The Role of Endothelial Nitric Oxide Synthase in the Pathogenesis of a Rat Model of Hepatopulmonary Syndrome

GASTROENTEROLOGY 1997;113:606 614 The Role of Endothelial Nitric Oxide Synthase in the Pathogenesis of a Rat Model of Hepatopulmonary Syndrome MICHAEL B. FALLON,* GARY A. ABRAMS,* BAO LUO,* ZIHYING HOU,* JUN DAI, and DAVID D. KU Departments of Internal Medicine and Pharmacology, and *Liver Center, University of Alabama at Birmingham, Birmingham, Alabama See editorial on page 682. Recently, chronic common bile duct ligation (CBDL) in the rat has been recognized as a model of the hepatopulmonary syndrome. 4 Animals develop progressive arte- Background & Aims: The hepatopulmonary syndrome rial gas exchange abnormalities, similar to human disoccurs when intrapulmonary vasodilatation causes im- ease, which correlate with the development of paired arterial gas exchange in liver disease. The patho- intrapulmonary vasodilatation measured in vivo. 5 Partial genesis is poorly understood, although nitric oxide may portal vein ligation (PVL) results in a similar degree of be involved. Common bile duct ligation in the rat is a portal hypertension without hepatic injury and does not model of the hepatopulmonary syndrome, but no stud- induce intrapulmonary vasodilatation. Indirect evidence ies have evaluated NO in pulmonary vasodilatation in suggests that guanosine 3,5 -cyclic monophosphate acthis model. The aim of this study was to determine cumulation in the pulmonary vasculature may contribute whether NO contributes to intrapulmonary vasodilatato vasodilatation after CBDL, 4 but no studies have evalution after bile duct ligation. Methods: Endothelial and ated pulmonary NO production or its effect on pulmoinducible NO synthase (NOS) levels and localization nary vascular reactivity in this model. and NO activity in pulmonary artery rings were assessed after bile duct ligation. Results: Pulmonary en- In this study, we assess the potential importance of dothelial NOS levels increased and alveolar vascular NO in intrapulmonary vasodilatation after CBDL by staining was enhanced after bile duct ligation. No evaluating protein levels and regional localization of enchange in pulmonary inducible NOS levels or localizathase (inos) in lungs and NO activity in intralobar dothelial NO synthase (enos) and inducible NO syn- tion was detected. Increased endothelial NOS levels correlated with alterations in gas exchange and were pulmonary artery rings in CBDL animals, sham-operated accompanied by enhanced NO activity and a blunted controls, and PVL animals. We report a progressive in- response to phenylephrine, reversible by NOS inhibition, crease in pulmonary enos protein levels with increased in pulmonary artery rings. Portal-vein-ligated ani- localization in the alveolar vasculature beginning within mals, which do not develop intrapulmonary vasodilata- 2 weeks after CBDL. This increase in pulmonary enos tion, had no changes in pulmonary NOS production or is accompanied by evidence of enhanced production and in NO activity in pulmonary artery rings. Conclusions: activity of NO in intralobar pulmonary rings from ani- NO, derived from pulmonary vascular endothelial NOS, mals 5 weeks after CBDL. In contrast, PVL animals, contributes to intrapulmonary vasodilatation in animal which do not develop intrapulmonary vasodilatation, had hepatopulmonary syndrome. no increase in pulmonary NO synthase (NOS) levels and no alterations in NO production and activity in pulmonary artery rings. The hepatopulmonary syndrome occurs when intrapulmonary vasodilatation results in abnormal arte- Materials and Methods rial oxygenation in patients with chronic liver disease and/or portal hypertension. 1,2 This syndrome is observed Animal Models in as many as 15% of cirrhotics, 2 and a role for nitric Male Sprague Dawley rats (Charles River, Wilming- oxide in its pathogenesis has been considered based on ton, MA) weighing 200 250 g were used in all experiments. evidence that NO contributes to systemic vasodilatation Abbreviations used in this report: CBDL, common bile duct ligation; in chronic liver disease. 3 However, the lack of a model enos, endothelial nitric oxide synthase; inos, inducible nitric oxide system for hepatopulmonary syndrome has hampered in- synthase; L-NAME, N G -nitro-l-arginine methyl ester; NO, nitric oxide; vestigation in the pathogenesis of alterations in the pul- PAGE, polyacrylamide gel electrophoresis; PVL, portal vein ligation; monary microcirculation and the development of effective medical therapies. SDS, sodium dodecyl sulfate; TNF, tumor necrosis factor. 1997 by the American Gastroenterological Association 0016-5085/97/$3.00

