A role for asymmetric dimethylarginine in the pathophysiology of portal hypertension in rats with biliary cirrhosis


  • See Editorial on Page 1255.

  • Potential conflict of interest: Nothing to report.


Reduced intrahepatic endothelial nitric oxide synthase (eNOS) activity contributes to the pathogenesis of portal hypertension (PHT) associated with cirrhosis. We evaluated whether asymmetric dimethylarginine (ADMA), a putative endogenous NOS inhibitor, may be involved in PHT associated with cirrhosis. Two rat models of cirrhosis (thioacetamide [TAA]-induced and bile duct excision [BDE]-induced, n = 10 each), one rat model of PHT without cirrhosis (partial portal vein–ligated [PPVL], n = 10), and sham-operated control rats (n = 10) were studied. We assessed hepatic NOS activity, eNOS protein expression, plasma ADMA levels, and intrahepatic endothelial function. To evaluate intrahepatic endothelial function, concentration–effect curves of acetylcholine were determined in situ in perfused normal rat livers and livers of rats with TAA- or BDE-induced cirrhosis (n = 10) that had been preincubated with either vehicle or ADMA; in addition, measurements of nitrite/nitrate (NOx) and ADMA were made in perfusates. Both models of cirrhosis exhibited decreased hepatic NOS activity. In rats with TAA-induced cirrhosis, this decrease was associated with reduced hepatic eNOS protein levels and immunoreactivity. Rats with BDE-induced cirrhosis had eNOS protein levels comparable to those in control rats but exhibited significantly higher plasma ADMA levels than those in all other groups. In normal perfused liver, ADMA induced impaired endothelium-dependent vasorelaxation and reduced NOx perfusate levels, phenomena that were mimicked by NG-nitro-L-arginine-methyl ester. In contrast to perfused livers with cirrhosis induced by TAA, impaired endothelial cell-mediated relaxation in perfused livers with cirrhosis induced by BDE was exacerbated by ADMA and was associated with a decreased rate of removal of ADMA (34.3% ± 6.0% vs. 70.9% ± 3.2%). In conclusion, in rats with TAA-induced cirrhosis, decreased eNOS enzyme levels seem to be responsible for impaired NOS activity; in rats with biliary cirrhosis, an endogenous NOS inhibitor, ADMA, may mediate decreased NOS activity. (HEPATOLOGY 2005;42:1382–1390.)

Portal hypertension (PHT) is a major complication of cirrhosis, irrespective of its etiology. The initial event in the pathogenesis of PHT associated with cirrhosis is an increased intrahepatic vascular resistance to portal flow, whereas an increase in splanchnic blood flow maintains and exacerbates PHT.1, 2 Although increased intrahepatic vascular resistance is primarily caused by structural changes such as fibrosis, thrombosis, and nodule formation, a nonstructural, dynamic vascular factor has also been demonstrated.3 Increased hepatic vascular tone is mediated mainly by activated hepatic stellate cells and is the result of an imbalance between vasodilator and vasoconstrictor forces. More specifically, increased production of vasoconstrictors (e.g., endothelin-1, tromboxane A2, and angiotensin II) is unopposed due to reduced bioavailability of NO, a potent intrahepatic vasodilator.4–7 The molecular basis of intrahepatic NO deficiency has consistently been attributed to a decreased activity of endothelial NO synthase (eNOS).4–6, 8 However, the mechanisms of this endothelial dysfunction remain ill-defined and may vary depending on the cause of the underlying liver disease.8–10

In the early 1990s, Vallance and colleagues11 showed that NO synthesis could be inhibited by the endogenous circulating amino acid asymmetric dimethylarginine (ADMA). ADMA and its vasoinactive stereoisomer, symmetric dimethylarginine (SDMA), are synthesized via enzymatic methylation of L-arginine residues in proteins and are released during proteolysis. Unlike SDMA, which is entirely eliminated via renal excretion, ADMA is primarily metabolized to citrulline and dimethylamine by the liver enzyme dimethylarginine dimethylaminohydrolase.12, 13 Accordingly, it has been suggested that impaired liver function leads to increased plasma levels of ADMA.13, 14 Several recent studies have linked ADMA to endothelial dysfunction and cardiovascular disease, such as arterial hypertension, and multiorgan failure.12, 14 However, the specific intrahepatic hemodynamic effects of ADMA remain unclear.

The aim of the present study was to evaluate in 3 different rat models of PHT whether the pathophysiology of PHT might differ depending on the etiology of cirrhosis, and to determine whether ADMA, an endogenous NOS inhibitor, may be implicated in the pathophysiology of PHT.


