Endotoxic shock is a major complication of portal hypertension in patients with cirrhosis.1 In patients with cirrhosis, there is an excessive proinflammatory response to lipopolysaccharide (LPS) in comparison with patients without cirrhosis.2–4 High-density lipoproteins (HDLs) have an anti-inflammatory effect by neutralizing circulating LPS.5 Because plasma HDL concentrations are reduced in cirrhosis,6, 7 the excessive proinflammatory response may be a result of a reduction of LPS neutralization by HDL in those patients. In vivo, reconstituted high-density lipoprotein (rHDL) administration inhibits proinflammatory cytokine production8, 9 and reduces organ dysfunction10 or mortality rates in animal models of sepsis.8 HDL also reduces the plasma levels of proinflammatory cytokines and the clinical symptoms associated with LPS administration in humans.11 In addition to having anti-inflammatory properties, HDL increases endothelial nitric oxide synthase (eNOS) activity in 2 ways: by promoting the dissociation between eNOS and caveolin-1 (Cav-1),12 which counteracts caveolin's inhibitory effect on eNOS activity,13 and by activating the major eNOS activator, that is, protein kinase B (Akt),14 which activates eNOS by phosphorylation. In cirrhotic rats with ascites, a recent study has shown that the administration of rHDL reduced the effects of endotoxin on tumor necrosis factor α (TNFα) production and systemic hemodynamics.15 However, the exact mechanisms of the beneficial effects of HDL in this model remain unknown. Thus, the aim of this study was to assess the effects of rHDL administration on the proinflammatory response in cirrhotic rats through the assessment of the LPS levels neutralized by the HDL fraction after LPS challenge, the circulating levels and hepatic expression of lipopolysaccharide binding protein (LBP) and TNFα, and the hepatic expression of inducible nitric oxide synthase (iNOS), LBP, TNFα, and iNOS being 3 major markers of hepatic inflammation. In addition, as HDL has certain effects on eNOS, and because the reduction of the intrahepatic eNOS activity plays a role in increased vascular intrahepatic resistance in cirrhosis, the effects of rHDL on hepatic eNOS and its 2 major modulators, Cav-1 and Akt, were studied in cirrhotic rats. Because eNOS-derived nitric oxide (NO) may stimulate cyclic 3′,5′-guanosine monophosphate (cGMP) production, the effect of rHDL on hepatic cGMP levels was also measured. Moreover, the impact of rHDL administration on the degree of portal hypertension and intrahepatic vascular resistance was assessed.
In patients with cirrhosis, endotoxic shock is a major complication of portal hypertension, which is related partly to intrahepatic endothelial nitric oxide synthase (eNOS) down-regulation. High-density lipoproteins (HDLs), whose plasma levels are reduced in cirrhosis, have an anti-inflammatory effect by neutralizing circulating lipopolysaccharide (LPS), and they increase eNOS activity in endothelial cells. Therefore, the aim of this study was to assess the effects of reconstituted high-density lipoprotein (rHDL) administration on the LPS-induced proinflammatory response, intrahepatic eNOS regulation, and portal hypertension in cirrhotic rats. Cirrhotic and control rats were pretreated with rHDL or saline and challenged with LPS or saline. The neutralization of LPS in HDL was assessed by the measurement of HDL-bound fluorescent LPS levels. Plasma tumor necrosis factor α (TNFα) and lipopolysaccharide binding protein (LBP) levels were measured. The expression of hepatic TNFα, LBP, inducible nitric oxide synthase (iNOS), and caveolin-1 (a major eNOS inhibitor) and the activity of protein kinase B (Akt; a major eNOS activator) and eNOS were determined. The portal pressure was measured. The plasma HDL levels were significantly lower in cirrhotic rats than in control rats. In cirrhotic rats, the plasma levels of HDL-bound fluorescent LPS were 50% lower than those in controls, and they were restored after rHDL administration. The plasma TNFα levels were significantly higher in LPS-challenged cirrhotic rats than in controls and significantly decreased after rHDL administration. rHDL administration decreased hepatic TNFα, LBP, iNOS, and caveolin-1 expression, restored hepatic eNOS and Akt activity, and significantly lowered the portal pressure and intrahepatic vascular resistance. Conclusion: In cirrhotic rats, rHDL administration decreases the hepatic proinflammatory signals induced by LPS, restores the hepatic eNOS activity, and lowers the portal pressure. This suggests that the decrease in circulating HDL in cirrhosis plays a role in the excessive proinflammatory response and intrahepatic eNOS down-regulation. (HEPATOLOGY 2007.)
