In vitro and in vivo studies suggested that reduced astrocytic uptake of neuronally released glutamate, alterations in expression of glial fibrillary acid protein, a major component of the glial filament network, and aquaporin-4 (AQP-4), a water channel protein contributes to the astrocyte swelling leading to brain edema and intracranial pressure in acute liver failure (ALF). However, there is no evidence available to-date to suggest that these alterations occur in patients with ALF. In order to address this issue, dissected samples of cerebral cortex were obtained at autopsy from 8 patients with ALF due to either viral hepatitis or toxic liver injury and from 7 control patients with no evidence of hepatic or liver or other neurological disorders. Expression of GFAP, EAAT-1, EAAT-2 and AQP-4 mRNAs was investigated by real-time PCR and protein expression was assessed using as well as immunoblotting and immunohistochemistry analysis. Expression of GFAP at both the mRNA and protein levels was significantly decreased in ALF patients (P<0.001 and P<0.008) as compared to control patient's material. On the other hand, loss of EAAT-2 protein in ALF (P<0.039) samples was found to be post-translational in nature. EAAT-1 protein remained within normal limits. Immunohistochemistry analysis confirmed that the losses of EAAT-2 and GFAP were uniquely astrocytic in localization in all cases. AQP-4 mRNA expression was significantly increased but its protein level in total cell lysate remained unchanged compared to controls, however immunohistochemistry demonstrated continuous dense band of AQP-4 immunoreactivity in the glial end-feet process surrounding the microvessels. Conclusion:These findings provide the first direct evidence for selective alterations in expression of genes coding for key astrocytic proteins implicated in CNS excitability and for brain edema in human ALF.
Keywords: acute liver failure; aquaporin-4, astrocyte protein; brain edema; glial fibrillary acid protein; glutamate transporter;
Introduction:
Hepatic encephalopathy (HE) syndrome is a neuropsychiatric complication of both acute (ALF) and chronic liver failure (CLF). Brain edema and Hepatic encephalopathy ( HE) are the major neuropsychiatric complications of ALF (Bémeur 2012; Cauli et al, 2011).1,2 ALF has an extremely poor prognosis with a mortality rate of about 65% (Dhiman et al, 2007),3 and an emergency liver transplantation remains the only effective treatment option (Polson 2005).4 A The major cause of death in this condition is brain herniation caused by increased intracranial pressure (ICP) due to brain edema.4 Studies suggest that ammonia toxicity is mainly involved in the pathogenesis of HE and brain edema with associated dysfunction of astrocytes (Butterworth, 2002; Norenberg et al, 2005).5,6
Glutamate being the major excitatory neurotransmitter in the mammalian brain, alterations of glutamatergic pathways are is likely to affect multiple brain functions. The removal of ammonia in the brain due to liver failure leads to astrocytic alterations that could critically affect glutamate-related processes such as the uptake and clearance of glutamate from the synapse (Butterworth et al, 2000).7 Under physiological conditions, glutamate is cleared by both neuronal and glial transporters, although the bulk of functional glutamate uptake is mediated by the glial transporters glutamate aspartate transporter (GLAST) or excitatory amino acid transporter 1 (EAAT-1) and glial glutamate transporter-1 (GLT-1) or EAAT-2 (Rothstein et al 1996).8 Since the glutamate transporters are the only mechanism for detoxifying extracellular glutamate, inhibition of these proteins, either by direct action on the transporter proteins per se or indirectly through inhibition of their expression, may have pathogenetic implications (Danbolt 1994).9 If EAAT-1 or EAAT-2 expression is reduced or eliminated, glutamate increases in cerebrospinal fluid, thus enhancing susceptibility to ischemic insults (Rothstein et al 1996). 8
Glial fibrillary acid protein (GFAP), a cytoplasmic filamentous protein that constitutes a major component of the cellular cytoskeleton in differentiated astrocytes of the CNS (Eng, 2000),10 modulates astrocyte motility and shape by providing structural stability to astrocytic processes (Cornet et al, 1993).11 GFAP represents the most specific astrocytic marker under normal and pathological conditions. Previous studies in both experimental ALF (Bélanger et al. 2002)12 and in cultured astrocytes exposed to ammonia reveal that GFAP expression at both the mRNA and protein levels is significantly reduced (Norenberg et al. 1990).13
In the brain, the bi-directional transmembrane water channel proteins - aquaporin (AQP) 1, 4, and 9 assist in the regulation of cellular water homeostasis along osmotic gradients. Among various isoforms of aquaporins, AQP-4, a 32 kDa protein has been extensively studied in brain because of its abundant localization in astrocytes, particularly in their end-feet that are in close apposition to cerebral capillaries (Rash et al, 1998).14 Such a strategic localization and regulation of water entry at the level of blood-brain barrier perhaps assigns a prominent role for AQP-4 in the brain edema associated with various neurological conditions (Amiry-Moghaddam and Ottersen, 2003).15 Direct evidence for a causal role of AQP-4 in brain edema was shown in AQP-4 knockout mice, which develop resistance to edema following ischemia, and water intoxication (Manley et al, 2004).16 Results of previous studies in in vitro (Rama Rao et al, 2003)17 and in vivo (Rama Rao et al, 2010;18 Wright et al, 2010;19Eefsen et al, 201020) models of ALF18-20 have suggested a role for AQP-4 in the development of brain edema.