August 1997 NITRIC OXIDE AND HEPATOPULMONARY SYNDROME 607 samples were snap-frozen in ornithine carbamoyltransferase embedding compound (Miles Laboratory, Elkhart, IN) after lung inflation via intratracheal instillation of ornithine carbamoyltransferase, and 8-mm cryostat sections were prepared on gelatin-coated slides and stored at 070 C until use. Sections were postfixed in acid-ethanol before blocking, incubation with primary and secondary antibodies, and washing using the ABC staining technique (Vetor Laboratories, Burlingame, CA) ac- cording to the manufacturer s instructions. Sections were counterstained with hematoxylin and viewed and photographed using a Nikon axiophot microscope (Long Island, NY). Control sections were incubated with an irrelevant primary antibody before addition of secondary antibody or with secondary antibody alone. CBDL was performed as described by Easter et al. 6 and partial PVL was performed as described by Chojkier and Groszmann. 7 Sham-operated animals underwent mobilization of the common bile duct or portal vein without ligation. Five animals from each group (sham CBDL or PVL, 2-week CBDL, 5- week CBDL, and 3-week PVL) were used for Western blot experiments, 3 animals from each group were used for immunohistochemical studies, and 6 sham CBDL, 4 sham PVL, 6 CBDL, and 4 PVL animals were used for pulmonary artery ring studies. All animals had the presence or absence of hepatic injury and/or complete bile duct ligation assessed by hepatic biochemical and histological studies and direct measurement of portal pressure and arterial gas exchange evaluated with arterial blood gases previously measured and published. 5 Endotoxin treatment was accomplished by intraperitoneal injection of 4 mg/kg of lipopolysaccharide (Sigma Chemical Co., St. Louis, MO) followed by harvesting of tissues 12 hours after injection. 8 The study was approved by the University of Alabama at Birmingham Animal Care and Use Committee and conforms to National Institutes of Health guidelines on the care and use of laboratory animals. Pulmonary Artery Ring Studies On the day of the experiment, rats were anesthetized with 10 mg/kg ketamine plus 1.5 mg/kg xylazine intraperito- neally. The thoracic cavity was opened, and both left and right main lobes of lung were quickly excised. The intralobar pulmonary arteries, approximately 1.5 cm in length, were carefully dissected from the surrounding connecting tissues under Protein Extraction and Immunoblotting a stereomicroscope. The arteries were cut into 3 4-mm long Lung samples for sodium dodecyl sulfate (SDS)-polyacryl- rings, with the outside diameter ranging from 0.5 to 1.2 mm. amide gel electrophoresis (PAGE) was obtained after a 30-second Each ring was then mounted as previously described 12,13 using in situ perfusion with phosphate-buffered saline, ph 7.4, con- two triangular-shaped 30-gauge stainless steel needles in jacktaining 130 mmol/l NaCl, 2 mmol/l NaH 2 PO 4, and 7 mmol/ eted tissue chambers containing Krebs Henseleit solution at LNa 2 HPO 4 at 30 ml/min. Tissue samples were rapidly diced, 37 C, gassed with 95% O 2 and 5% CO 2. Krebs Henseleit immediately frozen in liquid nitrogen, and maintained at 070 C solution consisted of (in mmol/l): NaCl, 118; KCl, 4.6; until use. Alveolar macrophages were isolated by bronchoalveolar NaHCO 3, 27.2; MgSO 4, 1.2; KH 2 PO 4, 1.2; CaCl 2, 1.75; Na 2 lavage using standard techniques. 9 Stimulation of isolated macro- ethylenediaminetetraacetic acid, 0.03; and glucose, 11.1. The phages was performed by 4 8 hours of incubation with 600 U/ upper needle of each arterial ring was attached with a silk ml tumor necrosis factor (TNF)-a (Sigma) and 1000 U/mL suture to a force-displacement transducer (Grass FT.03C), and interferon gamma (Gibco BRL, Gaithersberg, MD). Tissue and changes in isometric force were recorded on a Grass Polygraph alveolar macrophage homogenates were prepared using Dounce (model 7C). homogenization in RIPA buffer containing 1% NP-40, 0.1% All vessels were passively stretched to generate a resting SDS, 0.5% sodium deoxycholate, 50 mmol/l Tris-HCl, 150 tension of 2.0 g for isometric contraction recording. After mmol/l NaCl, 1 mmol/l Na 2 ethylenediaminetetraacetic acid, 1 hour of equilibration, pulmonary rings were exposed to a and 0.1 mmol/l ethylene glycol-bis(b-aminoethyl ether)- maximum depolarizing 70 mmol/l KCl. When contractile N,N,N,N,-tetraacetic acid in the presence of