobolewski et al., demonstrated in rat pulmonary smooth muscle cells (PSMCs) which were exposed to cicaprost, another PGI₂ analogue, a significant reduction in forskolin-induced cAMP production 60 as well as a decrease in adenylate cyclase isoform 2 and 5/6 expression after one hour of exposure to this prostanoid. However, both isoforms were expressed at control levels when PSMCs were incubated for longer time periods with cicaprost. It is possible that the agonists employed, as well as species differences are responsible for these conflicting data.

It is well known that iloprost, the clinically approved compound for treatment of thromboangiitis obliterans 20 and pulmonary arterial hypertension 5, is not highly selective for the IP receptor but also activates the EP1_{receptor which is coupled to Gq and signals via IP3 generation and increased intracellular calcium (Ca²⁺) 61. To investigate the effects of possible involvement of the EP1receptor we employed the EP1receptor antagonist AH 6809 and found a dramatic increase in the efficacy of intravenous iloprost. In contrary, the AH 6809 application alone had no effect on U46619-induced tension. Thus, the agonistic effects on EP1receptor explain the differences between iloprost and selective IP agonists in inhibition of smooth muscle cell response 2640.}

There is evidence that the initial desensitization process is dependent on phosphorylation of the IP receptor by PKC 4162. After phosphorylation, the IP receptor is sequestered to clathrin-coated vesicles and by a dynamin-dependant process internalized into endosomes 41. This receptor sequestration is, unlike desensitization, PKC-independent 27 and a time of approximately 18 h is needed for a resensitization in human fibroblasts 2730. However, several studies including our work employed inhibitors of PKC (GF-109203X, BIM) and could not or could only partially prevent receptor desensitization, suggesting a minor role of PKC in the desensitization process.

In conclusion, this study demonstrates for the first time in an intact lung model that the PGI₂ analogue iloprost caused desensitization of the IP receptor within 2 to 3 hours, resulting in complete loss of responsiveness after this time period. Desensitization was not due to increased PDE activity or loss of adenylate cyclase activity, but may involve PKC function and co-stimulation of the EP1 receptor in addition to the IP receptor by this PGI₂ analogue.