Materials and Methods
Male Sprague-Dawley rats (Charles River Laboratories, Saint-Aubin-les-Elbeuf, France) were divided into 2 groups. One group included rats with cirrhosis. Cirrhosis was induced by bile duct ligation as described.16 Briefly, under ketamine anesthesia (100 mg/kg body weight intraperitoneally), a midline abdominal incision was made, and the common bile duct was occluded by double ligatures. Then, the common bile duct was sectioned and resected between the 2 ligatures. The abdominal incision was closed. Studies were performed 4 weeks after bile duct ligation to allow the development of secondary biliary cirrhosis. Cirrhosis was confirmed by a macroscopic liver examination. Eighty-four rats had bile duct ligation. Seven rats died before the fourth week (first week, n = 2; fourth week, n = 5). One rat did not develop cirrhosis and was not included. A second group included rats that underwent sham operations (control rats). Studies were performed on rats weighing between 300 and 380 g. The protocols performed in this laboratory were approved by the French Agriculture Office in accordance with European legislation for research involving animals.
Effects of HDL Administration on LPS Distribution in Blood.
This protocol included 12 cirrhotic rats and 15 controls (Fig. 1A). The animals were studied after overnight fasting. They were anesthetized with ketamine (50 mg/kg bw), and a catheter was inserted into the left femoral artery for rHDL and LPS administration and for blood samples. Blood samples were taken to measure the serum levels of cholesterol, HDL, and aminotransferase activity before the administration of any other products. The blood samples were centrifuged (5000g for 10 minutes at 4°C) and immediately stored at −80°C. Either rHDL (80 mg/kg over 4 minutes; 6 cirrhotic rats and 6 controls) or an equivalent volume of saline (6 cirrhotic rats and 9 controls) was administered and followed by an intravenous injection of 50 μg/300 g of fluorescein isothiocyanate (FITC)-LPS (Escherichia coli 0111:B4) 5 minutes later. This dose of rHDL has been shown to decrease mortality and cytokine production after endotoxin exposure in mouse and rabbit models.8, 9, 17, 18 rHDL was prepared from human plasma by cholate dialysis by ZLB Behring (Bern, Switzerland) and contained apolipoprotein A1 and phosphatidylcholine (kindly supplied by Dr. Peter G. Lerch). The reconstitution of HDL was performed by dissolution in isotonic sterile saline (3 mL/100 mg of rHDL). FITC-labeled LPS was purchased from Sigma-Aldrich (St. Louis, MO) and was injected and protected from daylight. The rats were sacrificed 5 minutes (5 controls and 4 cirrhotic rats), 20 minutes (5 controls and 4 cirrhotic rats), or 1 hour (5 controls and 4 cirrhotic rats) after LPS administration. They were sacrificed by decapitation. Immediately after they were sacrificed, blood was obtained from the neck, collected into dry and heparinized tubes, centrifuged (5000g for 10 minutes at 4°C), and stored at −80°C until the fluorescence was measured.
Effects of HDL Administration on the Alanine Aminotransferase (ALT) Levels, TNFα and LBP Production, Hepatic TNFα, LBP, iNOS, eNOS, Akt, and Cav-1 Expression, Hepatic cGMP Levels, and Akt and eNOS Activity.
This protocol included 75 rats (38 cirrhotic rats and 37 controls; Fig. 1B). The rats were anesthetized, and a left arterial catheter was placed as previously described (see the previous protocol). Blood samples were taken before the rHDL/saline and LPS/saline were administered and when the rats were sacrificed to determine the serum ALT, plasma TNFα, and LBP levels at 4 hours. The blood samples were centrifuged (5000g for 10 minutes at 4°C) and immediately stored at −80°C to measure the plasma TNFα and LBP levels. rHDL at a dose of 80 mg/kg or an equal volume of saline was administered, and 5 minutes later, LPS at a dose of 0.5 mg/kg (E. coli LPS serotype 0111:B4) or saline was administered. The rats were sacrificed by decapitation at 4 hours. After they were sacrificed, blood samples were collected as previously described. The livers were removed and flash-frozen in liquid nitrogen.
Three rats died during anesthesia (2 cirrhotic rats and 1 control) and were not included. The remaining rats were divided into 8 groups: 4 controls received only saline (control + saline group), 15 controls received saline and LPS (control + LPS group), 13 controls received rHDL and LPS (control + rHDL + LPS group), and 4 controls received only rHDL (control + rHDL group). In the cirrhotic group, 8 rats received only saline (cirrhotic + saline group), 7 rats received saline and LPS (cirrhotic + LPS group), 8 rats received rHDL and LPS (cirrhotic + rHDL + LPS group), and 13 cirrhotic rats received only rHDL (cirrhotic + rHDL group).
The mortality at 4 hours was also studied for these rats through comparisons of cirrhotic rats and controls and of cirrhotic rats that were pretreated with rHDL and cirrhotic rats that were not.
Effects of rHDL Administration on the Mean Arterial Pressure (MAP) and Portal Pressure.