protease inhibitors, responses plateaued, ranging from 0.4 to 0.7 g active tension, followed by the addition of a final volume of 31 SDS-PAGE the pulmonary rings were rinsed with Krebs Henseleit solusample buffer (11 is 1.25 mmol/l Tris, ph 6.8, 1% SDS, 2% tion and allowed to equilibrate for an additional 60 minutes mercaptoethanol, and 2% sucrose) and boiling for 3 minutes. in the presence of 5 mmol/l indomethacin before the start of Samples were subjected to SDS-PAGE by the method of La- the experiment. At appropriate times, submaximal tone was emmli 10 on 7.5% acrylamide gels, electrophoretically transferred elicited with 0.1 1 mmol/l phenylephrine, during which to nitrocellulose paper (Scleicher & Schuell, Keene, NH). Incuba- times various pharmacological agents were added to the bathtion with enos or inos antibodies was followed by washing ing medium. In all cases, effects of only one dose-response curve and incubation with an appropriate horseradish peroxidase con- of acetylcholine was determined in each pulmonary segment, jugated secondary antibody and development by enhanced chemi- except for the cases when repeated testings of acetylcholine luminescence (Amersham, Arlington Heights, IL). Resulting au- were indicated after pretreatment of NOS inhibitor, N G -nitrotoradiographic signals were quantitated using Molecular Analyst L-arginine methyl ester (L-NAME). Data are expressed as software for the GS-670 densitometer (Bio-Rad, Hercules, CA). percentage of relaxation or percentage of decrease in phenylephrine-induced constrictory tone. For the determinations of Immunohistochemistry vasoconstrictory responses, cumulative dose responses to phenylephrine were performed while vessels were maintained at Immunohistochemical staining of lung was performed using a modification of the protocol by Kobzik et al. 11 Tissue resting tension. For the determination of the role of basal NO

608 FALLON ET AL. GASTROENTEROLOGY Vol. 113, No. 2 Table 1. Arterial Blood Gases After CBDL and PVL Sham CBDL a CBDL b PVL (n Å 4) (n Å 4) (n Å 4) (n Å 5) ph 7.45 { 0.01 7.45 { 0.02 7.47 { 0.01 7.45 { 0.01 pco 2 (mm Hg) 41.85 { 0.57 36.88 { 1.28 c 35.28 { 0.68 c 38.18 { 1.23 po 2 (mm Hg) 91.65 { 0.71 87.88 { 1.44 83.47 { 3.16 c 96.04 { 1.77 A-a gradient (mm Hg) 6.04 { 0.76 16.03 { 1.51 c 22.43 { 2.55 c 6.23 { 1.36 NOTE. Values are means { SE. a Two-week CBDL. b Five-week CBDL. c P õ 0.05 compared with SHAM. production on pulmonary vasomotor tone, a maximum dose of 0.3 mmol/l L-NAME was added in the present submaximal constrictory tone and the extent of contraction was then recorded and expressed as a percentage of maximum KCl-induced contraction in the same vessel. Antibodies and Reagents Commercially available monoclonal antibodies to human enos and inos, which cross-react with rat tissue, were used in this study (Transduction Laboratories, Lexington, KY). Homogenates of human aortic endothelial cells supplied by the manufacturer and rat alveolar macrophages isolated by tracheal lavage and stimulated with interferon gamma and TNF-a were used as controls and to document specificity of each antibody. A sheep anti-mouse horseradish peroxidase conjugated secondary antibody (Amersham) was used for Western blotting, and a horse anti-mouse biotinylated secondary antibody was used for immunohistochemistry (Vector Laboratories). For pulmonary artery ring studies, acetylcholine chloride, phenylephrine, and L-NAME were purchased from Sigma Co., and all other reagents were purchased from Fisher Scientific (Norcross, GA). Acetylcholine was prepared in sodium acetate buffer (ph 4.0) and stored at 4 C to ensure stability. All other solutions were prepared immediately before use. Statistics Data were analyzed using Student s t test, analysis of variance, and multiple comparisons between groups with Bonferroni correction and linear correlation when appropriate. All measurements are expressed as mean { SE. Statistical significance was designated as P õ 0.05. Results Arterial Blood Gases After CBDL To document the development of hepatopulmonary syndrome in animals after CBDL, microsphere measurements of the pulmonary microcirculation and arterial blood gases have been performed as previously published. 