Increased Sensitivity to Endotoxemia in theBile Duct–Ligated Cirrhotic Rat

DAVID HARRY,1 RADHI ANAND,1 STEPHEN HOLT,1 SUSAN DAVIES,2 RICHARD MARLEY,1 BIMBI FERNANDO,1

DAVID GOODIER,3 AND KEVIN MOORE1

Sepsis is a common complication of cirrhosis with a highmortality. In this study, we have investigated some of thepathways that may be involved in tissue injury and death.Bile duct–ligated (BDL) cirrhotic and control rats werechallenged with lipopolysaccharide (LPS). Sensitivity toLPS was markedly enhanced in the BDL group, and wasassociated with increased liver injury and mortality. Therewas a 5-fold constitutive activation of nuclear factor k B(NFkB) in the liver of BDL rat controls (P F .001), and thiswas activated further, but to a similar extent, in the liver ofboth sham and BDL rats after injection of LPS. Plasmatumor necrosis factor a (TNF-a) increased more markedlyin the BDL cirrhotic rats (2,463 6 697 pg/mL in BDL ratsversus 401 6 160 pg/mL in the controls at 3 hours; P F.01). Plasma nitrite/nitrate concentrations were increasedin the BDL controls at baseline, and increased further afterLPS (P F .05), but did not differ from sham controls at 6hours. Plasma F2-isoprostanes increased 6-fold in the cir-rhotic rats and 2-fold in the controls (P F .01) indicative oflipid peroxidation. Esterified F2-isoprostanes in the liverincreased 2- to 3-fold at 1 hour in control and BDL rats, butreturned to baseline levels by 3 hours. Esterified F2-isoprostanes in the kidney increased by 2-fold in the BDLrats after LPS administration, but remained unchanged insham controls. We conclude that there is a marked increasein sensitivity to LPS in BDL cirrhotic rats. This is associatedwith an enhanced TNF-a response and increased lipidperoxidation. These may be directly and causally related tomortality. (HEPATOLOGY 1999;30:1198-1205.)

Sepsis and associated endotoxemia occur in approximately40% of hospitalized patients with cirrhosis and is a majorcause of death.1,2 The increased risk of infection is secondaryto impairment of several of the host defense mechanismsincluding impaired neutrophil function. Furthermore, lowgrade endotoxemia may occur because of impaired Kupffer

cell removal of gut-derived endotoxin, which in cirrhosis maybe released into the peripheral circulation because of portalsystemic shunting. Complete diversion of bile (containingendotoxin-binding bile salts and immunoglobulin A) fromthe gut lumen changes the bacterial flora, causes loss ofmucosal integrity, and decreases endotoxin inactivation lead-ing to portal bacteremia and endotoxemia. This occurs in thebile duct–ligated cirrhotic rat, in which the endotoxin concen-tration in the portal blood is increased approximately 7-foldat 3 weeks after bile-duct ligation (BDL). Increased systemicendotoxin levels have been reported in patients with cirrho-sis.3-5

Lipopolysaccharide (LPS) is a component of the outermembrane of gram-negative bacteria. Macrophage and endo-thelial cells of the hepatic sinusoids are the major sites for theclearance of LPS from the circulation, most of which appearsto originate from the digestive tract. Interaction of hepaticmacrophages (Kupffer cells) with LPS results in enhancedactivity of a variety of parameters including increased produc-tion of reactive oxygen intermediates and of proinflammatorycytokines.6,7

Endotoxemia in the context of liver disease was firstdescribed in 1970 in patients with biliary obstruction.8 Sincethis original description, endotoxemia and infections havebeen shown to correlate with mortality and complications ofcirrhosis or acute liver disease. For example, endotoxemiaand an elevated cytokine response occurs in patients develop-ing the hepatorenal syndrome.9 In acute alcohol-inducedhepatitis, levels of endotoxin (LPS) are markedly elevatedand correlate with increased mortality and with levels of theproinflammatory cytokines tumor necrosis factor a (TNF-a),interleukin-1b (IL-1b), and IL-8.2,10 The hepatic reticulo-endothelial system is responsible for the initial response toLPS, and injected LPS is primarily removed by Kupffer cells.11

This results in the induction of various proinflammatorycytokines including TNF-a, IL-1b, IL-6, and g-interferon.9,10

These, together with up-regulation of adhesion molecules,cause a marked influx of activated polymorphonuclear neu-trophils into the liver, which are sources of reactive oxygenspecies.12 The action of LPS on cells to induce the productionof cytokines involves activation of the redox-sensitive, ubiq-uitous nuclear transcription factor nuclear factor k B (NFkB),13

which activates the genes involved in the inflammatory andimmune response.14 In the liver, LPS also triggers theproduction of reactive oxygen and nitrogen species,15 whichmay cause lipid peroxidation and disturb the integrity ofcellular membranes.7 In this study we have examined theeffect of LPS in both normal and BDL cirrhotic rats on thepathways leading to tissue injury and subsequent mortality.

Abbreviations: BDL, bile-duct ligation; LPS, lipopolysaccharide; TNF-a, tumornecrosis factor a; IL, interleukin; NFkB, nuclear factor k B; NO, nitric oxide; EMSA,electromobility shift assay; AST, aspartate transaminase.

From the Departments of 1Medicine, 2Histopathology, and 3Chemical Pathology,Royal Free and University College Hospital School of Medicine, Royal Free Campus,Pond Street, London, UK.

Received March 1, 1999; accepted August 19, 1999.Supported by The Medical Research Council, Great Britain.Address reprint requests to: Kevin Moore, M.D., Department of Medicine, Royal Free

and University College Medical School, Royal Free Campus, Pond Street, London NW32QG, UK. E-mail: kmoore@rfhsm.ac.uk; fax: (44) 171 794 3472.