This protocol included 10 cirrhotic rats (cirrhotic + saline, n = 5; cirrhotic + rHDL, n = 5) and 4 control rats (Fig. 1C). The animals were studied after overnight fasting. They were anesthetized with ketamine (50 mg/kg bw), and a catheter was placed in the left femoral artery. A dose of 80 mg/kg of rHDL or an equal volume of saline was administered intravenously. The MAP and portal pressure were measured at 4 hours: the arterial catheter was connected to a pressure transducer (pressure processor, Gould, Cleveland, OH). For portal pressure measurements, the spleen was exposed by retraction of the perisplenic fat after a midline laparotomy. The splenic pulp pressure, an indirect measurement of the portal pressure, was determined by the insertion of a fluid-filled 20-gauge needle into the splenic parenchyma. The needle was connected to the pressure transducer, which was calibrated for venous pressures.19 Previous studies have shown an excellent correlation between the splenic pulp pressure and portal pressure.20 Once this surgical procedure was completed, the rats were allowed to stabilize for 20 minutes. The splenic pulp pressure was then monitored. The rats were sacrificed by decapitation immediately after.
Effects of rHDL Administration on the Intrahepatic Resistance (Isolated Perfused Liver).
The protocol included 8 cirrhotic rats (cirrhotic + saline, n = 4; cirrhotic + rHDL, n = 4; Fig. 1D). The animals were studied after overnight fasting. They were anesthetized with ketamine (50 mg/kg bw), and a catheter was placed in the left femoral artery. A dose of 80 mg/kg of rHDL or an equal volume of saline was administered. After a laparotomy, the portal vein was cannulated as described.21, 22 The inferior vena cava below the kidneys was cut to allow the outflow of the perfusate, and the hepatic arteries were ligatured. The liver was then flushed clear of blood with saline. The liver was then removed and placed above the abdomen but kept wet. The pressure was recorded continuously from the catheter supplying the portal vein with a Statham strain gauge (Statham Instruments, Inc., Oxnard, CA). The zero was determined with no perfusion. The recorded pressure was considered the portal pressure. The flow rate (20 mL/min) was perfused for 5 minutes. The portal pressure was recorded. For each measurement, the hepatic vascular resistance was calculated as follows: hepatic vascular resistance = portal pressure/flow rate.
Serum Cholesterol and HDL Levels
The total serum cholesterol concentration was determined enzymatically with a Konelab kit on a Konelab 30i analyzer.23 The HDL-cholesterol was determined with a phosphotungstic acid/MgCl2 reagent (Boehringer, Mannheim, Germany) to precipitate the apolipoprotein B–containing lipoproteins,24 and the cholesterol was measured in the supernatant, as described earlier in the serum.
Isolation and Analysis of the Lipoprotein Fractions
Two lipoprotein fractions were isolated from the serum by sequential ultracentrifugation in a Beckman Optima XL 100 centrifuge (rotor NVT 100, Beckman, Roissy, France) after the densities were adjusted with potassium bromide: very low density and low-density lipoproteins (density < 1.070) after 3 hours of centrifugation at 10°C and 90,000 rpm and HDLs (1.070 < density < 1.21) after 4 hours of centrifugation at 10°C and 560,196 g. The cholesterol was enzymatically determined on each lipoprotein fraction with a final assessment on a microplate. The FITC fluorescence was measured in serum and each lipoprotein fraction with a Spectra Max Gemini electron microscope (emission wavelength = 490 nm, excitation wavelength = 520 nm) with microplates for the measurement of the fluorescence and was compared with the fluorescence in the total serum.
Serum ALT Activity
The serum ALT activity was measured with a Hitachi 717 automatic analyzer (Roche Diagnostic, Meylan, France)
Plasma TNFα and LBP Concentrations
The plasma TNFα and LBP levels were measured with a commercially available solid-phase enzyme-linked immunosorbent assay according to the protocol supplied by the manufacturer (R&D Systems, Minneapolis, MN, for TNFα and Cell Sciences, Canton, MA, for LBP). The standards and samples were incubated with a TNFα antibody–coated or LBP antibody–coated microplate. An enzyme-linked polyclonal antibody specific for rat TNFα or LBP was then added after washing. The intensity of the color was measured with a microplate reader (Molecular Devices, Sunnyvale, CA).