5 A subset of animals described in the initial report underwent NOS analysis in the present study, and their arterial blood gas results are presented in Table 1. After CBDL, animals develop progressive hyperventilation, hypoxemia, and widened alveolar-arterial oxygen gradi- ents, which correlate with pulmonary microcirculatory vasodilatation. PVL animals, which do not develop pul- monary microcirculatory changes, do not develop gas exchange abnormalities. Pulmonary NOS Levels After CBDL To assess whether altered pulmonary levels of NOS develop in the lung after CBDL, we performed Western blotting for enos and inos on equal quanti- ties of lung homogenates from sham, 2-week CBDL, 5-week CBDL, and PVL animals (Figure 1A C). The specificity of the antibodies to enos and inos was confirmed by blotting experiments using control ho- mogenates of human endothelial cells and rat alveolar macrophages stimulated with TNF-a and interferon gamma. enos was detectable in lung homogenates from all groups. A significant progressive increase in enos content relative to sham samples was observed in lung homogenates from 2- and 5-week CBDL ani- mals. Pulmonary homogenates from PVL animals, without intrapulmonary vasodilatation, had no significant increase in lung enos content. In some enos tissue homogenates, a faint diffuse signal was observed at approximately 90 kilodaltons. This signal could be completely eliminated by reducing the concentration of primary antibody or increasing the concentration of the blocking solution and was consistent with nonspe- cific binding of the primary antibody in lung. In contrast, inos was not detected in measurable quantities in pulmonary homogenates from any group. inos tissue homogenates did contain a single band at approximately 55 kilodaltons, which was detectable only after prolonged development using enhanced chemiluminescence (ú15 minutes). This likely represents low- level cross-reactivity between the inos antibody and an unidentified antigen in lung. These results show a progressive significant increase in lung enos content after CBDL.

August 1997 NITRIC OXIDE AND HEPATOPULMONARY SYNDROME 609 Figure 2. Correlation of alveolar-arterial oxygen gradients and lung enos content in animals after CBDL and PVL. enos values are expressed relative to sham control values set at 1. Sham alveolararterial gradients ranged from 4.6 to 7.7 mm Hg. The linear correlation coefficien (r) was highly statistically significant *P õ 0.0004., 2- week CBDL;, 5-week CBDL;, PVL. Correlation Between enos Levels and Gas Exchange Abnormalities After CBDL To assess whether there is a correlation between the development of gas exchange abnormalities and the increase in pulmonary enos content after CBDL, we compared alveolar-arterial oxygen gradients with lung enos protein content in CBDL and PVL animals (Figure 2). There is a strong positive linear correlation between pulmonary enos levels and gas exchange abnormalities in our animals, suggesting that the two may be causally related. In this study, 7 of 8 CBDL animals had enos Figure 1. enos and inos protein analysis in pulmonary homogenates levels greater than control and PVL values and each anifrom sham, 2-week CBDL, 5-week CBDL, and PVL animals. Equal mal had an abnormal alveolar-arterial gradient. Alveolarloading of protein in each lane was confirme by Coomassie blue arterial gradients in sham control animals ranged from staining and Lowry protein determinations. (A) Representative enos 4.6 to 7.7 mm Hg. immunoblot from sham and 5-week CBDL animals (30 mg of homogenate protein per lane). Lane E is a homogenate of human aortic endothelial cells that produce only enos and shows a single band of Ç140 kilodaltons consistent with enos. Lane AL is a homogenate of isolated stimulated rat alveolar macrophages that produce only inos and shows no signal when probed with the enos antibody. Note Pulmonary Immunohistochemical Localization of NOS After CBDL To localize changes in NOS levels after CBDL, the significan increase in enos in pulmonary homogenates after we performed immunohistochemical staining for enos CBDL. A faint diffuse band seen at 90 kilodaltons was eliminated by using lower concentrations of primary antibody and is consistent with and inos in lung from sham, CBDL, and PVL animals nonspecifi cross-reactivity of the antibody in lung tissue. (B) Representative inos immunoblot from sham and 5-week CBDL animals (30 irrelevant primary antibody or with secondary antibody (Figures 2 and 3). Control samples incubated with an mg of homogenate protein per lane). A single band of Ç130 kilodalalone showed no staining for enos or inos. enos tons consistent with inos is seen in lane AL. No signal is seen in lane E. No measurable inos signal is detectable in either sham or staining was detected in the vascular endothelium of 5-week CBDL lanes. A single band at approximately 55 kilodaltons arteries and in ciliated bronchial epithelium in sham was observed after prolonged development using enhanced chemiluanimals as previously described. 