Copyright r 1999 by the American Association for the Study of Liver Diseases.0270-9139/99/3005-0013$3.00/0

1198

MATERIALS AND METHODS

Animals

Male, Sprague-Dawley rats (body weight 230-260 g) were ob-tained from the comparative biological unit at the Royal Free andUniversity College Medical School (Royal Free Campus, London,UK). All animals were housed in the comparative biological unit andgiven free access to normal rodent chow (expanded SDSRM 1;Special Diet Services, Witham, UK) and water. A midline abdominalincision was made under anesthesia. In the BDL group, the commonbile duct was isolated, triply ligated with 3-0 silk, and sectionedbetween the ligatures. After BDL all animals continued to gainweight (251 6 4 g to 330 6 5 g) and were comparable with shamcontrols (246 6 3 g to 358 6 5 g). The overall mortality in bothgroups was less than 10% and occurred within 36 hours of theoperation. All studies were performed at 25 to 28 days post BDL orsham procedure, and 5 to 9 animals were used at each time point asindicated in the results. Animals were injected with either saline, orLPS (Salmonella typhimurium) dissolved in saline. To determine theeffect of LPS on mortality, doses of 0.5, 1, and 2 mg/kg werecompared. LPS was administered at the specified dose by intraperito-neal injection in the morning, and 24-hour survival was determined.Subsequently all animals were injected with 0.5 mg/kg LPS intraper-itoneally and then allowed free access to food and water. At the timepoints indicated, blood was withdrawn from the inferior vena cavauntil full exsanguination into plain or ethylenediaminetetraaceticacid containing tubes, centrifuged, and the serum or plasma storedat 280°C until assay. Because the group of animals that survived for24 hours are a self-selecting group of survivors (there was a 50%mortality at 24 hours), all studies were restricted to the first 6 hoursfor biochemical measurements. Liver and kidney tissue was snapfrozen in liquid nitrogen and stored at 280°C until measurement oftissue isoprostanes. Liver tissue was also fixed overnight in formalin,embedded in paraffin, and stained with hematoxylin-eosin forhistological examination by a liver pathologist. All animals withBDL had macroscopic evidence of cirrhosis at the time of study, andthis was confirmed histologically.

Plasma and urine biochemistry were determined by a standardautoanalyzer (Hitachi, Tokyo, Japan).

Measurement of Lipid Peroxidation

Measurement of F2-Isoprostanes. Free F2-isoprostanes in plasmawere extracted and quantified using a stable isotope dilution gaschromatography-mass spectrometry method as previously de-scribed,16 but using [D4] 8-iso-PGF2a as the internal standard(Caymen Chemical Company, Ann Arbor, MI). In brief, plasmasamples were extracted on C18 and silica Sep Pak cartridges(Waters, Watford, UK), derivatized to the pentafluorobenzyl ester,and after thin layer chromatography, reacted with bis(trimethylsilyl)-trifluoroacetamide to give the tri-methylsilyl derivative. Ions weremonitored at 569 and 573 using a VG TR100 1000 (Fisons,Manchester, UK) mass spectrometer coupled to a Carlo Erba GC8000 (Fisons). Liver and kidney samples for F2-isoprostane determi-nation were homogenized in ice cold Folch solution (chloroform/methanol 2:1 vol:vol) containing butylated hydroxytoluene (5mg/100 mL) and triphenylphosphine (50 mg/100 mL) to inhibit exvivo lipid peroxidation. The lipid-containing layer was dried undernitrogen and hydrolyzed with 15% methanolic potassium hydroxidefor 30 minutes at 37°C, and the free F2-isoprostanes were quantifiedas described previously after extraction. Esterifed isoprostanes wereexpressed as a ratio to the concentration of phospholipid in theFolch extract (described later). This is necessary to overcomevariable lipid content and extraction efficiency from sample tosample.

Tissue phospholipids were determined on the Folch17 extractsusing a commercially available enzymatic method (Alpha Laborato-ries Ltd., Eastleigh, UK). Briefly, liver was homogenized in Folchsolution as described previously and the lower lipid-containing layerremoved. The choline containing phospholipids (95% of phospho-

lipids) were enzymatically hydrolyzed to free choline, and thereleased choline was oxidized, with the simultaneous production ofhydrogen peroxide, which reacts with a chromophore producing achromogen that maximally absorbs at 505 nm.

Measurement of TNF-a

Plasma TNF-a concentrations were determined using the Gen-zyme rat enzyme-linked immunosorbent assay kit (Genzyme Di-agnsotics, Cambridge, MA).

Nitrite/Nitrate Assay

Nitrite and nitrate are stable decomposition products of nitricoxide (NO). Their plasma levels were determined by chemilumines-cence using an NO analyzer (Sievers Research Inc., Boulder, CO).Before measurement, plasma samples were filtered through 30K mwcut-off filters (Millipore, Bedford, MA) and incubated for 1 hour at37°C with nitrate reductase, fladin adenine dinucleotide (FAD), andreduced nicotinamide adenine dinucleotide phosphate (NADPH), toconvert nitrate to nitrite. Nitrite was reduced to NO by refluxingwith acetic acid and potassium iodide. The resulting NO was purgedfrom the refluxing solution by nitrogen gas and reacted with ozonebefore analysis by chemiluminescence. Standard curves of sodiumnitrite and nitrate were used for calibration. Conversion of nitrate tonitrite was greater than 90%.

Preparation of Nuclear Extracts for NFkB Assay

The nuclear proteins were extracted according to a modifiedmethod by Manning et al.18 In brief, 100 to 200 mg of liver andkidney tissue were each rinsed in Ca21/Mg21-free phosphate-buffered saline. The tissue was homogenized with a blade homog-enizer in 3 mL buffer A (10 mmol/L HEPES, pH 7.9, 1.5 mmol/LMgCl2, 10 mmol/L KCl, 0.5 mmol/L dithiothreitol, 0.5 mmol/Lphenylmethylsulphonyl fluoride, 0.1% Nonidet P-40). The homog-enized tissue was incubated on ice for 10 minutes and centrifuged at850g for 10 minutes at 4°C. The resulting pellet was washed twice inbuffer A prepared without Nonidet P-40. The nuclear pellet wasthen resuspended in 50-µL to 200-µL of buffer B (20 mmol/LHEPES, pH 7.9, 25% (vol/vol) glycerol, 0.42 mol/L NaCl, 1.5mmol/L MgCl2, 0.2 mmol/L ethylenediaminetetraacetic acid, 0.5mmol/L dithiothreitol, 0.5 mmol/L phenylmethyl sulfonyl fluoride)and incubated on ice for 30 minutes. A final spin of 100,000g for 20minutes at 4°C in an ultracentrifuge removed the remaining cellulardebris, leaving the nuclear proteins in the supernatant. All buffersand tissue were kept on ice or at 4°C. The protein concentrationswere determined using the Bradford reagent,19 and samples werethen diluted to a uniform concentration of 0.5 mg/mL.