PHT, portal hypertension; eNOS, endothelial NO synthase; ADMA, asymmetric dimethylarginine; TAA, thioacetamide; BDE, bile duct excision; NOx, nitrate/nitrite; SDMA, symmetric dimethylarginine; PPVL, partial portal vein ligation; L-NAME, NG-nitro-L-arginine-methyl ester; DDAH, dimethylarginine dimethylaminohydrolase.

Materials and Methods


Male Wistar rats (Animal House, University of Leuven, Leuven, Belgium) weighing 200–250 g were used. All experiments and procedures were undertaken in accordance with local animal care guidelines.

Models of PH.

Biliary cirrhosis was induced via bile duct excision (BDE).15 At laparotomy, the bile duct was cut between a ligature close to the hilum of the liver and one close to the duodenum.

Thioacetamide (TAA)-induced cirrhosis, a model of nonbiliary cirrhosis, was induced using the method described by Li and colleagues.16 Briefly, TAA was administrated orally at weekly intervals for 18 weeks. The dose was varied between 0.015% and 0.06% according to weekly changes in the animal's body. The aim was to keep body weight between the preset limits of 200 and 250 g.

Partial portal vein ligation (PPVL) was performed by placing a 21-gauge needle on the portal vein. A nonabsorbable surgical thread ligature was placed around the vein and needle, and the needle was then withdrawn.17

For sham operation, the duodenum, portal vein, and bile duct were exposed during laparotomy. The abdomen was closed 15 minutes later.

Experiments were performed within 1 week of the last dose of TAA and 4 weeks and 3 weeks after BDE and PPVL, respectively.

Hemodynamic Assays.

Measurements were performed under pentobarbital anesthesia (100 μL/100 g body weight) to allow cannulation of the right carotid artery and the portal vein, as previously described.8, 18

Biochemical Analysis.

Blood taken via cardiac puncture was collected in Vacutainer tubes (Becton-Dickinson, Erembodegem, Belgium). Urine was collected from rats 24 hours before hemodynamic studies for determination of creatinine clearance. Biochemical assays were undertaken using standardized automated procedures.

Aminopyrine Breath Test.

Hepatocyte microsomal function was assessed using an aminopyrine breath test, which was undertaken as previously described.18 The ABT-k represents the aminopyrine breath test constant; it is the regression coefficient (multiplied by 1,000) of the least square fit of radioactivity measurements in samples versus time.

NOS Activity Assay.

The conversion of L-3H-arginine to L-3H-citrulline was used as an index of NOS activity. A previously described method19 was applied with slight modifications. Briefly, fresh liver was homogenized in a sucrose buffer at a final concentration of 0.2 g wet weight of liver/mL. Samples containing 100 μL of homogenate were incubated with 100 μL of buffer (40 mmol/L HEPES [pH 7.4], 1 mmol/L reduced β-NADPH, 2 mmol/L CaCl2, 12 μmol/L arginine, containing 0.45 μCi L-2,3-3H-arginine [Dupont Products, Boston, MA]) at 37°C. Samples were processed in duplicate and were incubated in the presence or absence of NG-nitro-L-arginine-methyl ester (L-NAME) (20 mmol/L). The reaction was stopped by adding 1 mL stop buffer (10 mmol/L EGTA, 1 mmol/L citrulline, 100 mmol/L PIPES [pH 5.5]). The reaction mixture was applied to a cation-binding Dowex W50 column, and the counts per minute in generated L-citrulline were measured. L-NAME–inhibited NOS activity was determined and was considered to be specific NOS activity.

Western Blot Analysis for eNOS.

Fresh livers were homogenized in Chaps Buffer (40% glycerol, 0.2 mol/L K2PO4 × 7H2O [pH 7.2], 20 mmol/L Chaps [Sigma, St. Louis, MO]). The final concentration was 0.25 g wet weight of liver/mL. The protein concentration was measured according to Bradford.20 Equal amounts of protein (100 μg) from each sample were separated by 7% SDS-PAGE. The separated proteins underwent electrophoretic transfer onto a nitrocellulose Protran membranes (Schleicher & Schuell, Dassel, Germany) and were blocked in blotto-Tween overnight. They were subsequently incubated with mouse monoclonal immunoglobulin G against eNOS (Transduction Labs, Lexington, KY) (1:500), then incubated for 90 minutes with a secondary antibody (peroxidase-conjugated rabbit anti–mouse immunoglobulin, Dako). The product was visualized using chemiluminescence (ECL, Amersham, Rosendaal, The Netherlands). Membranes were stained with a Ponceau stain to confirm equal protein loading and transfer between lanes. Densitometry quantification of Western blot signal intensity was undertaken using Un-Scan-IT Gel Software (Silk Scientific, Orem, UT) and was expressed as a percentage of the mean value for sham-operated livers.

