Oxidative stress markers in the brain of patients with cirrhosis and hepatic encephalopathy^†

Hepatic encephalopathy (HE) defines a neuropsychiatric manifestation of chronic or acute liver disease with disturbances of psychomotoric, intellectual, cognitive, emotional/affective, behavioral, and fine motor functions of varying severity (see Häussinger and Blei1 for a review). HE in cirrhosis has been reported to reflect the clinical manifestation of low-grade cerebral edema,2, 3 as shown by in vivo studies using magnetic resonance spectroscopy2-4 and quantitative water imaging5 of the brain in vivo from patients with cirrhosis.

Ammonia, which is considered a key toxin in the pathogenesis of HE, can induce astrocyte swelling in vitro.6-8 Because astrocyte swelling is also induced by inflammatory cytokines, hyponatremia, or benzodiazepines, the action of long-known and heterogeneous precipitating factors of HE may integrate at the level of astrocyte swelling.2, 3 In cultured rat astrocytes, ammonia, tumor necrosis factor-α, diazepam, and hypo-osmotic cell swelling produce an oxidative/nitrosative stress response9-12 through activation of nicotinamide adenine dinucleotide phosphate-oxidase.8 A self-amplifying signaling loop between astrocyte swelling and oxidative stress has been proposed, because astrocyte swelling produces reactive oxygen species, and reactive oxygen species trigger astrocyte swelling.11, 13 The ammonia-induced, tumor necrosis factor α–induced, and swelling-induced oxidative/nitrosative stress response has multiple effects on astrocyte function, such as modulation of gene transcription through Zn⁺⁺ mobilization with the consequence of up-regulation of the peripheral benzodiazepine receptor.14 Other consequences of the reactive oxygen species/RNS formation are protein tyrosine nitration (PTN)10 and RNA oxidation through formation of 8-hydroxyguanosine.15 These covalent modifications of astrocytic proteins and astrocytic and neuronal RNA occur not only in vitro in response to ammonia,10, 15 benzodiazepines,11, 15 inflammatory cytokines,15 and experimental, hypo-osmotic astrocyte swelling,12, 15 but also in rat brain in vivo following ammonia, diazepam, or lipopolysaccharide intoxication or portocaval shunting.10, 11, 15 Astrocytes located near the blood–brain barrier exhibit especially high levels of PTN, with potential consequences for blood–brain barrier permeability.10 RNA oxidation in neurons involves RNA species contained in NOVA-2–positive RNA granules along the dendrites and in postsynaptic regions.15 These findings suggest that HE-associated oxidative stress modifies RNA species, which are involved in local synaptic protein synthesis16 and therefore may impair synaptic plasticity and memory formation.15 Because RNA oxidation in other systems can impair translational efficacy and accuracy,17, 18 such effects may underly the multiple and inhomogeneous derangements of neurotransmitter/receptor systems in HE1 and the disturbances of oscillatory networks in the brain of HE patients, as shown in magnet encephalography studies on the human brain in vivo.19

Although cell culture studies in vitro and rodent models in vivo suggest an important role of oxidative/nitrosative stress in the pathogenesis of ammonia toxicity, it is unknown whether oxidative stress is also involved in the pathogenesis of HE in humans with cirrhosis. Furthermore, no data are available on the occurrence of PTN and RNA oxidation in the brains of humans with cirrhosis and HE. Whereas noninvasive nuclear magnetic resonance spectroscopy (NMR) and tracer techniques allow to pick up brain water, several cerebral metabolites and receptor expression in humans in vivo, no in vivo techniques on the detection of oxidative modifications of brain constituents in humans are available. We therefore analyzed human post mortem cortical brain samples from patients with cirrhosis dying with or without HE in comparison to brain samples from patients without evidence for preexisting liver disease. The analysis shows that PTN and RNA oxidation are significantly increased in HE, which shows for the first time that oxidative/nitrosative stress is a hallmark of HE in patients with cirrhosis.