Western Blot Analysis
The expression of TNFα, LBP, iNOS, eNOS, phosphorylated eNOS, Cav-1, Akt, and phospho Akt was determined in the liver. Liver tissue was homogenized in a buffer containing 320 mmol/L sucrose, 50 mmol/L trishydroxymethylaminomethane (Tris)-HCl (pH 7.4), 1 mmol/L ethylene diamine tetraacetic acid, 0.1% sodium dodecyl sulfate (SDS), 1% Nonidet P-40, 1 mmol/L dithiothreitol, 10 μg/mL trypsin inhibitor, 2 μg/mL aprotinin, 100 μg/mL phenyl methane sulfonyl fluoride, and a protease inhibitor cocktail tablet (Roche Diagnostic, Mannheim, Germany). The samples were sonicated 3 times for 20 seconds. The homogenates were centrifuged at 12,000g for 5 minutes at 4°C to remove tissue debris. The protein concentration of the supernatants was determined with a Bio-Rad protein determination assay,25 and an aliquot of the homogenate was suspended in an SDS sample buffer (2% SDS, 2% β-mercaptoethanol, and 10% glycerol in 300 mmol/L Tris-HCl, pH 6.8). Equal amounts of the total protein (25–50 μg) were then subjected to SDS-polyacrylamide gel electrophoresis. Gels were transferred to nitrocellulose. Nitrocellulose filters were blocked by 1 hour of incubation in 5% nonfat dry milk, 1% bovine serum albumin, 0.1 mol/L NaCl, 0.01 mol/L Tris-HCl (pH 7.5), and 0.1% Tween 20. The blots were then incubated overnight at 4°C with anti-TNFα, anti-LBP (Santa Cruz Biotechnology Inc., Santa Cruz, CA), anti-iNOS, anti-eNOS, anti–phosphorylated eNOS at serine 1177 (Cell Signaling Technology, Beverly, MA), anti–Cav-1, anti-Akt, or anti–phospho Akt antibodies. The blots were washed in 0.1 mol/L NaCl, 0.01 mol/L Tris-HCl (pH 7.5), and 0.1% Tween 20 and incubated for 1 hour at 25°C with a horseradish peroxidase–conjugated secondary antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA). The blots were washed again in 0.1 mol/L NaCl, 0.01 mol/L Tris-HCl (pH 7.5), and 0.1% Tween 20. The immunoreactivity was assessed with an enhanced chemiluminescent western blot detection system according to the manufacturer's instructions (Amersham International, Buckinghamshire, England).
The protein bands were quantified by videodensitometry. The values are given as arbitrary units. To ensure that the same amounts of proteins were loaded, we confirmed that the same densitometric intensities of actin with an anti-actin antibody (Sigma, St Louis, MO) were observed.
Hepatic cGMP Levels
Frozen livers were homogenized in cold 6% (wt/vol) trichloroacetic acid at 4°C to give a 10% (wt/vol) homogenate and centrifuged at 2000g for 15 minutes at 4°C. The supernatant was recovered, and the pellet was kept for Bradford measurements. The supernatant was washed 4 times with 5 volumes of water-saturated diethyl ether. The upper ether layer was discarded after each wash. The remaining aqueous extract was lyophilized. The dried extract was dissolved in 450 μL of an assay buffer prior to its analysis with cGMP Direct Biotrak EIA (Amersham Biosciences, Buckinghamshire, United Kingdom). The enzyme immunoassay was performed according to the manufacturer's instructions.
The results are reported as mean values and standard deviations, except where specifically noted, and the categorical data are presented as counts and percentages. Comparisons involved the use of a 1-way analysis of variance, a Student t test, and Fischer's exact test. Repeated measurements were handled with a repeated-measure analysis of variance. Statistical significance was accepted for P < 0.05. All statistical analyses were performed with NCSS (Kaysville, UT) software.
Effects of rHDL Administration on LPS Neutralization
The baseline serum cholesterol levels were significantly higher in the cirrhotic rats than in the controls (1.16 ± 0.25 versus 0.72 ± 0.14 g/L, respectively, P < 0.05). In contrast, the baseline serum cholesterol HDL levels were significantly reduced in the cirrhotic rats compared with the controls (0.25 ± 0.09 versus 0.44 ± 0.02 g/L, respectively, P < 0.05).
As expected, when the rats were sacrificed, the total cholesterol levels were higher in the rats that received rHDL than in the controls receiving saline (0.87 ± 0.06 versus 0.65 ± 0.02 g/L, respectively, P < 0.05) and in the cirrhotic rats (1.37 ± 0.19 versus 1.02 ± 0.11 g/L, respectively, P < 0.05). The mean HDL-cholesterol values were significantly higher in the rats that received rHDL, in the cirrhotic group (0.97 ± 0.14 versus 0.47 ± 0.02 g/L, respectively, P < 0.05), and in the control group (0.72 ± 0.09 versus 0.46 ± 0.06 g/L, respectively, P < 0.05) in comparison with the rats that received saline. The LDL-cholesterol values were similar in the cirrhotic rats and controls after rHDL administration.
