INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Pulmonary ischemia-reperfusion (I/R) injury is characterized by increased vascular resistance and permeability and edema formation (1). Numerous studies demonstrated that inhibition of oxygen radicals or neutrophil-endothelial adherence prevented the I/R lung injury, indicating that toxic oxygen metabolites and/or neutrophil activation mediates I/R-induced lung injury (1). Reactive oxygen species or activated neutrophils increase pulmonary artery pressure and stimulate lipid peroxidation of the cell membrane and increase microvascular membrane permeability (11). The mechanisms by which reactive oxygen species or activated neutrophils alter endothelial barrier function are not well understood, but they probably involve, at least in part, the modulation of phospholipase activities such as phospholipase A₂, C, and D, resulting in altered signaling pathways (12). There is a substantial body of evidence which indicates that the I/R of the lungs increases the synthesis of cyclooxygenase metabolites during reperfusion in intact animals and isolated perfused preparations, suggesting that cyclooxygenase metabolites of arachidonic acid are involved in I/R lung injury (1, 9, 10, 13). However, the role of cyclooxygenase products in I/R-induced lung vascular injury has not been extensively investigated, although thromboxane and prostaglandin I₂ (PGI₂) may increase the pulmonary microvascular permeability to fluid and protein (14) and enhance the susceptibility to lung edema formation in intact animals and isolated perfused lungs (15).

Furthermore, cytokines may function as mediators of I/R lung injury. A recent study reported that pulmonary artery occlusion and reperfusion resulted in transient generation of plasma tumor necrosis factor- (TNF-) in anesthetized rabbit lungs, which might mediate neutrophil sequestration in the lungs at the onset of reperfusion (18).

Alternatively, agents that increase intracellular adenosine 3',5'-cyclic monophosphate (cAMP) levels such as -adrenergic agonist, activator of adenylate cyclase, phosphodiesterase inhibitor, and cAMP analogue, have been shown to prevent or reverse the increase in vascular permeability associated with I/R to isolated rat and rabbit lungs (6, 7). Although the mechanism by which cAMP enhances the endothelial barrier integrity is uncertain, cAMP-dependent protein kinase may regulate paracellular permeability by relaxing the endothelial cells through the phosphorylation and inhibition of myosin light chain kinase activity (8). However, conflicting evidence exists regarding whether vasodilator prostaglandin, PGI₂, inhibits the lung vascular injury associated with I/R (19).

This study was designed to assess the effects of low-dose Beraprost sodium (BPS; Kaken Pharmaceutical Co., Tokyo, Japan), a stable PGI₂ analogue, on I/R-induced pulmonary hemodynamic changes and vascular injury when given just before reperfusion. BPS is a synthetic PGI₂ analogue that is chemically and metabolically more stable than the native compound and has the properties of vasodilation and inhibition of leukocyte activation and platelet aggregation (23, 24). Lung vascular injury was evaluated by the microvascular filtration coefficient (K_f), a sensitive index of vascular permeability. In addition, we tested the efficacy of low-dose BPS treatment on plasma levels of cyclooxygenase products and TNF- as proinflammatory mediators in I/R rabbit lungs (10, 18).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In Situ Perfused Rabbit Lung Preparation

Male Japanese white rabbits weighing 1.8 to 2.2 kg were given 3,000 U of heparin sodium via an ear vein and anesthetized with pentobarbital sodium (25 mg/kg). The chest was opened, and the animal was rapidly exsanguinated from the left ventricle. After right and left parasternal incisions were made, both lungs were fully exposed and plastic cannulas were placed in the main pulmonary artery and the left atrium. A tracheostomy was performed and the lungs were mechanically ventilated with room air at 20 breaths/min (AR-300; Acoma, Tokyo). The tidal volume was set to achieve a peak airway pressure of 10 mm Hg and the end-expiratory pressure was set at zero mm Hg. The lungs were perfused in situ at a constant flow of 40 ml/min and covered with Saran Wrap to prevent evaporation. The closed perfusion system included a roller pump (Mera HAD-100; Senko Ika Kogyo, Tokyo), a heat exchanger (Mera MHE-3-P), a bubble trap and filter (CX-BT15; Terumo, Tokyo), an electromagnetic blood-flow meter (MFV-3100; Nihon Kohden, Tokyo) and a reservoir. Changes in lung weight were continuously recorded as the converse of the change in the weight of the perfusate reservoir. The total volume of the perfusion system was 400 ml. The perfusion medium was 330 ml of Krebs-Henseleit buffer containing 70 ml of autologous blood and it was continuously mixed with a magnetic stirrer.