Cell culture studies and animal models suggested that reduced astrocytic uptake of neuronally released glutamate, as well as alterations in expression of GFAP and AQP-4 contribute to the astrocyte swelling leading to brain edema and ICP in ALF. However, there is no evidence available to date to suggest that these alterations occur in patients with ALF. In order to address this issue, we, for the first time, analyzed human post-mortem cortical brain samples from patients with ALF dying with HE compared to brain samples from patients without evidence for pre-existing liver disease.
Methods
The study was approved by the Institutes Ethical Review Committee, Postgraduate Institute of Medical Education and Research (PGIMER), Chandigarh. Informed written consent for autopsy and use of brain tissue for research was obtained either from the legal next of kin of each patient or their close relatives.
Patients
Post-mortem human brain tissue (frontal cortex) was obtained from 8 autopsied patients with ALF and 7 disease controls who had no evidence of liver disease or neurological disorders at the time of death. The patients with neurological disorder other than HE were excluded. For the purpose of this study, ALF was defined according to the criteria as mentioned in our previous study (Dhiman et al, 2007).3A detailed patient history for identifying the cause of ALF, physical examination for evaluating evidence of HE and chronic liver disease and clues to the underlying cause of hepatic injury were obtained from the patient's case records. Age, gender, post-mortem delay (duration from estimated time of death to storage of brain tissues in RNA later solution) and diagnosis of each patient from different groups are summarized in Table 1.
Table 1: Characteristics of Control Patients and Patients with ALF
Patient No
Age at death (years)
Sex
PM delay (minutes)
Diagnosis
Control patients
1
56
Female
135
Ulcerative colitis
2
43
Male
240
Ventricular tachyarrhythmia
3
62
Female
300
Right common iliac vein thrombosis
4
78
Female
300
Coronary artery disease
5
54
Male
870
Sudden cardiac death
6
40
Male
240
Coronary artery disease
7
16
Female
195
Right ventricular hypertrophy
ALF patients
1
23
Female
220
ALF (acute hepatitis B related)
2
45
Male
300
ALF (Phenytoin related)
3
46
Male
250
ALF (ATD related)
4
25
Female
240
ALF (ATD related)
5
13
Male
180
ALF (acute hepatitis A related)
6
27
Female
210
ALF (acute hepatitis E related)
7
19
Male
415
ALF (acute hepatitis B related)
8
29
Female
270
ALF, anti HCV- positive but exact etiology undetermined
Abbreviations: ALF-acute liver failure; HCV- hepatitis C virus; PM, post-mortem; ATD, antituberculosis drugs (isoniazid, rifampicin and pyrazinamide)
Post-mortem brain tissue slices from frontal cortex, 5 mm or less in thickness, were collected in RNAlater (Ambion, New York, USA) and immediately stored at -80°C. These tissues were further used for the gene and protein expression analyses. Paraffin embedded blocks were prepared for histological and immunohistochemical analyses.