Immunohistochemical Staining.

For eNOS immunohistochemistry, a three-step indirect immunoperoxidase method was applied as previously described.8 Intensity of sinusoidal staining was scored blindly using a semiquantitative scale: 0, no; +1, weak; +2, moderate; and +3, strong reactivity. For each specimen, 10 different high-power fields were scored; the mean value was taken as the final score. This scoring system, which had been validated in a previous study,8 was applied by our pathologist (T. R.).

In Situ Liver Perfusion System.

The hepatic vascular response to ADMA was evaluated using a flow-controlled liver perfusion system.7 After laparotomy, a ligature was passed around the suprarenal portion of the inferior vena cava, which was injected with heparin (1,400 U/kg body weight). The portal vein was cannulated with a 14-gauge catheter and the liver was perfused with Krebs solution, which was pumped (Masterflex, Vernon Hills, IL) from a reservoir into an overflow chamber via a PALL Ultipor filter (Pall, East Hills, NY). The perfusate was oxygenated with a silastic tubing oxygenator (Sophysa, Nijvel, Belgium) at 37 ± 0.5°C. The abdominal aorta and inferior vena cava were cut below the ligature, allowing exsanguination of the animal and escape of the perfusate. Following thoracotomy, the right atrium was cannulated (PE-240; Portex, Kent, UK), and the catheter was advanced to the outlet of the hepatic vein. When the perfused effluent was clear, the system was closed and recirculation was established using a volume of 150 mL at a constant flow rate of 35 mL/min. An ultrasonic flow probe (T201; Transonic System, Ithaca, NY) and a pressure gauge were positioned in line, immediately upstream to the inlet cannula, so that portal flow and perfusion pressure could be continuously monitored. Another pressure transducer was placed immediately downstream of the outlet cannula for measurement of ouflow pressure.

The criteria for liver viability included a healthy gross appearance of the liver, a stable perfusion pressure (starting value ± 1 mm Hg), and a stable buffer pH (7.4 ± 0.1) during an initial stabilization period of 30 minutes. If viability criteria were not met, the experiment was terminated.

To evaluate the effect of ADMA and other substances on intrahepatic vascular resistance in healthy rats, the liver was preconstricted with the α1-adrenergic agonist methoxamine (Sigma, St. Louis, MO) at a concentration of 10−4 mol/L. Five minutes later, ADMA 10−3 mol/L (Merck, Darmstadt, Germany), vehicle (saline), SDMA 10−3 mol/L (Merck), L-NAME 10−3 mol/L (Sigma), or a combination of ADMA 10−3 mol/L and L-NAME 10−3mol/L (dissolved in equal volumes) were administered over a 1-minute period. Subsequently, concentration–effect curves in response to cumulative doses of acetylcholine (10−7 to 10−5 mol/L, one log increase every 1.5 minutes) (Sigma) were determined for each experimental paradigm. Similar experiments were undertaken using rats with BDE-induced cirrhosis and rats with TAA-induced cirrhosis, in which the effect of 10−3 mol/L ADMA (n = 5) was compared with that of vehicle (n = 5).

The concentration of ADMA chosen was based on a previously determined dose–response relationship using concentrations between 10−6 and 10−3 mol/L (data not shown). L-NAME is an inactive prodrug that upon hydrolization is bioactivated to NG-nitro-L-arginine, a direct inhibitor of NOS.21 The response to cumulative doses of acetylcholine was expressed as percent change in perfusion pressure, as previously described.5, 7

Determination of Dimethylarginines and NOx.

ADMA was determined in rat plasma, bile, and perfusates using solid phase extraction, subsequent derivatization with phenylisothiocyanate, separation via liquid chromatography, and UV detection. For solid phase extraction, the samples were added to a polymeric cationic exchanger (Oasis-MCX; Waters, Milford, MA). Following elution, subsequent derivatization with phenylisothiocyanate was undertaken in accordance with the manufacturer's instructions (Waters). For calibration and identification, arginine, ADMA, and SDMA were applied. In general, derivatization of an amino acid with phenylisothiocyanate leads to the formation of phenylthiocarbamyl-amino acids. Separation of these amino acids was achieved via reversed phase high-performance liquid chromatography according to Teerlinck and colleagues.13, 14, 22 Samples were spiked with known amounts of these amino acids for identification of peaks, recovery, and quantitation. A calibration curve (over the range 1–20 nmol ADMA) was constructed; the detection limit was 200 pmol/L.