The HDL fluorescence/total serum fluorescence ratio was highest when the rats were sacrificed at 20 minutes in all 4 groups of rats. In the rats that did not receive rHDL, this ratio was lower in the cirrhotic rats than in the controls (0.14 ± 0.01 versus 0.27 ± 0.06, respectively, P < 0.05). The HDL fluorescence/total serum fluorescence ratio increased after rHDL administration in the cirrhotic rats (0.30 ± 0.18 versus 0.14 ± 0.01, respectively, P < 0.05). The HDL fluorescence/total serum fluorescence ratio was not modified after rHDL administration in the normal rats (Fig. 2).
Effects of rHDL Administration on Serum ALT, Plasma TNFα, and LBP Production; Hepatic TNFα, LBP, and iNOS Expression; and Hepatic cGMP Levels
Serum ALT (Fig. 3A).
At 4 hours, the serum ALT levels were significantly different between the cirrhotic and control rats (147 ± 28 versus 64 ± 4 IU/L, respectively, P = 0.03). In the rats that did not receive LPS, the administration of rHDL did not significantly modify the ALT levels in the controls (62 ± 7 versus 64 ± 4 IU/L, not significant) or in the cirrhotic rats (73 ± 69 versus 147 ± 28 IU/L, not significant). In the LPS-challenged control rats, rHDL administration did not modify the ALT levels. In the LPS-challenged cirrhotic rats, rHDL administration significantly decreased the serum ALT at 4 hours (72 ± 5 versus 213 ± 93 IU/L, respectively, P < 0.05).
Plasma TNFα Levels (Fig. 3B).
In the LPS-challenged rats, the plasma TNFα levels were significantly different between the cirrhotic rats and controls at 4 hours (3021 ± 890 versus 70 ± 38 pg/mL, respectively, P < 0.05). After the injection of rHDL and the LPS challenge, the plasma TNFα levels were significantly decreased in the cirrhotic rats at 4 hours in comparison with the cirrhotic rats that did not receive rHDL (565 ± 208 versus 3021 ± 890 pg/mL, respectively, P < 0.05). rHDL administration did not modify the plasma TNFα levels in the control rats.
Serum LBP Levels (Fig. 3C).
The serum LBP levels were significantly increased in the cirrhotic rats (receiving LPS or not) in comparison with the controls (155 ± 27 versus 33 ± 1 μg/mL, respectively, in the LPS-challenged rats, P < 0.05). rHDL injection did not modify the LBP levels in the cirrhotic or control rats.
Hepatic TNFα Protein Expression.
After LPS administration, TNFα protein expression was detectable in the livers of all groups. Hepatic TNFα protein expression was significantly greater in the cirrhotic rats than in the control rats. In the cirrhotic rats, hepatic TNFα expression was significantly increased in the rats that received LPS and significantly decreased after a pretreatment with rHDL (Fig. 4).
Hepatic LBP Protein Expression.
Hepatic LBP protein expression was detectable in the livers of the controls. In the cirrhotic rats, hepatic LBP protein expression was significantly greater than in the controls and greater in the rats that received LPS. However, a pretreatment with rHDL nearly abolished hepatic LBP protein expression (Fig. 5).
Hepatic iNOS Protein Expression.
iNOS protein expression was detectable in the livers of all groups but was significantly greater in the cirrhotic rats than in the controls. It was increased in the cirrhotic rats that received LPS and significantly reduced by a pretreatment with rHDL (Fig. 6).
Hepatic eNOS Expression and Activity.
Hepatic eNOS protein expression was similar in the cirrhotic and control rats and not modified by LPS or rHDL administration. The expression of phosphorylated eNOS at serine 1177 (activated eNOS) was highly detectable in control rat livers and was not modified by LPS administration. It was less detectable in the cirrhotic rats that received saline and in the cirrhotic rats that received LPS. However, a pretreatment with rHDL restored phosphorylated eNOS expression in cirrhotic livers (Fig. 7).
Hepatic Cav-1, Akt, and Phosphorylated Akt Expression.
Hepatic Cav-1 protein expression was hardly detectable in control rat livers. It was significantly greater in cirrhotic rat livers and highest in cirrhotic rat livers that received LPS. A pretreatment with rHDL significantly decreased hepatic Cav-1 protein expression in the cirrhotic rats challenged or not challenged with LPS. Akt was detectable in all livers and not modified by LPS or rHDL administration. Phosphorylated Akt (activated Akt) was detectable in control rat livers. It was significantly decreased in cirrhotic rat livers and nearly abolished in the cirrhotic rats that received LPS. A pretreatment with rHDL restored phosphorylated Akt in the cirrhotic rats challenged or not challenged with LPS (Figs. 8 and 9).
Hepatic cGMP Levels.
The hepatic cGMP levels were significantly higher (5.36 ± 0.21 fmol/mg) in the cirrhotic rats pretreated with HDL (n = 5) than in the rats pretreated with saline (4.02 ± 0.11 fmol/mg, n = 5, P = 0.01).