RNA isolation and cDNA synthesis:
Total RNA was extracted from cortical tissue samples by TRI Reagent solution (Ambion, New York, USA) as per the manufactures instructions. The RNA was quantitated using GenQuant (GE Healthcare, Hong Kong). The integrity of isolated RNA was assessed by electrophoresis through denaturing 1% agarose gels stained with ethidium bromide to visually verify the absence of genomic DNA contamination and integrity of 28S and 18S bands. Reverse transcription of RNA to cDNA was done using 6 µg of total RNA using random hexamer primers and Moloney murine leukemia virus (M-MuLV) reverse transcriptase using the RevertAidTM first strand cDNA synthesis kit (Fermentas, Pittsburgh, USA).We extracted total RNA from human post-mortem brain with an OD260/OD280=1.8-1.95 (Heinrich M. et al 2007).21
Quantitative Real-Time PCR:
Real-time PCR (relative quantification) was performed by LightCycler® 480 (Roche, Indianapolis, USA) using LightCycler® 480 SYBR Green I master (Roche, Indianapolis, USA detection method to check the quantitative expression of astrocyte genes (EAAT-1, EAAT-2, GFAP and Aquaporin-4). GAPDH was used as internal standard to monitor loading variations. Reaction mixture containing 1ml (20-30ng) of purified cDNA, 10ml of 2X master mix (SYBER Green1 Dye), 10 picomoles of each primers of different genes, was prepared and total volume was made up to 20 ml in sterile water. Human-specific primer pair sequences were constructed using NCBI Primer-BLAST (Genetics computer group, University of Wisconsin, Madison). Primer pairs were obtained from Operon Biotechnologies (Huntsville, Alabama, USA). Real-time primers for humans were as follows: EAAT-1-for, 5' ACCGTGACTGTCATTGTTGGCT -3'; EAAT-1-rev, 5'-AAATGGCAGCCAAAGCCTCA -3' ; EAAT-2-for, 5'- CTGAAACCTTTCCGGTGTGT-3'; EAAT-2-rev, 5'- GCATGGCAAGG GAGAAATAA-3'; GFAP-for, 5'- ATCTGGGAACTGTGCCTTTG -3'; GFAP-rev, 5'- GCATG GTGGCTCCAATCTAT -3'; AQP-4-for, 5'- AAGATCAGCATCGCCAAGTCT -3'; AQP-4-rev, 5'- AGCCAGTGACATCAGTCCGT -3'; GAPDH-for, 5'-AAGGCACAGACATGGTTGGT -3' and GAPDH-rev, 5'-TGGAAAGCAAACTGCCCTGA -3'. Finally relative quantification was done using the formula 2-ΔΔCp (Livak et al, 2001).22
Preparation of cerebral cortex lysates
10% lysate were prepared by homogenizing 100mg of brain tissue in 1 ml of RIPA buffer [50mM Tris pH7.5, 1mM EDTA, 150mM NaCl, 1% Triton X-100, 0.5% NaDOC, 0.1% SDS] and containing 100 ïl protease inhibitor cocktail (Complete EDTA free; Roche, Mannheim, Germany), using a motorized tissue homogenizer in ice chilled conditions. The lysate was centrifuged at 10000Xg for 20 min at 4oC and supernatant was used for further protein analysis.
Western-blot analysis
Western-blot analysis was performed as described elsewhere (Rothstein, 1995).23 Briefly, protein concentration of cortical lysate was determined using bicinchoninic Acid (BCA) method (Sigma-Aldrich, St. Louis, USA. Equal quantities of tissue lysates was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis using 12% gels (Tris-HCl, pH 7.4) and then electrophoretically transferred to polyvinylidenedifluoride (PVDF) membranes (Amersham Biosciences, Buckinghamshire, UK). Blots were blocked with nonfat dry milk in Tris-buffered saline (TBS) containingTween 20 overnight at 4oC and then incubated with mouse anti-EAAT-1 antibody (1:250, Abcam, Cambridge, USA), rabbit anti-EAAT-2 antibody (1:500, Abcam, Cambridge, USA), rabbit anti-AQP4 antibody(1:500, Abcam, Cambridge, USA), rabbit anti-GFAP antibody (1:50000, Abcam, Cambridge, USA),and rabbit anti-β-Actin (1:1000) for 2 hours at 37oC. PVDF were washed with TBS-T and incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies for 2 hours at room temperature (RT). After washing, the membrane are visualized using 3,3'-Diaminobenzidine tetrahydrochloride (GENEI, Bangalore, India) till the color development. Band intensity was measured by densitometric analysis using Scion image 4.0.3.2 software.