NOx content, an index of NO production, was assayed in perfusates by applying a fluorometric method that uses 2,3-diaminonaphtalene.23

In the perfusion experiments using normal rats, 2-mL samples of perfusate were obtained just before the administration of methoxamine and immediately after determining the dose–response relationship to acetylcholine. The difference in NOx production between these two sampling times was calculated for each experimental paradigm. At these same time points, ADMA levels were determined in perfusates from normal rats, and rats with BDE-induced cirrhosis and TAA-induced cirrhosis, that were incubated with either vehicle or ADMA 10−3 mol/L. Subsequently, removal rates (RR) were calculated by applying the equation RR = 100 × (1 − Cafter/Cbefore), where Cafter is the perfusate level at the end of the experiment and Cbefore is the injected dose of ADMA 10−3 mol/L.

Statistical Analysis.

Data are expressed as the mean ± SEM. For comparison of multiple groups, ANOVA or ANOVA on ranks was applied when appropriate. When significant, subsequent multiple comparisons were undertaken. If only two conditions were compared, an unpaired Student t test was used. P values less than .05 were considered significant.


In the first part of the experiments, differences in intrahepatic eNOS protein and activity and ADMA plasma concentrations were determined in the different rat models of PHT. In the second part, the hemodynamic effect and metabolic fate of ADMA was studied further in the perfused liver of normal rats and the perfused liver of rats with biliary and nonbiliary cirrhosis.

Baseline Characteristics of the Different Experimental Groups

Portal pressure was significantly increased in the rat models of PHT (Table 1). In rats with PHT, mean arterial pressures were lower than in sham-operated controls, suggesting the presence of a hyperdynamic state. Creatinine clearances in all groups were comparable. Rats with cirrhosis exhibited increased serum aminotransferase, bilirubin, and γ-glutamyltransferase levels and impaired hepatocyte microsomal function.

Table 1. Baseline Characteristics of the Different Portal Hypertensive Models and the Sham-Operated Control Group
 Sham (n = 10)TAA (n = 9)BDE (n = 10)PPVL (n = 10)
  • NOTE. Data are expressed as the mean ± SEM.

  • Abbreviations: AST, aspartate aminotransferase; GGT, γ-glutamyltransferase; ABT-k, aminopyrine breath test constant.

  • *

    P < .05 versus sham-operated animals.

  • P < .05 versus BDE and PPVL rats.

  • P < .05 versus sham, TAA, and PPVL rats.

  • §

    P < .05 versus sham, BDE, and PPVL rats.

  • P < .05 versus sham and PPVL rats.

Portal pressure (mm Hg)5.7 ± 0.112.1 ± 0.8*13.5 ± 0.9*12.5 ± 0.9*
Mean arterial pressure (mm Hg)127 ± 4.788 ± 8*,110 ± 4.7*109 ± 3.8*
Serum creatinine (mg%)0.6 ± 0.030.6 ± 0.010.4 ± 0.030.7 ± 0.03
Serum Na+ (mEq/L)143.2 ± 0.7143.4 ± 1.5141.1 ± 1.0143.7 ± 1.2
Creatinine clearance (mL/hr/100g)9.0 ± 0.49.9 ± 0.510.2 ± 0.59.0 ± 0.3
AST (U/L)56.7 ± 5.3192.3 ± 14.4§99.7 ± 22.660.0 ± 25.1
GGT (U/L)2.0 ± 0.018 ± 2*19.5 ± 4.2*3.2 ± 1.2
Bilirubin (mg%)0.1 ± 0.00.24 ± 0.049.8 ± 3.20.1 ± 0.0
ABT-k−8.7 ± 0.8−4.8 ± 0.8−6.9 ± 0.7−9.7 ± 1.6

Hepatic NOS Activity

In rats with biliary and nonbiliary cirrhosis, specific hepatic NOS activity was substantially less than that in sham-operated animals (P < .05). The corresponding activity in rats with PPVL was similar to that in sham-operated rats (Fig. 1).

Figure 1.