Electrophoretic Mobility Shift Assay

NFkB was detected by electrophoretic mobility shift assay (EMSA)using a gel shift assay system (Promega, Madison, WI). However, 7%and 3 mm thick (instead of 4% and 1 mm thick) gels were used. Thebinding reaction mixture containing 6 µL of nuclear extract (3 µg ofprotein), 1 µL of water, and 2 µL 5x gel shift binding buffer wasincubated at room temperature for 10 minutes. One microliter of32P-labeled NFkB consensus oligonucleotide probe was added andthe reaction kept on ice for 1 hour. The reaction was terminated byadding 1 µL of 10x gel loading buffer. For competition experiments1 µL unlabeled oligonucleotide was added to the reaction mixtureinstead of water. After electrophoresis, the gels were vacuum driedand exposed to radiograph film at 280°C. Autoradiographic imageswere quantified by densitometric analysis (Model GS-670, Bio-Rad,Hercules, CA).

Chemicals

All additional chemicals were purchased from Sigma ChemicalCo. (Poole, Dorset, UK).

Statistics

All the results are expressed as mean 6 SEM. Statistical analysiswas performed using ANOVA and where appropriate the unpaired

HEPATOLOGY Vol. 30, No. 5, 1999 HARRY ET AL. 1199

student’s t test. P values less than .05 were considered statisticallysignificant.

RESULTS

Effects of LPS on Mortality

All normal rats survived intraperitoneal injection of LPS atdoses of 0.5, 1.0, and 2.0 mg/kg. Previously we have observedan LD50 of greater than 20 mg/kg in normal rats at 24 hours(Unpublished observations, Harry and Moore, December,1997). In the BDL cirrhotic rats there was 100% mortality by6 hours after 2 mg/kg intraperitoneal LPS. For BDL rats givenLPS at 1 mg/kg, the mortality rate was 33% (1 of 3) at 6 hours,and 66% (2 of 3) at 24 hours. In view of this high mortality,more extensive studies were performed at a dose of 0.5 mg/kgof LPS. The mortality rate after LPS administration at 0.5mg/kg in the BDL group was 10% (1 of 10 ) at 6 hours and53% (8 of 15) at 24 hours. For all subsequent experiments, asingle dose of 0.5 mg/kg of LPS was used.

Effects of LPS on Liver and Renal Function

Plasma aspartate transaminase (AST) activity was signifi-cantly elevated in the BDL rats at baseline (533 6 72 U/L)compared with the controls (157 6 51 U/L, P , .01),consistent with liver injury. AST values increased afterinjection of LPS to 486 6 52 U/L and 2,556 6 456 U/L at 3hours in shams and BDL cirrhotic rats, respectively (P , .01)with respect to baseline values (Table 1). There was nochange in serum creatinine values in control rats after LPSinjection, but a significant increase was observed from 1 houronwards in the BDL cirrhotic group after LPS administration(P , .05) (Table 1). Baseline serum albumin concentrationwas decreased in the BDL group compared with controlsconsistent with impaired synthetic function. Although albu-min levels increased significantly in the BDL group after LPSadministration (P , .05) they remained significantly lowercompared with the albumin values of the sham group(Table 1).

On histological examination, there were occasional singledegenerate hepatocytes, but no confluent necrotic foci pres-ent in the liver of BDL control animals (i.e., no LPS, n 5 6). Afew neutrophils were observed around bile ductules but wereinconspicuous within sinusoids and no microabscesses wereseen. At 3 hours after injection of LPS (0.5 mg/kg) the liversfrom both BDL cirrhotic and normal animals showed diffusesinusoidal hypercellularity with a neutrophilic infiltrate withoccasional focal clusters in both normal and BDL animals thatresembled microabscesses. Degenerate hepatocytes with pyk-notic nuclei, cell shrinkage, and nuclear debris were seen inall of the BDL animals studied (n 5 6), but were only present

in a single sham control animal (n 5 5) given LPS. Hepato-cyte degeneration was least evident within the peripheralhepatocytes of the cirrhotic nodules (Fig. 1). Kupffer cellprominence was evident in the BDL cirrhotic controls, andwas similar after injection of LPS.

Effects of LPS on Lipid Peroxidation

Plasma F2-Isoprostanes. Measurement of F2-isoprostanes inbiological fluids is now accepted as the most reliable methodfor the assessment of oxidative stress in vivo.20 PlasmaF2-isoprostane levels were determined before and up to 6hours after LPS administration. Levels of F2-isoprostaneswere similar in the 2 groups before LPS injection, althoughthe levels in the sham group were significantly higher than inthe BDL group (P , .02) before LPS injection (Fig. 2). Therewas a significant increase in plasma F2-isoprostanes 1 hourafter LPS administration in both groups indicative of lipidperoxidation or hydrolysis and release of preformed tissueF2-isoprostanes. Levels increased from 116 6 11 to 315 6 38pg/mL for controls, and from 79 6 5 to 368 6 50 pg/mL forBDL cirrhotic rats (P , .01 compared with basal values, Fig.2). However, at 3 hours, plasma F2-isoprostanes remainedmarkedly elevated only in cirrhotic rats (511 6 56 pg/mL)compared with sham (235 6 31 pg/mL) and the differencebetween the 2 groups was now significant (P , .01)indicative of persistent lipid peroxidation in the cirrhotic rats.The significant difference in plasma F2-isoprostanes contin-ued throughout the study period so that at 6 hours after LPStreatment the mean plasma F2-isoprostane levels in the BDLanimals was 408 pg/mL compared with 131 pg/mL in thesham animals (P , .03) (Fig. 2).

