INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

HYPERACUTE REJECTION (HAR) of pig organs by humans is generally believed to be caused primarily by activation of human complement by a "natural" antibody bound to porcine endothelium (37). Supporting this view is the fact that pig hearts and kidneys can sustain the life of primates when complement activation is prevented with soluble complement receptor type 1, preempted using depletion with cobra venom factor, or downregulated as a consequence of expression of human complement regulatory proteins (HCRPs) in organs from transgenic pigs (7, 20, 25, 33, 44).

Among various organs from HCRP-transgenic pigs studied to date, porcine lung is poorly protected from acute injury during ex vivo perfusion with human blood and when transplanted into baboons (31, 43). Furthermore, neither absorption of antispecies antibody nor treatment with soluble complement receptor type 1 (R. Pierson, unpublished observations) fully protects the lung from injury, whereas hearts and kidneys function well in analogous systems after similar manipulations of the complement activation pathway (30, 32, 33). Lung HAR (HALR) is characterized by rapid, marked elevation in pulmonary vascular resistance (PVR) and profuse capillary leak (30). The elevation in PVR observed with acute lung injury by cobra venom factor or endotoxin and after intravenous administration of xenogeneic blood in sheep is associated with thromboxane A₂ (TxA₂) elaboration; TxA₂ has also been implicated as a direct cause of increased capillary permeability (2, 10, 15, 22, 26, 39, 43). These considerations led us to investigate thromboxane's contribution to elevated PVR, increased vascular permeability, and elaboration of inflammatory mediators during HAR of pig lung by human blood. Furthermore, we investigated the primary cellular source of thromboxane in this model.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Ex vivo lung perfusion model. We used an ex vivo perfusion system to model HAR of pig lung by human blood (3) (Fig. 1). Median sternotomy was performed in juvenile piglets (3-5 kg) under general endotracheal anesthesia with isofluorane. After heparinization (1,000 U/kg) and adequate circulation time for administered drugs, two flanged stainless steel cannulas (3.5 mm inner diameter) were secured with ligatures: one in the pulmonary artery and the second introduced through the left ventricular apex and across the mitral valve, with its tip in the left atrium. The ascending aorta and superior vena cava were ligated, and bronchial collateral flow was recovered from the right atrium via thin-walled plastic tubing introduced through the inferior vena cava. In all experiments, residual pig blood was gravity flushed from the lungs using 100 ml of room temperature 5% human albumin in 0.9% saline administered through the pulmonary arterial cannula. The lungs were then connected to the perfusion circuit via the pulmonary arterial and left atrial cannulas (Fig. 1). Lung warm ischemic times were typically <15 min.

View larger version (16K): [in this window] [in a new window]   Fig. 1.   Ex vivo lung perfusion apparatus. The roller pump is set at 100 ml · min1 · kg1 piglet wt. Peak pulmonary arterial perfusion pressure is limited by the height of the overflow tubing (O); under conditions of increased pulmonary vascular resistance (PVR), pulmonary arterial perfusion pressure is limited to 50 cmH2O. During periods of low pulmonary blood flow, reservoir blood is continually warmed and recirculated through the "pop off" (P). Transpulmonary blood flow and pulmonary arterial, left atrial, and airway pressures are monitored continuously (not shown).

1

1

1

1

1

Perfusate preparation. Human blood was collected by trained phlebotomists from volunteer donors in the Vanderbilt Clinical Research Center, in accordance with protocols approved by the Vanderbilt Committee for the Protection of Human Subjects. Approximately 400 ml of fresh human blood from one donor were collected into 10,000 USP IU heparin (Elkins-Sinn, Cherry Hill, NJ) and expanded with ~180 ml of blood-type matched fresh-frozen plasma, which was heparinized (5,000 IU) before being recalcified with 450 mg calcium chloride [to neutralize CPDA-1 (citrate, phosphate, dextrose, adenine) anticoagulant and restore ionized calcium levels to the normal range].

