Inflammation and hepatic encephalopathy: Ibuprofen restores learning ability in rats with portacaval shunts

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

One of the neurological alterations in patients with minimal or overt hepatic encephalopathy is cognitive impairment. This impairment is reproduced in rats with chronic liver failure due to portacaval shunt (PCS). These rats show decreased ability to learn a conditional discrimination task in a Y-maze, likely due to reduced function of the glutamate–nitric oxide (NO)–cyclic guanosine monophosphate (cGMP) pathway in brain. It has been proposed that inflammation exacerbates the neuropsychological alterations induced by hyperammonemia, suggesting that inflammation-associated alterations may contribute to cognitive impairment in hepatic encephalopathy. This study assessed whether treatment with an anti-inflammatory drug, ibuprofen, is able to restore the function of the glutamate-NO-cGMP pathway in cerebral cortex in brain in vivo and/or learning ability in PCS rats. We show that PCS rats have increased levels of interleukin-6 and increased activities of cyclooxygenase and of inducible NO synthase in cerebral cortex, indicating the presence of inflammation. Chronic treatment with ibuprofen normalizes cyclooxygenase and inducible NO synthase activities but not interleukin-6 levels. Moreover, ibuprofen normalizes the function of the glutamate-NO-cGMP pathway in cerebral cortex in vivo and completely restores the ability of rats with chronic liver failure to learn the Y-maze task. This supports that inflammation contributes to the cognitive impairment in hepatic encephalopathy. Conclusion: the results reported point to the possible therapeutic utility of decreasing inflammation in the treatment of the cognitive deficits in patients with minimal or overt hepatic encephalopathy. (HEPATOLOGY 2007.)

Hepatic encephalopathy (HE) is a complex neuropsychiatric syndrome present in patients with liver disease, which covers a wide range of neuropsychiatric disturbances including impairment of intellectual and cognitive function.

Learning ability is also impaired1 in rats with chronic liver failure due to portacaval shunt (PCS). In this model the function of the glutamate-NO-cGMP pathway is impaired in cerebellum2 and cerebral cortex3in vivo. Moreover, increasing cGMP in brain by administration of phosphodiesterase inhibitors restores learning ability of PCS rats,1 indicating that reduced cGMP formation is responsible for impaired learning. However, the factors and mechanisms by which chronic liver failure impairs the function of this pathway and learning ability remain unclear.

Shawcross et al.4 proposed that inflammation exacerbates the neuropsychological alterations induced by hyperammonemia. Chung et al.5 also showed that a nonsteroidal anti-inflammatory drug (NSAID), indomethacin, prevents the development of ammonia-induced brain edema in rats after portacaval anastomosis. These reports support the idea that hyperammonemia and inflammation cooperate in the cerebral alterations present in hepatic encephalopathy. There is also increasing evidence that inflammatory mechanisms are involved in the cognitive impairment in patients with Alzheimer's disease. Epidemiologic studies show that patients (e.g., with arthritis) treated with anti-inflammatory drugs such as ibuprofen have lower risk of suffering Alzheimer's disease.6 Moreover, in animal models of Alzheimer's disease, treatment with ibuprofen improves the cognitive deficit.7 This suggests that inflammation-associated alterations and hyperammonemia would contribute to cognitive impairment in hepatic encephalopathy.

We wanted to assess whether inflammation-associated alterations are involved in learning impairment in PCS rats. This study explored the hypothesis that ibuprofen normalizes the function of the glutamate-NO-cGMP pathway in cerebral cortex in vivo and learning ability in PCS rats.

Abbreviation

cGMP, cyclic guanosine monophosphate; COX, cyclooxygenase; HE, hepatic encephalopathy; iNOS, inducible nitric oxide synthase; NMDA, N-methyl-D-aspartate; NO, nitric oxide; NOS, nitric oxide synthase; nNOS, neuronal nitric oxide synthase; PCS, portacaval shunt.

Materials and Methods

Portacaval Anastomosis.

It was constructed according to Lee and Fisher8 using male Wistar rats (220-240 g). Control rats were sham operated. Experiments were performed 4 weeks after surgery. Animal procedures were approved by the center and met the guidelines of the European Union for care and management of experimental animals.

Ibuprofen Treatment.

Male Wistar rats were treated daily with S-(+) Ibuprofen (Fluka, Seelze, Germany) or saline. Treatment started 10 days after surgery and was maintained during all experiments. Ibuprofen in sterile saline was injected intraperitoneally at 30 mg/kg per day in 0.5 ml/100 g of body weight.