Abbreviations:

GAPDH, glyceraldehyde 3-phosphate dehydrogenase; GLAST, glutamate/aspartate cotransporter; Glt-1, glial glutamate transporter-1; GS, glutamine synthetase; HE, hepatic encephalopathy; Hsp27, heat shock protein-27; iNOS, inducible nitric oxide synthase; nNOS, neuronal nitric oxide synthase; PTN, protein tyrosine nitration.

Patients and Methods

Detailed information about the materials used in this study can be found in the Supporting Information.

Results

Markers for Oxidative Stress Are Increased in the Cerebral Cortex of HE Patients.

As shown recently, ammonia exposure induces oxidative/nitrosative stress in rat astrocyte cultures and mouse brain slices in vitro.8-10 In addition, acute ammonia intoxication of rats in vivo induces oxidative stress in the brain.22 Consequences of this oxidative/nitrosative stress response are PTN10 and oxidation of RNA, as evidenced by increased 8-hydroxyguanosine formation.15 Whether such alterations also occur in the brain of patients with hepatic encephalopathy is unknown. Therefore, oxidative stress markers were analyzed in post mortem human brain tissue. As shown by western blot analysis in Fig. 1, PTN of individual proteins within the molecular mass range between 20 and 75 kDa was significantly increased in brain samples from HE patients as compared with controls. Furthermore, total RNA extracted from the cerebral cortex of HE patients showed about two-fold higher levels of oxidized RNA as revealed by northwestern blot analysis when compared with control patients (Fig. 2). These data suggest the presence of oxidative stress in the brain from patients with cirrhosis and HE.

In line with this and as shown by western blot analysis, protein expression of the small heat shock protein-27 (Hsp27), a well-established and frequently used marker for oxidative stress23 was significantly and about four-fold up-regulated in the cerebral cortex of HE patients compared with controls (Fig. 3A,B). On the other hand, protein expression levels of neuronal nitric oxide synthase (nNOS) and of antioxidative enzymes, such as copper/zinc-dependent and manganese-dependent superoxide dismutase were not significantly increased in brains from HE patients when compared with the control group (Fig. 3A,B). Up-regulation of the inducible nitric oxide synthase (iNOS) has been reported in ammonia-treated astrocytes in vitro10 and in portocaval anastomized rats in vivo.24 However, in all cortical brain samples from controls and HE patients iNOS protein was not detectable (for representative blots, see Fig. 3C).

Astrocytic GS was identified as one target of PTN in response to ammonia,10 and further studies have shown that nitration of tyrosine residues in the adenosine triphosphate–binding pocket, similar to mutations in this region of the enzyme are associated with a loss of enzymatic activity of GS.25, 26 By this means, ammonia intoxication lowers GS activity in the brain without affecting the amount of GS protein.10 As shown in Fig. 4A, also in the cerebral cortex from HE patients GS activity, but not the GS protein level was significantly decreased. This was accompanied by a significantly increased PTN of GS (Fig. 4B). A decrease of GS activity had also been described in rats after portocaval anastomosis.27

HE but Not Cirrhosis Triggers Oxidative Stress in the Cerebral Cortex.

In order to determine whether increased cerebral oxidative stress is a feature of HE or of cirrhosis, brain samples from four patients with cirrhosis without HE were analyzed in comparison with four controls without cirrhosis. These eight brain samples were obtained from the Australian Brain Donor Programs NSW Tissue Resource Centre (see Table 1 for patient characteristics). As shown in Fig. 5A-E and Supporting Fig. 1, no significant differences with respect to tyrosine nitration of cerebral proteins, levels of RNA oxidation, GLAST and Hsp27 expression or GS activity were detectable in brain samples from patients with cirrhosis without HE compared with controls. These findings indicate that oxidative/nitrosative stress is a hallmark of HE, but not of cirrhosis itself.

Correlational Analysis.

The expression of oxidative stress markers, glutamate transporters, and GS in post mortem brain tissue was correlated to in vivo laboratory data, which were obtained during the 24-hour period before death. Such data were available in six to seven patients from the European cohort (Table 2). Results of this analysis are given in Supporting Table 1. Highly significant correlations were found between serum bilirubin concentration and GS activity (Fig. 6A), the serum sodium concentration and the extent of RNA oxidation (Fig. 6B), and serum urea concentration and GLAST expression (Fig. 6C). Neither plasma ammonia nor C-reactive protein in serum or the blood leukocyte count correlated with oxidative stress parameters in post mortem brain cortex (Supporting Table 1).

