Endotoxemia produces coma and brain swelling in bile duct ligated rats


  • Potential conflict of interest: Nothing to report.


This study explores the hypothesis that the inflammatory response induced by administration of lipopolysaccharide (LPS) exacerbates brain edema in cirrhotic rats; and if so whether this is associated with altered brain metabolism of ammonia or anatomical disturbance of the blood-brain barrier. Adult Sprague-Dawley rats 4 weeks after bile duct ligation (BDL)/Sham-operation, or naïve rats fed a hyperammonemic diet (HD), were injected with LPS (0.5 mg/kg, intraperitoneally) or saline, and killed 3 hours later. LPS administration increased brain water in HD, BDL, and sham-operated groups significantly (P < 0.05), but this was associated with progression to pre-coma stages only in BDL rats. LPS induced cytotoxic brain swelling and maintained anatomical integrity of the blood-brain barrier. Plasma/brain ammonia levels were higher in HD and BDL rats than in sham-operated controls and did not change with LPS administration. Brain glutamine/myoinositol ratio was increased in the HD group but reduced in the BDL animals. There was a background pro-inflammatory cytokine response in the brains of cirrhotic rats, and plasma/brain tumor necrosis factor alpha (TNF-α) and IL-6 significantly increased in LPS-treated animals. Plasma nitrite/nitrate levels increased significantly in LPS groups compared with non-LPS controls; however, frontal cortex nitrotyrosine levels only increased in the BDL + LPS rats (P < 0.005 versus BDL controls). Conclusion: Injection of LPS into cirrhotic rats induces pre-coma and exacerbates cytotoxic edema because of the synergistic effect of hyperammonemia and the induced inflammatory response. Although the exact mechanism of how hyperammonemia and LPS facilitate cytotoxic edema and pre-coma in cirrhosis is not clear, our data support an important role for the nitrosation of brain proteins. (HEPATOLOGY 2007.)

Hepatic encephalopathy (HE) is an important complication of cirrhosis, the severity of which can vary widely from mild sleep disturbance through neuropsychological impairment to coma, brain herniation, and death. Though a common and defining characteristic of acute liver failure (ALF),1 the development of advanced cerebral edema in cirrhosis is thought to be rare. However, severe brain swelling and intracranial hypertension has been reported in patients with cirrhosis who have severe precipitating illness, such as uncontrolled gastrointestinal bleeding, insertion of a transjugular intrahepatic shunt, or severe sepsis.2–4

Although the exact pathogenesis of HE is not clear, ammonia is thought to play a central role, and its levels have been shown to correlate with the severity of HE in patients with cirrhosis.5 Furthermore, arterial ammonia levels predict brain herniation in patients with acute liver failure.6 Ammonia is thought to produce astrocytic brain edema through the accumulation of glutamine, which is the detoxification product of ammonia metabolism resulting in increased osmotic stress.7, 8 Indeed, extracellular glutamine levels have been shown to correlate with the severity of intracranial hypertension.9 In patients with ALF, the severity of inflammation (often secondary to infection) correlates with the progression to advanced stages of encephalopathy and intracranial hypertension.10, 11 Recently, our group reported a correlation between the severity of increased intracranial pressure and the circulating levels of pro-inflammatory cytokines in ALF patients.12 We also observed brain production of cytokines in a group of ALF patients with uncontrolled intracranial hypertension.12, 13 In patients with cirrhosis, we demonstrated that induced hyperammonemia resulted in significant worsening of neuropsychological function in those patients with evidence of an inflammatory response due to bacterial infection. This effect was not found after resolution of the infection after antibiotic therapy.14

These data strongly indicate a possible synergy between hyperammonemia and inflammation in the development of HE; however, whether the background cirrhotic state or the associated hyperammonemia predisposes to the effects of the superimposed inflammation is not clear. The aim of this study was to explore the hypothesis that the inflammatory response induced by administration of bacterial cell wall lipopolysaccharide (LPS) exacerbates brain edema in cirrhotic rats, and if so whether the edema is associated with altered brain metabolism of ammonia or anatomical disturbance of the blood-brain barrier. We chose to perform this study using our well-characterized bile duct–ligated rat (BDL) model of cirrhosis and diet-induced hyperammonemic rats.


ALF, acute liver failure; BDL, bile duct ligation; HD, hyperammonemic diet; HE, hepatic encephalopathy; LPS: lipopolysaccharide (bacterial cell wall); NT, nitrotyrosine; TNF-α, tumor necrosis factor alpha.