Immunohistochemistry
Immunohistochemistry was performed as described elsewhere (Rothstein, 1995).23 Briefly, sections (3-5 ïm) from paraffin embedded blocks were made and mounted on poly-L-lysine coated slides (Sigma-Aldrich, St. Louis, USA), fixed at 56oC for 30 min and stored for subsequent staining. Before staining, the sections were deparaffinized by heating them at 60oC, followed by serial passages through few changes of xylene and graded alcohol (100%, 95% and 70%). Endogenous peroxidase activity was blocked by incubating the sections with the blocking solution (0.5-3% H2O2 in methanol) for 30 min. and subsequently washed 3 times with water and once in a Tris solution for 5 min. The sections were subjected to high-temperature antigen unmasking for 15 min at 95oC. The primary antibody against EAAT-1 (Abcam), EAAT-2 (Abcam, Cambridge, USA), AQP-4 (Millipore, Billerica, USA) and GFAP (Dako, Glostrup, Denmark), with a dilution of 1:35, 1:20, 1:250 and 1:1000 in 0.02% BSA was applied onto the section and the sections were incubated overnight at 4oC under moist conditions, followed by a 2x5 min washing with PBS-solution. The sections were incubated with secondary antibodies against the primary antibodies (Envision1, K4001, Dako, Glostrup, Denmark) for 40 min followed by washing with a PBS-solution for 5 min and water for 2x4 min, respectively. After PBS washings, sections were incubated in peroxidase substrate solution i.e. 3-3' diaminobenzidinetetrahydrochloride (DAB) (Envision1, K4001, Dako, Glostrup, Denmark) for 5 min. The sections were then rinsed in tap water, counter stained with hematoxylin and mounted with DPX. The brown color obtained was visualized and scored by light microscopy. The primary antibody was omitted for the negative controls. The immunoreactivity was scored by the intensity of the staining pattern as follows: no staining (-), weak staining (+), moderate staining (++), and strong staining (+++).
Statistical Analysis
The data are expressed as median with range or as mean ± SD. The Student's t-test or Mann Whitney U test was used for comparison between two groups as appropriate. For comparing more than two groups ANOVA was performed and for the data that did not followed normal distribution, nonparametric Kruskal-Wallis was used. Correlation between quantity of total RNA isolated and post-mortem delay was evaluated by Spearman rank correlation coefficient with Bonferroni correction. Statistical analysis was performed with SPSS software for Windows, version 17.0 (SPSS Inc., Chicago, IL). In all the cases a probability value of P <0.05 was considered to be significant.
RESULTS
Demographic and Etiological Data
The demographic, post-mortem delay and etiological data from various groups are summarized in Table 2. Age showed significant difference between control subjects and ALF subjects; as expected most of the subjects who have had died of ALF were young. The etiology of ALF was viral in 5 [HAV in 1 (12.5%), HBV in 2 (25%), HEV in 1 (12.5%), drug induced in 3 (37.5%) and undetermined in 1 (12.5%)] (Table 2). All the patients with ALF were in grade 3 or 4 hepatic encephalopathy at the time of death.
Table 2: Demographic and Etiological Profile in Different Groups
Parameter
Controls
(n=07)
Acute liver failure (n=08)
P value*
Age (years)
54 (16-78)
26 (13-46)
0.003
Sex (Male:Female)
3:4
4:4
0.89
PM delay (minute)
240 (135-870)
245 (180-415)
0.593
(hour)
4 (2.25-14.5)
4.08 (3-6.91)
Etiology
Hepatitis A virus
1 (12.5%)
Hepatitis B virus
-
2 (25.0%)
Hepatitis C virus
-
-
Hepatitis E virus
-
1 (12.5%)
Drug induced
-
3 (37.5%)
Undetermined
-
1 (12.5%)
Data are expressed as median and range or number (percentage), *ANOVA, analysis of variance
Gross and Microscopic Changes of the Brain
All the control brains on gross observation showed unremarkable changes. On gross observation 2 of 8 brains from the ALF patients showed the flattening of gyri and the uncal notching at the base (cerebral edema), (Supplementary Figure 1). Normal microscopic neuropathology was observed in cortex from control patients and all the ALF patients showed significant cerebral edema and astrocyte swelling (Supplementary Figure 2). Normal histopathology of liver was demonstrated in controls, while all patients with ALF showed massive hepatocyte necrosis (Supplementary Figure 3).