Ammonia Determination.

Plasma ammonia was measured as in Hermenegildo et al.9

Determination of the function of the glutamate-NO-cGMP pathway in cerebral cortex in vivo was performed by microdialysis. Rats were anesthetized using halothane, and a microdialysis guide was implanted in the cerebral cortex (coordinates: AP +3.7 mm, ML −0.8 mm, and DV −1.0 mm), according to Paxinos and Watson.10 After 48 hours, a microdialysis probe was implanted. Probes were perfused (3 μl/minute) with artificial cerebrospinal fluid consisting of: 145 mM NaCl; 3.0 mM KCl; 2.26 mM CaCl2; buffered at pH 7.4 with 2 mM phosphate buffer. After a 2-3 hour stabilization period, 30-minute samples were collected. Four samples were taken to determine the basal levels of extracellular cGMP, then N-methyl-D-aspartate (NMDA) (0.5 mM) was perfused through the microdialysis probe as indicated in Fig. 2. NMDA activates NMDA receptors and the glutamate-NO-cGMP pathway. Part of the cGMP formed is released to the extracellular space. The increase in extracellular cGMP induced by NMDA, measured in the subsequent fractions, is a good measure of the function of the glutamate-NO-cGMP pathway in vivo.9 Samples were made to concentrations of 4 mM in ethylene diamine tetraacetic acid (EDTA) and stored at −80°C until analysis of cGMP content was conducted.

Figure 2.

Treatment with ibuprofen restores the function of the glutamate-NO-cGMP pathway in cerebral cortex of PCS rats. The function of the pathway was assessed by in vivo brain microdialysis. NMDA (0.5 mM) was administered for 20 minutes (indicated by arrow) and extracellular cGMP was measured. Values are the mean ± SEM of 7-8 rats per group and are expressed as percentage of basal values. Basal values were 101 ± 11, 149 ± 11*, 146 ± 12*, and 149 ± 12* pM for sham, sham + ibuprofen, PCS, and PCS + ibuprofen rats. Values significantly different from controls (sham) are indicated by asterisks. Values significantly different from PCS rats not treated with ibuprofen are indicated by a. * P < 0.05; ** P < 0.01; aP < 0.01. PFCx, prefrontal cortex.

cGMP Determination.

cGMP was measured by using the Biotrak cGMP enzyme immunoassay kit from Amersham (Amersham Pharmacia Biotech, Buckinghamshire, UK).

Interleukin-6 and Tumor Necrosis Factor Alpha Measurement.

Interleukin-6 (IL-6) was measured in microdialysis samples using the Biotrak enzyme-linked immunosorbent assay system (Amersham Biosciences, UK). Tumor necrosis factor alpha (TNF-α) was determined using the OptEIA test (BD Biosciences, San Diego, CA). The detection limits were 16 pg/ml for IL-6 and 13 pg/ml for TNF-α. The interassay coefficient of variation were 7%-10% for IL-6 and 5%-6% for TNF-α.

iNOS activity.

Inducible NO synthase (iNOS) activity was determined as the activity of NOS that is calcium-independent. Calcium-independent NOS activity was measured by the conversion of [14C]arginine to [14C]citrulline as described in Rodrigo et al.3 Briefly, rats were killed by decapitation and cerebral cortex was homogenized in 5 volumes of ice-cold buffer A (20 mM 4-(2-hydroxyethyl)-1-piperazine ethanesulfonic acid, 0.32 M sucrose, 1 mM dithiothreitol, 10 mg/l leupeptin, 10 mg/l pepstatin A, and 1 mM ethylene glycol tetraacetic acid to chelate endogenous calcium; pH 7.4). Homogenates (25 μl) were incubated for 60 minutes at 37°C with 100 μl of reaction mixture containing 200 μM reduced nicotinamide adenine dinucleotide phosphate, 100 μM tetrahydro-biopterin, 2 mM EDTA, 1.2 mM ethylene glycol tetraacetic acid and 1 μCi/ml L-[14C]arginine monohydrochloride in buffer A and 75 μl distilled water. The reaction was stopped by adding 1 ml ice-cold buffer containing 30 mM 4-(2-hydroxyethyl)-1-piperazine ethanesulfonic acid and 3 mM EDTA (pH 5.5). [14C]Citruline was separated from [14C]arginine by chromatography through a Dowex AG50WX-8 (Na+ form) column. Protein concentration was quantified by the bicinchonic acid method.

Cyclooxygenase Activity.