11 In these regions, 2- minescence and may represent low-level cross-reactivity between the inos antibody and an unidentifie antigen in lung. (C ) Summary of week and 5-week CBDL and PVL animals also showed lung enos protein levels in 2-week CBDL, 5-week CBDL, and PVL enos staining. However, after CBDL, there was a proanimals expressed relative to sham values. A significan progressive gressive increase in staining in the region of small alveoincrease in lung enos content is observed in 2- and 5-week CBDL animals. Data are expressed as mean { SE (n Å 5 for each group). lar capillaries and a qualitative increase in staining in *P õ 0.05. pulmonary arterial branches, which was not observed in

610 FALLON ET AL. GASTROENTEROLOGY Vol. 113, No. 2

August 1997 NITRIC OXIDE AND HEPATOPULMONARY SYNDROME 611 sham or PVL animals. inos staining was not detected significant 77% increase in contraction after L-NAME in lung sections from sham, CBDL, or PVL animals, nor administration (maximum L-NAME contraction in was it detectable in alveolar macrophages isolated from CBDL rings 0.69 { 0.09 vs. sham rings 0.39 { 0.04, 5-week CBDL animals. However, staining was easily detected n Å 12 in each group, P õ 0.001). In contrast, PVL in lung from animals 24 hours after intraperitoneal rings did not contract differently from sham after L- injection of endotoxin and in alveolar macrophages from NAME administration. These findings provide evidence CBDL animals after stimulation with TNF-a and inter- that pulmonary artery ring enos activity is increased feron gamma. These results show an increase in enos after CBDL but not PVL and support our protein deter- staining after CBDL, partly in the region of small alveolar mination and immunohistochemical localization of vessels, which correlates with increased enos protein enos in lung tissue. levels in lung homogenates. inos protein was not de- To evaluate whether increased basal NO activity alters tected in lung either by immunoblotting or immunohistochemistry the responsiveness of the pulmonary vasculature to vaso- in sham, CBDL, or PVL animals, despite its coactive agents, the cumulative dose-response to the a- ready detection in positive controls. adrenergic receptor agonist phenylephrine was assessed Pulmonary Artery Ring Studies After CBDL in endothelium intact pulmonary artery rings from CBDL and PVL animals in the presence and absence of To determine whether increased pulmonary enos L-NAME (Figures 6 and 7). Pulmonary artery rings from levels are accompanied by evidence of enhanced NO prosham 5-week CBDL animals responded significantly less than duction or activity in the pulmonary vasculature, we rings. The maximum phenylephrine-induced (3 assessed the contractile function of intralobar pulmonary mmol/l) contraction in CBDL rings was 0.51 { 0.09 g artery rings from sham, 5-week CBDL, and PVL animals (96.3% { 12% of the maximum KCl contraction) com- (Figures 4 6). pared with 0.72 { 0.05 g (133.8% { 4% of the maxi- To assess the endothelial response to acetylcholineendothelial mum KCl contraction) in sham rings. Inhibition of basal induced NO production, we administered 0.001 3 NO production by pretreatment with L- mmol/l acetylcholine to indomethacin-pretreated, phenduced NAME significantly potentiated the phenylephrine-inylephrine-preconstricted pulmonary artery rings (Figure contraction in CBDL rings, normalizing the de- 4). In pulmonary artery rings from sham animals, the pressed contractile response toward the sham values. The addition of acetylcholine resulted in a dose-dependent maximum phenylephrine-induced contraction in the L- relaxation reaching a maximum of approximately 080%. NAME pretreated CBDL pulmonary artery rings was The IC50 of the acetylcholine-induced relaxation was not significantly different from sham pulmonary artery 0.1 mol/l, similar to previous studies in other vessel rings (CBDL, 0.75 { 0.05 g, 137.5% { 12% of the preparations, including coronary arteries and aorta. Inhi- maximum KCl contraction vs. sham, 0.78 { 0.1 g, bition of enos with L-NAME completely abolished this 151% { 8% of the maximum KCl contraction). In con- relaxation, confirming the obligatory role of NO syntherings trast, the contractile response of PVL pulmonary artery sis in this process. CBDL and PVL did not alter the to phenylephrine was not significantly different responsiveness of pulmonary artery rings to acetylcholine, from sham