RNA Extraction from Autopsied Brain Tissues
All the samples from the frontal cortex yielded intact RNA (28S and 18S rRNA bands) as judged by formaldehyde gel electrophoresis and no sample showed RNA degradation (Supplementary Figure 4).The quantification of total RNA ranged from 860 to 1960 µg/ml. There is no correlation between the quantity of total RNA recovered and the interval from death to tissue retrieval (r = ï€0.084; P = 0.677).
Glial Glutamate Transporters
mRNA and Protein Expression Level of Glial Glutamate Transporters in Brain Cortex
EAAT-1 mRNA level was significantly up-regulated in ALF patients by 4.37 fold (P = 0.023) compared to controls. However, EAAT-2 mRNA levels did not differ between ALF patients and control subjects (Fig. 1). The present study revealed no statistically significant difference in the expression of EAAT-1 protein in patients with ALF patients compared to controls. However, we have observed a significant decrease in the protein expression of EAAT-2 in ALF patients compared to controls (Fig. 2).
Fig.1. Quantification of EAAT-1, EAAT-2, GFAP and AQP-4 mRNA by real-time PCR. Bars represent the fold change.
Fig.2: Expression levels of EAAT-1, EAAT-2, GFAP and AQP-4 by western blot analysis. ß-actin used as internal control. (A) Western blot results from controls and ALF patients. (B) Densitometric quantification of western blot. Data represents mean ± SD.
Immunohistochemistry of Glial Glutamate Transporters
At higher magnification, EAAT-1 expression was mainly localized in the astrocyte processes; the intensity and distribution of staining pattern was shown to be similar in the disease groups as comparable to controls (Fig. 3).
Fig.3: EAAT-1 immunoreactivity in the frontal cortex. Higher magnification shows localization of EAAT-1 immunoreactivity in the astrocyte process (arrow).
EAAT-2 immunoreactivity was particularly abundant in layer II forming a band of strong immunoreactivity (Fig. 4a). The staining was faint in layer 1 and sparse in layers III, IV and V, where single cells or groups of cells with morphology of astrocytes were easily identifiable at high magnification. The amount of immunostaining was again higher around layer VI before dropping in the white matter. The EAAT-2 immunoreactivity in deep sulci showed intensely stained astrocytes with patchy and bushy pattern and uniformly distributed throughout. The staining pattern was weak in ALF patients as compared to controls but the distribution of the immunoreactivity was similar to that of normal cortex (Fig 4a). In the sulci, the immunoreactivity was mild and the remaining part of deep sulci showed complete absence of immunoreactivity both in cirrhosis and ALF patients compared to control cortex (Fig 4b). The white matter showed complete absence of EAAT-2 immunoreactivity in all the groups (Fig. 4c). At higher magnification, cortical layer from control showed the intense staining of the astrocyte processes (bushy like appearance) and in the diseased group of ALF cortical brains showed weak staining of the localized astrocyte processes (Fig. 5b).
In general, staining in all layers was most intense over the crests of gyri and weaker in the sulci. This laminar pattern of EAAT-2 expression suggests that, in addition to any activity related specifically to synaptic scavenging of glutamate, the transporter acts as barrier to the diffusion of glutamate into the CSF through the pial membrane and out of the deep aspect of the cortex into white matter.
Fig.4: Immunohistochemical expression EAAT-2 in the frontal cortex. Control: (a) The superficial cortex shows a continuous strongly diffused pattern of staining in the neuropil (arrow) of the gray matter; (b) shows a continuous band of intensely reactive astrocytes in the deep sulci of the cortex (arrow); (c) absence of EAAT-2 immunoreactivity in the white matter. ALF: (a) The superficial cortex shows a continuous diffuse weak staining of neuropil (arrow); (b) Patchy diffuse weak staining of the neuropil and some isolated immunoreactive astrocytes in the deep sulci (arrow); (c) No staining in the white matter.