Tissue F2-Isoprostanes. The levels of esterified F2-isopros-tanes in the liver and kidney membrane lipids were similar inboth groups of animals before the administration of LPS.However, LPS caused a marked increase in hepatic esterifiedF2-isoprostane content at 1 hour. Hepatic F2-isoprostanesincreased from 134 to 398 pg/mg phospholipid in the BDLcirrhotic rats, and from 100 to 192 pg/mg phospholipid in thesham group (P 5 .058). This difference did not reachstatistical significance because of a low outlying value in asingle BDL rat. If this animal is excluded from the analysis thedifferences become highly significant (P , .01). However, by3 hours after LPS administration, esterified F2-isoprostanes inthe liver had returned to baseline levels in both BDL andcontrol animals. A similar pattern was observed in the kidneyof BDL cirrhotic rats. The content of esterified F2-isopros-tanes increased in 116 612 to 242 6 22 pg/mg phospholipidin the kidney of BDL cirrhotic rats, with the levels returning

TABLE 1. Effect of LPS on Liver and Renal Function Tests

Time (h) AST (U/L) Creatinine (mmol/L) Bilirubin (mmol/L) Albumin (g/L) No.

BDL 0 533 6 72 51 6 2 81 6 10 15 6 1 61 768 6 189 58 6 3* 117 6 11* 17 6 1 93 2,556 6 456* 66 6 3* 125 6 17* 21 6 1* 96 909 6 234* 62 6 5* 87 6 11 20 6 1* 8

Sham 0 157 6 51 47 6 2 2 6 0.1 30 6 0.3 61 517 6 58* 43 6 3 3 6 0.7 28 6 1 63 486 6 52* 43 6 3 3 6 0.5 28 6 1 66 336 6 49* 46 6 4 2 6 0.1 30 6 1 6

*Significantly different from baseline values in the same group (P , .05).

1200 HARRY ET AL. HEPATOLOGY November 1999

to baseline values by 3 hours (P , .05). There was, however,no change in the sham kidneys (Fig. 3).

Effects of LPS on Plasma TNF-a Concentrations

Mean baseline plasma TNF-a levels were higher in the BDLgroup (52 6 22 pg/mL versus 9 6 1 pg/mL in the BDL andsham groups respectively, P , .05). After LPS administrationthere was a marked increase in TNF-a levels with peakconcentrations in both groups occurring at 1 hour (sham,2,988 6 454 vs. BDL, 3,621 6 498 pg/mL); differencesbetween the 2 animal groups were not significant. However,at 3 hours after LPS administration there was a persistent andsignificant (P , .01) elevation of plasma TNF-a in cirrhoticanimals (2,463 6 697 pg/mL) compared with controls(401 6 160 pg/mL); and levels remained significantly ele-vated for up to 6 hours in cirrhotic animals after LPS injection(Fig. 4).

Effects of LPS on NFkB Activation

Nuclear extracts prepared from the liver of the normal andBDL rats were analyzed with a 32P-labeled NFkB oligonucleo-

tide probe, and the 32P-labeled protein-oligonucleotide com-plex formed was determined using EMSA. The profile ofNFkB activation in normal and BDL rats before LPS injectionis shown in Fig. 5. NFkB was constitutively activated in BDLcirrhotic rats before LPS administration. Quantitation bydensitometry showed a 5-fold elevation in band intensity inBDL rats compared with normal controls (P , .001) at 0hours (Fig. 6). After injection of LPS, the activation of NFkBincreased to similar maximum levels after 1 hour in bothnormal and BDL cirrhotic animals, and there was now nosignificant difference between the 2 groups. NFkB levelsdecreased in both BDL and sham rats at 3 hours andcontinued to fall towards pre-LPS values 6 hours after LPSadministration (Fig. 6).

Effects of LPS on Plasma Nitrite/Nitrate Concentrations

The production of NO was estimated by measuring thelevels of nitrite/nitrate in the plasma. Plasma nitrite/nitratewere significantly increased in the BDL cirrhotic rats atbaseline (82 6 19 µmol/L vs. 29 6 4 µmol/L in the shamanimals; P , .03, Fig. 7). LPS injection caused plasmanitrite/nitrate levels to increase, and these were still signifi-cantly higher in the BDL group compared with the sham ratsat 0, 1, and 3 hours after LPS administration (Fig. 7).However, at their peak concentration (6 hours), levels weresimilar in the sham (172 6 23 µmol/L) and BDL rats (179 639 µmol/L).

DISCUSSION

There have been very few reports on the increased lethalityto endotoxin in animal models of cirrhosis. In healthyanimals the sensitivity to the toxic effects of LPS variesconsiderably across animal species. The rat is normally betterable to tolerate large doses of LPS (10-50 mg/kg), given eitherparenterally or by intraperitoneal injection, with only moder-ate adverse effects.21 In the current study it was observed thatsham (control) rats could tolerate 2 mg/kg of LPS intraperito-neally without any obvious adverse reactions, and we have

FIG. 1. Representative histological sections (hematoxylin-eosin stain) ofnormal rat liver (A) and BDL cirrhotic liver (B) at 3 hours after intraperito-neal injection of LPS. The normal liver shows cellular sinusoids with focalneutrophilic aggregates (arrow). Normal vesicular nuclei are present in thehepatocytes. After injection of LPS, examination of the BDL cirrhotic liver(B) shows focally degenerate hepatocytes with dark shrunken nuclei presentto the extreme left and to the right. Neutrophils and debris are also present(arrows). (Original magnification 3400.)

FIG. 2. Plasma F2-isoprostane concentration after the administration ofLPS (0.5 mg/kg) to BDL rats (solid bars) and control rats (hatched bars). Dataare means 6 SEM (n 5 6) for each time point. aP , .01, for values comparedwith baseline (t 5 0 hours) values. *P , .05, for control rats vs. BDL rats atthe corresponding time point. **P , .01, for control rats vs. BDL rats at thecorresponding time point.

