Previous Article | Next Article 
The Journal of Neuroscience, April 15, 2003, 23(8):3394
Coordinate Regulation of Glutathione Biosynthesis and
Release by Nrf2-Expressing Glia Potently Protects Neurons from
Oxidative Stress
Andy Y.
Shih1,
Delinda A.
Johnson3,
Gloria
Wong1,
Andrew D.
Kraft3,
Lei
Jiang1,
Heidi
Erb1,
Jeffrey A.
Johnson3, 4, 5, 6, and
Timothy H.
Murphy1, 2
Kinsmen Laboratory of Neurological Research and Departments of
1 Psychiatry and2 Physiology, University of
British Columbia, Vancouver, British Columbia, V6T 1Z3, Canada, and
3 School of Pharmacy, 4 Molecular and
Environmental Toxicology Center, 5 Waisman Center, and
6 Center for Neuroscience, University of Wisconsin,
Madison, Wisconsin 53705-2222
 |
ABSTRACT |
Astrocytes have a higher antioxidant potential in
comparison to neurons. Pathways associated with this selective
advantage include the transcriptional regulation of antioxidant enzymes via the action of the Cap`n'Collar transcription factor Nrf2 at the
antioxidant response element (ARE). Here we show that Nrf2 overexpression can reengineer neurons to express this glial pathway and
enhance antioxidant gene expression. However, Nrf2-mediated protection
from oxidative stress is conferred primarily by glia in mixed
cultures. The antioxidant properties of Nrf2-overexpressing glia
are more pronounced than those of neurons, and a relatively small
number of these glia (< 1% of total cell number added) could protect fully cocultured naive neurons from oxidative glutamate toxicity associated with glutathione (GSH) depletion. Microarray and
biochemical analyses indicate a coordinated upregulation of enzymes
involved in GSH biosynthesis (xCT cystine antiporter, 
-glutamylcysteine synthetase, and GSH synthase), use (glutathione S-transferase and glutathione reductase), and export
(multidrug resistance protein 1) with Nrf2 overexpression, leading to
an increase in both media and intracellular GSH. Selective inhibition of glial GSH synthesis and the supplementation of media GSH indicated that an Nrf2-dependent increase in glial GSH synthesis was both necessary and sufficient for the protection of neurons, respectively. Neuroprotection was not limited to overexpression of Nrf2, because activation of endogenous glial Nrf2 by the small molecule ARE inducer,
tert-butylhydroquinone, also protected against oxidative glutamate toxicity.
Key words:
Nrf2; xCT; system
xc-; phase II detoxification enzymes; astrocyte; oxidative glutamate toxicity; glutathione; cystine
deprivation; tert-butylhydroquinone; antioxidant
response element; quinone reductase; oxidative stress; neuroprotection
| |
Introduction |
Glial cells (astrocytes) are known
to interact with surrounding neurons by nourishing, protecting, and
modulating growth and excitability (Travis, 1994
). Although it has been
shown repeatedly that astrocytes can improve neuronal survival, the
mechanisms of protection remain uncertain. Astrocytes have stronger
antioxidative potential than neurons (Raps et al., 1989; Makar et al.,
1994; Lucius and Sievers, 1996) and can protect neurons from oxidative stress induced by various compounds such as dopamine,
H2O2, and 6-hydroxydopamine
and nitric oxide (Desagher et al., 1996; Mena et al., 1996; Chen et
al., 2001). One potential defense against the toxicity of reactive
oxygen species (ROS) is the induction of a family of phase II
detoxification enzymes (Fahey et al., 1997; Kensler, 1997). Data from
our lab and others suggest that this response is expressed
preferentially in astrocytes, with considerably lower levels in neurons
(Ahlgren-Beckendorf et al., 1999; Eftekharpour et al., 2000; Murphy et
al., 2001; Johnson et al., 2002). Originally thought to be restricted
to promoting xenobiotic conjugation with endogenous ligands, such as
glutathione (GSH) (Hayes and Pulford, 1995; Primiano et al., 1997), the
observed functions of phase II enzymes have broadened recently. Now,
there are ~2 dozen known phase II enzyme genes, with more to be
elucidated via current technologies (Li et al., 2002). Sharing common
regulatory pathways, these enzymes possess chemically versatile
antioxidant properties and are inducible by various agents, including
those found in a normal diet (Fahey et al., 1997; Gao et al., 2001). Treatment of mammalian cells with electrophilic agents provokes a
cellular response leading to the coordinated transcription of phase II
genes (Prestera et al., 1993). A unique cis-acting
regulatory sequence, termed the antioxidant response element (ARE), is
essential for the constitutive and induced expression of many
antioxidant genes involved in the phase II pathway (Friling et al.,
1990; Rushmore et al., 1991; Nguyen et al., 2000). Several lines of evidence suggest that NF-E2-related factor 2 (Nrf2) is an important transcription factor responsible for upregulating ARE-mediated gene
expression (Itoh et al., 1997, 1999; Alam et al., 1999; Hayes et al.,
2000; Ishii et al., 2000). Studies that use knock-out mice have shown
that Nrf2 was part of a transcription factor complex required for
regulation of the mouse glutathione S-transferase (GST)
and NADPH:quinone oxidoreductase (NQO1) genes (Itoh et al., 1997; Hayes
et al., 2000). In addition, Nrf2/small Maf heterodimers bind to the ARE
sequence with high affinity during regulation of the GST and NQO1 genes
(Venugopal and Jaiswal, 1996). From these observations Nrf2 appears to
be the major transcription factor necessary for ARE activation and thus
essential for the induction of phase II detoxification enzymes. In this
study we use replication-deficient adenoviruses to overexpress Nrf2
protein in both neurons and glia to determine whether augmentation of the ARE-mediated antioxidant response might reduce neuronal
vulnerability to oxidative stress.
| |
Materials and Methods |
Materials. All chemicals were purchased from
Sigma-Aldrich Canada (Ontario, Canada) unless otherwise stated.