Autologous blood diluted with Krebs-Henseleit buffer was used as a perfusate, because studies using isolated blood-perfused rabbit lungs with a hematocrit level  10% may be complicated by spontaneous pulmonary hypertension (25). The diluted autologous blood contained a total protein content of only 0.5 to 0.8%, but exogenous serum albumin was not added to the perfusion medium to minimize osmotic buffering during the measurement of K_f.

Experimental Protocols

The experimental protocol is depicted in Figure 1. After the recirculating flow was established, the lungs were perfused for 10 min to ensure an isogravimetric state. If the rate of lung weight gain was more than 0.2 g/min, the lungs were discarded and not used in this study. Thirty rabbit lungs were divided into six groups (n = 5 each): (1) nonischemic control (control group), (2) I/R + saline (vehicle, saline group), (3) I/R + 3 pmol/L of BPS (BPS-1 group), (4) I/R + 150 to 300 pmol/L of BPS (BPS-2 group), (5) I/R + 900 pmol/L of BPS (BPS-3 group), and (6) I/R + 60 µmol/L of indomethacin (IND group). We used relatively low doses of BPS (3 to 900 pmol/L), because an intravenous injection of BPS (300 pmol/L or more) dose- dependently decreased systemic arterial pressure in anesthetized dogs and rats (23) and because high doses of PGI₂ may be limited for clinical application owing to systemic hypotension (26). In the control group, the lungs were perfused and ventilated for 1 h without interrupted perfusion or drug administration. I/R lungs were subjected to 2 h of ischemia by occluding the perfusion cannulas and stopping ventilation. The perfusion circuit included the bypass tubing interposed between the arterial and venous cannulas. So, we opened the bypass and directed flow directly from the arterial to the venous cannulas, thereby stopping inflow into the lungs. Thus, blood continued to recirculate in the perfusion circuit, although the lungs received no blood flow. Throughout the 2 h of ischemia the lungs were kept at room temperature. Ten minutes before the completion of ischemia, either an equivalent volume of physiological saline (vehicle), low doses of BPS or IND was added to the reservoir. Then, the ischemic lungs were reperfused and reventilated for 1 h.

View larger version (18K): [in this window] [in a new window]   Figure 1.   Experimental protocols. Lungs perfused for 1 h not subjected to ischemia served as a nonischemic control (control group). Ischemia-reperfusion lungs were subjected to a 2-h period with no ventilation and flow followed by reinstitution of ventilation and flow for 1 h. After 2 h of ischemia, saline as a vehicle (saline group), 3 pmol/L of Beraprost sodium (BPS-1 group), 150 to 300 pmol/L of Beraprost sodium (BPS-2 group) 900 pmol/L of Beraprost sodium (BPS-3 group), or 60 µmol/L of indomethacin (IND group) was added into the reservoir 10 min before reperfusion and reventilation. Vascular pressures and changes in reservoir weight were continuously monitored on a polygraph recorder. Kf and myeloperoxidase (MPO) activity in lung tissue were measured at the end of the reperfusion. Asterisks indicate the time of perfusate sampling.

Measurements of Vascular Pressures and Resistances

Pulmonary arterial (Ppa), left atrial (Pla), and airway pressures were continuously monitored by pressure transducers (UK215; Baxter, CA) referenced to the top of the lung and recorded on a multichannel recorder (RMC-1100; Nihon Kohden, Tokyo). Pulmonary capillary pressure (Pdo) was estimated using the double-occlusion method. During simultaneous occlusion of both inflow and outflow cannulas, Ppa and Pla rapidly equilibrate to Pdo, which has been shown to be an excellent estimate of the existing capillary pressure (2, 4). Double occlusion was performed rapidly and maintained for 3 to 4 s. The respirator was turned off at end of expiration before each occlusion. Because the flow for a given lung was constant, changes in Ppa were proportional to changes in total pulmonary vascular resistance (Rt). The Rt was partitioned into arterial (Ra) and venous (Rv) resistances by the Pdo. Measurements of the Pdo, Rt, Ra, and Rv were obtained at baseline and at each end of the reperfusion.

