The eNOS was the only NOS isoform detected in livers of all groups. The amount of eNOS protein in liver homogenates was similar in normal and PPVL rats (Fig. 2A), but was lower in CCl₄ cirrhotic livers (Fig. 2A), although some overlap existed with healthy controls. This is in accordance with the variable severity of the cirrhosis in this model. The iNOS protein content was below the limit of detection in livers of healthy controls and the two groups with portal hypertension (Fig. 2B). The nNOS protein was not detected in any liver homogenate (Fig. 2C). In Western blots of iNOS and nNOS but not in those of eNOS, some atypical bands of smaller proteins were observed (data not shown).

A large variation of caveolin-1 mRNA values was present in all groups. Levels in the two portal hypertensive groups were not significantly different from healthy control values (Table 2).

Levels of endothelin-1 mRNA in the PPVL group were comparable to those of healthy control rats (Table 2), but values of CCl₄ cirrhotic livers were significantly and approximately 40-fold higher (p < 0.05 vs controls) (Table 2).

SOD activity in liver homogenates of healthy controls was 15 (7) U/mg protein and it was 14 (3) U/mg protein in PPVL rats (Table 2). In CCl₄ cirrhotic livers, SOD activity was significantly reduced to 10 (3) U/mg protein (p < 0.05 vs normal livers) (Table 2).

Malondialdehyde, a marker of lipid peroxidation, ranged in normal livers from 2 to 20 pmol/mg liver (median 15) and similar values were measured in PPVL rats. In CCl₄ cirrhotic livers, malondialdehyde levels were significantly elevated, with a median of 26 pmol/mg liver (range 6 to 130) (p < 0.05 vs normal livers) (Table 2).

Male Sprague-Dawley rats (Charles River Wiga, Germany) were used either as healthy controls (n = 14), for prehepatic portal hypertension (n = 6) or for CCl₄ induced cirrhosis (n = 11) (Table 1). In later experiments, male inbred Wistar rats (Animal House Leuven, Belgium) were used similarly as healthy controls (n = 9), for prehepatic portal hypertension (n = 5) or CCl₄ induced cirrhosis (n = 9) (Table 2). Prehepatic portal hypertension was achieved by partial portal vein ligation (PPVL) 31 and haemodynamic measurements were carried out 2 weeks later. CCl₄ induced cirrhosis was obtained by 12 weekly inhalations (Table 1) or ingestion (Table 2) of the hepatotoxin CCl₄, together with phenobarbital 350 mg/l in the drinking water 39. Rats were studied 2 weeks after the last CCl₄ administration. Under pentobarbital anaesthesia (50 mg/kg intraperitoneally), portal venous pressure was measured in all rats, the liver was removed and 2 g of liver tissue were homogenised in 8 ml ice-cold buffer I consisting of 250 mM sucrose, 5 mM MgCl₂.6H₂O and 50 mM Tris/HCl pH 7.4. Homogenates were divided into aliquots and stored at -20°C until further processing. A small slice of liver tissue was put in guanidinium buffer on ice for 30 minutes, snap frozen in liquid nitrogen and stored at -80°C until further processing. Additionally, a small liver sample was fixed and used for haematoxylin-eosin stained paraffin-embedded sections; only those CCl₄ rats with micronodular cirrhosis were maintained for analysis.

SDS/PAGE 7.5 % gel electrophoresis was run with diluted homogenates containing 30 μg of protein and with marker proteins (Sigma, St. Louis, USA) including a 120 kD protein, E. coli β-galactosidase. All protein concentrations were measured using the Bradford method (Bio-Rad Labs, Hemel Hempstead, UK) and with bovine serum albumin as standards. As positive controls were taken: a lysate of human aorta endothelial cells (Transduction Labs, Lexington, USA) for eNOS; a homogenate of rat brains for nNOS; and a liver of a rat given LPS 800 μg/kg IV (Sigma, St. Louis, USA) 6 hours before harvesting for iNOS. Samples of both portal hypertensive conditions and healthy controls were run simultaneously on the same gel. Two liver homogenate samples per liver were run. After blotting on a nitrocellulose membrane, blots were blocked overnight at 4°C. Blots were incubated for 2 hours with a mouse monoclonal antibody respectively against eNOS, nNOS or iNOS (Transduction Laboratories, Lexington, USA), dissolved at 1:1000 in buffer II (10 mM Tris-HCl pH 7.6, 0.1 M NaCl, Tween-20 at 0.1 %). Subsequently, blots were incubated with sheep anti-mouse IgG, horseradish peroxidase-labelled (Amersham, Bucks, UK), at 1:3000 diluted in 5 % skimmed milk powder blocking solution for one hour. After washing, detection reagents (ECL Western blotting system, Amersham, Bucks, UK) were added and blots were shortly exposed to an autoradiography film (Nen Life Science Products, Boston, USA) (Figs. 2-4) 8. To check for adequate protein loading and blotting, all blots were stained afterwards with Ponceau S red dye (Sigma, St. Louis, USA).