Materials and Methods

All animal experiments were conducted according to Home Office guidelines under the UK Animals in Scientific Procedures Act 1986. Male Sprague-Dawley rats (body weight 230–280g) were obtained from the comparative biological unit at the Royal Free and University College Medical School, University College London. All rats were housed in the unit and given free access to standard rodent chow and water, with a light/dark cycle of 12 hours, at a temperature of 19°C to 23°C and humidity of approximately 50%.

Animal Models

Bile Duct Ligation.

Under anesthesia (diazepam 1mg/kg administered intravenously, followed by Hypnorm 150 μl/kg administered intramuscularly, Janssen Pharmaceutica, Belgium), all rats underwent bile duct ligation to induce biliary cirrhosis, or a sham operation as described previously.15 One BDL and one sham-operated rat died within 36 hours of the operation because of complications of anesthesia.

Noncirrhotic Hyperammonemic Animals.

Naïve rats (n = 14) were administered a high protein/ammoniagenic diet (HD) for 7 days before administration of LPS/saline 3 hours before termination (as outlined previously). The diet consisted of a liquid rodent feed (Bioserve, Frenchtown, NJ) and a tailor-made mixture mimicking the amino-acid composition of the hemoglobin molecule (4 g/kg/d; Nutricia, Cuijk, The Netherlands) as described previously,16, 17 mixed with commercially available gelatin to prevent sedimentation. This regimen produces chronic hyperammonemia to levels that are similar to that observed in BDL animals.

Study Design

Twenty-eight days after surgery, the operated rats (BDL and sham) were randomized into four groups, with BDL or sham-operated rats injected with LPS (Salmonella typhimurium, 0.5 mg/kg intraperitoneally made up to 0.5 ml in saline or saline alone: (1) sham-operated + saline (n = 7); (2) sham-operated + LPS (n = 7); (3) BDL + saline (n = 6); and (4) BDL + LPS (n = 6). The HD rats were also randomized to receive LPS (n = 7) or saline (n = 7).

As per protocol, the rats were allowed free access to food and water for a period of 3 hours after intervention in a temperature-controlled environment and were then killed by exsanguination under anesthesia (Hypnorm 200 μl/kg intramuscularly), 20 minutes after diazepam (1 mg/kg intraperitoneally). Blood was withdrawn from the descending aorta and immediately put into ice-cold heparin-/EDTA-containing tubes (until full exsanguination), centrifuged at 4°C, and the plasma collected and stored at −80°C until assayed.

Assessment of Level of Consciousness

The conscious levels of rats used in this study were rated using an established neurological scale.18 It was rated as either normal, with loss of the scatter reflex (c.f. grade 1 encephalopathy) and ataxia (c.f. grade 2 encephalopathy), together representing pre-coma stages, or loss of the righting reflex (c.f. grade 3-4 encephalopathy) representing the coma stage. A previous study15 established that 3 hours after administration of LPS, BDL rats all reached pre-coma or early coma stages. Therefore, in the current experiments all rats were killed 3 hours after injection of LPS or saline.

Brain Water Measurement

Immediately after death, the whole brain was rapidly removed, and 2-mm2 samples were dissected from the frontal cortex (gray matter) and the cerebellar cortex. Brain tissue water content (as a measure of cerebral edema) was determined using a previously described gravimetric technique.19 Tissue samples were coded with separate investigators collecting tissue samples or performing gravimetry (in adjoining rooms).

Measurement of Plasma Biochemistry

Plasma samples (200 μl) were analyzed for ALT, albumin, total protein, bilirubin, urea, and electrolytes, using a Cobas Integra 400 multianalyzer with the appropriate kits (Roche-diagnostics, Burgess Hill, West Sussex, UK).

Measurement of Ammonia


Ammonia was measured in the deproteinized plasma, using an indophenol detection method as described previously.20


The cerebral cortex (100 μg) was homogenized and deproteinized (using a glass tube Teflon pestle homogenizer) in 300 μl ice-cold cell lysis buffer solution. After centrifugation at 12,000g for 10 minutes at 4°C, the supernatants were collected and ammonia measured using absorbance photometry (Cobas integra 400 multianalyzer; Roche Diagnostics).