Fig.5: EAAT-2 immunoreactivity in the frontal cortex. Control: (a) patchy and diffused staining of neuropil (arrow), and (b) intense staining of the reactive astrocyte process (bushy like) (arrow) in the control cortex. ALF: (a) patchy weak staining of neuropil (arrow) and (b) weak staining of the astrocyte processes (arrow).
Glial Fibrillary Acidic Protein
GFAP mRNA and Protein Expression in Brain Cortex
This study demonstrated a significantly lower GFAP mRNA level in cortical brain tissues from the ALF patients (8.22 fold) compared to controls (Fig 1). This study also demonstrated a significant decrease in the protein expression of GFAP in the cortical tissues of ALF (P = 0.008) patients when compared to controls (Fig. 2).
Immunohistochemistry of GFAP:
There was decrease in the number of astrocytes and the intensity of staining pattern of astrocytes in ALF patients compared to controls (Fig.6a). At higher magnification, the proportion of GFAP-positive glial cells decreased in the cortex of ALF patients compared to controls (Fig. 6b). Microvessels showing a continuous dense band of GFAP reactive glial foot processes in normal control cortex, where as in ALF patients showed a significant loss of GFAP reactive glial processes around the blood vessels (Fig.7c). The GFAP positive glial cells in the white matter were unchanged in diseased groups compared to controls, restricting the reduction in GFAP positivity to the grey matter.
Fig.6: Immunohistochemical expression GFAP in the frontal cortex. Control: (a & b) Intensely reactive astrocytes in the cortex (arrow); (c) Continuous dense band of GFAP reactive glial foot processes around the microvessels (arrow). ALF: (a & b) loss of GFAP immunoreactvity of glia processes (arrow); (c) weak staining of glial foot process around microvessels (arrow).
Aquaporin - 4
AQP-4 mRNA and Protein Expression
Our findings demonstrated significant up-regulation of AQP-4 mRNA expression in ALF patients by 3.38 fold (P = 0.003) when compared to controls (Fig.1). However, AQP-4 protein expression in total protein cell lysate was not significantly different ALF patients (P=0.30) when compared to controls (Fig.2).
Immunohistochemistry of AQP-4
The AQP-4 protein was expressed in both the grey and white matter of human cerebral cortices. Its expression was prominent around the blood vessels, consistent with localization to the perivascular astrocytic end-feet (plasma membrane staining), in the grey matter, and was also present in the neuropil. Neuropil immunoreactivity was particularly intense beneath the pial surface (Fig.7a). In the white matter, AQP-4 labelling was prominent around the vessels, and outlined the proximal processes of cells that have the morphological features of astrocytes. In comparison with the control cortex, AQP-4 immunoreactivity in the ALF specimens was characterized by more intense perivascular (microvessels) staining and forming a continuous perivascular sheath (Fig.7b) in both the grey and white matter.
Fig.7: AQP-4 immunoreactivity in the frontal cortex. Control: Patchy and diffused staining of neuropil (a) (10X), continuous (weak) band of AQP-4 reactivity in the glial end feet process surrounding the microvessels (arrow) (b) (40X). ALF: Patchy and diffused staining of the neuropil (a) (10X), continuous (dense) band of AQP-4 reactivity in the glial end feet process surrounding the microvessels (arrow) (b) (40X).
Discussion
This study provides the first direct evidence for selective alterations in expression of genes coding for key astrocytic proteins implicated in CNS excitability and brain edema in human ALF.
Astrocyte swelling is a common neurohistological feature in patients with liver failure as well as an important event in cytotoxic edema. Astrocyte swelling not only induces an increase in the water content of brain tissue, leading to cerebral edema and cerebral herniation but may also neuronal function (Kato et al, 1992; Thumburu KK et al. 2012).24,25This study demonstrated significant cerebral edema and astrocyte swelling on neurohistopathology in all patients with ALF.