Specific NOS activity on liver homogenates in the different models of PHT (n = 8 per group). Both groups with cirrhosis—TAA-induced and BDE-induced—exhibited less NOS activity than sham-operated rats. *P < .05. NOS, NO synthase; TAA, thioacetamide; BDE, bile duct excision; PPVL, partial portal vein ligation.

eNOS Western Blot Analysis and Immunohistochemistry

Western blotting undertaken on liver homogenates from rats with TAA-induced cirrhosis exhibited decreased amounts of eNOS protein: densitometry analysis indicated that eNOS protein in livers of TAA-treated rats was 75% ± 5% of that in sham-operated controls (P < .05) (Fig. 2A). In the two other models of PHT, the degree of eNOS protein expression was comparable to that in control rats (Fig. 2A). Immunoreactivity for eNOS in normal liver and in the models of PHT was homogeneously distributed in endothelial cells lining portal tract vessels and sinusoids. Only TAA-induced livers with cirrhosis exhibited decreased sinusoidal expression. Semiquantitative scores of eNOS immunohistochemical expression in the different models are shown in Fig. 2B.

Figure 2.

(A) Top panel: Densitometric analysis of Western blots from sham-operated rats, rats with TAA-induced cirrhosis, rats with BDE-induced cirrhosis, and rats with PPVL (n = 5 per group). *P < .05 versus sham-operated rats. Bottom panel: Representative blot of eNOS protein for the different experimental groups. (B) Semiquantitative evaluation of eNOS immunohistochemistry in liver biopsies from the different rat models: 0, none; 1, weak; 2, moderately strong; 3, strong, 4, reactivity (vertical arrow: P < .001 vs. other groups). TAA, thioacetamide; BDE, bile duct excision; PPVL, partial portal vein ligation.

Plasma Levels of ADMA and SDMA

The plasma concentrations of ADMA in sham-operated rats (0.69 ± 0.37 μmol/L) and in rats with TAA-induced cirrhosis (0.58 ± 0.10 μmol/L) were less than those in rats with BDE-induced cirrhosis (2.92 ± 0.59 μmol/L; P < .001 vs. other groups) (Fig. 3). Plasma SDMA levels were 0.42 ± 0.13 μmol/L in sham-operated rats, 0.33 ± 0.17 μmol/L in rats with TAA-induced cirrhosis and 0.46 ± 0.22 μmol/L in rats with biliary cirrhosis (P value not significant). ADMA and SDMA levels in rats with PPVL were below the respective detection limits.

Figure 3.

Plasma levels of ADMA. Rats with BDE-induced cirrhosis exhibited significantly higher levels than all other groups (n = 8 per group). *P < .001. B.D., below detection limit; TAA, thioacetamide; BDE, bile duct excision; PPVL, partial portal vein ligation.

Hepatic Vascular Response to ADMA in the Perfused Rat Liver

Normal Rats.

In normal rat livers after methoxamine-induced vasoconstriction and injection of vehicle (n = 5 for each experimental paradigm), an incremental vasorelaxation occurred in response to cumulative doses of acetylcholine. In contrast, livers after incubation with ADMA exhibited less relaxation in response to acetylcholine (10−7 mol/L, −0.6 ± 1.6% vs. −7.0 ± 0.7% [ADMA vs. vehicle], P = .006; 10−6 mol/L, −2.9 ± 1.8% vs. −11.7 ± 1.3%, P = .016; and 10−5 mol/L, −2.5 ± 2.4% vs. −16.9 ± 2.3%, P = .002) (Fig. 4A). SDMA, the vasoinactive stereoisomer of ADMA, was associated with a similar concentration–effect curve as vehicle (10−7 mol/L, −10.6 ± 1.9% vs. −7.0 ± 0.7%; 10−6 mol/L, −15.3 ± 1.7% vs. −11.7 ± 1.3%; 10−5 mol/L, −18.6 ± 1.6% vs. −16.9 ± 2.3%; all P values not significant) (Fig. 4A). Comparison of ADMA with L-NAME revealed no significant differences (10−7 mol/L, −0.6 ± 1.6% [for ADMA] vs. −2.3 ± 1% [for L-NAME]; 10−6 mol/L, −2.9 ± 1.8% vs. −5.7 ± 1%; 10−5 mol/L, −2.5 ± 2.4% vs. −8.4 ± 2.2%; all P values not significant) (Fig. 4B). As with ADMA, a similar significant difference was found between L-NAME and vehicle (10−7 mol/L, P = .005; 10−6 mol/L, P = .006; 10−5mol/L, P = .028). When preincubation with a combination of ADMA and L-NAME was undertaken, the increase in perfusion pressure was similar for all concentrations of acetylcholine studied (10−7 mol/L, 4.3 ± 0.8; 10−6 mol/L, 5.6 ± 1.7; 10−5 mol/L, 5.9 ± 2.6; P = .590) (Fig. 4B). These increases in perfusion pressure were significantly greater than those associated with L-NAME or ADMA alone for each concentration of acetylcholine studied; P < .001).