Brain Proton Magnetic Resonance Spectroscopy (1H-MRS)

Snap-frozen cortical brain samples were processed and analyzed by 1H-NMR using a previously described technique.211H-NMR spectra were recorded on a Bruker WB 360 spectrometer using a 5-mm QNP probe, 100 to 200 accumulations, repetition time 16 seconds, spectral width 3,623 Hz, data size 16 K, zero filling to 32 K. Chemical shifts were referenced to lactate at 1.33 ppm. Total metabolite concentrations (μmol/g ww) were analyzed from 1H-NMR spectra of PCA extracts using (trimethylsilyl) propionic-2,2,3,3d4-acid as external standard.22

Measurement of Nitrite/Nitrate

Combined NO2/NO3 levels were determined from the heparinized plasma samples by a modified Greiss test, using methods we have described previously.23–25 Nitrite concentrations were measured in the filtrate by chemiluminescence as described previously.15

Measurement of Brain Tissue Nitrotyrosine Levels

Brain tissue was homogenized in buffer and proteins precipitated by mixing with methanol/chloroform (1:2). Brain tissue homogenates were analyzed for nitrotyrosine (NT), and tyrosine levels were quantitated using a stable isotope-dilution gas chromatography/negative ion chemical ionization mass spectrometry method.26–28 This method prevents the artifactual nitration of tyrosine that occurs during acidic hydrolysis conditions. This method has an intra-assay and inter-assay variation of less than 5%. All results are expressed in relation to dry weight of protein as measured by direct weighing before hydrolysis.

Measurement of Plasma and Brain Cytokine Levels

Snap-frozen and stored (−80°C fridge) 100-μg cortical brain samples were homogenized in cell lysis buffer solution. After protein concentration quantification of brain homogenates, equilibrated brain protein samples (50 μl) were loaded onto tumor necrosis factor alpha (TNF-α) (Bender MedSystems; Vienna Austria), IL-1β (Biosource International Inc., Nivelles, Belgium) and IL-6 (Biosource International Inc., Nivelles, Belgium) enzyme-linked immunosorbent assay 96-well plate kits and read using a 96-well plate reader at 450nm (Sunrise, Tecan, Salzburg, Austria) according to the manufacturer's instructions.

Histopathological Assessment

Twenty-four additional rats (4 BDL + LPS, 4 BDL + saline, 4 sham-operated + LPS, 4 sham-operated + saline, 4 HD, and 4 HD + LPS), were anesthetized for brain tissue collection (Diazepam intraperitoneally 1 mg/kg: Hypnorm, intramuscularly 200 μl/kg) (for details of methodology, see Appendix, online). After removal of the blood using a heparinized saline flush, fixative containing 2% lanthanum nitrate (Agar Scientific Ltd, UK), a low molecular weight (MW 433) ionic tracer used to evaluate the integrity of the blood-brain barrier was injected through the left ventricle to perfuse (100-110 mmHg) and fix the brain. Two-cubic-millimeter sections of cerebellar and frontal were processed into Spur's resin (Agar Scientific Ltd, UK). Semi-thin (1-2 μm) sections were cut and viewed in a Hitachi 7100 transmission electron microscope at 75 kV. Digital images were captured with a Gatan column-mounted CCD camera at a resolution of 1024 × 1024 pixels and archived on a personal computer.

All materials and reagents were of a suitable laboratory grade and obtained from the Sigma Chemical Company (Poole, UK) unless otherwise stated.


Data are expressed as mean (±SEM). Significance of difference was tested with ANOVA, the unpaired t test, and the Mann-Whitney test, indicated in figure legends; P < 0.05 was taken to be statistically significant. Software used included Microsoft Excel 2003 (Microsoft Corp., Redmond, WA) and GraphPad Prism 4.0 (GraphPad Software, Inc., San Diego, CA).


All rats continued to gain weight after surgery. From the final weight taken immediately before termination, BDL rats (mean of 320 g; n = 12) were found to be marginally heavier compared with sham-operated controls (mean of 296 g; n = 14); P = 0.01. This difference was attributed to the presence of ascites. At 28 days, all BDL rats were deeply jaundiced.

Any potential shock was reduced because the study rats were used to handling and because of the anesthetic protocol used. This was indicated by the maintenance of systemic hemodynamics seen within our study groups (MAP, mean ± SEM: sham-operated, 118 ± 3.83; BDL, 104 ± 6; HD, 132 ± 7; P > 0.5), with the effect of LPS on hemodynamics (over 3 hours) already well characterized by our group.29

Conscious Level and Brain Water

All rats were still alive 3 hours after injection of LPS or saline. After injection of LPS, the BDL rats had a progressive and marked deterioration in their conscious level starting within 30 to 60 minutes, reaching pre-coma stages at the 3-hour killing time. The BDL, sham-operated, and HD rats administered saline and the sham-operated and HD rats administered LPS remained fully alert. A significant increase was seen in the water content in the frontal cortex of rats administered LPS compared with the corresponding saline-administered controls (sham-operated, P = 0.038; BDL, P = 0.004; and HD, P = 0.012: Table 1). No significant difference was seen in the brain water content between saline-treated sham-operated and BDL groups (P = 0.73); however, in saline-treated controls, HD rats had significantly raised frontal brain water compared with sham-operated (P = 0.001) and BDL (P = 0.0007) groups. No significant difference was seen in the brain water content between the LPS-administered groups (P = 0.7) (Fig. 1). No effect of LPS on cerebellar water content was seen in any of the groups studied (data not shown).