Hepatic caveolin-1 or endothelin-1 mRNA levels were assessed semi-quantitatively with RT-PCR, using serial dilutions of cDNA as a measure for the amount of specific mRNA in the different livers. Briefly, total RNA was extracted in a single step procedure 40. The precipitated RNA was dissolved in 20-μl DEPC-treated water and the concentration was measured using the Ribogreen RNA quantitation kit (Molecular Probes, Eugene, USA), with ribosomal RNA as standard. One μg of this RNA was used for cDNA synthesis with M-MLV reverse transcriptase (GibcoBRL, Life Technologies, Merelbeke, Belgium) and random primers (Amersham Pharmacia Biotech, Little Chalfont, UK) in a volume of 20 μl, 1 hour at 37°C. The reaction was stopped by heating in boiling water for 1 min.

The PCR primer set used for the detection of rat caveolin-1 mRNA (access number Z 46614) was:

P18: 5'-CCG.GGA.ACA.GGG.CAA.CAT.CTA.CAA.GCC-3' positions 82–108;

M28: 5'-GCC.GTC.RAA.ACT.GTG.TGT.CCC.TTC.TGG-3' positions 251–277, resulting in a fragment of 195 bp. Note that R stands for [A,G].

The PCR primer set used for the detection of rat endothelin-1 mRNA (preproendothelin-1) (access number NM 612548) was:

P1: 5'-CAG.GTC.CAA.GCG.TTG.CTC.CTG.CTC.CTC.C-3' positions 328–355;

M2: 5'-CAC.CAC.GGG.GCT.CTG.TAG.TCA.ATG.TGC.TCG-3' positions 782–811, resulting in a fragment of 483 bp.

PCR determinations were performed on a dilution series of each sample. The first sample contained cDNA equivalent to 0.25 μg total RNA; each following sample was diluted to half the concentration of the previous one. The PCR mixture contained 5 μl of cDNA solution, 6.25 pmol of each primer, 0.2 μM of each dNTP, 1 U of Taq DNA polymerase adjusted by PCR buffer (10 mM Tris-HCl pH 8.3, 50 mM KCl, 2.5 mM MgCl₂ and 0.01 % gelatin) in a final volume of 50 μl. Samples were overlaid with 100 μl mineral oil. PCR conditions were identical for both primer sets: denaturing 5 min at 95°C; 45 cycles of 1 min at 95°C, 45 sec at 58°C and 45 sec at 72°C; and a final step for 5 min at 72°C, after which the samples were stored at 4°C. Samples were analysed on a 2 % agarose gel. Samples of both portal hypertensive conditions and healthy controls were separated simultaneously on the same gel. The majority of the samples were amplified and analysed at least in duplicate. Results are given as the highest dilution that gave a positive signal on the gel (Table 2).

SOD activity in liver homogenates in buffer I was diluted 400 times with buffer I and measured with a RANSOD kit (Randox Laboratories, Crumlin, UK) according to the manufacturer's instructions. In brief, xanthine oxidase generates O₂^-, which reacts with a chromogen to form a red formazan dye that is photometrically quantified. One SOD unit was defined as 55 % inhibition of dye formation. SOD activity was expressed as U/mg protein (Table 2).

Determination of malondialdehyde was performed as published before 41. Briefly, liver homogenates in buffer I were run together with 1,1,3,3-tetraethoxypropane as standard and buffer I as blanks. After the addition of phosphoric acid and thiobarbituric acid, samples were heated at 80°C for 15 min. Longer and more intense heating created too much interference of sucrose 42 (own personal observation). Ice cooled reaction products were further separated by high performance liquid chromatography-reverse phase technique 41. The latter step is necessary to eliminate other substances that had reacted with thiobarbituric acid 43. Results are expressed as pmol malondialdehyde/mg liver wet weight (Table 2).

Data are given as mean (SD) or as median [range] for respectively normally and non-normally distributed data. We made comparisons of the 3 groups (normal, PPVL, CCl₄ cirrhosis) by one-way-analysis of variance in case of normally distributed data with equal variances. If other cases, we used analysis of variance-on-ranks, where the sum of ranks of each group was compared. When significant differences between groups means were found, the Scheffe's post hoc test was performed to identify the groups. Significance level was always taken at α = 0.05. Statistical analyses were carried out with Sigma STAT 2.0 (Jandel Corporation, San Rafael, USA).

Written approval for the present experiments was obtained from the Ethical Committees for Animal Research of the Catholic University of Leuven, Belgium, and of the University of Berne, Switzerland.