Plasmids and adenoviruses. pEF mammalian expression plasmids
carrying cDNA encoding mouse Nrf2 and Nrf2DN were a generous gift from
Dr. Jawed Alam (Alton Ochsner Medical Foundation, New Orleans, LA; Alam
et al., 1999). Rat ARE sequences were obtained from the NQO1 promotor.
To make the human placental alkaline phosphatase (hPAP) reporter
construct (rQR51), we excised a 51 bp ARE/electrophile response
element (EpRE) fragment by using restriction sites (XhoI and
HindIII) flanking the ARE sequence from a parent luciferase expression vector (Moehlenkamp and Johnson, 1999) and subcloned it into
the pGEM-7zf vector [American Type Culture Collection (ATCC), Manassas, VA] upstream of the hPAP reporter gene. A mutant rQR51 (rQR51mut) was made by replacing the original core ARE sequence with a cassette encoding a mutant ARE core sequence. The
BglII/NheI excisable cassette was constructed
with the following oligonucleotides (Sigma-Genosys, The
Woodlands, TX). The 10 bp mutation is underlined, and the region
corresponding to the ARE core sequence is shown in bold:
5'-CTAGCTCGAGATCCTCAGAGATTTCAGTCTAGAGTCACACGCAAACAGGAAAATCA-3' and
3'-GAGCTCTAGGAGTCTCTAAAGTCAGATCTCAGTGTGCGTTTGTCCTTTTAGTCTAG-5'.
Recombinant adenoviral vectors were constructed by using the Cre-lox
system (Canadian Stroke Network core facility, University of Ottawa,
Ottawa, Canada; Hardy et al., 1997). The Nrf2 and Nrf2DN cDNAs were
excised from the pEF vector by using restriction enzymes NotI, and NotI and HindIII,
respectively. All viruses were titered on human embryonic kidney 293 (HEK293) cells.
Mammalian cell culture. Mixed neuronal-glial cultures were
prepared from the cerebral cortex of Wistar rat fetuses embryonic day
17-18 (E17-E18), using the papain dissociation method (Murphy et al.,
1990). Viable cells were plated at 1 × 106 cells/ml on
poly-D-lysine-coated plastic culture plates
(Costar, Pleasanton, CA) in B27-supplemented Neurobasal
medium (Invitrogen, San Diego, CA). After 1 d
in vitro (1 DIV) the medium was changed to MEM
(Invitrogen) supplemented with 5.5 gm/l
D-glucose, 2 mM glutamine,
10% fetal bovine serum (FBS; HyClone, Logan, UT), 1 mM Na+-pyruvate, 100 U/ml penicillin, and 0.1 mg/ml streptomycin (MEM-pyr). This medium
change was required to reduce excessive antioxidant levels from the B27
medium. Enriched neuronal cultures used for Western blot analysis were
prepared by culturing cortical cultures in the presence of 10 µM 5-fluoro-2'-deoxyuridine/uridine that was
replaced every 4 d until the cultures were harvested at 14 DIV.
Enriched glial cultures were prepared from 1-2 d postnatal rat pups.
Cortices were dissected, minced, and used in the papain dissociation
method. Cells were plated in MEM supplemented with 10% FBS, 5%
heat-inactivated horse serum (HyClone), 2 mM glutamine, 100 U/ml penicillin, and 0.1 mg/ml
streptomycin in noncoated 10 cm plates (2 plates per brain). After 1 DIV the medium was replaced, and the glia were allowed to grow for
3 d. Adherent neurons were removed from the glial culture by
repeated pipetting of the medium. All glial cultures were used within
10 DIV because older, quiescent cultures appeared to lack the
antioxidant response. These conditions for isolating glial cells mainly
result in a population of type I and II astrocytes. Anti-glial
fibrillary acid protein (GFAP) staining suggests that the glial
cultures are mainly of the astrocyte phenotype. For a simple
neuron-glial coculture setup virus-infected glia were trypsinized and
transplanted directly into naive (no contact with virus) cortical
culture after 24 hr of transgene expression (coculture; Fig.
1A). Glia were
transplanted with concentrations ranging from 0.1 × 104 to 2 × 104 cells/ml, a plating density of
0.1-2% of the total number of cells per well. Cocultures with glia
that were separated physically from neurons (membrane-delimited
coculture) consisted of naive cortical cultures prepared in 24-well
plates (neuronal compartment) and infected glia separated in culture
plate inserts with 0.4 µm pore size (glial compartment;
Millipore, Bedford, MA) (Fig. 1B).
Infected glia were trypsinized 6 hr after infection and plated at
2 × 104 cells per insert, a plating
density of 2% of the total cell number, into the collagen-coated
inserts. For depletion of glial glutathione the glial compartment was
pretreated with 200 µM
L-buthionine sulfoximine (BSO) for 24 hr, then
preincubated for 10 min in fresh medium, and washed twice before
transfer into the neuronal compartment. For both setups the
transplanted glia were washed three times with PBS to prevent the
transfer of virus to the neuronal compartment.
View larger version (17K):
[in this window]
[in a new window]
|
Figure 1.