Cyclooxygenase (COX) activity was determined by monitoring the appearance of oxidized N,N,N′,N′-tetramethyl-p-phenylenediamine at 590 nm wavelength.11 Cortical tissue was homogenized in 5 ml cold 100 mM Tris-HCl, pH 7.8, containing 1 mM EDTA and centrifuged at 10,000g for 15 minutes at 4°C. COX activity was measured in the supernatant using a kit from Cayman Chemical Co. (Ann Arbor, MI).

Y-Maze Learning Test.

Learning ability was tested as described in Aguilar12 in a Y-maze. Rats were trained for 10 trials per day, until the completion of a criterion of 10 correct responses in 10 consecutive trials or until a maximum of 250 trials.

Determination of Urea, Creatinine, and Sodium.

Blood samples were obtained at death from 6 animals in each group. Urea, creatinine, and sodium were determined with a Kodak Ektachem 700 autoanalyzer (Orthoclinical Diagnostics, Rochester, NY).

Statistical Analysis.

Data were analyzed by analysis of variance (ANOVA) followed, when appropriate, by Bonferroni post-hoc test. When only 2 values were compared, the Student t test was used. Significance levels were set at α = 0.05.

Results

Chronic liver failure impairs the ability of rats to learn the Y-maze task. Control rats need 51 ± 4 trials to learn the task whereas PCS rats need more trials (80 ± 5, p < 0.001). Treatment with ibuprofen did not affect learning ability in control rats. However, ibuprofen completely restored learning ability in PCS rats that needed 54 ± 4 trials to learn (Fig. 1).

Figure 1.

Treatment with ibuprofen restores learning ability in PCS rats. Control (SHAM) and PCS rats treated with vehicle (veh) or ibuprofen (IBU) were subjected to the conditional discrimination learning test in the Y-maze. The tests were initiated after 12 days of treatment. Values are the mean ± standard error of the mean (SEM) of 13-15 rats per group and are given as the number of trials needed to learn. Values significantly different from control rats are indicated by asterisks. Values significantly different from PCS rats not treated with ibuprofen are indicated by a. *** P < 0.001; aP < 0.001

We proposed that the ability of rats to learn the Y-maze depends on the function of the glutamate-NO-cGMP pathway,1 which is impaired in cerebral cortex of PCS rats.3 We therefore tested whether ibuprofen-induced restoration of learning is associated with restoration of the function of the pathway in cerebral cortex in vivo (Fig. 2). In control rats, administration of NMDA through the microdialysis probe induces an increase of 23 ± 4 fmol cGMP in 60 minutes (fractions 5-6). In PCS rats, NMDA did not increase extracellular cGMP (-0.1 ± 0.3 fmol). Treatment with ibuprofen did not affect the pathway in control rats, NMDA induced an increase of 16 ± 4 fmol of cGMP. However, ibuprofen completely restores the function of the pathway in PCS rats, NMDA induced an increase of 19 ± 4 fmol of cGMP (Fig. 2). These data show that ibuprofen restores both the function of the pathway and learning ability in PCS rats.

We then analyzed the effects of chronic liver failure and of ibuprofen on different inflammatory factors in cerebral cortex. IL-6 is significantly (P < 0.05) increased in PCS rats. Ibuprofen does not normalize IL-6 (Table 1). We did not find any change in TNF-α in cortex of PCS rats (Table 1).

Table 1. Effects of Chronic Liver Failure and of Treatment with Ibuprofen on Inflammatory Parameters in Cerebral Cortex
ParameterControl RatsControl + IbuprofenPCS RatsPCS + Ibuprofen
  • NOTE. Values are the mean ± SEM of the number of rats indicated in parenthesis. Values significantly different from controls are indicated by asterisks. Values significantly different from PCS rats non treated with ibuprofen are indicated by a.

  • *

    P < 0.05.

  • **

    P < 0.01.

  • ***

    P < 0.001.

  • a

    P < 0.001.

Blood ammonia (μM)74 ± 8 (19)102 ± 9 (17)203 ± 14*** (21)185 ± 17** (22)
IL-6 (pg/ml)57 ± 4(6)65 ± 3 (6)77 ± 7* (6)86 ± 7** (6)
TNF-α (pg/ml)26 ± 4 (5)23 ± 4 (5)17 ± 3 (5)33 ± 2a (5)
COX activity (U/mg protein)2.2 ± 0.4 (7)1.9 ± 0.4 (7)20 ± 3*** (7)3.0 ± 0.9a (7)
iNOS activity (cpm/mg prot.min)14 ± 1 (16)16 ± 1 (14)18 ± 1** (17)14 ± 1a (14)

The activity of iNOS is significantly (P < 0.01) increased (by 29%) in PCS rats. Ibuprofen normalized iNOS activity, which returned to normal levels (Table 1). COX activity is strongly increased (about 9-fold, P < 0.001) in cerebral cortex of PCS rats. Treatment with ibuprofen completely reversed this effect, returning COX activity to control levels (Fig. 3).