Table 1. Brain Levels of Nitrated Proteins and Brain Water, and Plasma Levels of Nitrate and Nitrite
  • NOTE. Values are given as mean ± standard error of mean (SEM) of duplicate samples from at least (n = 6) rats per group. Statistical analysis was determined using the Mann-Whitney U test (for brain water and nitrite/nitrate studies) or an unpaired t test (nitrite alone) to compare corresponding LPS- and saline-treated groups. Intergroup differences for NT/T ratios were calculated using the Newman-Keuls multiple comparison test.

  • *

    P < 0.05: compared with sham-operated rats.

  • P < 0.01 compared with BDL rats.

  • P < 0.05 compared with HD rats.

  • §

    P < 0.05

  • P < 0.05 compared with BDL + LPS rats.

  • Abbreviations: BDL, bile duct ligation; HD, high-ammonia diet; LPS, lipopolysaccharide.

Frontal brain water (%)79.8 ± 0.2980.9 ± 0.24*79.9 ± 0.2780.8 ± 0.1881.1 ± 0.1481.8 ± 0.15
Nitrite/nitrate39.10 ± 5.52139.7 ± 38.3*59.49 ± 4.9124.7 ± 13.943.86 ± 9.086.22 ± 19.3
Nitrite (μM)1.2 ± 0.281.9 ± 0.18*1.5 ± 0.142.9 ± 0.51§1.2 ± 0.022.2 ± 0.38
NT/T26 ± 9.660 ± 3512 ± 1.8139 ± 5613 ± 1.19.3 ± 0.53
Plasma ammonia (μmol/L)134 ± 12127 ± 13168 ± 14*172 ± 37231 ± 47*180 ± 38
Figure 1.

Frontal cortex brain water content. There was a significant increase in percentage water content with LPS administration in BDL, sham-operated rats, and HD rats (BDL + LPS compared with BDL + saline rats, #P < 0.005; Sham-operated + LPS compared with sham-operated + saline rats, *P < 0.038; and HD versus HD + LPS rats, $P < 0.013, respectively), as determined using the Mann-Whitney test to compare groups. Values are given as mean ± standard error of mean (SEM).


In BDL rats, ALT, bilirubin and albumin levels were consistent with cirrhosis (Table 2), and the bilirubin was significantly elevated compared with sham-operated and HD groups (P < 0.001), respectively. No significant impact was seen after LPS administration in BDL, sham-operated, or HD rats except for an expected, and previously well-described, increase in ALT and urea seen in the BDL rats. Plasma albumin was lower in the BDL animals, but the renal function, sodium, potassium, and plasma osmolarity were not significantly different between groups (Table 2).

Table 2. Plasma Biochemical Measurements and Cerebral Ammonia
  • NOTE. Values are given as mean ± standard error of mean (SEM) of duplicate samples from at least (n = 6) rats per group. Statistical significance was calculated using Newman-Keuls multiple comparison test.

  • *

    P < 0.01 compared with BDL rats.

  • P < 0.001 compared with sham-operated rats.

  • P < 0.001 compared with HD rats.

  • Abbreviations: BDL, bile duct ligation; HD, high-ammonia diet; LPS, lipopolysaccharide.

Plasma ALT (U/l)70 ± 690 ± 1298 ± 15168 ± 54*31 ± 237 ± 4
Plasma bilirubin (μmol/l)10 ± 324 ± 6180 ± 18181 ± 1428 ± 332 ± 3
Plasma creatinine (μmol/l)22 ± 124 ± 318 ± 0.0$27 ± 539 ± 232 ± 2
Plasma urea (mmol/l)5 ± 0.25 ± 0.65 ± 0.69 ± 0.4%7 ± 0.66 ± 0.9
Plasma albumin (g/l)35 ± 238 ± 126 ± 1#$24 ± 239 ± 0.740 ± 0.7
Plasma total protein (g/l)46 ± 348 ± 254 ± 348 ± 351 ± 0.951 ± 1
Plasma sodium (mmol/l)139 ± 2133 ± 3141 ± 3141 ± 3134 ± 5141 ± 5
Plasma potassium (mmol/l)4 ± 0.24 ± 0.66 ± 0.56 ± 0.55 ± 0.45 ± 0.1
Plasma chloride (mmol/l)104 ± 197 ± 2106 ± 2105 ± 397 ± 0.999 ± 0.9
Plasma osmolarity (mOsm/l)283 ± 22270 ± 22280 ± 23282 ± 23280 ± 41289 ± 40