rings, regardless of the presence or absence showing that the endothelial response to acetylcholineenos of L-NAME. These findings show that increased basal induced NO production was preserved in these models. and NO activity in CBDL pulmonary artery rings, To assess whether basal NO activity is altered in pulconstrictor but not PVL rings, decreases responsiveness to the vasomonary artery rings after CBDL and PVL, we measured phenylephrine, which is reversible by NOS the contractile response after inhibition of endogenous inhibition. NOS with L-NAME. Figure 5 shows the extent of L- NAME induced contraction in sham, CBDL, and PVL Discussion pulmonary artery rings, an indirect measure of basal en- In this study, we have documented that the development dothelial NO production. Compared with sham, pulmonary of intrapulmonary vasodilatation and the hepatodothelial artery rings from 5-week CBDL animals showed a pulmonary syndrome in CBDL animals is accompanied Figure 3. Immunohistochemical localization of enos in lung sections from sham, 2-week CBDL, 5-week CBDL, and PVL animals. (A) Negative control shows no staining (original magnificatio 401). (B I) Shown are 201 original magnificatio (low power, left panel) and 401 original magnificatio (high power, right panel) from sham (B, C), 2-week CBDL (D, E), 5-week CBDL (F, G), and PVL (H, I) lung sections. Brown enos staining was seen in large arteries (a) and bronchial epithelium (b) in all sections. In 2- and 5-week CBDL sections, a progressive increase in staining in small vessels in alveolar regions was observed (arrows ), which was not seen in sham or PVL sections (arrowheads ).

612 FALLON ET AL. GASTROENTEROLOGY Vol. 113, No. 2 Figure 4. Immunohistochemical localization of inos in lung. (A) Positive control lung section from an endotoxin-treated animal showing extensive brown staining predominately in alveolar macrophages (arrows). (B) A 5-week CBDL lung section shows no inos staining. (C ) Alveolar macrophages, isolated from a 5-week CBDL animal and then stimulated with TNF-a and interferon gamma, show intense cytoplasmic inos staining. (D) Alveolar macrophages isolated from a 5-week CBDL animal and not stimulated have no inos staining (original magnificatio 401). by an increase in pulmonary enos protein content and enhanced enos localization in the region of small alveolar vessels. This increase in pulmonary enos correlates with the severity of gas exchange abnormalities and is accompanied by evidence of enhanced basal NO activity and an NO-mediated decrease in the sensitivity to vasoconstrictors in intralobar pulmonary artery rings. In contrast, PVL animals that develop a similar degree of portal hypertension without intrapulmonary vasodilatation or gas exchange abnormalities do not show changes in pulmonary enos protein or alterations in NO activity in intralobar pulmonary artery rings. We were unable to detect significant changes in pulmonary inos content or localization in any animals. These findings support that changes in pulmonary enos protein and pulmonary Figure 5. Comparison of acetylcholine-induced endothelial NO-dependent relaxation in sham ( ), 5-week CBDL ( ), and PVL ( ) rat intralo- of intrapulmonary vasodilatation in an animal model of vascular NO activity are important in the development bar pulmonary arteries. All arteries were pretreated with 5 mmol/l hepatopulmonary syndrome. indomethacin and precontracted with phenylephrine before acetylcholine testing. Twelve pulmonary rings were isolated and studied from Our observation that there is a significant increase in 6 sham (circles), while 10 and 8 rings were studied from 5 CBDL enos levels in pulmonary homogenates after CBDL is (squares) and 4 PVL (triangles) rats, respectively. For sham ( ), CBDL consistent with previous observations suggesting that en- ( ), or PVL ( ) plus L-NAME pretreatment, 0.3 mmol/l L-NAME was added 20 minutes before the repeated acetylcholine testing in the same rings. Data are expressed as mean { SE. hanced NO production and NOS levels are important in the development of the systemic hyperdynamic circula-