Schematic explanation of neuron-glial
coculture setups used in this study. A, For a simple
neuron-glial coculture setup the virus-infected glia were transplanted
directly into naive (no contact with virus) mixed neuronal-glial
cultures after 24 hr of transgene expression. B, Some
experiments required a setup by which glia were separated physically
from neurons (membrane-delimited coculture). This system consisted of
naive cultures prepared in 24-well plates; infected glia were separated
by a culture plate insert. Both cocultures were maintained for 24 hr
and then exposed to 3 mM glutamate (Glu) for a further 24 hr, followed by quantitation of neuronal viability.
|
|
HEK293 (ATCC) were plated at a density of 1 × 105 cells/ml on 10 cm plates
(Costar). Culture medium was prepared from MEM
supplemented with 10% FBS, 1 mM
Na+-pyruvate, 2 mM
L-glutamine, and 100 U/ml penicillin, 0.1 mg/ml streptomycin. After plating, the cells were allowed to grow for ~16
hr before commencement of transfection. All cultures were maintained at
37°C in a humidified 95% O2/5%
CO2 incubator.
Transfections and infections. HEK293 cells were transiently
transfected by the calcium-phosphate method (10 µg/10 cm plate) (Chen
and Okayama, 1987) or with Polyfect reagent (Qiagen,
Chatsworth, CA) according to the manufacturer's protocol. The
transfection efficiency was typically 60-70% as assessed by 
-gal
staining. Mature (14 DIV) primary cultures of mixed cortical neurons
and glia were transiently transfected with the ballistic gene transfer method by using a Helios gene gun (Bio-Rad, Hercules, CA)
firing 0.6 µm gold particles coated with DNA (1 µg DNA/mg gold
loading ratio). For adenovirus infection the immature cortical cultures were infected at 1 DIV by virus diluted to a multiplicity of infection of 200 in MEM-pyr. The cultures were allowed to express transgenes for
48 hr before usage. All infected cultures were examined for adequate
infection efficiency (25% of neurons, 80% of glia) as assessed by
green fluorescent protein (GFP) fluorescence.
Toxicity treatment. For all toxicity studies the cortical
cultures were used in their immature state (1-4 DIV). MEM-pyr was replaced with MEM supplemented with 5.5 gm/l
D-glucose, 2 mM glutamine, 5% FBS, 100 U/ml penicillin, 0.1 mg/ml streptomycin (MEM/5% FBS) containing the indicated concentrations of
L-glutamate,
H2O2, or staurosporine.
Cells were exposed to all toxins for 24 hr. No inhibitors of GSH
oxidation were added to the medium during the toxicity experiments.
Determination of neuronal viability. Neuronal viability was
evaluated by two methods: (1) manual counting of cells labeled with
fluorescent antibodies to neuron-specific enolase (NSE) or GFP, and (2)
3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) cell viability assay. For cell counting after
immunocytochemistry, 10 pictures of each experimental group were taken
with a 20× objective (Zeiss, Oberkochen, Germany), using
wide-field fluorescence microscopy (Zeiss Axiophot) with a
color CCD camera (Sony powerHAD, model DXC-950, Tokyo,
Japan). The imaged areas were chosen randomly from at least three
different wells per experimental group. All counting was performed with
the rater blinded to the experimental conditions that were used. To
confirm cell counts, we scanned the 96-well plates that were used for
counting with a Fluoroskan fluorescence plate reader (excitation, 530 nm; emission, 620 nm; Labsystems, Helsinki, Finland) to
measure red fluorescence from NSE-positive cells, although not all
experiments could be processed this way; absolute levels of NSE
fluorescence could vary because of background staining or cell density.
All measurements made with the Fluoroskan were background-subtracted,
where background was the fluorescence measured from ad-GFP-infected
wells treated with 3 mM glutamate (see Fig.
6B,C), 30 µM
H2O2 (see Fig.
5A), or 10 µM staurosporine (see
Fig. 5B) for 24 hr. For the MTT assay the cultures were
incubated with 0.45 mg/ml MTT diluted in MEM/5% FBS with no phenol red
for 2 hr at 37°C after the toxicity studies. Then the MTT staining
solution was removed and replaced with PBS and an equal volume of MTT
lysis buffer consisting of 20% SDS in 50% dimethylformamide, pH 4.7. The formazan crystals were allowed to dissolve overnight before the
plate was read on a Multiskan plate reader (absorbance, 560 nm;
Labsystems). In mixed neuron-glia cortical cultures
glial-mediated MTT turnover accounted for approximately one-half of the
total MTT reduction. To eliminate the glial component of MTT reduction,
we used measurements from control wells treated with 3 mM glutamate for 24 hr, in which nearly all
neurons had died, as a background that was subtracted from all other
groups (see Figs. 7C, 8A).
Western blot analysis and immunocytochemistry. Cells were
washed in PBS and harvested in PBS containing 2 µg/ml aprotinin plus
(in mM) 1 phenylmethylsulfonyl fluoride, 1 EGTA,
and 1 EDTA; the cells were sonicated for 10 sec to make crude lysate.
Protein concentration was measured with the bicinchoninic acid method (BCA; Pierce, Rockford, IL). Samples prepared in loading
buffer (7 mg/ml DTT, 6% SDS, 30% glycerol, and 0.38 M Tris, pH 6.8, and pyronin Y) were denatured by
boiling for 2 min before loading. For SDS-PAGE 30% acrylamide gels
were used to run HEK293 protein lysates (10 µg loaded) and enriched
neuron or glial lysates (30 µg loaded), respectively. Antibody
reactivity was detected by enhanced chemiluminescence substrate
(Amersham Biosciences, Arlington Heights, IL). For
immunocytochemistry after toxicity experiments the cultures were washed
twice with 37°C PBS to remove dead cells and cellular debris, were
fixed with 2% paraformaldehyde (PFA) for 10 min, and then were
incubated with primary and secondary antibodies. Immunostained cells
were mounted in Fluoromount-G (Southern Biotechnology
Associates, Birmingham, AL). Antibodies used in this study
include anti-Nrf2 from rabbit (1:200 dilution; Santa Cruz
Biotechnology, Santa Cruz, CA), anti-GFP from mouse (1:1000;
Boehringer Mannheim, Indianapolis, IN), anti-NSE from rabbit (1:2000;
Polysciences, Warrington, PA), anti-GFAP from rabbit
(1:100; Sigma-Aldrich Canada), anti-mouse Alexa Fluor 488 from goat (1:2000; Molecular Probes, Eugene, OR),
anti-actin from goat (1:1000; Santa Cruz Biotechnology), anti-heme
oxygenase 1 from rabbit (1:500; Stressgen Biotechnologies,
San Diego, CA), anti-rabbit horseradish peroxidase (HRP) from sheep
(1:5000; Amersham Biosciences), and anti-goat HRP from
donkey (1:5000; Santa Cruz Biotechnology).