Mortality at 4 Hours.
In rats pretreated with saline, the mortality at 4 hours was not significantly higher in the cirrhotic rats than in the controls (8% versus 4%, respectively, not significant). After the injection of LPS, the mortality of the cirrhotic rats was significantly increased (29% versus 8%, respectively, P = 0.01). However, a pretreatment with rHDL significantly decreased the mortality in the LPS-challenged cirrhotic rats (14% versus 29%, respectively, P = 0.02). A pretreatment with rHDL did not significantly change the mortality in the cirrhotic rats that were not challenged with LPS.
Effects of HDL Administration on MAP and Splenic Pulp Pressure
At 4 hours, MAP was significantly lower in the cirrhotic rats than in the control rats (100 ± 32 mm Hg versus 120 ± 42 mm Hg, respectively, P = 0.02). MAP was significantly lower in the cirrhotic rats challenged with LPS than in those challenged with saline (75 ± 13 versus 100 ± 32 mm Hg, respectively, P = 0.02). However, a pretreatment with rHDL significantly increased MAP after LPS challenge in the cirrhotic rats (98 ± 20 versus 75 ± 13 mm Hg, respectively, P = 0.02).
The splenic pulp pressure was significantly higher in the cirrhotic rats than in the control rats (135 ± 23 versus 23 ± 4 mm Hg, respectively, P = 0.01). The administration of rHDL significantly decreased the splenic pulp pressure in the cirrhotic rats (90 ± 17 versus 135 ± 23 mm Hg, respectively, P = 0.02; Fig. 10). MAP was not modified by the administration of rHDL.
Effects of rHDL Administration on the Intrahepatic Vascular Resistance (Isolated Perfused Liver)
The rats with cirrhosis pretreated with rHDL showed significantly lower perfusion pressures than the rats pretreated with saline, and this suggested decreased intrahepatic vascular resistance (0.15 ± 0.03 versus 0.32 ± 0.02 mm Hg min/mL, respectively, P = 0.02). Individual calculated values of the mean hepatic vascular resistance for each rat are shown in Fig. 11.
The main purpose of this study was to determine the mechanisms of action of rHDL administration on the excessive proinflammatory response in cirrhotic rats. The results show that a pretreatment with rHDL significantly neutralized circulating LPS in LPS-challenged rats. rHDL administration decreased plasma TNFα and liver proinflammatory signals, that is, hepatic TNFα, LBP, and iNOS expression, and mortality after LPS challenge. rHDL also restored hepatic eNOS and lowered the portal pressure and intrahepatic vascular resistance in non–LPS-challenged cirrhotic rats. This suggests that rHDL, in addition to its anti-inflammatory properties, has an effect on hepatic microcirculation, thus providing a double benefit in cirrhosis.
Previous studies have shown the high LPS binding capacity of lipoproteins, especially HDL,8, 26 which seems to be related to the abundance of phospholipids on the HDL surface.27 In this study, in the absence of an rHDL pretreatment, the amount of LPS trapped by HDL decreased in cirrhotic rats in comparison with controls. This is probably related to the low levels of HDL in cirrhosis.6, 7 As expected, a pretreatment with rHDL increased the fraction of HDL-bound LPS. No significant difference in the trapping of LPS by rHDL in control rats was observed, but the administered dose of LPS was very low for normal rats. However, the HDL fluorescence/total fluorescence ratio was also higher after rHDL administration in noncirrhotic rats (34% versus 27%, respectively, P = 0.29).
In noncirrhotic animal models, certain studies have already highlighted the beneficial role of rHDL administration after LPS challenge.8–10 In cirrhotic rats, Ramirez et al.15 showed that the administration of rHDL reduced the effects of LPS on TNFα production and systemic hemodynamics. This study confirms the beneficial effects of rHDL on proinflammatory cytokine production in cirrhotic rats. In agreement with this hypothesis, the serum ALT levels decreased following a pretreatment with HDL in LPS-challenged cirrhotic rats. Hepatic TNFα production was also reduced in cirrhotic rats, and this suggested a reduction in liver inflammation, which has already been observed in noncirrhotic rats.10 It is interesting to note that the serum ALT activity and TNFα levels did not change in LPS-challenged control rats pretreated with rHDL. In previous studies, the reduction of cytokine levels in noncirrhotic rats was not systematic.10 Moreover, in this study, the dose of LPS was low for noncirrhotic rats (0.5 mg/kg bw), and this may explain the absence of an effect of rHDL on cytokine production in noncirrhotic rats.