August 1997 NITRIC OXIDE AND HEPATOPULMONARY SYNDROME 613 tory abnormalities in chronic liver disease. 3,14 Our immunohistochemical studies document that a component of the increased pulmonary enos is found in the region of alveolar capillaries and probably alters NO production and vascular tone in the pulmonary microcirculation as we have documented in intralobar pulmonary artery segments. There is controversy over which NOS isoforms are involved in altering vascular NO production in chronic liver disease. Our present finding that inos was not present in measurable quantities by immunoblotting and immunohistochemistry in the lung in any of our animals, despite its ready identification in lung in positive controls, suggests that enos is responsible for altering NO production in the lungs in this model. This observation is consistent with our previous finding showing no detectable influx of inflammatory cells as a source for inos in lung sections from CBDL animals 5 and with other studies that have failed to detect inos protein expression in systemic or splanchnic vessels in models of cirrhosis and portal hypertension. 15 18 However, it is possible that small increases in inos protein in localized populations of cells within the lung could have escaped detection; therefore we cannot completely exclude that inos has a minor role in altering pulmonary vascular tone in this model. The finding that intralobar pulmonary artery rings from 5-week CBDL animals showed enhanced basal NO activity and an NO-mediated impairment in vasoconstrictive response to phenylephrine is similar to what has been observed in the systemic and splanchnic circulation in other models of cirrhotic portal hypertension. 3,14 This observation suggests that the pathogenetic mechanisms resulting in altered aortic and splanchnic vascular reactiv- ity in models of cirrhotic portal hypertension may be L-NAME. (B), Sham;, PVL;, sham plus L-NAME;, PVL plus L- NAME. Figure 7. Comparison of phenylephrine-induced contraction in the presence and absence of L-NAME in (A) 5-week CBDL and (B) PVL rat intralobar pulmonary arteries relative to sham. Pulmonary rings were pretreated with 5 mmol/l indomethacin before cumulative additions of phenylephrine. For the L-NAME pretreated group, 0.3 mmol/l L- NAME was added 20 minutes before the phenylephrine testing. The contractile responses to phenylephrine were normalized as percentage of maximum 70 mmol/l KCl contraction obtained from each pulmonary ring preparation. Data are expressed as mean { SE. *P õ 0.05. (A), Sham;, CBDL;, sham plus L-NAME;, CBDL plus similar to those in the pulmonary arteries after CBDL. The changes in the effects of NO in the pulmonary vasculature after CBDL are temporally correlated with changes in enos levels in lung homogenates, suggesting the two are causally related. We have performed our studies in the presence of indomethacin to isolate our analysis on the effects of NO in the pulmonary arterial system. Thus, we cannot exclude that changes in prostanoid or eicosanoid metabolism may contribute to pulmonary vascular changes after CBDL. Previously, we have shown that PVL animals develop a similar degree of portal hypertension compared with CBDL animals, without developing intrapulmonary vasodilatation or gas exchange abnormalities. 5 In this study, we document that PVL animals do not develop changes Figure 6. Comparison of L-NAME induced contraction in sham, 5- week CBDL, and PVL rat intralobar pulmonary arteries. Pulmonary rings were pretreated with 5 mmol/l indomethacin and precontracted with a submaximal dose of phenylephrine before 0.3 mmol/l L-NAME addition. Changes in contractile force, after L-NAME addition, are plot- ted for 8 12 sham rings, 10 CBDL rings, and 8 PVL rings. Data are expressed as mean { SE. *P õ 0.05.

614 FALLON ET AL. GASTROENTEROLOGY Vol. 113, No. 2 in NOS levels in lung homogenates or immunolocalizashunting using gamma-labelled microspheres. Am J Physiol 7. Chojkier M, Groszmann RJ. Measurement of portal systemic tion in lung sections and that the reactivity of pulmonary 1981; 240:G371 G375. artery rings is unchanged from sham animals. These 8. Knowles RG, Merrett M, Salter M, Moncada S. Differential induction of brain, liver and lung nitric oxide synthase by endotoxin in findings document that changes in pulmonary NOS levthe rat. Biochem J 1990; 270:833 836. els and pulmonary artery ring reactivity correlate with 9. Kamp DW, Dunn MM, Sbalchiero JS, Knap AM, Weitzman SA. the development of intrapulmonary vasodilatation. In ad- Contrasting effects of alveolar macrophages and neutrophils on dition, PVL animals develop a hyperdynamic circula- asbestos-induced pulmonary epithelial cell injury. Am J Physiol 1994; 266:L84 L91. tion 19 as do CBDL animals, 20 yet our PVL animals have 10. Laemmli UK. Cleavage of structural proteins during the assembly no detectable alteration in the pulmonary vasculature. of the head of bacteriophage T4. Nature 1970; 227:680 685. This observation supports that the development of portal 11. Kobzik L, Bredt DS, Lowenstein CJ, Drazen J, Gaston B, Sughypertension and a hyperdynamic circulation alone is not arbaker D, Stamler JS. Nitric oxide synthase in human and rat lung: immunocytochemical and histochemical localization. J sufficient to cause pulmonary vascular alterations in our Resp Cell Mol Biol 1993; 9:371 377. models. 