Densitometry. Densitometric analysis of Western blots was
performed with the Scion Imaging program (version 4.0.2, Scion, Frederick, MD). Band intensities were
measured by taking the mean pixel intensity. All band measurements were
background-subtracted.
[3H]-glutamate uptake assay.
L-[3H]-glutamate
uptake was measured as described previously (Shih and Murphy, 2001).
Briefly, infected immature cortical cultures were preincubated in
Na+-free HBSS for 10 min at 37°C before
being incubated with Na+-free HBSS
containing 38 nM
L-[3H]-glutamate
(Amersham Biosciences) and the indicated unlabeled compounds (1 mM) for a further 20 min at 37°C.
Then the cells were washed three times with ice-cold
Na+-free HBSS and lysed with 0.5% Triton
X-100 in 0.1 M potassium phosphate buffer.
Radioactivity was determined by liquid scintillation counting and
normalized to protein concentration for each sample.
Placental alkaline phosphatase and quinone reductase
staining. For hPAP staining (Murphy et al., 2001) the cultures
were rinsed with PBS and fixed with 4% PFA and 0.2% glutaraldehyde
for 10 min. Preparations were rinsed again with PBS and incubated at 65°C for 30 min to inactivate any endogenous heat-labile alkaline phosphatase activity. The preparations were stained with 1 mg/ml nitroblue tetrazolium (NBT) and 1 mg/ml 5-bromo-4-chloro-3-indoyl phosphate dissolved in 0.1 M Tris buffer, pH 10. The staining reaction was performed at 37°C for ~30-40 min and was
terminated by washing with PBS. Pictures were taken with a 60×
objective (Zeiss) that used conventional light microscopy
with a color CCD camera. For NQO1 staining (Murphy et al., 1998) the
cells were fixed with 2% PFA for 10 min, washed with PBS, and
preincubated in buffer A (25 mM Tris, 0.08%
Triton X-100, 2 mg/ml BSA, pH 7.4) for 30 min. The preincubation
solution was replaced with buffer containing 100 µM NBT and 100 µM LY
83583 (Alexis Pharmaceuticals, San Diego, CA), and
staining was initiated with the addition of NADPH (1 mM final). The reaction was incubated at 37°C
for 20-30 min and terminated by washing in buffer A. Differential
interference contrast (DIC) and phase-contrast pictures were taken
before and after staining, respectively, using a 63× oil immersion
objective (Zeiss) on an Axiovert 200M microscope
(Zeiss) equipped with an AxioCam HRm digital camera
(Zeiss). DIC and phase-contrast images were overlaid for
final presentation.
Total intracellular GSH assay and effluxed GSH assay.
Total GSH was quantified by the method of Tietze (1969). Briefly, cells were collected in PBS and sonicated for 10 sec on ice. The acid-soluble fraction was obtained by adding perchloric acid to a final
concentration of 3%, followed by centrifugation at 14,000 × g for 10 min. The acid-soluble fraction was neutralized to
pH 7 with 0.5 M KOH/50 mM
Tris. After the removal of precipitate (potassium perchlorate) by a
second centrifugation, 50 µl aliquots of sample were combined with
100 µl of reaction mixture consisting of 2.5 ml of 1 mM 5,5',dithiobis-(2-nitrobenzoic acid) (DTNB),
2.5 ml of 5 mM NADPH, and 2.65 ml of
phosphate buffer (100 mM
NaPO4, pH 7.5, 1 mM EDTA),
glutathione reductase (5 U/ml final). The increase in
A412 from GSH-mediated reduction of DTNB was
measured at 30 sec intervals over 30 min. GSH content among treatment
groups was normalized to protein. For GSH efflux assays the MEM/5% FBS
with no phenol red was incubated with cultures for 24 hr and used in
the same procedure minus the removal of proteins via perchloric acid.
In some experiments the proteins were removed by filtration through a 3 kDa molecular weight cutoff membrane (Nanosep3K Omega, Pall
Gelman Laboratory, Sydney, Australia). Monochlorobimane (mCBi;
Molecular Probes) staining was performed by incubating
adenovirus-infected cultures with 60 µM mCBi
for 20 min (Tauskela et al., 2000). Stained cultures were washed twice with PBS to remove excess mCBi and were fixed with 4% PFA with 0.2%
glutaraldehyde for 20 min. At this stage the cells either were viewed
directly for fluorescence (excitation, 360 nm; emission, 410 nm)
or were stained further for GFP and GFAP, using immunocytochemistry. Quantitative measurements of mCBi/GSH adduct fluorescence were made
with Photoshop (Adobe Systems, Mountain View, CA) by
taking mean intensities from individual cells. All values were
background-subtracted.