Table 2. Blood Test Results of European Patients with Cirrhosis and HE

Patient No.	Sodium (135-145 mmol/L)	Urea (18-55 mg/dL)	Creatinine (<1.2 mg/dL)	Albumin (3.3-4.7 g/dL)	Bilirubin (<1.0 mg/dL)	Leukocytes (4.0-11.0 × 1,000/μL)	C-Reactive Protein (<0.5 mg/dL)	Quick (70-130%)	Ammonia (<60 μmol/L)
1	131	146	2.6	3.1	19.0	—	—	—	—
2	122	258	2.9	2.4	0.8	30	14.9	37	97
3	138	148	1.5	1.8	20.0	22.9	4.6	10	154
4	—	—	—	—	—	—	—	—	—
5	120	226	7.9	1.8	29.0	2.5	1.0	13	148
6	158	310	5	2.0	36.8	27.3	1.6	29	86
7	126	94	2.4	2.1	11.6	15.8	5.8	26	48
8	141	159	4.1	—	1.98	2.9	1.1	59	343

• Laboratory parameters were measured on the day of (patient 8) or the day before death (patients 1, 2, 3, 5, 6, and 7). Reference values for each parameter are included in brackets.

Discussion

Cell culture studies and animal experiments suggest that the induction of oxidative/nitrosative stress is closely related to the pathogenesis of cerebral ammonia toxicity8-10, 14, 22 due to chemical modifications of proteins10 and RNA15 as well as effects on cerebral microcirculation,34 zinc homeostasis,14 and gene transcription.35, 36 However, it is unknown whether oxidative/nitrosative stress and its sequelae are also found in the brain of patients with cirrhosis with HE. This important question was addressed in the present study using post mortem brain tissue obtained at autopsy from eight European patients suffering from liver cirrhosis and accompanying HE and from eight European control patients without a record of hepatic, neurological, or psychiatric disease. In addition, brains from four Australian control patients and from four Australian patients with cirrhosis without HE could be analyzed. Both the European and Australian cohorts were analyzed separately in order to rule out population and sample handling differences. Several lines of evidence indicate for the first time an increased oxidative/nitrosative stress in the brains of HE patients. First, the expression of Hsp27 was found to be increased four- to five-fold in the brains of HE patients. Hsp27 is known to confer resistance toward oxidative stress37 and is frequently used as marker for tissue exposure to oxidative stress.23 In line with this, up-regulation of Hsp27 also occurs in ischemia/reperfusion injury and perilesional ischemic brain regions38 and multiple sclerosis.39 Furthermore, brain samples from HE patients showed significantly increased levels of tyrosine-nitrated proteins and of oxidized RNA, both parameters being well-established surrogate markers for oxidative stress.40, 41 All these changes were not found in brain samples from patients with cirrhosis without HE, indicating that oxidative/nitrosative stress is a feature of HE, but not of cirrhosis. The decrease of GS activity, but not GS protein in brains from HE patients reflects the known inactivating tyrosine nitration of the enzyme.21 GS inactivation through nitration was shown recently to occur in ammonia-intoxicated rat brain10 and endotoxin-treated rat liver,25 in the brains of epilepsia42 and HE patients (present study), and after in vitro treatment of GS with peroxynitrite.21 Here, nitration involves tyrosine residues close to the adenosine triphosphate–binding pocket of the enzyme, triggering activity loss of the enzyme,25 as it is also found in genetic defects affecting this region.26 In the setting of HE, down-regulation of GS activity may reflect a beneficial adaptation toward hyperammonemia, because astrocytic glutamine accumulation was identified as one trigger of low-grade cerebral edema in patients with cirrhosis and HE.2 It should be emphasized that the increase in protein tyrosine nitration and RNA oxidation in the brain of HE patients was highly discriminative between HE patients and controls: most of the HE brains showed RNA oxidation and PTN levels, which were above the highest values found in the control group. This contrasts the results obtained in a recent study using 14 out of the 16 brains of the present study for multireceptor autoradiography.20 Here, the HE cases did not build a homogeneous group, but differed concerning direction and intensity of binding site density divergence.20 Thus, neurotransmitter alterations may be less likely primary key factors underlying HE pathogenesis than RNA oxidation and protein tyrosine nitration.