HEPATOLOGY Vol. 30, No. 5, 1999 HARRY ET AL. 1201

previously used doses of up to 20 mg/kg without anyappreciable short-term mortality (Unpublished observation,Harry and Moore, December, 1997). By contrast the BDLcirrhotic rats had a marked increase in mortality at the lowdose of 0.5 mg/kg of LPS. Chang and Ohara22 observedincreased mortality rates in BDL rats after as little as 0.01mg/kg S. enteriditis LPS given intravenously. They attributedthe increased mortality to the appearance of pulmonaryintravascular phagocytes after bile duct ligation, which en-hanced the sensitivity to endotoxin in this model, leading tolung vascular injury and death.

In our study the liver function tests, already abnormal,deteriorated further in the BDL group after intraperitonealLPS. Serum AST levels increased markedly in the cirrhoticanimals within 3 hours of LPS administration indicative ofenhanced hepatic injury compared with controls. Renaldysfunction also developed in the BDL group, and serumcreatinine increased by approximately 30%. These observa-tions, together with the work of Chang and Ohara22 show

that there is multiorgan involvement after injection of endo-toxin in this model. Some of these changes could besecondary to hypotension and shock. The hemodynamicchanges evoked by LPS in the cirrhotic animals and controlswere not, however, evaluated in this study. Animals withcirrhosis may be more susceptible to shock because theycharacteristically have vascular dysfunction, in part causedby increased synthesis of NO. Basal synthesis of NO wasincreased in the BDL group. The reason for this increasedbasal synthesis is unknown, but could be linked to changes inantioxidant status or the endotoxemia that is known to occurafter biliary obstruction.3-5 The enhanced levels of nitrite/nitrate (NOx) observed in the first 3 hours in the BDLcirrhotic animals may be secondary to the considerableelevation of plasma TNF concentrations evident at 3 hours inthis group after injection of LPS23,24 (see later). Others,however, have suggested that TNF-a is not required forLPS-mediated induction of inducible nitric oxide synthase(iNOS) in the rat.25 The early differences in plasma NOxconcentrations were no longer evident at 6 hours suggestingthat NO synthesis is similar to controls at a time when themortality is high in the BDL cirrhotic rats. This suggests thatincreased NO formation alone is not the major cause of deathin the BDL cirrhotic rats. Indeed, some studies have shownthat administration of L-nitro-arginine methyl ester to normalrats given intravenous LPS markedly accelerated death,indicating that NO had a protective role after endotoxemia.26

The literature concerning the role of NO in endotoxic shockis, however, confusing with reports of both protective anddeleterious effects, for example, although infusion of L-nitro-monomethyl arginine may transiently improve blood pres-sure, there may be a marked reduction of survival.27 Severalstudies have shown that both Kupffer cells and hepatocytesthemselves synthesize NO.28,29 In LPS treated rats portal andsystemic plasma nitrite and nitrate levels (considered mark-ers of NO production) are increased over 40-fold comparedwith normal rats as a result of iNOS induction by LPS.30

Chronic exposure to sublethal levels of endotoxin, as occurs

FIG. 4. Plasma TNF-a concentrations after the administration of LPS (0.5mg/kg intraperitoneally) to BDL rats (solid bars) and control rats (hatchedbars). Data are means 6 SEM (n 5 6) for each time point. *P , .05, for BDLvs. sham controls at baseline. **P , .01, for BDL rat values vs. control ratvalues at the indicated time points.

FIG. 3. Liver and kidney esterified F2-isoprostane levels after theadministration of LPS (0.5 mg/kg intraperitoneally) to BDL rats (solid bars)and control rats (hatched bars). Data are means 6 SEM (n 5 5-6) for eachtime point except at 6 hours where n 5 3. aP , .01, for control rats vs. BDLrats at the corresponding time point. *P , .05, for values compared withbaseline (t 5 0 hours) values **P , .01, for BDL rat values at 1 hour vs. allother time points.

1202 HARRY ET AL. HEPATOLOGY November 1999

in biliary cirrhosis, may prime the liver parenchymal cells forthe production of NO when exposed to increased levels ofendotoxin or TNF-a.31

After injection of LPS there is a rapid accumulation ofneutrophils into the liver, which is maximal at 3 to 4hours.32,33 Neutrophils are major sources of reactive oxygenspecies and reactive nitrogen species (e.g., peroxynitrite),which can initiate lipid peroxidation and cell injury. Hostdefense against oxidant injury is impaired in BDL cirrhoticrats. Singh et al.34 observed decreased activity of hepaticglutathione peroxidase, catalase, and glutathione transferaseand decreased levels of plasma vitamin E and selenium, all ofwhich are important in limiting free radical-mediated injury.Thus, one might expect increased lipid peroxidation in theBDL cirrhotic animals. Many studies have used measurementof tissue malondialdehyde as a marker of oxidant injury.However, the measurement of malondialdehyde in tissues isgenerally considered to be a poor technique to accuratelyassess lipid peroxidation, and measurement of tissue orplasma F2-isoprostanes is being increasingly used to monitoroxidative stress in vivo.20 In the current study, plasma levels ofF2-isoprostanes increased after LPS presumably as a conse-quence of release from the liver and kidneys and other organs.However, there was an exaggerated response in the BDLcirrhotic rats with sustained elevation of plasma F2-isoprostanes and increased levels in the liver and kidney. Therapid decrease in levels of tissue esterified F2-isoprostanesmost probably occurs as a consequence of activation of a

phospholipase, which has been observed by others afterbiliary cirrhosis in the rat35 and after exposure of cells to LPS36

or injection of LPS into the perfused liver.37 Hatch et al.37

showed that Kupffer cell phospholipase A2 activity increasedby 4-fold in the rat within 2 hours of LPS injection, andactivity continued to increase thereafter. Molecular modelingstudies have shown potential deleterious effects of esterifiedF2-isoprostanes on membrane function,38 and rapid removalwould be essential for the maintenance of a functional cellmembrane in the presence of uncontrolled lipid peroxidation.This study shows that F2-isoprostanes may be formed rapidly

FIG. 5. A representative EMSA autoradiograph of nuclear extracts fromnormal rat liver (lane A), BDL cirrhotic liver (lane B), BDL liver 1 coldoligonucleotide (lane C), BDL liver 1 P50 antibody (lane D), BDL liver 1P65 antibody (lane E), and a Hela cell nuclear extract as a control (lane F).Lane B shows that there is activation of NFkB in the unstimulated BDLcirrhotic liver. The complete competition by cold specific oligonucleotide(lane C), and the presence of a supershift with the P50 (lane D) and P65 (laneE) antibodies confirms the identification of this band as NFkB, confirmingconstitutive activation.