RT-PCR and microarray analysis. Total RNA was isolated from
infected cultures with Trizol reagent (Invitrogen)
according to the manufacturer's instructions. RNA (1 µg) was
reverse-transcribed by using Oligo-dT 15 primer in accordance with the
RT System (Promega, Madison, WI). Then the resulting cDNA
was PCR-amplified by using primer sets for the genes: GFP, 5'-GAG CTG
TTC ACC GGG GTG GTG-3' and 5'-GAG CTC GAG ATC TGA GTC CGG-3'; mouse
Nrf2, 5'-TGA AGC TCA GCT CGC ATT GAT CC-3' and 5'-AAG ATA CAA
GGT GCT GAG CCG CC-3'; rat xCT, 5'-TTG CAA GCT CAC AGC AAT TC-3' and
5'-CGT CAG AGG ATG CAA AAC AA-3'; actin, 5'-CCC AGA GCA AGA GAG GTA
TC-3' and 5'-AGA GCA TAG CCC TCG TAG AT-3'. PCR conditions were as
follows: initial denaturing step that used 1 cycle at 95°C for 3 min,
followed by 35 cycles at 95°C for 30 sec, and the appropriate primer
annealing temperature (ranging from 53.4 to 56.6°C) for 1 min, then
72°C for 1.5 min, and a final cycle at 72°C for 5 min. The PCR
products were separated on a 1.4% agarose gel containing ethidium
bromide. Stained cDNA was visualized by using an ultraviolet light
source. Microarray analysis was performed as previously described (Li and Johnson, 2002; Li et al., 2002; Stein and Johnson, 2002), using
total RNA extracted from primary cultures with Trizol reagent (Invitrogen). Briefly, cDNA was synthesized from total RNA
by reverse transcription with T7-(dT)24 primer
incorporating a T7 RNA polymerase promotor, followed by a DNA
polymerase reaction (MessageAmp kit, Ambion, Austin, TX).
Biotin-labeled cRNA was prepared by an in vitro
transcription reaction that used the cDNA from above (MessageAmp kit,
Ambion; biotin-labeled nucleotide triphosphates,
Enzo Biochem, New York, NY). Labeled cRNA was fragmented by incubation at 94°C for 35 min in the presence of (in
mM) 40 Tris-acetate, pH 8.1, 100 potassium
acetate, and 30 magnesium acetate. Fragmented cRNA (15 µg) was
hybridized for 16 hr at 45°C to a rat genome U34A array
(Affymetrix, Santa Clara, CA). After hybridization the
gene chips were washed automatically and stained with
streptavidin-phycoerythrin by using a fluidics station. Probe arrays
then were scanned at 3 µm resolution, using the Genechip System
confocal scanner made for Affymetrix by Agilent
Technologies (Palo Alto, CA). Affymetrix Microarray
Suite 5.0 was used to scan and analyze the relative abundance of each
gene from the average difference of hybridization intensities. Analysis
parameters used by the software were set to values corresponding to a
moderate stringency (
1, 0.04; 2, 0.06; 
, 0.015; 1L and
1H, 0.0025; 2L and 2H, 0.003). Output from the microarray
analysis was stored as an Excel data spreadsheet. The definition of
statistically significant increase or decrease for individual genes was
based on the following three criteria, in order, as previously
described (Li and Johnson, 2002; Li et al., 2002; Stein and Johnson,
2002). (1) Rank analysis of the difference call (a measure of the
direction and magnitude of change) was made from three intergroup
comparisons (3 × 3, matrix analysis of three replicate samples)
for neuronal cultures and two intergroup comparisons (2 × 2) for
astrocyte cultures; no change was given a value of zero, marginal
increase/decrease was given a value of 1/-1, and increase/decrease was
given a value of 2/-2. The final rank referred to summing up the nine
values for the neuronal cultures and four values for the astrocyte
cultures corresponding to the difference calls. These values
varied from 18 to -18 and from 9 to -9 for the neuronal and astrocyte
cultures, respectively. The cutoff value for the determination of
increase/decrease for neuronal cultures was set as 9/-9 and for
astrocyte cultures was set as 4/-4. (2) For the coefficient of
variation the cutoffs were set at 1.20. (3) For the fold change the
cutoffs were set at 1.3 or greater for increased genes or 
1.3 or
lower for decreased genes.
Statistical analysis. All experiments were repeated at least
three times unless otherwise stated. Results are presented as the
mean ± SE. Statistical analysis of raw data was performed with
GraphPad Prism 2.0 (San Diego, CA). Experimental groups
were compared by one-way ANOVA, two-way ANOVA followed by Bonferroni's posttest, Kruskal-Wallis test, or Student's t test as
appropriate. A statistical probability of p < 0.05 was
considered significant.
| |
Results |
Neurons express lower levels of Nrf2 protein than astrocytes
We previously demonstrated that cortical astrocytes have a higher
basal and stimulated level of ARE-mediated gene expression than neurons
(Murphy et al., 2001; Johnson et al., 2002). One explanation for this
observation would be that astrocytes express higher levels of the
transcription factor Nrf2 than neurons. To test this hypothesis, we
prepared enriched neuronal and astrocyte primary cortical cultures and
probed their extracts with an anti-Nrf2 antibody by Western blot. A 105 kDa Nrf2-specific band was identified in enriched glial cultures that
comigrated with recombinant Nrf2 overexpressed in HEK293 cells (Fig.
2A,B). The same band
was detected at very low levels in neuron-enriched cultures.
Densitometry analysis revealed that glial-enriched cultures express
~12-fold more Nrf2 protein than neuron-enriched cultures, implying
that astrocytes may have more pronounced ARE-mediated gene expression
because they express higher levels of Nrf2 than neurons (Fig.
2B). The 105 kDa band was not detectable
in HEK293 cells overexpressing a -galactosidase control. The Nrf2
antibody, raised against the Nrf2 C-terminal region, also recognized
the dominant-negative N-terminal-deleted form of Nrf2 migrating at 29 kDa (Nrf2DN). Although the Nrf2 antibody was used for Western blots,
its nonspecific binding activity made it unsuitable for
immunocytochemistry.
View larger version (80K):
[in this window]
[in a new window]
|
Figure 2.