Figure 4.

Dose–response curves in response to cumulative doses of acetylcholine in rat livers perfused in situ (n = 5 for each condition). (A) Comparison of normal livers incubated with vehicle (●), ADMA 10−3 mol/L (▪), or SDMA 10−3 mol/L (○). *P < .001 (▪) versus (●) and (○) (ANOVA). (B) Comparison of normal livers incubated with ADMA 10−3 mol/L (▪), L-NAME 10−3 mol/L (▿), and a combination of ADMA 10−3 mol/L and L-NAME 10−3 mol/L (+). *P < .05 (+) versus (▪) and (▿); **P < .001 (+) versus (▪) and (▿); ***P < .05 (+) versus (▿) (ANOVA). (C) Comparison of rat livers with biliary cirrhosis with (♦) or without ⋄) preincubation with ADMA versus normal rat livers preincubated with vehicle (●). *P < .001 (♦) versus (●); **P < .001 (♦) versus ⋄) and (●). (D) Comparison of TAA-induced preascitic livers with cirrhosis with (▾) or without (▿) preincubation with ADMA versus normal rat livers preincubated with vehicle (●). *P < .05 (▾) and (▿) versus (●). AcH, acetylcholine.

Rats With BDE-Induced Cirrhosis.

In BDE-induced cirrhosis, the effect of acetylcholine was less than that in normal livers preincubated with vehicle (10−7 mol/L, −1.7 ± 0.9% vs. −7.0 ± 0.7%, P = .002; 10−6 mol/L, −5.7 ± 0.8% vs. −11.7 ± 1.3%, P = .005; and 10−5 mol/L, −8.1 ± 0.9% vs. −16.9 ± 2.3%, P = .007) (Fig. 4C). The impaired relaxation in rats with BDE-induced cirrhosis (n = 5 for each experimental paradigm) was even greater after preincubation with ADMA 10−3 mol/L and resulted in paradoxical vasoconstriction (10−7 mol/L, 0.6 ± 0.1%; 10−6, 5.4 ± 1.8%; 10−5, 6.6 ± 1.2%; all P < .05 vs. rats with BDE-induced cirrhosis incubated with vehicle).

Rats With TAA-Induced Cirrhosis.

Like rats with BDE-induced cirrhosis, rats with preascitic TAA-induced cirrhosis (n = 5 for each experimental paradigm) exhibited endothelial dysfunction compared with normal livers preincubated with vehicle (10−7 mol/L, −0.8 ± 0.6% vs. −7.0 ± 0.7%, P = .008; 10−6 mol/L, −1.9 ± 1.6% vs. −11.7 ± 1.3%, P = .001; and 10−5 mol/L, −5.0 ± 1.8% vs. −16.9 ± 2.3%, P = .004) (Fig. 4D). Addition of ADMA to perfused TAA-induced livers with cirrhosis was associated with a similar response to acetylcholine to that associated with TAA-induced livers with cirrhosis incubated with vehicle (10−7 mol/L, 0.6 ± 1.1%; 10−6, −1.2 ± 1.6%; 10−5, −2.6 ± 2.2%; all P values not significant vs. TAA-induced livers with cirrhosis in the absence of ADMA).

NOx in Perfusates.

Comparison of NOx concentration before methoxamine administration and after the dose–response to acetylcholine revealed a comparable decrease in NOx levels in healthy livers incubated with ADMA or L-NAME (−29.6 ± 3.1 nmol/L and −29 ± 10.9 nmol/L, respectively; P < .05 vs. SDMA and vehicle). Normal livers incubated with both L-NAME and ADMA exhibited a further decrease in NOx levels (−52.9 ± 5.4 nmol/L; P = .062 vs. normal rat livers incubated with ADMA or L-NAME alone) (Fig. 5).

Figure 5.