In saline-treated BDL and HD rats, the plasma ammonia concentration was significantly higher compared with the respective saline-treated sham-operated rats (P = 0.035 and P = 0.038, respectively; Fig. 2, Table 1). However, no significant difference was seen between the hyperammonemic saline-treated BDL compared with HD rats (P = 0.63). After injection of LPS in each group of rats, no change was observed in plasma ammonia concentrations compared with the saline-treated controls. The brain ammonia concentration was found to be elevated in BDL and HD rats compared with sham-operated rats. LPS administration did not alter brain ammonia concentrations significantly from the respective saline-treated animals (Table 3).

Figure 2.

Plasma ammonia levels. Plasma ammonia levels as determined by the indophenol detection method. BDL + saline rats were found to have a higher plasma ammonia level compared with sham-operated controls, #P = 0.035. HD + saline rats were also found to have significantly higher plasma ammonia levels compared with the sham + saline group, $P = 0.038. No significant difference was found between the BDL and HD saline-treated groups (P = 0.067), and no differences were found in any group after the administration of LPS. Values are given as mean ± standard error of mean (SEM).

Table 3. Brain Osmolyte Profile as Analyzed by Proton NMR Spectroscopy and Brain Ammonia
  • NOTE. Values are given as mean ± standard error of mean (SEM) of duplicate samples from at least (n = 6) rats per group.

  • *

    P < 0.05 compared with HD rats.

  • P < 0.001 compared with HD rats.

  • Abbreviations: BDL, bile duct ligation; HD, high-ammonia diet; LPS, lipopolysaccharide.

Brain ammonia (μmol/g ww)0.27 ± 0.081.2 ± 0.171.0 ± 0.362.0 ± 0.660.61 ± 0.150.80 ± 0.26
Glutamine (μmol/g ww)4.4 ± 0.21*4.4 ± 0.123.5 ± 0.263.8 ± 0.165.4 ± 0.315.6 ± 0.24
Myoinositol (μmol/g ww)5.6 ± 0.315.6 ± 0.314.7 ± 0.285.5 ± 0.194.4 ± 0.234.8 ± 0.10
Creatine (μmol/g ww)7.4 ± 0.117.3 ± 0.236.6 ± 0.326.9 ± 0.346.7 ± 0.217.5 ± 0.26
NAA (μmol/g ww)4.1 ± 0.454.1 ± 0.522.8 ± 0.143.2 ± 0.136.1 ± 0.266.3 ± 0.26
Glutamine/Myinositol ratio0.80 ± 0.060.81 ± 0.060.73 ± 0.060.69 ± 0.091.2 ± 0.091.2 ± 0.03

Proton NMR Spectroscopy

No statistical difference was found in brain tissue osmolytes after administration of LPS to any of the test groups compared with the saline control (Table 3, Fig. 3). A significant decrease was seen in brain glutamine levels in BDL compared with sham (P < 0.03) and HD (P = 0.01) saline-treated rats. A significant decrease in brain myoinositol levels was found in HD (±LPS/saline) rats compared with sham-operated rats (HD + LPS, P = 0.005, and HD, P = 0.014, respectively). Also, the brain glutamine/myoinositol ratio was significantly higher in our HD (±LPS/saline) rats compared with sham/BDL ± LPS/saline rats (P < 0.001 for each case).

Figure 3.

1H-NMR spectra of brain extracts: Segments of 1H-NMR spectra of brain extracts obtained from a sham-operated control and from rats with BDL and HD. Peak assignment abbreviations: Asp, aspartate; Cho, choline-containing compounds; Cr, creatine; Gln, glutamine; myo-Ins, myo-inositol; NAA, N-acetyl-aspartate; Tau, taurine.

Plasma and Brain Cytokines

After administration of LPS in sham-operated and BDL rats, a significant increase was seen in the plasma levels of the proinflammatory cytokines TNF-α and IL-6 (P < 0.001 compared with each corresponding saline control group), which was also observed in HD rats (P < 0.05, Table 4). In saline-treated BDL rats, the levels of brain IL-6 and TNF-α were increased 1.6- and 1.5-fold, respectively, compared with the sham-operated group; this only reached significance for measured IL-6 concentrations, most likely because of intragroup variation (P = 0.05, unpaired t test). This background inflammatory response, in which brain TNF-α was augmented by fivefold after administration of LPS in BDL rats, was found to be not statistically significant.