Determination of the Microvascular Filtration Coefficient

K_f was used as an index of endothelial permeability. The K_f was determined from the rate of lung weight gain induced by the elevation of left atrial pressure to 10 to 13 mm Hg over a 10-min period. The rate of weight change during the 5- to 10-min interval was plotted on a semilogarithmic scale as a function of time and extrapolated to time zero. The K_f was calculated by dividing the time 0 filtration rate by the change in Pdo and was expressed in ml/min/mm Hg /100 g wet lung (2).

Measurements of the Effluent Perfusate Concentrations of Thromboxane B₂, 6-Keto-Prostaglandin F₁, TNF-, and cAMP

Plasma cyclooxygenase products and TNF- have been shown to increase in the reperfused lungs as a proinflammatory mediator (10, 18, 27). Thus, effluent perfusate concentrations of thromboxane B₂ (TxB₂) and 6-keto-prostaglandin F₁ (6-keto-PGF₁), stable metabolites of thromboxane A₂ and PGI₂, and TNF- were measured during the baseline and after reperfusion by radioimmunoassay using commercial kits (New England Nuclear, Boston, MA and Otsuka Pharmaceutical Co., Tokyo, Japan). Moreover, several recent studies demonstrated that vascular endothelial permeability is regulated by the intracellular adenosine 3',5'-cyclic monophosphate (cAMP) content (6). Thus, the perfusate concentration of cAMP was also measured during the baseline and after reperfusion using a commercially available kit (Yamasa, Tokyo, Japan). The detection limits of assays for TxB₂, 6-keto-PGF₁, TNF-, and cAMP were < 3.0 pg/ml, < 3.0 pg/ ml, < 5.0 pg/ml, and < 1.6 pmol/ml, respectively.

Lung Tissue Myeloperoxidase Assay

In order to ascertain whether BPS influences the pulmonary leukostasis during reperfusion, we measured the myeloperoxidase activity of the lung tissue at the end of reperfusion. Myeloperoxidase was extracted from the right lung tissues as described (18). Myeloperoxidase activity was determined at 460 nm for 3 min using a standard spectrophotometric technique and expressed as absorbance changes per gram of lung tissue.

Lung Histology

To obtain qualitative information about the histological effect of BPS, the middle portion of the right lower lobe was taken at the end of each experiment. Tissue sections were fixed with 10% formaldehyde, stained with hematoxylin-eosin, and then processed for light microscopy.

Statistics

Data are expressed as means ± SE. Differences among the groups were analyzed by one-way analysis of variance (ANOVA). If the F ratio indicated a statistical difference among the groups, the Dunnett test was used to compare between-group means. Within-group comparisons were analyzed by ANOVA for repeated measurements followed by the Dunnett test. A p value < 0.05 was accepted as indicating statistical significance.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Effects of Low-Dose BPS on Vascular Pressures and Resistances

The Ppa remained stable over the 1 h of perfusion in nonischemic control lungs (Figure 2). Transient pulmonary hypertension at the onset of reperfusion was found in all five ischemic groups with a maximum Ppa reached within the first 3 min of reperfusion. This rise in Ppa was followed by a gradual decline toward baseline with time and Ppa increased again after 30 min in the saline group (Figure 2). The rise in Ppa observed during the reperfusion period was significantly smaller in the three groups in which BPS-2, BPS-3, and IND were added. At the end of reperfusion, there were no significant differences in Ra and Rv among BPS-2, BPS-3, and IND groups (Figure 3), suggesting that the rise in Ppa during reperfusion is from pulmonary vasoconstriction caused by thromboxane. The BPS-1 did not affect the bimodal pattern of changes in Ppa associated with I/R.

View larger version (31K): [in this window] [in a new window]   Figure 2.   Pulmonary artery pressure during the reperfusion period. Bimodal pattern of pulmonary artery pressure was observed in saline group. The pressor responses during reperfusion were attenuated in the BPS-2, BPS-3, and IND groups. Open circles: nonischemic control group; closed circles: saline group; open triangles: BPS-1 group; closed squares: BPS-2 group; closed triangles: BPS-3 group; open squares: IND group. Values are means ± SE. *p < 0.05 versus control.

View larger version (18K): [in this window] [in a new window]   Figure 3.   Effects of Beraprost sodium on arterial and venous resistances after 1 h of reperfusion. Open columns: arterial resistance; closed columns: venous resistance. Arterial and venous resistances were increased in the saline group. The increases in vascular resistances were attenuated in the BPS-2, BPS-3, and IND groups. Values are means ± SE. *p < 0.05 versus control, BPS-2, BPS-3, and IND groups.