Differences in NOx concentration in normal perfused rat livers before the administration of methoxamine and immediately after the end of the determination of the dose–response relationship to acetylcholine in the different experimental paradigms: vehicle, SDMA, ADMA, L-NAME, and L-NAME combined with ADMA. *P < .05 versus SDMA and vehicle. NOx, nitrate/nitrite; SDMA, symmetric dimethylarginine; ADMA, asymmetric dimethylarginine; L-NAME, NG-nitro-L-arginine-methyl ester; FYS, physiological saline or vehicle.

ADMA in Bile and Perfusates.

In a subset of rats with BDE-induced cirrhosis (n = 4), bile was obtained via puncture of the extrahepatic bile duct. ADMA was undetectable in bile.

In perfusate samples obtained either just before methoxamine administration or at the end of the experiments, ADMA was undetectable in normal livers as well as livers with TAA-induced cirrhosis or BDL-induced cirrhosis that had been infused with vehicle. When incubated with ADMA 10−3 mol/L at the end of the perfusion experiment, 264 ± 53.8 μmol/L and 290.7 ± 32.3 μmol/L were retrieved in the perfusate of normal rats (n = 5) and rats with TAA-induced cirrhosis (n = 5), respectively. In rats with BDE-induced cirrhosis (n = 5), the recovery was significantly higher (656.8 ± 59.9 μmol/L; P = .007). The associated removal rates are shown in Fig. 6.

Figure 6.

Removal rates at the end of determinations of the dose–response relationship to cumulative doses of acetylcholine in normal rats, rats with TAA-induced cirrhosis, and rats with BDE-induced cirrhosis preincubated with ADMA 10−3 mol/L. *P < .001 versus normal and TAA-treated rats. TAA, thioacetamide; BDE, bile duct excision.


Decreased eNOS-derived NO production is presumed to play a pivotal role in the increased intrahepatic vascular resistance associated with PHT with cirrhosis.4–6, 24 However, different morphological types of cirrhosis and PHT exist depending on the etiology.25 We investigated whether the pathophysiology of PHT might differ according to different conditions. For this purpose, we used three rat models of PHT; namely, TAA-induced cirrhosis, BDE-induced cirrhosis and PPVL; these models are representative of different conditions associated with PHT in humans.

Upon characterization of the different models with regard to eNOS, we observed decreased eNOS activity in both animal models with cirrhosis, in agreement with what is generally accepted.4 However, eNOS protein in endothelial cells lining portal vessels and sinusoids, assayed via immunohistochemistry, and in liver homogenates, assayed via Western blot analysis, was decreased only in rats with TAA-induced cirrhosis. This finding suggests decreased eNOS enzyme levels in this model and is consistent with a previous report.26 In rats with CCl4-induced cirrhosis, another model of nonbiliary cirrhosis, we previously documented a similar decrease in eNOS expression and showed that restoration of hepatic eNOS by gene transfer led to a decrease in PHT, a finding that further supports the hypothesis of decreased eNOS enzyme levels.8 Other groups have shown decreased eNOS activity in rats with CCl4-induced cirrhosis in the presence of unaltered eNOS expression and have suggested that NO deficiency is caused by dysfunctional cofactors of eNOS, such as caveolin and Akt.9, 10 These findings illustrate that intrahepatic eNOS regulation is complex and depends on different factors.

In biliary cirrhosis, eNOS protein expression was normal as assessed via both immunohistochemistry and Western blot analysis, but NOS activity was decreased. These findings suggest that the pathogenesis of impaired intrahepatic NO production might differ depending on the etiology of cirrhosis. Further support for this hypothesis is provided by the observation that eNOS gene transfer in rats with BDE-induced cirrhosis did not restore normal vasodilatory responsiveness, whereas it did in rats with CCl4-induced cirrhosis.8, 27

Because the level of eNOS enzyme was unaltered in our rats with biliary cirrhosis, the cause of the associated decreased NOS activity remains to be explained. Theoretically, eNOS malfunction might result from either a direct inhibition of eNOS or indirect through dysfunctional cofactors. Shah and colleagues19, 27 provided support for the latter hypothesis by demonstrating that enhanced hepatic caveolin-1 protein levels in rats with BDE-induced cirrhosis may mediate reduced NOS activity, since NOS activity was partially restored when excess calmodulin, a protein that competitively binds eNOS, was added to the liver.