Experimental groups. Positive control (heterologous) experiments were performed with unmodified fresh human blood; negative controls consisted of isolated piglet lungs perfused with autologous pig blood. OKY-046 [10 mg (2-3.3 mg/kg); Kissei Pharmaceutical, Matsumoto, Japan], a TxA₂ synthase inhibitor, was administered intravenously to the piglet 5 min before surgical isolation of the lungs; 10 mg (16 µg/ml) were added as a bolus to the human blood perfusate 5 min before initiation of lung perfusion (n = 5). Interaction between thromboxane and its receptor was prevented by SQ-30741, a competitive TxA₂-receptor antagonist [5 µg (1-1.6 µg/kg) to the piglet and 6.5 µg to the blood perfusate; n = 5; Squibb and Sons, Princeton, NJ]. Papaverine hydrochloride [15 mg (3-5 mg/kg) to the piglet and 30 mg (50 µg/ml) to the blood; YorPharm, Buffalo Grove, IL], a smooth muscle relaxant, was employed to prevent vasoconstriction by a mechanism that does not affect prostaglandin endoperoxide metabolism (n = 6) (27). Doses of OKY-046, SQ-30741, and papaverine added to blood were chosen based on literature review, extensive experience in endotoxic shock models (R. E. Parker), and limited pilot experiments. Doses administered to piglets were chosen based on similar considerations.

Eicosanoid quantitation. Samples for eicosanoid determination were collected with meclofenamate and EDTA from the precirculated reservoir (time 0) and pulmonary venous effluent at 1, 5, 10, 20, 30, and 60 min and hourly thereafter until a survival endpoint was met. Blood samples (1 ml/sample) were promptly placed on ice and centrifuged at 4°C, and plasma was stored at 70°C. Eicosanoid measurements in all experiments were conducted by stable isotope dilution gas chromatography-mass spectrometry. Plasma TxA₂ and prostacyclin production were assayed by measurement of their stable metabolites, TxB₂ and 6-keto-PGF₁ (8). Respective tetradeuterated internal standards (Cayman Chemical) were added to each sample. Samples for TxB₂ analysis were acidified to pH 3 and extracted in ethyl acetate. Samples for 6-keto-PGF₁ analysis were acidified to pH 3 and applied to a Sep-Pak C₁₈ (Waters, Milford, MA), washed with H₂O (pH 3), and eluted in ethyl acetate. Samples for both analyses were concentrated under vacuum (Speed-Vac, Savant, Farmingdale, NY) and incubated overnight at 25°C in 0.5% methoxyamine · HCl (Sigma Chemical, St. Louis, MO) in pyridine. The 6-keto-PGF₁ samples were further purified via thin-layer chromatography (Whatman LK6DF silica obtained from VWR, Oak Ridge, TN) in the organic portion of H₂O-ethyl acetate-hexane-acetic acid (7.5:6.75:3.12:1.5 vol/vol/vol/vol) and extracted in methanol. Both TxB₂ and 6-keto-PGF₁ samples were concentrated under vacuum and treated with acetonitrile, pentafluorobenzyl bromide (Sigma Chemical), and diisopropylethylamine (Aldrich Chemical, Milwaukee, WI) (4:3:2 vol/vol/vol) for 30 min at 37°C. The methoxime-pentafluorobenzyl esters were then concentrated and purified by thin-layer chromatography with a mobile phase of ethyl acetate-methanol (98:2 vol/vol). Derivitization of each sample was completed by trimethylsilation with bis (trimethylsilyl)trifluoroacetamide (Supelco, Bellefonte, PA) in pyridine. Gas chromatography-mass spectrometry analyses were conducted on a 15-m fused silica capillary column SPB-1 in a Varian Vista 6000 gas chromatograph coupled to a Nermag R10-10 mass spectrometer operated in the negative ion chemical ionization mode. Selected ions monitored included mass-to-charge ratio of 614/618. Eicosanoid levels were normalized to internal standard and expressed as nanograms per milliliter human plasma.

Histology. Lung biopsies for histology were gently inflated with dilute Tissue-Tek optimal cutting temperature compound (Sakura, Tokyo, Japan) instilled into a visible airway, snap frozen in liquid nitrogen, and stored at 80°C until study. Twelve-micrometer cryostat sections were employed for staining with either a Cy3-conjugated anti-COX-2 monoclonal antibody (no. 160112, Cayman Chemical) or hematoxylin and eosin after in situ terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling assay (Boehringer Mannheim, following manufacturer's suggested methods) for detection of apoptotic PIMs after Clodronate pretreatment. PIMs were identified in hematoxylin-and-eosin-stained sections of lung of a piglet killed minutes after intravenous injection of a 1% solution of Monastral blue pigment in 0.9% saline; this lung was not exposed to heterologous or other perfusion.