FIG. 6. NFkB activation in the liver after the administration of LPS (0.5mg/kg intraperitoneally) to BDL rats (solid bars) and control rats (hatchedbars). NFkB was detected by EMSA (see text) and quantified by scanninglaser densitometry; results are expressed as relative optical density (R.O.D.)with respect to a known standard run with each gel. Data are means 6 SEM(n 5 6) for each time point. **P , .01, for BDL rat baseline values (t 5 0hours) vs. control rat baseline values.

FIG. 7. Plasma nitrite/nitrate levels after the administration of LPS (0.5mg/kg intraperitoneally) to BDL rats (solid bars) and control rats (hatchedbars). Data are expressed as the mean 6 SEM (n 5 5) for all time pointsexcept control rats at 3 hours and BDL rats at 6 hours where n 5 4. *P , .05,for BDL rat values vs. control rat values at the times indicated. aP , .01, forBDL rat values and control rat values in respect of their pre-LPS values.

HEPATOLOGY Vol. 30, No. 5, 1999 HARRY ET AL. 1203

during endotoxemia and that processes are activated to cleavethese compounds rapidly from the cell membrane in vivo.

Our results clearly have implications for the measurementof tissue F2-isoprostanes as a marker of lipid peroxidation inconditions in which phospholipases become activated. Theobservation that tissue levels of F2-isoprostanes were de-creased under basal conditions in the BDL rat liver andkidney could reflect a decrease in the availability of thesubstrate, arachidonic acid, or on-going activation of phospho-lipase that occurs in this model. Immediately after bile ductligation there is a marked increase in urinary excretion ofF2-isoprostanes, with a gradual decrease towards baselinevalues by 4 weeks (Unpublished observations, Harry andMoore, September, 1995). This suggests that there is contin-ued activation of a phospholipase in the BDL model asreported by Vishwaneth et al.,35 and this ultimately decreasesthe pool of arachidonate available for oxidation. Recentstudies have suggested that synthesis of F2-isoprostanes ismediated through a cyclo-oxygenase pathway in the rat.39,40

However, our observation that esterified F2-isoprostanesrapidly increased in the liver and kidney, after intraperitonealinjection of LPS, indicates that they are formed, at least inthese organs, before activation of phospholipases and there-fore through a non–cyclo-oxygenase dependent pathway.Three hours after LPS, esterified levels of F2-isoprostanes hadreturned to basal values. A similar pattern was observed inthe kidneys of the BDL rats but interestingly there was nochange in the kidney levels of F2-isoprostanes in the controlgroup. Tissue-specific lipid peroxidation has been reported tooccur within 90 minutes of endotoxin infusion in ratssuggesting that it contributes to the pathophysiology ofendotoxemia.41 These data also indicate that there is rapidup-regulation of phospholipase A2 activity. Stoner et al.42 havepreviously shown a strong correlation between plasma levelsof phospholipase A2 (PLA2) and septic shock, whereas Lo etal.43,44 have shown that inhibition of PLA2 activity causeddown-regulation of the over-activity of endotoxin-challengedmacrophages. Phospholipase A2 is believed to play an impor-tant role in the maintenance of cell membrane integrityduring lipid peroxidation, and levels of this enzyme in theliver are known to increase in inflammatory disorders includ-ing septic shock.45 These observations lend support to theimportance of processes that remove esterified isoprostanesfrom cell membranes rapidly after they are formed. In ourstudy another striking observation was the sustained 6-foldto 10-fold increase in plasma TNF-a values in the BDLcirrhotic group at 3 to 6 hours. This greater and moreprolonged elevation of TNF-a concentrations in the BDLcirrhotic rats could be because of either an increase insynthesis or a decrease in clearance of TNF-a. However, theobservation by McClain and Cohen46 that monocyte synthe-sis of TNF-a is increased, as well as the magnitude of theplasma concentrations observed (.2,500 pg/mL), points toelevated production rather than decreased clearance as themajor reason for this elevation. Luster et al.47 have shown bykinetic analysis of cultured mouse liver that most TNF-a isreleased within 1 hour of LPS challenge. It is possible thatimpaired antioxidant defense in cirrhosis results in a dysregu-lation of TNF-a synthesis. This idea is supported by data thatshowed that pretreatment with antioxidants protect miceagainst the adverse effects of LPS and abolished the TNF-aresponse.48 Kupffer cells are considered to be the primarysites of TNF-a synthesis, but the epithelium of intrahepatic

bile ducts, which proliferate after bile duct obstruction, havealso been shown to produce TNF-a after LPS administra-tion.49 Spontaneous production of TNF-a by monocytes inculture is similar in both cirrhotic and healthy subjects.However, when monocytes from patients with alcohol-induced hepatitis were challenged with LPS there was agreater increase in TNF-a release by these cells comparedwith controls.46 This suggests that these cells have beenprimed in some way, e.g., by circulating endotoxin. Our initialobservations that there was constitutive activation of NFkB inthe liver led us to determine whether control of nucleartranscription and therefore TNF-a synthesis was dysregu-lated at the NFkB level. However, after injection of LPS therewas a similar degree of NFkB activation in the liver in bothcontrol and BDL cirrhotic animals.