Table 4. Plasma and Brain Cytokine Profile
  • NOTE. Values are given as mean ± standard error of mean (SEM) of duplicate samples from at least (n = 6) rats per group. Statistical

  • Newman-Keuls multiple comparison test.

  • P < 0.001 compared with sham with BDL + LPS rats.

  • P < 0.05 compared with HD + LPS Abbreviations: BDL, bile duct LPS, lipopolysaccharide.

Plasma TNF-α (nmol/L)0.03 ± 0.020.19 ± 0.00.05 ± 0.02*1.7 ± 0.130.02 ± 0.0010.84 ± 0.40
Plasma IL-6 (nmol/L)0.053 ± 0.0052.3 ± 0.270.094 ± 0.035*2.1 ± 0.220.077 ± 0.0420.89 ± 0.42
Brain TNF-α (nmol/L)0.008 ± 0.0010.065 ± 0.0120.011 ± 0.0020.055 ± 0.0460.057 ± 0.0190.060 ± 0.03
Brain IL-6 (nmol/L)1.3 ± 0.241.4 ± 0.132.1 ± 0.202.0 ± 0.501.8 ± 0.142.0 ± 0.23

Nitrite/Nitrate (NOx)

The plasma concentration of NOx (Fig. 4A, Table 1) was significantly higher in rats treated with LPS compared with the non–LPS-administered groups (sham-operated, P = 0.017; BDL, P = =0.008; and HD, P = 0.053, respectively). In saline-treated controls, BDL rats also had higher NOx levels compared with sham-operated control rats (P = 0.056). Similarly, the BDL/sham-operated and HD rats treated with LPS also had higher plasma nitrite values compared with the BDL/sham-operated and HD rats administered saline (sham-operated, P = 0.045; BDL, P = 0.038; and HD, P = 0.054: unpaired t test, respectively; Fig. 4B: Table 1).

Figure 4.

(A) Plasma nitrite/nitrate levels. A significant increase occurred in plasma nitrite/nitrate levels with administration of LPS in BDL, sham-operated, and HD rats compared with saline-treated BDL, sham-operated, and HD rats (#P < 0.008; $P < 0.017; and @P = 0.05, respectively, as determined using the Mann-Whitney test to compare groups). A significant difference was also found between BDL + saline rats compared with sham-operated + saline rats, *P = 0.05. (B) Plasma nitrite levels. A significant increase occurred in plasma nitrite levels in LPS-treated BDL rats versus saline-treated BDL rats (#P < 0.04), LPS-treated sham-operated rats compared with saline-treated sham-operated rats (*P < 0.045) and HD + saline compared with HD + LPS (@P = 0.019), as determined using the unpaired t test to compare groups. (C) Brain nitrotyrosine levels. A significant increase occurred in brain nitrotyrosine level in LPS-treated BDL rats compared with saline-treated BDL rats (*P < 0.005) using the Mann-Whitney test. Values are given as mean ± standard error of mean (SEM) in each case. A significance level of P < 0.05 found between the BDL + LPS and all other groups was tested using the Newman-Keuls multiple comparison test for nonparametric data.

Brain Nitrotyrosine/Tyrosine Ratio

The only significantly increased NT/T ratio was observed in the LPS-administered BDL rats compared with saline-treated BDL controls (P = 0.004), and these rats were also found to have a higher NT/T ratio when compared with all other groups (P < 0.05; Fig. 4C, Table 1). The administration of LPS had no significant effect on sham-operated or HD rats (Fig. 4C, Table 1).

Electron Microscopy

Figure 5A through F are transmission electron micrographs (Scale bar, 2 μm) showing representative cerebral microvessels and the perivascular astrocytes for each treatment group.

Figure 5.

Brain histology. Transitional electron micrographs (Scale bar, 1 μm), of frontal cortical brain sections from (A) sham-operated rat, showing normal microvessel architecture and lanthanum retained within the microvessel lumen; (B) BDL rat, showing a collapsed microvessel with moderate perimicrovessel edema and lanthanum retained within the microvessel lumen; (C) LPS-treated sham-operated rat, showing early perimicrovessel edema and lanthanum retained within the microvessel lumen; (D) LPS-treated BDL rat showing a significantly collapsed microvessel with severe perimicrovessel edema, yet lanthanum still retained within the microvessel lumen; (E) HD rat showing a partially collapsed microvessel, with intact endothelium shown by the retention of the lanthanum particles in the lumen; pronounced perivascular edema can be seen. (F) HD + LPS-treated rats, showing similar disruption of structure to that found in BDL + LPS and HD rats, but with greater perivascular edema and partial microvessel collapse. The blood-brain barrier appears intact with no leakage of the lanthanum nitrate from the vessel lumen. Abbreviations: MV, microvessel; E, perimicrovessel edema; Ln, the ionic tracer “lanthanum nitrate,” which is be visualized as small black densities opposing the luminal surface of the microvessel.