Figure 3.

Treatment with ibuprofen normalizes COX activity in cerebral cortex of PCS rats. COX activity was determined in cerebral cortex of control (SHAM) and PCS rats treated with vehicle (veh) or ibuprofen (IBU). Values are the mean ± SEM of 7 rats per group. Values significantly different from controls are indicated by asterisks. Values significantly different from PCS rats not treated with ibuprofen are indicated by “a”. *** P < 0.001; aP < 0.001.

Ibuprofen is an NSAID, which promote salt and water retention and can be potentially nephrotoxic. As an index of kidney function, we measured creatinine, urea and sodium in serum. Ibuprofen did not significantly affect sodium or urea levels and slightly increased (35%) creatinine levels both in control and PCS rats (Table 2). The increase was statistically significant only in PCS rats.

Table 2. Effects of Portacaval Shunt and of Treatment with Ibuprofen on Serum Creatinine, Urea and Sodium
ParameterControl RatsControl + IbuprofenPCS RatsPCS + Ibuprofen
  • NOTE. Values are the mean ± SEM of six rats per group. Values significantly different from controls are indicated by asterisks. Values significantly different from PCS rats non treated with ibuprofen are indicated by “a”.

  • *

    P < 0.05.

  • a

    P < 0.01.

Creatinine (mg/dl)0.49 ± 0.02 (6)0.66 ± 0.09 (6)0.42 ± 0.03 (6)0.57 ± 0.05a (6)
Urea (mg/dl)41 ± 5 (6)31 ± 5 (6)25 ± 3* (6)36 ± 4 (6)
Sodium (mEq/L)141 ± 1 (6)131 ± 4 (6)140 ± 5 (6)137 ± 2 (6)

Discussion

The results reported show, for the first time, that treatment with an anti-inflammatory improves cognitive function in an animal model of hepatic encephalopathy. PCS rats recover learning ability when treated with ibuprofen.

Shawcross et al.4 propose that systemic inflammation exacerbates the neuropsychological alterations induced by hyperammonemia. They induced hyperammonemia in cirrhotic patients and showed that hyperammonemia deteriorates neuropsychological test scores during inflammatory state but not after its resolution. This supports the idea that hyperammonemia and some factor related with inflammation cooperate to impair cognitive function.

Chung et al.5 also showed that the NSAID indomethacin prevents the development of ammonia-induced brain edema in rats after portacaval anastomosis. These reports support the idea that hyperammonemia and inflammation cooperate in the cerebral alterations present in hepatic encephalopathy.

Hyperammonemia increases the sensitivity to immune challenges. Marini and Broussard13 injected normal or hyperammonemic mice with lipopolysaccharide and found that cytokine production increased similarly in both groups. However, the cognitive deficits induced by lipopolysaccharide were stronger and long-lasting in hyperammonemic mice. This further supports that hyperammonemia and inflammation cooperate in inducing cognitive deficits.

The mechanisms underlying this cooperation in induction of encephalopathy and the possible therapeutic implications have not been tested. In this study, we addressed both aspects.

PCS rats present both hyperammonemia and inflammation as shown by the increase in IL-6, and in iNOS and COX activity in cerebral cortex. This hyperammonemia-inflammation state is associated with decreased learning ability. If cognitive impairment is consequence of the cooperation between hyperammonemia and inflammation it should be possible to improve cognitive function by reducing inflammatory factors. We show that treatment with the anti-inflammatory ibuprofen restores the ability of PCS rats to learn the Y-maze task. This is in agreement with the proposal4 that inflammation exacerbates the neuropsychological alterations induced by hyperammonemia. PCS rats continue being hyperammonemic after treatment with ibuprofen, but reduction of inflammatory factors improves learning ability.

The data shown here point to the possible therapeutic utility of decreasing inflammation in the treatment of minimal or overt hepatic encephalopathy. The possible mechanisms by which inflammation exacerbates the neurological alterations in hyperammonemia are just beginning to be studied. Also, the mechanisms by which anti-inflammatory agents could improve cognitive function are not clear. Our work provide some initial insight on these topics.