Effects of Low-Dose BPS on K_f

The K_f significantly increased in the saline group compared with that of the nonischemic control group, indicating that 2 h of total lung ischemia followed by 1 h of reperfusion causes an increase in vascular permeability (Table 1). The addition of BPS-3 and IND attenuated the increase in K_f associated with I/R. However, the addition of BPS-1 and BPS-2 to the perfusate tended to increase the K_f.

Effects of Low-Dose BPS on Effluent Perfusate Concentrations of TxB₂, 6-Keto-PGF₁, TNF-, and cAMP

There were no significant differences in the effluent perfusate concentrations of TxB₂ and 6-keto-PGF₁ during baseline among the groups (Table 1). Figure 4 shows the percent changes of perfusate TxB₂ and 6-keto-PGF₁ from respective baseline value in six groups. Percent changes of perfusate TxB₂ increased significantly after reperfusion in the saline group compared with the nonischemic control group, indicating that thromboxane increases in response to transient total lung ischemia followed by reperfusion. However, percent changes of perfusate 6-keto-PGF₁ did not differ among the groups (Figure 4). There was a linear relationship between the K_f (Y) and the TxB₂ (X) released after 1 h of reperfusion (Y = 1.01 + 0.00256X, r = 0.85, p < 0.05) (Figure 5). This means that the physiological marker of vascular permeability was correlated with evidence for formation of arachidonic mediators in the reperfused lungs, and that thromboxane may play a critical role in the pathogenesis responsible for increased vascular permeability due to I/R. In contrast, plasma TNF- was less than 5 pg/ml in all the lungs at the baseline and at the end of reperfusion, suggesting that TNF- is not involved in this model of I/R lung injury.

View larger version (16K): [in this window] [in a new window]   Figure 4.   Effluent perfusate thromboxane B2 and 6-keto-PGF1 after 1 h of reperfusion. Open columns: thromboxane B2; closed columns: 6-keto-PGF1. Perfusate concentrations of thromboxane B2 and 6-keto-PGF1 were expressed as percentage of baseline values. Thromboxane B2 increased significantly in the saline group, compared with the nonischemic control, BPS-3, and IND groups. Values are means ± SE. *p < 0.05 versus control, BPS-3, and IND groups.

View larger version (19K): [in this window] [in a new window]   Figure 5.   A relationship between the microvascular filtration coefficient and thromboxane B2 after reperfusion. The microvascular filtration coefficient was linearly correlated with perfusate concentration of thromboxane B2. Values are means ± SE.

There were no significant differences in the effluent perfusate concentrations of cAMP during baseline among the groups (Table 1). The I/R caused approximately a 3-fold decrease in the level of cAMP. After reperfusion, BPS-3 significantly increased the perfusate concentration of cAMP, whereas IND decreased the perfusate cAMP concentration, suggesting that BPS may require a threshold dose of about 1 nmol/L to elevate the perfusate cAMP levels in the pulmonary circulation and that BPS may enhance the endothelial cell integrity at doses capable of elevating the intracellular cAMP content.

Lung Tissue Myeloperoxidase

There were no significant differences in the lung tissue myeloperoxidase activities among the six groups (p = 0.61, Table 1). However, this does not mean that 900 pmol/L of BPS attenuates the K_f without inhibiting leukocyte sequestration, since lung tissue myeloperoxidase activity did not always follow the damage and protection, as demonstrated by Moore and coworkers (5).

Lung Histology

Histological examination after reperfusion demonstrated the swelling in the alveolar septum (Figure 6). In contrast, the alveolar septum was less edematous in the BPS-3 and IND groups than in the saline group.

View larger version (97K): [in this window] [in a new window]   Figure 6.   Light micrographs demonstrating structural alteration of alveolar septum. (A) Nonischemic control lungs; (B) lungs subjected to ischemia-reperfusion in the presence of vehicle (saline group); (C ) lungs subjected to ischemia-reperfusion in the presence of 900 pmol/L of Beraprost sodium (BPS-3 group); (D) lungs subjected to ischemia-reperfusion in the presence of 60 µmol/L of indomethacin (IND group). Original magnification ×200.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Arachidonic Mediators in I/R-induced Lung Vascular Injury