Sham-Operated Rats.

Figure 5A shows a fully intact and well-perfused microvessel with no surrounding astrocytic, perivascular changes. Lanthanum nitrate particles line the microvessel luminal wall, suggesting an intact blood-brain barrier to this tracer.

BDL Rats.

Figure 5B shows a partially collapsed microvessel, although the endothelium barrier appears intact because the lanthanum particles are clearly retained in the microvessel lumen. Only minimal brain water accumulation is observed in the astrocytic, perivascular tissue.

Sham-Operated + LPS Rats.

Figure 5C shows mild astrocytic, perivascular edema despite an otherwise intact blood-brain barrier as suggested by the retained lanthanum.

BDL + LPS Rats.

Figure 5D shows greater disruption of structure than found in the other sham and BDL experimental groups, evident by the massive astrocytic, perivascular edema and collapsed microvessel. However, the blood-brain barrier still appeared intact with no leakage of the lanthanum nitrate into either the endothelial cell cytoplasm or perivascular space.

HD Rats.

Figure 5E shows a partially collapsed microvessel, although the endothelium barrier appears intact because of the retention of the lanthanum particles in the microvessel lumen. There is massive brain water accumulation in the astrocytic, perivascular tissue.

HD + LPS Rats.

Figure 5F shows similar disruption of structure to that found in BDL + LPS and HD rats, but with more significant astrocytic, perivascular edema and collapsed microvessel. However, the blood-brain barrier still appeared intact with no leakage of the lanthanum nitrate into either the endothelial cell cytoplasm or perivascular space.


An important observation of our study was the finding that the administration of LPS to sham-operated, HD, and BDL rats resulted in marked brain water increases. The greatest increase in brain water was seen in the HD rats administered LPS, which occurred without any observed effect on the level of consciousness. Only the BDL animals administered LPS reached pre-coma stages at 3 hours after administration of LPS. This lack of correlation between increased brain water and preserved mental state, in all rats except the LPS-treated BDL rats, suggests that factors in addition to hyperammonemia and brain swelling contribute to the effects on consciousness found in HE.

The histological analysis (Fig. 5) confirms the data obtained from brain water measurements, which show significant worsening of brain edema in all groups after administration of LPS (Fig. 1). This increase in brain water was characterized by astrocytic edema that is indistinguishable from the well-documented brain edema seen in hyperammonemic liver failure models.30, 31 We used the ionic tracer lanthanum nitrate (MW 433), which is a highly reproducible method for EM detection of blood-brain barrier integrity to paracellular movement.32 Retention of lanthanum within the vessel wall in all the groups clearly indicates that the anatomical integrity of the blood-brain barrier is maintained. Our observation differs from earlier findings in the rat models of ALF, in which the blood-brain barrier was observed to be compromised.31, 32 The discrepancies between these earlier observations and our study may suggest a specific phenomenon of cytotoxic edema in cirrhotic rats compared with a combined cytotoxic and vasogenic edema in ALF.

Astrocytic edema in the non-cirrhotic, hyperammonemic (HD) rats was associated with an increase in glutamine, reduction in myoinositol, and a marked increase in the glutamine/myoinositol ratio, in keeping with the ammonia-glutamine-brain water hypothesis. Of note is the observation that the severity of astrocytic edema seen in the saline-treated BDL rats was markedly less despite degrees of plasma and brain ammonia levels similar to those of the LPS-administered BDL group. Our observation adds to the existing debate about the role of glutamine in the development of astrocytic edema during hyperammonemia.

The mechanism of the observed astrocytic swelling in BDL rats is therefore unclear and possibly regulated by the effects of inflammation. The increase in brain edema after LPS administration in each of the groups does not directly correlate with plasma and brain ammonia levels, or brain ammonia metabolism, because there was no difference in the ammonia, glutamine, or myoinositol levels between the rats administered LPS and the respective control groups. We have previously shown that administration of LPS resulted in a sustained increase in TNF-α and F2-Isoprostane levels in the BDL rats 3 hours post-injection compared with sham-operated rats administered LPS.15 In support of this argument, we observed that in the BDL, sham-operated, and HD rats, administration of LPS led to higher NO levels. To explore whether this altered peripheral inflammatory response was reflected in the brain, we measured plasma and brain proinflammatory cytokine levels. Our results show that LPS administration not only results in an increase in systemic cytokine response but also increases brain tissue TNF-α, confirming our previous observation that a systemic inflammatory response may initiate brain inflammation during liver failure despite retained anatomical barrier integrity.