Histamine, tumor necrosis factor, and C3a assay. Sera stored in EDTA were assayed by commercial ELISA [histamine: Immunotech, Marseille, France; tumor necrosis factor (TNF)-: R&D Systems, Minneapolis, MN; and C3a: Quidel, San Diego, CA] according to the manufacturer's instructions.

Animal care and use. All experiments were conducted under protocols approved by the Vanderbilt Animal Care and Use Committee, and in accordance with guidelines from the "Guide for the Care and Use of Laboratory Animals" [DHEW Publication No. (NIH) 85-23, Revised 1985, Office of Science and Health Reports, DRR/NIH, Bethesda, MD 20205].

Statistical analyses. All values are expressed as means ± SE. Comparisons within groups were conducted by Wilcoxon signed rank test, and those among groups were conducted by a Mann-Whitney rank sum test, except where otherwise noted. Statistical significance was defined as P < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Thromboxane's role in HALR. Porcine lungs perfused under physiological conditions with fresh human blood (heterologous blood, positive controls) exhibited a rapid fall in transpulmonary blood flow within the first minutes of perfusion, associated with a rise in PVR from 0.07 ± 0.01 cmH₂O · ml¹ · min at baseline to 5.6 ± 1.3 cmH₂O · ml¹ · min at 10 min (Fig. 2A). Lungs in this group met the loss of flow survival endpoint at 10.0 ± 2.7 min (Fig. 2B) and often exhibited tracheal edema as well when ventilation was subsequently discontinued. In contrast, lungs perfused with autologous blood universally survived with preserved gas exchange and low airway pressures to the 240-min endpoint, demonstrating that the physiological perturbations observed with human blood are not an artifact of the experimental preparation (Fig. 2B). In lungs perfused with unmodified human blood, TxB₂ levels in pulmonary vein effluent rose to 22.8 ± 3.9 ng/ml at 1 min (not shown) and plateaued at 44.4 ± 7.9 ng/ml at 10 min (P < 0.05 vs. autologous controls, <1.5 ng/ml at 10 min) (Fig. 2C). Administration of a competitive inhibitor of thromboxane synthase (OKY-046) or blockade of the thromboxane receptor (SQ-30741) markedly blunted the rise in PVR otherwise seen with heterologous blood (Fig. 2A). SQ-30741-pretreated lungs maintained PVRs between 0.08 ± 0.02 and 0.09 ± 0.02 cmH₂O · ml¹ · min for the duration of perfusion; OKY-046-pretreated lungs exhibited PVRs between 0.05 ± 0.01 and 0.06 ± 0.01 cmH₂O · ml¹ · min (P = 0.027 for SQ-30741 and OKY-046 experiments vs. heterologous controls at 10 min). OKY-046 prevented thromboxane production (<2 ng/ml) through the first hour (Fig. 2C) and beyond (not shown), whereas receptor blockade (SQ-30741) led to levels (~40 ng/ml) that tended to be higher than those of human blood controls (P = 0.07 at 10 min). Survival was significantly improved with either approach to thromboxane blockade (P < 0.01 for both vs. control; P = 0.1 for OKY-046 vs. SQ-30741). Although all OKY-046- and SQ-30741-treated lungs exhibited gradual fluid sequestration, some survived beyond 4 h in each group (Fig. 2C), and tracheal edema was generally not seen. Throughout the first 10 min of perfusion, prostacyclin production in the OKY-046 group was significantly elevated with respect to all other groups (P < 0.02) (Fig. 2D).

View larger version (27K): [in this window] [in a new window]   Fig. 2.   PVR, thromboxane, survival, and 6-keto-PGF1 results for autologous or heterologous blood perfusion or with 2 approaches to selective thromboxane blockade. A: thromboxane synthase inhibition (OKY-046, n = 5; ) or receptor antagonism (SQ-30741, n = 5; ) is associated with PVR elevation analogous to levels that were seen in autologous controls (n = 5; ), whereas a profound rise in PVR is observed in heterologous controls (n = 5; ). B: thromboxane synthase inhibition (OKY-046) or receptor antagonism (SQ-30741) prolongs survival of pig lungs perfused with untreated human blood (). P < 0.01 for each vs. heterologous controls. Autologous perfusion yields consistent 240-min survival. C: thromboxane B2 (TxB2) is rapidly elaborated during perfusion with human blood (). TxB2 levels with receptor inhibition (SQ-30741) tend to be higher than in heterologous controls (P = 0.07). Low-level TxB2 production is observed in piglet lungs perfused with autologous blood or after thromboxane synthase inhibition (OKY-046; P < 0.01 for each vs. heterologous controls). D: prostacyclin rises with autologous and heterologous blood perfusion and with thromboxane inhibition. Levels are higher at 5 and 10 min with OKY-046 than with SQ-30741 or in other groups. Note that time scales are nonlinear for A, C, and D.