In this study, no significant change in the plasma LBP concentrations was observed after rHDL administration. However, although hepatic LBP expression was increased 4 hours after LPS administration, it was nearly abolished after a pretreatment with rHDL. The discrepant response between the plasma and hepatic LBP levels might be due to the delay between the rHDL administration and the time of sacrifice. In a bacterial infection, LBP concentrations may reach 200 μg/mL. However, the role of LBP is still being debated. Some authors argue that LBP is involved in the neutralization of LPS.28, 29 For others, LBP is a prerequisite for triggering a proinflammatory response of monocytes/macrophages.30 In support of this hypothesis, the recent findings of Albillos et al.31 show that increased LBP is independently associated with the severity of infection in patients with cirrhosis and ascites. Our study also supports this hypothesis, as hepatic LBP expression was increased after LPS administration and nearly abolished after a pretreatment with rHDL.
One important original result in this study is that hepatic iNOS expression, which was increased after LPS administration, was dramatically reduced by the pretreatment with rHDL. iNOS is synthesized only de novo in vascular smooth muscle cells, macrophages, hepatic stellate cells, and hepatocytes after induction by LPS and proinflammatory cytokines (TNFα). Once it is expressed, iNOS synthesizes large amounts of NO for long periods independently of any mechanical or hemodynamic stimuli.32 iNOS-derived NO reacts with free radicals as a superoxide anion and undergoes autoxidation, thus forming nitrogen oxide species and promoting lipid peroxidation and protein nitrosation.33 Therefore, we could speculate that rHDL administration, by decreasing iNOS expression, could reduce the production of reactive oxygen species (ROS). To date, no data are available concerning the effects of rHDL on ROS in patients with cirrhosis. However, the effects of rHDL on ROS production have been well studied in animals and humans without cirrhosis. One study34 showed that the exposure of human aortic endothelial cells to glucose-modified HDL attenuated the expression of eNOS, but not of iNOS, and stimulated the production of ROS. Furthermore, in rats challenged with oxidized HDL, a pretreatment with rHDL attenuated the production of ischemia/reperfusion-induced free radicals, and this suggested that rHDL could have a negative effect on the production of ROS.35 In another study, in a model of stroke, the incubation of brain cells with rHDL decreased the ROS levels.36 In smooth muscle cells, HDL counterbalanced the proinflammatory effect of oxidized HDL by inhibiting intracellular ROS formation.37 All those studies suggest that rHDL administration could decrease the production of ROS in cirrhosis as well.
Moreover, in the liver, up-regulated iNOS may cause the inhibition of eNOS via NO autoinhibition, as described in the kidneys.38 Therefore, the reduced expression of hepatic iNOS after rHDL administration could explain its beneficial effect by decreasing oxidative stress and restoring eNOS activity (discussed later).
The hepatic eNOS activity was decreased in the livers of cirrhotic rats. The alteration of hepatic eNOS-derived NO synthesis seems to be the first step toward hemodynamic abnormalities and complications in cirrhosis. Alterations in eNOS-derived NO synthesis in the hepatic microcirculation result in a microcirculatory NO deficiency, leading to increased intrahepatic vascular resistance. As a result, the splanchnic and systemic vasculature exhibits marked endothelial NO overproduction, playing a crucial role in the development of the hyperdynamic circulation.39 Among the mechanisms involved in abnormal eNOS regulation in the endothelial cells of cirrhotic livers, the up-regulation of Cav-1 seems to be a major mediator of reduced eNOS activity.13, 40, 41 Indeed, Cav-1 binds with eNOS and directly inhibits NO production.42 In agreement with this hypothesis, Cav-1 was increased in the livers of cirrhotic rats in comparison with controls in this study. In addition, the increased hepatic cGMP levels following rHDL administration support the production of intrahepatic NO by eNOS in cirrhotic livers.
LPS was shown to suppress hepatic eNOS activity in cirrhosis. Conversely, bacterial translocation has been found to activate eNOS in the splanchnic vasculature and in the lungs of cirrhotic rats.43, 44 Moreover, Cav-1 hepatic expression was increased in cirrhotic rats that were challenged with LPS. This result was expected, as Cav-1 has already been shown to be up-regulated by LPS in noncirrhotic animals.45 This supports the hypothesis that eNOS activity was reduced in LPS-challenged cirrhotic livers because of caveolin overexpression. Another explanation could be the inhibition of eNOS by the up-regulated iNOS via NO autoinhibition, as described earlier.