Data supporting direct inhibition of intrahepatic NOS have not been reported yet, but a role for ADMA has been proposed.13, 14 In this study, we investigated whether ADMA may be related to intrahepatic endothelial dysfunction. Plasma ADMA levels were increased only in rats with biliary cirrhosis. However, increased plasma levels of ADMA do not necessarily imply inhibition of intrahepatic NOS and subsequent hemodynamic effects. We therefore assessed the vasomotor endothelial function by measuring vasorelaxation in response to acetylcholine. Acetylcholine couples to its endothelial muscarinic M3 receptor and stimulates production of vasodilatory NO from eNOS.7, 28 We demonstrated an impaired endothelium-dependent vasodilator reaction known as “endothelial dysfunction” when normal livers were perfused with ADMA. In contrast to ADMA, SDMA, its vasoinactive stereoisomer, did not induce impaired vasomotor function, consistent with its vasoinactive properties. The effect of ADMA was similar to that of L-NAME, a known inhibitor of NOS.21 ADMA induced a decrease in NOx levels, suggesting that ADMA induces intrahepatic endothelial dysfunction by inhibiting NOS. When ADMA was administrated together with L-NAME, a synergistic increase in perfusion pressure—or paradoxical vasoconstriction—occurred, which was associated with a further decrease in NOx levels. In this perfusion system, these findings probably result from concurrent vasoconstrictor response following endothelial stimulation by acetylcholine and loss of eNOS stimulation, since in rats that received an incubation with both ADMA and L-NAME, the dose-related response to acetylcholine was immediately lost and there was a steady-state positive perfusion pressure.7, 28 In perfused BDL-induced and TAA-induced livers with cirrhosis, endothelial dysfunction was present before preincubation with vehicle. This finding was expected in animals with chronic liver injury and has been documented by others.7, 19 However, when ADMA was added, in BDE rats the pre-existing endothelial dysfunction was further aggravated and resulted in paradoxical vasoconstriction, which is in agreement with an environment where vasoconstrictors highly upregulated1, 24, 29, 30 and where an already dysfunctional eNOS is impeded further or even completely inhibited. In contrast, in TAA-induced rats with cirrhosis the decreased vasorelaxing capacity was not impaired further, which supports the hypothesis of a decreased eNOS enzyme level as a causative factor.

The question remains why ADMA plasma concentrations are highest in biliary cirrhosis. Renal and hepatic function as well as the degree of PHT in BDE and TAA were comparable and therefore cannot explain the differences observed in our models. The presence of cholestasis, a major feature of rats with biliary cirrhosis, might suggest that constituents of bile cause elevated ADMA levels. Dimethylarginine dimethylaminohydrolase (DDAH), the enzyme responsible for the degradation of ADMA, is highly expressed in the liver and makes the liver a key organ in the degradation of ADMA.12, 31 We showed that the removal rate of ADMA in BDE-induced livers with cirrhosis is lower than that in normal livers or nonbiliary livers with cirrhosis. These findings support a decrease in activity of DDAH in rats with BDE-induced cirrhosis. Stuhlinger and colleagues32 have suggested that DDAH is oxidant-labile and is thus subject to inhibition by oxidants derived from endothelial superoxide. Recent evidence suggests that cholestasis causes liver injury through mechanisms that involve mitochondrial oxidative stress.33–36 Accordingly, a decrease in DDAH activity in cholestatic livers, caused by oxidative damage to the enzyme, might explain the increased plasma ADMA levels found in rats with BDE-induced cirrhosis. The possibility that ADMA accumulation is due to mechanical biliary obstruction in rats with BDE-induced cirrhosis, was ruled out by finding undetectable levels of ADMA in bile in this model.

The apparent discrepancy between measured plasma levels of ADMA (10−6 mol/L) in rats with biliary cirrhosis and the concentration of ADMA required (10−3 mol/L) in healthy perfused rat livers can be explained by the high Michaelis constant (KM) of DDAH (>100 μmol/L)—that is, the substrate concentration at which the reaction rate is half of its maximal value.12 Under certain conditions, inhibition of DDAH is known to lead to high local concentrations of ADMA, further feeding the issue of whether plasma levels adequately reflect biologically active ADMA—or, indeed, whether they represent a marker of high intracellular levels.12, 37, 38

In conclusion, we have demonstrated that the pathogenesis of impaired intrahepatic NO production in cirrhosis depends on the etiology of cirrhosis. Whereas in TAA-induced cirrhosis decreased intrahepatic NO bioavailability results from reduced eNOS enzyme levels and activity, in rats with biliary cirrhosis eNOS enzyme levels were unaltered. In rats with biliary cirrhosis, decreased NOS activity could, therefore, arise from eNOS dysfunction. ADMA might be one of the molecules responsible for inhibition of NOS in this context.


The authors appreciate helpful discussions with Dr. David Cassiman.