12. Ku DD. Mechanism of thrombin-induced endothelium-dependent The mechanisms responsible for triggering intrapulactivity. J Cardiovasc Pharmacol 1986; 8:29 36. coronary vasodilatation in dogs: role of its proteolytic enzymatic monary vasodilatation after CBDL and in human hepato- 13. Ku DD, Winn MJ, Grisby T, Caulfiel JB. Human coronary vascular pulmonary syndrome are unknown. Our results show smooth muscle and endothelium-dependent response after storage at 075 C. Cryobiology 1992; 29:199 209. that a combination of hepatic injury and portal hypertension are needed, over the time frame studied, to induce 14. Sogni P, Moreau R, Gadano A, Lebrec D. The role of nitric oxide in the hyperdynamic circulatory syndrome associated with portal pulmonary vascular dilatation. In addition, they indicate hypertension. J Hepatol 1995; 23:218 224. that increased enos levels and enhanced NO reactivity 15. Niederberger M, Gines P, Martin P, Tsai P, Morris K, McMurthy in the pulmonary vasculature correlate with the onset of I, Schrier RW. Comparison of vascular nitric oxide production and systemic hemodynamics in cirrhosis versus prehepatic portal hypulmonary vascular dilatation. These findings are consispertension in rats. Hepatology 1996; 24:947 951. tent with the concept that hepatic injury after CBDL 16. Atucha NM, Shah V, Garcia-Cardena G, Sessa WE, Groszman RJ. results in either increased or decreased levels of mediators Role of endothelium in the abnormal response of mesenteric vessels in rats with portal hypertension and liver cirrhosis. Gasthat influence pulmonary vascular tone by altering pultroenterology 1996; 111:1627 1632. monary vascular NOS production and NO activity. Fu- 17. Heinemann A, Stauber RE. The role of inducible nitric oxide synthase in vascular hyporeactivity of endotoxin-treated and portal ture studies will focus on identifying mediators that might alter pulmonary NOS expression and on determinhypertensive rats. Eur J Pharmacol 1995; 278:87 90. 18. Weigert AL, Martin PY, Niederberger M, Higa EMS, McMurty IF, ing whether NOS inhibition in the pulmonary vascula- Gines P, Schrier RW. Endothelium-dependent vascular hyporeture improves intrapulmonary vasodilatation and gas ex- sponsiveness without detection of nitric oxide synthase induction change after CBDL. in aortas of cirrhotic rats. Hepatology 1995; 22:1856 1862. 19. Sikuler E, Kravetz D, Groszman RJ. Evolution of portal hyperten- References sion and mechanisms involved in its maintenance in a rat model. Am J Physiol 1985; 248:G618 G625. 20. Lee SS, Girod C, Braillon A, Hadengue A, Lebrec D. Hemodynamic characterization of chronic bile duct-ligated rats: effects of pento- barbital sodium. Am J Physiol 1986; 251:G176 G180. 1. Abrams GA, Jaffe CC, Hoffer PB, Binder HJ, Fallon MB. Diagnostic utility of contrast echocardiography and lung perfusion scan in patients with hepatopulmonary syndrome. Gastroenterology 1995; 109:1283 1288. 2. Lange PA, Stoller JK. The hepatopulmonary syndrome. Ann Intern Med 1995; 122:521 529. 3. Groszmann RJ. Hyperdynamic circulation of liver disease 40 years later: pathophysiology and clinical consequences. Hepatology 1994; 20:1359 1363. 4. Chang S-W, O Hara N. Pulmonary circulatory dysfunction in rats with biliary cirrhosis. Am Rev Respir Dis 1992; 148:798 805. 5. Fallon MB, Abrams GA, McGrath JW, Hou Z, Luo B. Common bile duct ligation in the rat: a model of intrapulmonary vasodilatation and the hepatopulmonary syndrome. Am J Physiol 1997; 272: G779 G784. 6. Easter DW, Wade JB, Boyer JL. Structural integrity of hepatocyte tight junctions. J Cell Biol 1983; 96:745 749. Received January 3, 1997. Accepted March 17, 1997. Address requests for reprints to: Michael B. Fallon, M.D., Liver Center and Division of Gastroenterology and Hepatology, University of Alabama at Birmingham, 425 Lyons-Harrison Research Building, 701 South 19th Street, Birmingham, Alabama 35294-0007. Fax: (205) 975-6381. Supported in part by Physician-Scientist Award NIDDK DK-02030 (to M.B.F.), a grant-in-aid from the Alabama American Heart Associa- tion (to M.B.F.), a grant from the PureGar Company (to M.B.F.), and grant HL-50483 from the National Heart, Lung, and Blood Institute (to D.D.K.).