A pretreatment with rHDL restored hepatic eNOS activity. This finding is consistent with a previous study showing that HDL stimulated eNOS activity in endothelial cells by binding to a high-affinity HDL receptor, scavenger receptor class B type I (SR-BI).46 In noncirrhotic endothelial cells, the binding of HDL to SR-BI caused marked stimulation of eNOS activity by decreasing eNOS–Cav-1 binding.47 In accordance with this hypothesis, in our study, Cav-1 expression was nearly abolished by a pretreatment with rHDL. This suggests that in cirrhotic rats, rHDL restores eNOS activity by disrupting Cav-1. The binding of HDL to SR-B1 also causes rapid activation of the nonreceptor tyrosine kinase Src, leading to phosphoinositide 3 kinase activation and downstream activation of Akt, which enhances eNOS activity by phosphorylation.14 Indeed, we found that Akt activity was restored with the administration of rHDL in cirrhotic rats. Recently, Liu et al.48 showed that G protein-coupled receptor kinase 2 played a crucial role in the inhibition of Akt phosphorylation in portal hypertensive rats. Therefore, the role of HDL administration in G protein-coupled receptor kinase 2 expression needs to be explored in subsequent studies. Another explanation of the beneficial effect of rHDL on eNOS activity could be that the down-regulation of iNOS after rHDL administration suppresses the NO autoinhibition of eNOS, thus restoring its activity.
A major result of this study was that the administration of rHDL decreased portal pressure and intrahepatic vascular resistance in cirrhotic rats. This is consistent with the finding that rHDL restored eNOS hepatic activity. Subsequently, NO synthesis in the hepatic microcirculation would be increased, and intrahepatic vascular resistance would be reduced. This finding is especially important for cirrhosis, for which pathways leading to a decrease in portal hypertension are actively studied.
This study is the first in which rHDL was directly administered to cirrhotic rats in order to see the effects on eNOS hepatic production. This suggests that rHDL could have therapeutic potential for severe cirrhosis, in which portal hypertension is the main complication. Statins, which inhibit 3-hydroxy-3-methyl-glutaryl-coenzyme A reductase, have already been shown to decrease hepatic vascular resistance.49 This effect seems to be related to an increase in NO production in endothelial cells, being mediated through the phosphatidylinositol 3 kinase–dependent activation of protein kinase Akt, which leads to eNOS phosphorylation at serine 1177, with subsequent increased activity.50 However, whether this effect resides in a reduction of the total cholesterol levels and in an increase of the HDL levels remains to be proven. Indeed, the regulation of eNOS expression by statins is complex and can be attributed to noncholesterolic effects (for example, the inhibition of Rho geranylgeranylation). Moreover, if statins decreased the hepatic vascular resistance, the portal pressure was not lowered because of an increase in the liver output.
Our study has several limitations. First, we chose a model of biliary cirrhosis. The lipid profile of bile duct–ligated rats may be very different from that in humans. In particular, the total cholesterol levels in this animal model are higher than in controls.51, 52 However, the HDL-cholesterol serum concentrations were markedly decreased in our rats, as already described in the literature for this model and other rat models15, 51 and in accordance with our hypothesis. Another drawback of this animal model is that rHDL clearance might be reduced because of biliary obstruction. Indeed, HDL is known to transport endotoxin to the liver, at which it is detoxified and excreted via bile by hepatocytes. Of more concern, some authors have reported increased mortality and cytokine production after rHDL infusion in LPS-challenged bile duct–ligated rats, suggesting that Küpffer cells may have increased exposure to endotoxin because of biliary obstruction. However, in that study, the rHDL doses were higher than those in our study, the rats were bile duct–ligatured only 1 week before the experiment, and they were not cirrhotic.52 Another limitation of this study is that we administered rHDL before injecting endotoxin, whereas in the clinical setting, it would be used in patients who are already infected. Therefore, rHDL may be effective in the prevention but not the treatment of the deleterious consequences of endotoxemia. Nevertheless, in a rat model of hemorrhagic shock and organ dysfunction, in which cytokine production was also increased, rHDL was administered 90 minutes after hemorrhage and reduced the rate of multiorgan failure.53 This study suggests that rHDL may have beneficial effects not only on prophylaxis, which will have to be demonstrated in our model.
Future studies are needed to evaluate the beneficial effects of the administration of rHDL in patients with cirrhosis. A randomized controlled study testing the effect of a treatment with HDL in humans without cirrhosis and with sepsis is ongoing. In patients with cirrhosis, the administration of HDL should be studied not only during sepsis but also in patients with portal hypertension without sepsis to reduce portal hypertension.
In conclusion, this study demonstrates that, in LPS-challenged cirrhotic rats, rHDL decreases the TNFα-mediated proinflammatory response and the hepatic expression of TNFα, LBP, and iNOS by neutralizing circulating LPS. Moreover, rHDL restores hepatic eNOS activity and decreases the portal pressure and intrahepatic vascular resistance in cirrhotic rats. Thus, the beneficial effects of rHDL in cirrhosis are not only a result of their anti-inflammatory effects but also a result of the decrease in the portal hypertension. Because portal hypertension and bacterial infections are the major causes of death in patients with cirrhosis, the administration of rHDL should be considered a useful strategy for the treatment of cirrhosis and its complications.