In the current study we did not attempt to determine thecellular origin of the constitutively activated NFkB, becausetechniques to isolate cells from cirrhotic or normal liver mayactivate cellular NFkB. However, it is likely that NFkB isactivated in either the Kupffer cells, e.g., because of endotox-emia, or in the hepatic stellate cells. Constitutive activation ofNFkB has recently been described in the hepatic stellate cellsduring their transformation to myofibroblasts.50

We conclude that LPS has several effects in the BDLcirrhotic rat, including increased mortality, increased oxidantinjury, and increased cytokine synthesis. The marked increasein sensitivity to LPS in BDL cirrhotic rats is similar to thatseen in patients with cirrhotic liver disease in whom infec-tions are an important cause of mortality. Because thepathways involved are sensitive to reactive oxygen or nitro-gen species, strategies to limit oxidant injury may down-regulate the cytokine response51,52 and subsequent tissueinjury. Furthermore, inhibiting lipid peroxidation by antioxi-dants may decrease tissue injury and improve survival in thismodel.

REFERENCES

1. Nolan JP. The role of endotoxin in liver injury. Gastroenterology1975;69:1346-1356.

2. Bode C, Kugler V, Bode JC. Endotoxemia in patients with alcoholic andnon-alcoholic cirrhosis and in subjects with no evidence of chronic liverdisease following acute alcohol excess. J Hepatol 1987;4:8-14.

3. Scott-Conner CE, Grogan JB. The pathophysiology of biliary obstructionand its effect on phagocytic and immune function [review]. J SurgicalRes 1994;57:316-336.

4. Lumsden AB, Henderson JM, Kutner MJ. Endotoxin levels measured bya chromogenic assay in portal, hepatic and peripheral venous blood inpatients with cirrhosis. HEPATOLOGY 1988;8:232-236.

5. Grinko I, Geerts A, Wisse E. Experimental biliary fibrosis correlates withincreased numbers of fat-storing and Kupffer cells and portal endotox-emia. J Hepatol 1995;23:449-458.

6. Manthay CL, Vogel SN. Interactions of lipopolysaccharide with macro-phages. Immunol Series 1994;60:63-81.

7. Peavy DL, Fairchild Jr ET. Evidence for lipid peroxidation in endotoxin-poisoned mice. Infect Immunol 1986:52:613-616.

8. Wardle EN, Wright NA. Endotoxin and acute renal failure associatedwith obstructive jaundice. British Med J 1970;4:472-474.

9. Burke GW, Cirocco R, Roth D, Fernandez J, Allouche M, Markou M,Reddy R, et al. Activated cytokine profile in hepatorenal syndrome: fallin levels after successful orthotopic liver transplantation. Trans Proc1993;25:1876-1877.

10. Khoruts A, Stahnke L, McClain CJ, Logan G, Allen JI. Circulating tumornecrosis factor, interleukin-1 and interleukin-6 concentration in chronicalcoholic patients. HEPATOLOGY 1991;13:267-276.

11. Mathison JC, Ulevitch RJ. The clearance, tissue distribution and cellularlocalization of intravenously injected lipopolysaccharide in rabbits. JImmunol 1979;123:2133-2143.

12. Levy E, Ruebner BH. Hepatic changes produced by a single dose of

1204 HARRY ET AL. HEPATOLOGY November 1999

endotoxin in the mouse. Light microscopy and histochemistry. Am JPathol 1967;51:269-285.

13. Schreck R, Alberman K, Baeuerle PA. Nuclear factor Kappa B; anoxidative stress-responsive transcription factor of eukaryotic cells [re-view]. Free Radical Res Commun 1992;17:221-237.

14. Baeuerle PA, Henkel T. Function and activation of NFkB in the immunesystem. Ann Rev Immunol 1994;12:141-179.

15. Hewett JA, Schultze AE, VanCise S, Roth RA. Neutrophil depletionprotects against liver injuries from bacterial endotoxin. Lab Invest1992;66:347-361.

16. Morrow JD, Roberts LJ. Mass spectrometry of prostanoids F2-isoprostanes produced by non-cyclooxygenase free radical-catalyzedmechanism. Methods Enzymol 1994;233:163-174.

17. Folch J, Lees M, Sloane Stanley GH. A simple method for the isolationand purification of total lipides from animal tissues. J Biol Chem1957;226:497-509.

18. Manning AM, Bell FP, Rosenbloom CL, Chosay JG, Simmons CA,Northrup JL, Shebuski RJ, et al. NF-kappa B is activated during acuteinflammation in vivo in association with elevated endothelial celladhesion molecule gene expression and leukocyte recruitment. J In-flamm 1995;45:283-296.

19. Bradford MM. A rapid and sensitive method for the quantitation ofmicrogram quantities of protein utilizing the principle of protein-dyebinding. Anal Biochem 1976;72:248-254.

20. Moore KP, Roberts LJ Measurement of lipid peroxidation. Free RadicalResearch 1998;28:659-671.

21. Berizi I, Bertok L, Berezani T. Comparative studies on the toxicity ofEscherichia coli lipopolysaccharide endotoxin in various animal species.Can J Microbiol 1966;12:1070-1071.

22. Chang S, Ohara N. Chronic biliary obstruction induces pulmonaryintravascular phagocytosis and endotoxin sensitivity in rats. J ClinInvest 1994;94:2009-2019.

23. Aono K, Isobe K, Kiuchi K, Fan ZH, Ito M, Takeuchi A, Miychi M, et al.In vitro and in vivo expression of inducible nitric oxide synthase duringexperimental endotoxemia: involvement of other cytokines. J CellBiochem 1997;65:349-358.

24. Spitzer JA. Cytokine stimulation of nitric oxide formation and differen-tial regulation of hepatocytes and non-parenchymal cells of endotox-emic rats. HEPATOLOGY 1994;19:217-228.

25. Xie J, Joseph KO, Bagby GJ, Giles TD, Greenberg SS. Dissociation ofTNF-a from endotoxin-induced nitric oxide and acute-phase hypoten-sion. Am J Physiol 1997;273:H164-H174.

26. Wolkow PP, Bartus JB, Gryglewski RJ. Pneumotoxicity of lipopolysaccha-ride in nitric oxide-deficient rats is limited by a thromboxane synthaseinhibitor. J Physiol Pharm 1997;48:645-653.