In the present study we demonstrated significant decrease in the protein expression of EAAT-2 but not EAAT- in frontal cortex of ALF patients. These findings add to a growing body of evidence suggesting that ALF results in altered glutamatergic synaptic regulation in brain. Knecht and co-workers (1997)26 showed significant down-regulation of EAAT-2 protein by 31% and EAAT-2 mRNA by 60% in frontal cortex of rats with ALF at coma stages of encephalopathy compared to sham operated controls.27A significant decrease of EAAT-2 mRNA levels by 21% and 20% respectively was demonstrated in brain tissues from rats with thioacetamide-induced ALF and of hyperammonemic rats by northern blot analyses (Norenberg, 1997).27 Prolonged treatment of cultured rat astrocytes with ammonia has been shown to induce an increase in extracellular glutamate (Ohara et al. 2009).28 Together, these findings offer a convincing case for a role of ammonia in the EAAT-2 down-regulation in ALF. A theoretical alternative would relate to the neuroinflammation shown to occur in ALF (Jiang et al, 2009)29 . However, pro-inflammatory stimuli are known to cause an increase, rather than a decrease of EAAT-2 expression in brain (Persson et al., 2005)30. An increase in extracellular brain glutamate has been consistently demonstrated in different animal models of ALF (Rose et al, 2000; Michalak et al, 1996)31,32, a finding that is consistent with decreased glutamate uptake .
In the present study we did not observe any significant difference in EAAT-2 mRNA expression compared to controls. The mechanism underlying EAAT-2 protein loss is unclear, but likely to be due to inhibitory effect of ammonia at the transcriptional level of the EAAT-2 protein. The differences between measures of mRNA and the corresponding protein could result from the differential regulation of splice variants of these transporters. Several studies have demonstrated that the steady state levels of functional glutamate transporters do not necessary correlate with the levels of corresponding mRNAs (Bristol and Rothstein, 1996; Torp et al, 1995; Li et al, 1997).33-35 For example, in amyotrophic lateral sclerosis patients, dramatic loss of EAAT-2 protein was not matched by down-regulation of corresponding mRNA (Bristol and Rothstein, 1996).33 These results suggest that the metabolic turnover rates of glutamate transporters (i.e., translation versus degradation rates) and turnover rates of corresponding mRNAs (transcription versus degradation) do not necessarily change in concert and are probably regulated differentially depending on cell phenotype, environmental cues and signaling pathways used.
The EAAT-1 and EAAT-2 expression was restricted to grey matter and the immunoreactivity was exclusively localized in the glial cells. Moreover, the distribution of EAAT-1 in the control frontal cortex appears to be rather homogeneous with only slight gradual attenuation towards the cortex-white matter junction. Our observations are in coherence with observations of Rothstein et al (1995).23 EAAT-2 immunoreactivity distribution in this study showed heterogeneous pattern of expression throughout human cortical lamina, with areas of high expression corresponding to regions of greatest synaptic density (cortical layer I and II) in the control cortex. These observations were in complete accordance with previous findings in the human control cortex (Rothstein et al, 1995; Milton et al, 1997).23,36 To our knowledge this is the first study to describe changes in EAAT-1 and EAAT-2 in brains from ALF patients using immunohistochemistry.
Down regulation of glial glutamate transporters has been described in several neurological disorders, such as ischemia (Torp et al, 1995),34 amyotrophic lateral sclerosis (Rothstein et al, 1995)23 and Alzheimer's disease (Li et al, 1997).35 The demonstration of a significant loss of expression of EAAT-2 in brain in human ALF is consistent with overactivity of porstsynaptic glutamatergic transmission as has been proposed based upon studies in experimental animal models and suggests that glutamate(NMDA) receptor antagonists such as memantine (that is currently undergoing evaluation in the treatment of Alzheimer's diasease) could also be of benefit for the treatment of the neurological complications of ALF.
The present study, for the first time, demonstrated a significant loss in expression of GFAP mRNA and protein in frontal cortical extracts from patients with ALF and confirms the previous observations demonstrated in culture and animal models (Belanger et al. 2002;12 Neary et al, 199437 ). Belanger et al. (2002)12 observed significant decrease in mRNA and protein expressions in the brain extracts from rats with ALF due to hepatic devascularization. Neary et al, 1994,37 also reported that treatment of cultured astrocytes with NH4C1 reduces GFAP mRNA by up to 85% without inhibiting total RNA synthesis. Concurrent treatment with extracellular ATP prevented the loss of GFAP mRNA, possibly by activation of purinergic receptors. More recently, ammonia and proinflammatory mediators (IL-1β) led to decreased expression of GFAP in rat cultured astrocytes (Chastre et al, 2010).38 Haghighat et al (2000)39 demonstrated NH4Cl induced reduction of both ATP and GFAP in C6-glioma cells in a dose dependent manner, indicating that NH4Cl interrupts oxidative metabolism and the decrease in GFAP was probably a consequence of diminished ATP.