View larger version (24K): [in this window] [in a new window]   Fig. 3.   PVR, thromboxane, survival, and 6-keto-PGF1 with Pall filtration or papavarine treatment of human blood or acetylsalicylic acid (ASA) pretreatment of the piglet. A: nonspecific vasodilation (papaverine, n = 6; ) abrogates the rise in PVR but does not prolong graft survival. Partial reperfusion is seen in both Pall filtration (n = 5; ) and ASA pretreatment groups (n = 5; ). B: nonspecific vasodilation (papaverine) does not confer significant survival advantage over heterologous controls (P = 0.15). Platelet and leukocyte depletion from the human perfusate (Pall filtration) or COX inhibition limited to the pig lung (ASA) also prolongs survival (P < 0.01 for each vs. heterologous controls). C: nonspecific vasodilation (papaverine) is associated with TxB2 levels similar to heterologous controls (P = 0.4 at 10 min). Inhibition of lung cyclooxygenase (COX) (ASA) blocks TxB2, whereas Pall filtration tends to reduce TxB2 production (P = 0.01 and 0.07, respectively, vs. heterologous controls). That TxB2 production is not an artifact of the perfusion system is shown by pump-circulated human blood in the absence of a pig lung (blood only, n = 3; ) and with autologous blood perfusion (, Fig. 2C). D: pulmonary COX inhibition (ASA) blocks prostacyclin production during the first hour of perfusion; whereas, with papaverine treatment and Pall filtration, levels parallel those observed with untreated blood or thromboxane-receptor blockade. Note that time scales are nonlinear for A, C, and D.

View larger version (99K): [in this window] [in a new window]   Fig. 4.   Lymphatic dilatation during hyperacute lung rejection. Biopsies of pig lung 10 min after initiation of perfusion with autologous (A) or unmodified heterologous (B) blood. Marked dilatation of a lymphatic channel (*) is evident. Patchy engorgement of interalveolar septae with erythrocytes and focal hemorrhage into alveoli and perivascular sheaths (B, top left) is also demonstrated.

The source of thromboxane in HALR. The primary source of thromboxane could reside either in the formed elements of the human blood perfusate or within the pig lung, where macrophages are the most likely source. Depletion of platelets and leukocytes (95% reduction) from the human blood by Pall filtration was associated with a rapid rise in PVR to 1.24 ± 0.52 cmH₂O · ml¹ · min by 10 min (Fig. 3A), a level intermediate between heterologous controls (P = 0.02 vs. heterologous controls) and groups with thromboxane blockade (P = 0.12 vs. OKY-046; P = 0.09 vs. SQ-30741) (Fig. 2A). The early rise in PVR partially resolved between 30 and 60 min, with subsequent transpulmonary blood flows averaging about one-third of those observed with autologous blood. Lung survival with Pall filtration was prolonged relative to unmodified human blood, with a mean survival of 222 ± 11 min (Fig. 3B). Pall filtration reduced thromboxane production by 75% to 11.11 ± 3.45 ng/ml at 10 min (Fig. 3C) (P = 0.07 vs. heterologous controls).

1

View larger version (87K): [in this window] [in a new window]   Fig. 5.   Cy3-labeled anti-COX-2 monoclonal antibody staining of ASA-treated pig lung. A: preperfusion biopsy of ASA-pretreated lung (×20 magnification) demonstrating minimal COX-2 positivity. B: ASA-pretreated lung after 4 h of perfusion with unmodified human blood demonstrates marked COX-2 expression (×40 magnification).