We found that ibuprofen does not normalize IL-6 levels in cerebral cortex, suggesting that IL-6 does not contribute to cognitive impairment. Ibuprofen normalizes both iNOS and COX activities in cerebral cortex of PCS rats. Ibuprofen would act by reducing COX activity. This supports the idea that increased COX activity would be involved in cognitive impairment in hepatic encephalopathy.

The fact that ibuprofen normalizes iNOS activity in cerebral cortex of PCS rats suggests that the increase in iNOS could be subsequent to the increase in COX activity. Increased iNOS could mediate the effects of COX on the function of the glutamate-NO-cGMP pathway in PCS rats.

Concerning the mechanisms by which these inflammatory factors would lead finally to learning impairment, the above data indicate that a main step would be impairment of the glutamate-NO-cGMP pathway function, which is completely restored by treatment with ibuprofen. We proposed that impaired function of this pathway and reduced formation of cGMP in response to activation of NMDA receptors is responsible for the decrease in the ability to learn the Y-maze task in rats with chronic liver failure or hyperammonemia. Moreover, oral or intracerebral administration of phosphodiesterase inhibitors increases cGMP in brain and restores learning ability.1, 14 These data clearly supports that decreased formation of cGMP by the glutamate-NO-cGMP pathway is responsible for the decreased ability to learn this task.

The mechanisms by which ibuprofen restores the function of this pathway in portacaval shunted rats remain unclear. In cerebral cortex of PCS rats, the impairment of the pathway is mainly due to impaired activation of neuronal NOS (nNOS).3 It is not yet clear which is the mechanisms by which portacaval shunt results in reduced activation of nNOS following activation of NMDA receptors, and it is therefore difficult to discuss how ibuprofen may normalize this activation. The mechanisms by which chronic liver failure affects modulation of nNOS activity by calcium may include alterations in phosphorylation or in the content or function of the proteins that modulate nNOS localization. The results reported here suggest that in PCS rats the increased formation of prostaglandins as a consequence of increased COX activity would be involved in the mechanism leading finally to impairment of nNOS activation. Ibuprofen normalizes COX activity and the function of the pathway. The steps or pathways by which modulation of nNOS activation would be affected by prostaglandins remain unclear. There are conflicting reports in the literature concerning the possible modulation of nNOS activity or content by prostaglandin E2 in different systems. Some reports indicate that prostaglandin E2 increases NO production by nNOS and/or nNOS expression,15, 16 whereas other reports indicate that prostaglandin E2 reduces nNOS expression,17 and still others say that control of nNOS and COX expression are not dependent.18 It has also been reported that COX inhibitors increase19 or decrease20 NO formation. Further studies are therefore necessary to clarify the mechanisms by which increased COX activity leads to impaired nNOS activation.

It is possible that increased iNOS activity could mediate the effects of increased COX on nNOS modulation. NO overproduction by iNOS may lead to some adaptive response in nNOS phosphorylation or location, resulting in reduced activation following NMDA receptors activation. Further studies are required to clarify the mechanism(s) by which portacaval shunt leads to reduced activation of nNOS. In any case, the results reported here clearly show that ibuprofen restores this activation.

Ibuprofen is an NSAID. Apart from being anti-inflammatory, it has other actions, particularly on the kidney. NSAIDs promote salt and water retention and can be potentially nephrotoxic. NSAIDs thus may have secondary effects and are relatively contraindicated in patients with liver cirrhosis. Treatment with ibuprofen slightly increased creatinine (but not urea or sodium) in PCS rats (Table 2). The data presented here, using ibuprofen as an experimental tool to decrease inflammation, show that a decrease in inflammation completely restores the ability to learn the Y-maze task in a rat model of hepatic encephalopathy. This supports the idea that reducing inflammation would improve cognitive function in patients with hepatic encephalopathy. It would be convenient to look for procedures to decrease inflammation without having the possible secondary effects of NSAIDs, and perhaps new specific inhibitors of COX with no or less secondary effects could have beneficial effects.

In summary, this work shows that treatment with ibuprofen completely restores the ability of rats with portacaval shunt to learn the Y-maze task. This is due to the normalization by ibuprofen of the function of the glutamate-NO-cGMP pathway. This supports the idea that inflammation and hyperammonemia cooperate in the induction of neurological alterations in minimal or overt hepatic encephalopathy. Moreover, these data point to the possible therapeutic utility of decreasing inflammation, by safe procedures, in the treatment of the cognitive deficits in patients with minimal or overt hepatic encephalopathy.

Ancillary