Cortical glia have higher basal Nrf2 expression
and ARE promotor activity than neurons. A, Western blot
of heterologously expressed Nrf2 (105 kDa) and Nrf2DN (28 kDa) in
HEK293 cells. B, A comigrating 105 kDa band
corresponding to Nrf2 is observed in enriched cortical glial, but not
neuronal, cultures. Densitometric analysis reveals an ~12-fold
difference in Nrf2 protein (n = 3);
*p < 0.05. C, Coexpression of Nrf2
cDNA with a hPAP-encoding reporter of ARE-mediated gene expression
(rQR51) greatly increases neuronal hPAP expression. Reporter constructs
carrying a mutation within the core ARE consensus sequence (rQR51Mut)
were not inducible. Coexpression with Nrf2DN cDNA suppresses both
neuronal and glial hPAP expression. *p < 0.05, neuron comparison to pEF (empty vector) control; #p < 0.05, glial comparison to pEF control. D, E,
Representative hPAP-stained astrocyte-like cells with coexpression of
pEF vector only. F, G, With Nrf2 overexpression cells of
both neuronal and glial morphology show ARE-driven hPAP expression.
Data from hPAP experiments represent the mean ± SEM number of
cells counted from triplicate coverslips over four independent
experiments.
|
|
If a lower basal level of Nrf2 protein is the reason for the lack of
ARE-mediated gene expression in neurons, then overexpression of Nrf2 in
neurons may be a strategy to boost their antioxidant gene expression.
To test this idea, we used a reporter construct carrying a heat-stable
hPAP gene driven by a minimal ARE-bearing promoter (rQR51). Previous
studies in our lab show that, in rQR51-transfected cortical cultures
and cultures derived from transgenic ARE-hPAP reporter mice, hPAP
staining is restricted mainly to the astrocyte population (Murphy et
al., 2001; Johnson et al., 2002). Consistent with our hypothesis, when
rQR51 was cotransfected with an expression vector for Nrf2, the number
of neurons staining for hPAP increased dramatically (Fig.
2C--G). Coexpression of rQR51 with the empty expression vector pEF did not increase neuronal staining, and coexpression with Nrf2DN not only prevented neuronal staining but also
suppressed basal astrocyte staining. Thus neurons are able to undergo
ARE-mediated gene expression but appear to be constrained because of
insufficient levels of Nrf2 protein. Induction of hPAP expression from
rQR51 was dependent on the presence of a wild-type consensus ARE
sequence and was abolished by mutation of this sequence (rQR51mut; Fig.
2C).
Overexpression of Nrf2 enhances antioxidant activity of neurons and
astrocytes in immature cortical cultures
We next evaluated whether Nrf2 overexpression could increase the
antioxidant properties of cortical neurons and glia. For these
experiments replication-deficient adenoviruses (Crocker et al., 2001)
were used to infect a large fraction of neurons efficiently. Three
different adenovirus constructs were made with CMV promoters driving
the expression of enhanced green fluorescent protein (eGFP) alone
(ad-GFP), Nrf2 and eGFP (ad-Nrf2), or Nrf2DN and eGFP (Ad-Nrf2DN). For
the latter two viruses each cDNA was driven by a separate CMV promoter.
When used to infect immature rat cortical cultures (1 DIV), all three
viruses were able to infect ~25% of the neurons and ~80% of the
glia, when used at a multiplicity of infection of 200.
To characterize the antioxidant activity of ad-Nrf2-infected cultures,
we evaluated the protein products of three known Nrf2 target genes:
heme-oxygenase-1 (HO-1) (Alam et al., 1999; Gong et al., 2002), system
xc
(xCT, cystine
transporter) (Ishii et al., 2000; Sasaki et al., 2002), and NQO1 (Bloom
et al., 2002). Western blot analysis showed that HO-1 protein levels
increased in parallel to Nrf2 overexpression in mixed cortical cultures
(Fig. 3A). Similarly, when
infected with ad-Nrf2, both neurons (Fig. 3B) and glia (data
not shown) exhibited robust increases in NQO1 staining. Glial staining
increased over and above normal basal levels (compared with ad-GFP
control). Staining also was observed selectively in ad-Nrf2-infected
neurons, but not in ad-GFP-infected neurons, further supporting the
idea that neurons are capable of having an enhanced antioxidant
response but are limited by low basal Nrf2 expression. Uninfected
neurons in the ad-Nrf2-infected culture did not stain for NQO1. In
addition, when infected with ad-Nrf2, xCT-mediated
L-[3H]-glutamate
uptake in mixed cortical cultures increased to 372 ± 82% of the
ad-GFP-infected group (Table 1).
Conversely, activity was suppressed to 75.7 ± 7.1% of the ad-GFP
control in ad-Nrf2DN-infected cultures. The Nrf2-induced increase in
L-[3H]-glutamate
uptake fit the pharmacological profile of system xc because it was
blocked by competitive inhibitors of cystine uptake such as quisqualate
and homocysteic acid (Table 1). Nrf2-dependent upregulation of xCT mRNA
also was verified by using RT-PCR with two independent primer sets
(Fig. 3C). Comparison of basal xCT mRNA levels between mixed
cultures and glial-enriched cultures with RT-PCR showed an
approximately threefold (n = 1) higher signal in
glial-enriched cultures (data not shown). However, this observation should be interpreted as a qualitative difference attributable to
potential nonlinearity in the RT-PCR method. RT-PCR for GFP confirmed
than an approximately equal amount of viral gene expression occurred in
ad-GFP- and ad-Nrf2-infected cultures as well as between the mixed and
glia-enriched culture types.
View larger version (73K):
[in this window]
[in a new window]
|
Figure 3.