The immunohistochemical analysis in the current study provides direct evidence that patients with ALF had reduction in GFAP immunoreactivity in astroglial cells and supports the observations reported earlier in experimental studies. Selective loss of GFAP expression in the astrocyte and cell swelling was observed when the primary astrocyte cultures were treated with NH4Cl in acute doses (10mM) (Norenberg et al, 1990).13It was suggested that loss of GFAP expression due to ammonia exposure was due to instability of GFAP mRNA (Neary et al. 1994)37 and that a loss of expression of GFAP due to ammonia could result in altered visco-elastic properties of the cell leading to cytotoxic brain edema (Bélanger et al. 2002).12 Transient loss of GFAP immunoreactivity has been observed in astroglial cells of the fish spinal cord in response to high ammonia concentrations (Bodega et al, 1992).40
AQP-4 is restricted to the glia limitans and astrocytic end-feet (Warth et al, 2007).41 We demonstrated significant elevation of AQP-4 mRNA in the cerebral cortex of patients with ALF, whereas its total protein levels remain unchanged; however immunohistology showed that AQP-4 was intensely expressed by perivascular astrocytic end-feet forming a continuous perivascular sheath around the microvessels in ALF cortices compared to control cortices. In a rat model of ALF induced by the hepatotoxin thioacetamide, Rama Rao et al (2010)18 demonstrated an elevation of AQP-4 in plasma membrane of cerebral cortex, whereas total cellular levels of AQP-4 remained unchanged. Additionally, rats with ALF showed increased expression of α-syntrophin, a protein that binds to AQP-4 and stabilizes it in the plasma membrane. These results are in agreement with findings by Eefsen et al. (2010)20 in which rats with liver failure induced by galactosamine also showed increased AQP-4 expression in the plasma membrane fraction. However, they have reported no significant change in the AQP-4 mRNA expression. Wright et al. (2010)19 reported findings on AQP-4 protein expression in brain in several models of liver failure including galactosamine-, and lipopolysaccharide-induced, hyperammonemia and bile duct ligated liver failure. Their studies found an increase in the AQP-4 protein in the total brain fraction in rats following bile duct ligation (chronic liver failure), but did not observe any such increase in other liver failure models, including rats treated with galactosamine (Wright et al, 2010).19 These investigators concluded that AQP-4 does not appear to play a role in the brain edema associated with liver failure. However, it may be argued that these investigators did not examine AQP-4 protein levels in the plasma membrane fraction, the site where AQP-4 mediates water transport into astrocytes.
The mechanism by which ALF results in an apparent increase in plasma membrane levels of AQP-4 in the absence of increase of total brain protein content is not known. In a thioacetamide-induced ALF model an increase in AQP-4 plasma membrane levels might be caused by conformational changes in AQP4 protein (Rama Rao et al 201018), interference with its degradation, or increased anchoring of AQP4 to the PM possibly related to an increase in α-syntrophin (Neely et al, 200142 Rama Rao et al 201018). The latter may also be involved in the polarization of AQP4 to the perivascular astrocytic end-feet. L-histidine, an inhibitor of mitochondrial glutamine transport, completely blocked the increase in AQP4 and α-syntrophin proteins, suggesting a key role of glutamine in the process by which AQP4 is overexpressed in plasma membrane (Rama Rao et al 201018). ALF is associated with oxidative/nitrosative stress (Norenberg et al, 2007),43 and that oxidative/nitrosative stress has been implicated in AQP-4 over expression in other neurological conditions (Kaur et al, 2007;44). Therefore, it is possible that oxidative/nitrosative stress resulting from ALF may also contribute to increased AQP-4 plasma membrane levels.
In summary, these findings provide the first direct evidence for selective alterations in expression of genes coding for key astrocytic proteins implicated in excitatory (glutamatergic) neurotransmission as well as astrocyte cell volume regulation/brain edema in human ALF.