A number of investigations have demonstrated that the restoration of blood flow to an ischemic lung causes increased vascular resistance and microvascular leakage of liquid and protein (1). Our results are consistent with previous studies which showed that 2 h of ischemia followed by 1 h of reperfusion resulted in significant increases in the vascular resistance and permeability compared with nonischemic lungs in isolated blood-perfused rabbit lungs (7). Ljungman and coworkers demonstrated in isolated rat lungs perfused with a physiological salt-Ficoll solution that IND and flubiprofen, cyclooxygenase inhibitors, and U 63557A, a thromboxane synthase inhibitor, given before an ischemic insult did not reduce the elevation of Ppa after reperfusion, but significantly inhibited the accumulations of ¹²⁵I-BSA in the lung parenchyma and bronchoalveolar lavage fluid, and inhibited the increase in the wet lung weight associated with untreated I/R lungs (10). This study showed that 60 µmol/L of IND, given just before reperfusion, markedly attenuated I/R-induced changes in the Ra, Rv, K_f, and the production of thromboxane, although protection by IND was not complete. Furthermore, a significant positive linear correlation was observed between the K_f and TxB₂ in all lungs, suggesting that thromboxane may have contributed, at least in part, to the pathogenesis of increased vascular permeability following I/R. Because IND was administered after a period of ischemia, its protective effect appeared to be related to injury associated with reperfusion. Other studies also found that mediators derived from arachidonic acid were of primary importance in oxidant-induced lung injury (11). Alternatively, some investigators have shown that plasma TNF- increased with I/R in anesthetized rabbit lungs (18), and it has been postulated to be important in I/R lung injury (27). However, it is unlikely that TNF- may have been involved in this model of reperfusion lung injury, because perfusate TNF- was within the detection limit (5 pg/ml) in all lungs.

Effects of BPS on I/R-induced Lung Vascular Injury

Agents that increase intracellular cAMP concentrations have been shown to prevent or reverse the increase in endothelial permeability in I/R lung injury to isolated rat and rabbit lungs (6). Alternatively, it has been shown that these agents protect against the pulmonary capillary injury in a variety of species and injury models of increased-permeability lung edema, treated with tert-butyl hydroperoxide, instillation of hydrochloric acid into the trachea, thrombin, intravenous air embolism, or endotoxin (28). Furthermore, Imai-Sasaki and coworkers reported that pretreatment with 30 to 1,000 nM of BPS inhibited the thrombin-induced increase in fluorescein isothiocyanate (FITC)-albumin permeability in cultured human umbilical vein endothelial cells (33). However, the vasodilator prostaglandin, PGI₂ may promote or inhibit the development of lung edema despite acting via the common second messenger of cAMP, because PGI₂ may increase blood flow in the injured area, probably by inhibition of thromboxane-induced vasoconstriction. We hypothesized that increasing intracellular cAMP concentrations in the vascular smooth muscle would promote edema via increased vascular surface area and blood flow, but increasing intracellular cAMP concentrations in the endothelium would suppress edema by enhancing the permeability barrier. Therefore, the tissue selectivity and the degree of the cAMP-elevating effect may be important in protecting or reversing the lung vascular injury. PGI₂ has been shown to increase intracellular cAMP levels, thereby leading to vasodilation, antiplatelet aggregation, anti-white cell adhesion. However, the contribution of PGI₂ to I/R-induced lung injury is still poorly understood. Okuda and colleagues demonstrated that pretreatment with OP-2507, a stable PGI₂ analogue, attenuated the increase in vascular resistance and lung weight gain associated with I/R in physiological salt solution/Ficoll perfused rat lungs (19). Hooper and coworkers reported that 30 ng/kg/min of PGI₂ infusion had a protective effect on decreased oxygenation, increased vascular resistance, and accumulation of neutrophils in the alveolar space in an acute canine model of ischemic injury (20). In contrast, Matsuzaki and coworkers reported in anesthetized rabbits with bilateral lung ischemia that PGI₂ did not prevent the malfunction in pulmonary gas exchange or the increase in the wet/dry weight ratio (21). Hooper and coworkers failed to demonstrate improved lung preservation with a single flush of Iloprost, a stable PGI₂ analogue (22).