Ad-Nrf2-infected cultures exhibit enhanced
antioxidant potential. A, Time course evaluation of Nrf2
protein overexpression with parallel induction of HO-1 expression.
B, Histochemical staining revealed increased NQO1
activity in ad-Nrf2-infected neurons, but not neighboring uninfected
neurons, visible under DIC optics (bottom panels). Neuronal NQO1
staining was not observed in the ad-GFP group (top panels). Scale bar,
20 µm. C, xCT mRNA levels increase with Nrf2
overexpression as shown by RT-PCR. Nrf2 mRNA derived from infection was
detected by using selective mouse primers. D, Nrf2
overexpression increases total intracellular GSH/GSSG levels. Sublethal
glutamate exposure (6 hr) leads to partial depletion of intracellular
GSH in all groups (open bars). Control groups represent a separate
group of cultures with no glutamate exposure but that were
vehicle-treated (filled bars). Culture were given a total of 48 hr for
expression before being harvested for GSH analysis. E,
Increase in mCBi staining is primarily enriched in glia of
ad-Nrf2-infected mixed cultures. Cultures are depicted in
phase-contrast (Phase), fluorescence immunostaining for anti-GFAP
(GFAP+) and anti-GFP (Infected), and 60 µM mCBi staining (mCBi). mCBi staining images show
selected areas with high numbers of glial clusters and are not
representative of actual mixed culture composition. HO-1 and NQO1
images are representative of at least three separate experiments. GSH
data represent mean ± SEM of four separate experiments performed
in duplicate; *p < 0.05, compared with GFP
control.
|
|
View this table:
[in this window]
[in a new window]
|
Table 1.
Evaluation of Na+-independent
L-[3H]-glutamate uptake by system
x in ad-Nrf2-infected mixed immature
cortical cultures: percentage increase and pharmacology of induced
uptake
|
|
Nrf2 overexpression in mixed cultures also generated an approximately
four- to fivefold increase in total intracellular glutathione [GSH + oxidized glutathione (GSSG)] as compared with GFP control (Fig.
3D). Treating the cultures with 3 mM
Glu for 6 hr before harvesting for GSH measurement reduced the GSH
levels in all groups without causing neuronal death, but the
ad-Nrf2-infected group still contained more GSH in comparison to
controls. Using mCBi to stain for GSH in ad-Nrf2-infected cultures, we
observed that the GSH/mCBi adduct fluorescence was enriched in
GFAP-positive glia cells (Fig. 3E). This staining pattern
was not observed in ad-GFP-infected cultures. A detailed quantitation
of mCBi fluorescence revealed that both infected and uninfected neurons
also were stained more heavily in the ad-Nrf2-infected mixed cultures
in comparison to neurons in ad-GFP-infected cultures (Table
2). It should be noted, however, that
increased mCBi staining can be indicative of both increased GSH levels
as well as enhanced GST activity (Tauskela et al., 2000).
Microarray analysis of Nrf2-overexpressing mixed cortical cultures
and enriched glial cultures
To define the gene set targeted by Nrf2 thoroughly, we performed
microarray analysis on mRNA derived from mixed neuron-glia cultures (3 DIV) and glial-enriched cultures (5-10 DIV) 48 hr after infection with
ad-GFP or ad-Nrf2. Affymetrix rat genome U34A arrays were
used to monitor the expression of ~7000 full-length mRNAs and ~1000
expressed sequence tag clusters. To define our culture systems better,
we initially assessed the basal expression of glial markers in
GFP-overexpressing cultures. As expected, the basal GFAP and vimentin
signals were approximately sevenfold higher than in glial-enriched
cultures in comparison to mixed cultures. The GFAP and vimentin signals
from the two different culture types were normalized to various
housekeeping genes, including lactate dehydrogenase,
glyceraldehyde-3-phosphate dehydrogenase, and -actin; similar
results were obtained for each. The basal levels of the housekeeping
genes varied by only ~10% between the glial-enriched and mixed
cultures. Conversely, the basal signals for the neuronal markers,
neurofilament M and tau microtubule-associated protein, in mixed
cultures were ~34- and ~62-fold higher, respectively, in comparison
to glial-enriched cultures. The relative GFAP signal was used further
to estimate that ~13% of the mRNA from mixed cultures was of glial
origin. Basal signals for rat Nrf2 mRNA indicated that glial-enriched
cultures typically contained ~31-fold more endogenous Nrf2 than the
mixed cultures. Interestingly, not all transcription factors were
enriched in glia because basal signals for CREB, Maf 1, Maf 2, and
c-Jun varied only slightly between the two culture types. However, the
basal signal for NF-
B also was highly enriched in glial cultures
(~15-fold). As expected from previous work, examination of basal mRNA
levels indicated preferential glial expression of antioxidant factors
such as NQO1, GST A3, GST P2, HO-1, catalase, thioredoxin reductase,
metallothionein 1 or 2, and peroxiredoxin 1 and 5 (Dwyer et al., 1995;
Murphy et al., 1998, 2001; Ahlgren-Beckendorf et al., 1999;
Eftekharpour et al., 2000).
Analysis of three separate comparisons between mRNA from GFP and
Nrf2-overexpressing cultures consistently indicated the selective upregulation of a number of known and previously unknown phase II
detoxification genes in response to acute Nrf2 overexpression (48 hr;
Tables 3,
4). Given the large number of
Nrf2-upregulated genes, only some genes were listed and categorized
into four groups on the basis of function: detoxification,
antioxidant/reducing potential/metabolism, signal transduction, and
inflammation. A full gene list including increased and decreased genes
is available online:
http://www.pharmacy.wisc. edu/facstaff/sciences/JohnsonGroup/microdata.cfm. Of the increased genes, 92 were common to both glial-enriched and mixed
cultures, the genes of interest of which are presented in Table 3 (fold
induction cutoff, 1.3; repression cutoff, 1.3). In addition, a subset
of genes was induced selectively in either the mixed cultures or
glial-enriched cultures. Genes of interest from this list are presented
in Table 4.