Effect of PIM depletion on thromboxane production. To selectively deplete PIMs from the lung, dichloromethylenebisphosphonate (Clodronate) in large multilamellar liposomes was given intravenously 2 days before study. Clodronate is an inorganic bisphosphonate that, when encapsulated in liposomes, selectively depletes macrophages in vivo (41). Six hours after intravenous infusion of Clodronate liposomes, positive terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling staining localized to large mononuclear cells in alveolar septae (Fig. 6, B and C); these apoptotic cells corresponded in quantity and location with PIMs previously identified in untreated lungs by their phagocytosis of Monastral blue pigment (Fig. 6D), suggesting that the apoptotic cells are PIMs. Lungs treated with liposomes containing only PBS demonstrated HAR similar in pace and in physiological and biochemical characteristics to that observed for lungs perfused with unmodified human blood, as described above (Fig. 7). Lungs pretreated with Clodronate liposomes exhibited a modest, transient rise in PVR to 0.11 ± 0.02 cmH₂O · ml¹ · min at 10 min, similar to that observed with autologous pig blood perfusion or thromboxane blockade (Fig. 7A). Median survival was 240 min (3 of 5 experiments), with gradual sequestration of reservoir volume in the lung, low airway pressures, and little edema fluid in the airways (Fig. 7B). Thromboxane production in the Clodronate-treated lungs was similar to autologous blood perfusion (P > 0.32 for all time points) and was markedly reduced relative to levels observed in PBS-liposome-treated or untreated control groups (P < 0.01 at 10 min) (Fig. 7C). In contrast to ASA treatment, prostacyclin elaboration (Fig. 7D) was not inhibited by Clodronate. Thus Clodronate pretreatment inhibits production of thromboxane but not prostacyclin, and PVR elevation and loss of microvascular barrier function were attenuated in the absence of any direct inhibition of antibody- or complement-mediated lung injury. Interestingly, only 40-50% of the human platelets in the perfusate were sequestered in the Clodronate-treated lungs, compared with >95% with OKY-046 or SQ-30741 treatment.

View larger version (122K): [in this window] [in a new window]   Fig. 6.   Localization of pulmonary intravascular macrophages (PIMs) and demonstration of apoptotic cells in lungs after Chlodronate-liposome treatment. A: apoptotic cells were not seen in lungs of piglets pretreated with PBS-loaded liposomes. B and C: apoptotic cells localized to alveolar septae were identified in biopsies of Clodronate liposome-pretreated lungs (B: ×40; C: ×63 magnification). Apoptotic cells are identified by in situ terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling assay. D: PIMs in normal pig lung, visualized by phagocytosis of monastral blue pigment given intravenously 15 min before biopsy.

View larger version (24K): [in this window] [in a new window]   Fig. 7.   PVR, thromboxane, survival, and 6-keto-PGF1 with dichlorobisphosphonate- or PBS-liposome pretreatment. After pretreatment with Clodronate liposomes (n = 5; +), pig lung exhibits physiological behavior and survival similar to those observed in lungs perfused with autologous blood (n = 5; ) with respect to PVR (A), graft survival (B), TxB2 (C), and 6-keto-PGF1 (D). In contrast, lungs pretreated with PBS-loaded liposomes (n = 4; ) behave like heterologous controls (see Fig. 2). Note that time scales are nonlinear for A, C, and D.

1

Thromboxane and mediators of inflammation. To investigate why thromboxane inhibition is associated with prolonged survival in the setting of endothelial interaction with heterologous antibody and complement, elaboration of several inflammatory mediators associated with pulmonary parenchymal injury were evaluated (Table 2). Histamine production, an index of mast cell activation, was significantly inhibited (50-90%) in each group in which thromboxane production was inhibited (OKY-046, Clodronate, ASA), or its receptor was blocked (SQ-30741). Produced primarily by activated neutrophils, lymphocytes, and macrophages, TNF- elaboration was significantly inhibited by Clodronate and OKY-046 relative to other human blood groups and tended to be lower with SQ-30741 (one outlier prevented this difference from reaching statistical significance). Pall filtration effectively eliminated TNF- production, suggesting that, in contrast to findings in an ex vivo liver perfusion model, the primary source of TNF- during hyperacute lung rejection is leukocytes in the human blood (29). We observed a consistent trend toward reduced C3a production with each thromboxane blocking strategy but not with papaverine or Pall filtration; a statistically significant reduction in C3a was demonstrated only for PIM depletion with Clodronate. Thus thromboxane inhibition is associated with diminished activation of the TNF--, histamine-, and complement-mediated proinflammatory cascades, all of which likely contribute to progression of lung injury. Papaverine effectively blocked both TNF- and histamine elaboration.