Expression levels of superoxide dismutase isoenzymes and of neuronal and iNOS were not significantly altered in the brain from HE patients when compared with controls. This does not contradict the strong increase in tyrosine-nitrated proteins in HE, because nNOS is also regulated by Ca²⁺ and peroxynitrite-dependent and independent pathways of tyrosine nitration are known.11, 43 Chronic hyperammonemia was reported to down-regulate nNOS in rat cerebellum,44 whereas no effect on nNOS and iNOS was reported in the cerebellum of rats with thioacetamide induced45 and up-regulation of iNOS and nNOS was reported in rats with portocaval anastomosis.24, 46 Similarly, opposing results regarding superoxide dismutase expression in rat brain have been reported. Whereas thioacetamide-induced fulminant hepatic failure is accompanied by an up-regulation of manganese-dependent superoxide dismutase,47 down-regulation of Cu/Zn-dependent superoxide dismutase was reported after portocaval anastomosis.36

Experiments with cultured rat astrocytes suggest that overstimulation of N-methyl-D-aspartic acid receptors plays an important role in triggering the oxidative/nitrosative stress response.10 Astrocyte swelling and ammonia trigger astrocytic glutamate release,30, 31 thereby amplifying N-methyl-D-aspartic acid receptor activation.10, 12 Glutamate reuptake is accomplished by the astrocytic glutamate transporters GLAST and Glt-1. A significant up-regulation of GLAST was found in the brains of HE patients, but not of patients with cirrhosis without HE. GLAST up-regulation was reported in long-term portocaval shunted rats,33 whereas ammonia intoxication for only a few days triggered down-regulation of GLAST protein and messenger RNA in rat astrocytes.48, 49 The latter may be a consequence of GLAST messenger RNA oxidation, which was recently described in acutely ammonia intoxicated rats.15

Although the correlations between in vivo laboratory data and post mortem brain characteristics need to be interpreted with caution due to the small sample size, a few observations merit comment. There was a significant correlation between serum sodium concentration and the extent of cerebral RNA oxidation, indicating that hypo-osmolarity and RNA oxidation are somehow interlinked. In line with this, hypo-osmotic exposure of rat astrocyte cultures and brain slices induces RNA oxidation,15 and there is good evidence that HE in patients with cirrhosis reflects the clinical manifestation of low-grade cerebral edema, which can exacerbate in response to hyponatremia.2 However, it is also possible that hyponatremia (hypo-osmolarity) is not the cause of HE, but rather its consequence through swelling of hypothalamic structures promoting inadaequate vasopressin secretion.50 In any case, the correlation depicted in Fig. 6A corroborates clinical data on the relationship between hyponatremia and HE severity.2, 51 Interestingly, neither plasma ammonia nor markers of infection (leukocyte count, C-reactive protein) showed any relationship to oxidative/nitrosative brain modification. This finding may be explained by the multifactorial pathogenesis of low-grade cerebral edema and consequently generation of oxidative/nitrosative stress in HE11 and corroborates the view that HE in patients with cirrhosis is not simply ammonia intoxication or a pure response to systemic inflammation. In line with this, no increased expression of iNOS was detectable in the brains of patients with cirrhosis and HE. No significant relationship was found between RNA oxidation and PTN in the brain, although both processes are consequences of oxidative stress. Differences between both processes may relate to different nitric oxide requirements and kinetics of repair and elimination mechanism of nitrated proteins and oxidized RNA, respectively.

In conclusion, the present study shows for the first time the involvement of oxidative/nitrosative stress in the pathobiology of HE in patients with cirrhosis. Oxidative modifications of cerebral proteins and RNA are found in the cerebral cortex of HE patients, but not of patients with cirrhosis without HE. This may shed a new light on the pathogenesis of HE and offer potential new treatment options.