In our I/R lung model, 900 pmol/L of BPS prevented increases in Ppa, K_f, and thromboxane production during reperfusion probably via the elevation of cAMP in both the vascular smooth muscle and endothelium. However, 150 to 300 pmol/L of BPS inhibited the increase in Ppa, but did not attenuate the increases in K_f and thromboxane production. The lack of attenuation of increased endothelial permeability with 150 to 300 pmol/L of BPS may be explained by the inability to increase endothelial cAMP concentrations. The deleterious vasodilatory effects of PGI₂ on the development of pulmonary edema have been reported in previous animals studies (15- 17). Krausz and colleagues reported that 1 µg/min PGI₂ infusion for 5 h caused significantly higher lung weight gain and wet-to-dry weight ratio than in the untreated controls in isolated perfused canine lung lobes (15). Ogletree reported that 1.25 µg/kg/min PGI₂ infusion increased the lung lymph flow associated with a tendency for the lymph-to-plasma protein concentration ratio to rise under a condition of mechanically increased left atrial pressure (16). Yoshimura and coworkers demonstrated that PGI₂ augmented the lung edema formation induced by thromboxane (17). Therefore, possible explanations for our results are that 150 to 300 pmol/L of BPS does not affect the hydraulic conductivity, but increases blood flow in the injured area by inhibiting thromboxane-induced vasoconstriction, leading to an increase in K_f.

Study Limitations

A potential mechanism for the beneficial effect of 900 pmol/L of BPS is not clear, because we did not establish the causal relationships, such as interaction of endothelial cell with circulating hormones, cytokines, and inflammatory mediators released during I/R. However, the protective mechanisms of BPS appear to be the attenuation of lipid peroxidation- induced release of arachidonic acid as well as the direct antipermeability effect of cAMP. Previous studies indicated that oxidants could stimulate the generation of cyclooxygenase and lipoxygenase products (11) and that cAMP-elevating drugs could inhibit the production of arachidonic mediators in several different cells (34). Although cyclooxygenase products can be generated and released from a number of different cell types including neutrophils, monocytes, platelets, vascular endothelium, and other parenchymal cells when stimulated, the major source of thromboxane in this study was not identified.

Alternatively, there were no significant differences in the lung tissue myeloperoxidase activity among the six groups. However, Seibert and coworkers demonstrated that the isolated buffer-perfused lung contained a substantial number of leukocytes, and that these leukocytes contributed to I/R lung injury (4). Furthermore, Moore and coworkers showed that the damage or protection of I/R lung injury was not always correlated to lung tissue myeloperoxidase activity (5). Thus, the role of BPS on "activated endogenous leukocytes" in the pathogenesis of I/R lung injury cannot be ruled out in this study.

Finally, K_f was used as an accurate measure of endothelial damage. However, K_f includes the hydraulic conductivity and vascular surface area. Thus, we may have overestimated or underestimated the effects of BPS or indomethacin on I/R lung injury by the increase or decrease of vascular surface area.

Clinical Implications

I/R lung injury is implicated in several clinical situations, such as pulmonary embolism, collapsed lung from pneumothorax or pleural effusion, acute respiratory distress syndrome, cardiopulmonary bypass, and heart-lung transplantation. In these clinical situations, oxygen radicals formed after reperfusion may damage tissue directly and initiate a cascade of events including neutrophil recruitment and activation, and release of proinflammatory mediators such as arachidonic acid metabolites and TNF-, and additional tissue damage may be induced. Previous animal studies noted that cyclooxygenase inhibitors had a beneficial effect on I/R lung injury (10, 13), probably by inhibiting synthesis or release of thromboxane. In contrast, conflicting evidence exists regarding whether PGI₂ has a protective effect on acute lung injury, including I/R injury (19, 26, 33). PGI₂ or PGI₂ analogue such as BPS is capable of elevating intracellular cAMP in both vascular smooth muscle and endothelial cells. In the studies of cultured bovine aorta, however, stimulatory effects of PGI₂ or PGI₂ analogue on cAMP content were greater in the vascular smooth muscle cells than in the endothelial cells (35). In fact, we observed that 150 to 300 pmol/L of BPS had a "deleterious" vasodilatory action without inhibiting production of thromboxane in I/R injury, although BPS-induced increase in endothelial cAMP may be downregulated by proinflammatory mediators. Furthermore, basic pharmacological studies on BPS indicated that an intravenous injection of BPS (300 pmol/L or more) dose-dependently decreased the systemic arterial pressure in anesthetized dogs and rats (23). Therefore, this study suggests that an optimal low dose of BPS may offer a potential strategy for prevention or treatment of I/R-induced lung vascular injury by the antipermeability effect of cAMP in certain circumstances. However, vasodilatory action of BPS may exacerbate the reperfused lung injury by increasing the flow through injured capillaries via inhibition of thromboxane- induced vasoconstriction.