View this table: [in this window] [in a new window]   Table 2.   Elaboration of inflammatory mediators (C3a, histamine, TNF-) during hyperacute lung rejection

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

These results demonstrate that thromboxane mediates the rise in PVR observed during HAR of the lung and influences biochemical indexes of inflammation and the pace of intraparenchymal fluid sequestration. Both increased PVR and inflammation are mediated at least in part through the thromboxane receptor, because results with SQ-30741 approximate those with OKY-046. The influence of thromboxane on inflammation and rate of increase in lung water can be dissociated from pure rheological effects, as shown by improved oxygen transport and protection from tracheal edema with thromboxane blockade relative to papaverine. We find that Clodronate-sensitive cells in the lung are critical to the production of thromboxane during the interaction of pig lung with human blood. We infer that these cells are PIMs activated in the pig lung during the initial interaction with human blood, as ASA therapy targeted selectively at the lung reduces thromboxane by 96%. However, Pall-filterable blood constituents, both human and porcine, contribute significantly to overall thromboxane production, because Pall filtration of the human blood before lung perfusion inhibits thromboxane production by 70% and additional transient pulmonary vein filtration effectively prevents thromboxane production. Together these data suggest that eicosanoids elaborated by the lung are required to trigger thromboxane production by blood elements, presumably by platelets and perhaps by activated monocytes and macrophages. Whether neutrophils, which are also effectively depleted by this intervention, contribute to this interaction remains to be defined. In aggregate, our experiments suggest that PIMs in the lung elicit TxA₂ synthesis in platelets or other human and porcine formed blood elements through COX- and TxA₂-dependent mechanisms. Prolific thromboxane production, accelerated by PIMs in the pig lung, may account for this organ's particular susceptibility to injury in this model of HAR.

Reduced thromboxane effect generally correlates with a blunted inflammatory response, as reflected in reduced elaboration of TNF-, histamine, and C3a in every experimental group in which thromboxane production was inhibited or its receptor blocked. Smooth muscle relaxation with papaverine also inhibits TNF- and histamine, suggesting that altered rheology associated with elevated PVR contributes significantly to these facets of the proinflammatory HALR milieu. In contrast, papaverine does not blunt C3a production, whereas thromboxane inhibition does have a modest effect that reaches significance with PIM depletion. Because C3a, C5a, and other membrane-bound components of the complement cascade are mechanistically linked to increased vascular permeability, we postulate that thromboxane's proinflammatory effect is mediated in part through activation of the alternative complement pathway. This hypothesis is supported by the association between thromboxane and alternative pathway complement activation in other models (2, 15, 26, 39) and by our demonstration of properdin deposition in pig lungs perfused with human blood (3).

When pulmonary COX activity is inhibited using ASA, the importance of endothelial prostacyclin elaboration to preservation of transpulmonary blood flow is apparent. In contrast to any selective antithromboxane strategy (OKY-046, SQ- 30741, or Clodronate), each of which results in preserved transpulmonary blood flow, inhibition of both prostacyclin and thromboxane is associated with significantly elevated PVRs. Recovery of perfusion was associated with de novo expression of COX-2 in the lung, delayed elaboration of prostacyclin, and a subsequent rise in thromboxane. Thus, in the setting of thromboxane blockade, pulmonary prostacyclin production appears to be required for physiological transpulmonary flow of human blood through the pig lung. Our finding is consistent with the observation in a canine model of oleic acid-induced acute lung injury, where inhibition of COX by ASA is associated with augmented PVR elevation (23). We infer that preventing synthesis of PGI₂ substrate by COX inhibition results in increased vascular smooth muscle tone or permits constriction in response to the endothelial activation associated with HALR (36, 40). Whether manipulation of the balance between eicosanoids can be used to protect the lung from PVR elevation or other manifestations of HALR will require further investigation. If thromboxane causes increased vascular permeability, we would predict that PGI₂ supplementation, even to supraphysiological levels, will probably fail to prevent lung injury in the absence of selective thromboxane blockade.

Our results support the idea that endoperoxide shunts can have important physiological consequences. Lungs pretreated with the thromboxane synthase inhibitor OKY-046 demonstrated the lowest PVRs and highest prostacyclin levels observed in this series of experiments. Substrate diversion from platelets to endothelium has been well documented in vitro and in vivo (28). We speculate that, in this model of pulmonary HAR, thromboxane synthase inhibition affords diversion of prostaglandin endoperoxide to prostacyclin pathway precursors at the interface among PIMs, formed blood elements, and endothelium. As prostacyclin is a vasodilator and potent antagonist of platelet aggregation, increased production may preserve perfusion of microvascular beds, and this may account in part for the trend toward improved survival with OKY-046 relative to SQ-30741.