The BDL rats administered LPS showed significantly higher levels of nitrosation of brain proteins compared with all other groups. Ammonia-induced nitrosation of astrocytic proteins has been demonstrated in isolated astrocytes and astroglial tissue in brain sections of portacaval anastomosed rats33; however, hyperammonemia is unlikely to be the sole mechanism for protein nitration because the NT/T ratio was not elevated in the hyperammonemic saline-treated BDL and HD rats. Similarly, LPS itself can lead to nitrosation of proteins in the brain34; however, the observation of only a minimal increase in the NT/T ratio in sham-operated and HD rats administered LPS also argues against this being the only mechanism for protein nitrosation. Hyperammonemia and LPS together may provide an environment for nitrosation of brain proteins. However, the observation that the NT/T ratio remained unchanged in the HD rats administered LPS argues against this hypothesis. The significant elevation of plasma NT/T and TNF-α levels in the LPS-treated BDL rats led us to hypothesize that the subliminal inflammation that is a feature of BDL35 may “prime” the animal to the effect of ongoing hyperammonemia and subsequent endotoxemic/inflammatory insult. This hypothesis is supported by the recent demonstration in astrocyte cell cultures that physiological levels of TNF-α seen in liver failure induce tyrosine nitration.36

As discussed, the differential progression to pre-coma observed in the LPS-treated BDL rats appeared not to directly correlate with any alteration in plasma and brain ammonia, electrolytes, brain osmolytes, glutamine, and myoinositol, or even the degree of astrocytic edema. It is possible that in the BDL rats hyperammonemia in some way acts synergistically with the systemic inflammatory response to alter the level of consciousness. This concept fits with our recent observation in patients with cirrhosis, in which induction of hyperammonemia resulted in worsening of neuropsychological tests when the patients were infected and showed evidence of an inflammatory response.14 After treatment of infection, similarly induced hyperammonemia did not result in an alteration of neuropsychological function, supporting the hypothesis that ammonia and inflammation may be synergistic.14 Our suggestion of synergy between hyperammonemia and inflammation is also supported by the recent observation that BDL rats that were fed a high-ammonia diet (and which showed evidence of low level of inflammation) developed increases in brain water.35 However, hyperammonemia and inflammation were present in the LPS-treated HD rats without any observed change in consciousness level. Given the extent of brain edema observed in the HD rats, with or without administration of LPS, that there was no impact on the level of consciousness is surprising. This would suggest that there is an as-yet-undetermined additional factor critical in the progression to coma in cirrhotic rats. On this background of cirrhosis, hyperammonemia, and superimposed inflammation may then facilitate nitrosation of brain proteins, which may be important to neurological function.

The LPS-administered BDL model used in the current study was developed to encapsulate the pathophysiological features of acute-on-chronic liver failure that is starting to be recognized as a distinct clinical and pathophysiological entity. Accordingly, chronic liver disease was induced by ligation of the bile duct which induces secondary biliary cirrhosis, and the superimposed injury was provided by the administration of LPS. This model contains the essential clinical features of cirrhosis—encephalopathy and mild renal dysfunction with exaggerated inflammatory response. Clinically, the animals are hemodynamically stable, and after LPS we observe the typical changes in blood pressure response during the 3-hour time frame of the study.29 Blood electrolytes and osmolarity are maintained, reducing confounding effects. Whether the increase in brain edema is associated with alterations in cerebral blood flow is not clear, but the electron microscopic appearances of collapsed blood vessels possibly indicate reduced cerebral blood flow, which is in keeping with the observation of increased cerebrovascular resistance in humans with advanced cirrhosis.37

In conclusion, this study provides further evidence to indicate that on a background of cirrhosis and hyperammonemia, superimposed inflammation has an important synergistic role in the development of HE. Further studies are needed to determine how ammonia and LPS facilitate cytotoxic edema and pre-coma in cirrhosis and to determine the role of nitrosation of brain proteins.


The authors thank R. F. Moss and C. Cope (The Division of Basic Medical Sciences, St George's University of London, UK) for their technical support with microscopy and photography, which contributed to this paper. This work was undertaken at UCLH/UCL who received a proportion of funding from the Department of Health's NIHR Biomedical Research Centre funding scheme. The views expressed in this publication are those of the authors and not necessarily those of the Department of Health.