27. Grylgelwski RJ, Wolkow PP, Uracz W, Janowska E, Bartus JB, Balbatun O,Patton S, et al. Protective role of pulmonary nitric oxide in the acutephase of endotoxemia in rats. Circulation Res 1998;82:819-827.

28. Curran RD, Billiar TR, Stuehr DJ, Hofmann K, Simmons RL. Hepatocytesproduce nitrogen oxides from L-arginine in response to inflammatoryproducts of Kupffer cells. J Exp Med 1989;170:1769-1774.

29. Billiar TR, Curran RD, Stuehr DJ, West MA, Bentz BG, Simmons RL. AnL-arginine dependent mechanism mediates Kupffer cell inhibition ofhepatocyte protein synthesis in vitro. J Exp Med 1989;169:1467-1472.

30. Frederick JA, Hasselgren PO, Davis S, Higashiguchi T, Jacob TD, FischerJE. Nitric oxide may up-regulate hepatic protein synthesis duringendotoxemia. Arch Surg 1993;128:152-157.

31. Pittner RA, Spitzer JA. Endotoxemia and TNFa directly stimulate nitricoxide formation in cultured rat hepatocytes from chronically endotox-emic rats. Biochem Biophys Res Comm 1992;185:430-435.

32. Nolan JP. Endotoxin, reticuloendothelial function and liver injury.HEPATOLOGY 1981;1:458-465.

33. Spitzer JA, Mayer AMS. Hepatic neutrophil influx: eicosanoid andsuperoxide formation in endotoxemia. J Surg Res 1993;55:60-67.

34. Singh S, Shackleton G, Ah-Sing E, Chakraborty J, Bailey ME. Antioxi-dant defenses in the bile duct-ligated rat. Gastroenterology 1992;103:1625-1629.

35. Vishwanath BS, Frey FJ, Escher G, Reichen J, Frey BM. Liver cirrhosisinduces renal and liver phospholipase A2 activity in rats. J Clin Invest1996;98:365-371.

36. Mayer AM, Brenic S, Stocker R, Glaser KB. Modulation of superoxidegeneration in in vivo lipopolysaccharide-primed rat alveolar macro-phages by arachidonic acid and inhibitors of protein kinase C, phospho-lipase A2, protein serine-threonine phosphatase(s), protein kinase(s)and phosphatase(s). J Pharm Exper Ther 1995;274:427-436.

37. Hatch GM, Vance DE, Wilton DC. Rat liver mitochondrial phospholi-pase A2 is an endotoxin-stimulated membrane-associated enzyme ofKupffer cells which is released during perfusion. Biochem J 1993;293:143-150.

38. Morrow JD, Awad JA, Boss HJ, Blair IA, Roberts LJ. Non-cyclooxygenase-derived prostanoids (F2-isoprostanes) are formed in situ on phospholip-ids. Proc Natl Acad Sci U S A 1992;89:10721- 10725.

39. Pratico D, Lawson JA, Fitzgerald GA. Cyclooxygenase dependentformation of the isoprostane, 8-epi prostaglandin F2a. J Biol Chem1995;270:9800-9808.

40. Delanty N, Reilly MP, Pratico D, Lawson JA, McCarthy JF, Wood AE,Ohnishi ST, et al. 8-epi-PGF2 alpha generation during coronary reperfu-sion. A potential quantitative marker of oxidant stress in vivo. Circula-tion 1997;95:2492-2499.

41. Clements NC, Habib MP. The early pattern of conjugated dienes in liverand lung after endotoxin exposure. Am J Respir Crit Care Med1995;151:780-784.

42. Stoner CR, Reik LM, Donohue M, Levin W, Crowl RM. Human group 11phospholipase A2 . Characterisation of monoclonal antibodies andimmunochemical quantitation of the protein in synovial fluid. J ImmunMeth 1991;145:127-136.

43. Lo CJ, Cryer HG, Maier RV. Prostaglandin E2 production by endotoxin-stimulated alveolar macrophages is regulated by phospholipase Cpathways. J Trauma 1996;40:557-562.

44. Lo CJ, Cryer HG, Fu M, Kim B. Endotoxin-induced macrophage geneexpression depends on platelet-activating factor. Arch Surg 1997;132:1342-1347.

45. Man WJ, Tomlinson P, Wilton DC. The effect of endotoxin on the activityof group 11 rat liver phospholipase A2. Biochem Soc Trans 1994;22:316S.

46. McClain C, Cohen DA. Tumor necrosis factor production by monocytesin alcoholic hepatitis. HEPATOLOGY 1989;9:349-351.

47. Luster MI, Germolec DR, Yoshida T, Kayama F, Thompson M. Endotoxin-induced cytokine expression and excretion in the liver. HEPATOLOGY

1994;19:480-488.48. Neihorster M, Inoue M, Wendel A. A link between extracellular reactive

oxygen and endotoxin-induced release of tumor necrosis factor a in vivo.Bioch Pharm 1991;43:1151-1154.

49. Beno DWA, Retsky JE, Davis BH. Lipid peroxidation-induced isopros-tane independently activates ITO cell MAP kinase and increases type 1collagen mRNA [Abstract]. HEPATOLOGY 1994;20(Suppl):290A.

50. Elsharakawy AM, Wright MC, Hay RT, Authur MJP, Hughes T, Bahr MJ,Degitz K, Mann DA. Persistent activation of nuclear factor kB in culturedrat hepatic stellate cells involves the induction of potentially novelrel-like factors and prolonged changes in the expression of IkB familyproteins. HEPATOLOGY 1999;30:761-769.

51. Hoffmann R, Grewe M, Estler HC, Schultze-Specking A, Decker K.Regulation of tumor necrosis factor-alpha-mRNA synthesizing cells inrat liver during experimental endotoxemia. J Hepatol 1994;20:122-128.

52. Villa P, Ghezzi P. Effect of N-acetyl-L-cysteine on sepsis in mice. Europ JPharm 1995;292:341-344.

HEPATOLOGY Vol. 30, No. 5, 1999 HARRY ET AL. 1205