Thromboxane is closely associated with experimental and clinical pulmonary dysfunction (4, 5, 9, 19, 39). Whether thromboxane directly affects pulmonary microvascular integrity is controversial (2, 17, 22, 34, 38); our data support the view that it does, at least in the context of antispecies antibody binding and complement activation, and that its effect on permeability is mediated through the thromboxane receptor. As a consequence of its postvenular vasoconstrictive effects, TxA₂ likely also increases microvascular fluid transudation by causing increased hydrostatic pressure in pulmonary capillary beds. The adherence of inflammatory cells and platelets to endothelial adhesion molecules is generally favored by low flow, and stasis associated with increased vascular resistance may enhance the interaction of antibody, complement, and formed blood elements with pulmonary endothelium, amplifying pathogenic autocrine and paracrine cytotoxic pathways (12, 21). Finally, thromboxane might have direct effects on microvascular barrier function. In support of this, rapid capillary leak occurred with papaverine administration, which both blocked low flow and increased capillary hydrostatic pressure, while thromboxane production, C3 formation, and accumulation of extravascular lung water proceeded rapidly. This observation argues for a direct effect of thromboxane on pulmonary vascular-epithelial integrity, in addition to indirect (hydrostatic pressure and stasis) mechanisms.

How thromboxane might directly influence permeability has not yet been defined at the molecular level, but plausible cellular mechanisms are suggested by several observations. There is compelling evidence for the presence of functional TxA₂ receptors on vascular endothelium, that TxA₂ can directly activate endothelium via its receptor, and that TxA₂ can cause alterations in endothelial physiology and structure likely to contribute to pulmonary microvascular leak (14, 18, 42). Blocking thromboxane in the context of endothelial activation by antibody and complement may thus attenuate structural and biochemical endothelial alterations crucial to the loss of microvascular barrier function. With regard to thromboxane's putative proinflammatory effects, myeloid lineage cells in blood and lung (monocytes, platelets, macrophages, mast cells, and neutrophils) bear thromboxane receptors. Our results with Pall filtration of human blood and of pulmonary vein effluent suggest that myeloid constituents of blood and lung contribute to intersecting positive feedback loops, which together determine the pace of microvascular injury and lung failure.

Based on these considerations and the observations reported here, our working hypothesis is that TxA₂-receptor-mediated endothelial and PIM activation occurs as an initial event during HAR of the pig lung by human blood, causing alterations in endothelial cytoskeletal structure that result in plasma and cellular extravasation between endothelial cells (6, 24). By inhibiting amplification of the complement cascade and prothrombotic pathways and blunting release of other inflammatory mediators (TNF-, histamine) at a proximal stage in the HAR response, selectively blocking thromboxane allows recruitment of anti-inflammatory cellular defenses to a degree sufficient to facilitate prolonged lung function in a "discordant" xenogeneic environment (1).

The trigger for thromboxane production remains to be elucidated. The thromboxane cascade might be driven by events specific to the immunological interaction between pig and humans (antibody binding, complement activation), physiological incompatibility in coagulation pathway regulation, or interaction of human formed blood elements with molecules constitutively expressed on pig lung (30, 31, 39, 43). The deposition of the complement membrane attack complex on macrophages is known to stimulate phospholipase A₂ activity, providing arachidonic acid for eicosanoid synthesis, which may in turn be the initiating event for thromboxane production in this model (13). Experiments wherein endothelial and systemic complement activation are optimally inhibited (e.g., perfusion of lungs with high-HCRP transgene expression with soluble complement receptor type 1-treated blood) will address this important issue.

The eicosanoid pathway, like the complement system, is a primitive yet potent component of the innate immune system. There exists circumstantial evidence that thromboxane may contribute to the increased vascular resistance and subsequent inflammation observed in organ xenografts other than the lung (6, 16, 33, 40). Indeed, direct manipulation of the eicosanoid balance has been used experimentally to facilitate short-term survival of kidney and lung xenografts in the setting of complement regulation (30, 44). If thromboxane's trigger proves to be independent of complement, pharmacological inhibition of thromboxane, or organ-specific approaches to donor macrophage depletion, may contribute to safe clinical use not only of lung xenografts but of other solid organ xenografts as well.