The increased expression of various GST isoforms NQO1, HO-1,
-glutamylcysteine synthetase (-GCS; only modifier light chain upregulation was observed), thioredoxin reductase, and malic enzyme by
Nrf2 overexpression in both cultures types corresponds to the gene
induction profile previously observed when electrophilic agents such as tert-butylhydroquinone (tBHQ) and
sulforaphane were used (Ahlgren-Beckendorf et al., 1999; Eftekharpour
et al., 2000; Li et al., 2002; Thimmulappa et al., 2002). Induction of these well characterized Nrf2 gene targets is, in part, supported by
our own immunoblot data for HO-1, histochemical stains for NQO1, and
measurements of GSH production (Fig. 3A,B,D). Of particular interest is the induction of many genes associated with the
biosynthesis, use, and export of GSH, including the following: (1)
uptake of cystine at the cell surface via the xCT cystine/glutamate
antiporter, shown by RT-PCR (see Fig. 9) (Bannai, 1986; Sato et al.,
1999); (2) synthesis of gGluCys by the rate-limiting enzyme for GSH
synthesis -GCS; (3) incorporation of Gly to gGluCys to make the
complete GSH tripeptide by GSH synthetase; (4) use of GSH by various
GSTs and glutathione reductase; (5) possible efflux of GSH via the multidrug resistance protein (MRP1), a mechanism previously described for astrocyte GSH release (Hirrlinger et al., 2001, 2002); and (6)
possible extracellular cleavage of GSH by -glutamyl transpeptidase (GT).
Enhancing antioxidant potential by Nrf2 overexpression protects
neurons from cell death caused by oxidative stress, but not
staurosporine-induced apoptosis
Can the increased antioxidant potential from Nrf2 overexpression
protect neurons in in vitro models of oxidative stress? For this experiment we used a well established NMDA receptor-independent oxidative glutamate toxicity paradigm in which neurons die from GSH
depletion (Murphy et al., 1989, 1990). When ad-GFP-infected cultures
were treated with 3 mM glutamate, only 12.5 ± 5.3% of the infected neurons were viable after 24 hr (Fig.
4A,B). With the same
treatment 103.7 ± 4.6% of the ad-Nrf2-infected neurons survived
the toxic exposure as compared with the ad-GFP control. Surprisingly,
88.0 ± 14.0% of the uninfected neurons in the same culture were
protected also (Fig. 4A,D). The extent of neuronal death in the ad-Nrf2DN-infected group was very similar to that of the
ad-GFP-infected group. Doubling the glutamate concentration to 6 mM was not able to overcome the protective effect
of Nrf2 (data not shown), suggesting that upregulated glutamate removal mechanisms such as enhanced glutamate metabolism, glutamine synthesis from glutamate, or sodium-dependent glial glutamate uptake were not the
main mechanisms for protection. Furthermore, after a 24 hr incubation
the glutamate levels in the medium were not significantly different
among the ad-GFP and ad-Nrf2 groups as assessed by a fluorescence-based
glutamate dehydrogenase cycling assay (data not shown) (Nicholls et
al., 1987). To ensure that the observed toxicity was indeed a result of
oxidative stress, we added 100 µM
-tocopherol in the presence of 3 mM glutamate,
and a nearly complete block of the neuronal death (80.3 ± 11.5%
reduction; n = 4) was observed. We further ruled out
the possibility of NMDA receptor-mediated excitotoxicity in our
paradigm by treating 3-4 DIV cultures with 200 µM NMDA overnight or with a short 500 µM NMDA or glutamate exposure for 10 min in
HBSS, followed by washout. Neither treatment caused appreciable
neuronal death in these cultures (data not shown). The viability of
glial cells was not affected by oxidative glutamate toxicity as
observed in previous studies (Murphy et al., 1990) (Fig.
4A,C).
View larger version (44K):
[in this window]
[in a new window]
|
Figure 4.
Nrf2 overexpression in a subpopulation of cells
confers widespread neuronal protection from oxidative glutamate
toxicity. A, Immunocytochemistry for eGFP (green,
identifying infected cells) and NSE (red marker, a neuron-selective
marker). Within a typical ad-GFP-infected culture the infected neurons
(yellow, red + green), uninfected neurons (red), and infected glia
(green) can be observed. B, Group data evaluating the
vulnerability of infected neurons to oxidative glutamate toxicity. Data
are expressed as a percentage of GFP+
NSE+ cells (presumed infected neurons) in the
indicated glutamate treatment group as compared with the ad-GFP control
group. VE, Vitamin E (-tocopherol), 100 µM.
C, Viability of GFP+
NSE cells (presumed infected glia) present per
image was not affected significantly with glutamate treatment.
D, Uninfected neurons within cultures containing
Nrf2-infected cells are more resistant to oxidative glutamate toxicity.
Data represent the mean ± SEM number of cells counted over
triplicate wells from at least three independent experiments;
*p < 0.05, compared with the GFP control
no-glutamate group.
|
|
Ad-Nrf2-infected cultures were also more resistant to direct
exposure to the pro-oxidant molecule
H2O2. Ad-GFP- or
ad-Nrf2-infected cultures were exposed to 0.3-30 µM
H2O2 for 24 hr. Under these conditions the ad-GFP-infected group began to show significant neuronal
toxicity at 3 µM
H2O2 and maximum cell death
at 30 µM H2O2
(Fig. 5A). Infecting the
cultures with ad-Nrf2 reduced neuronal toxicity by 53.1 ± 16.9%
of the ad-GFP control during exposure to 10 µM
H2O2. The protective effect
of Nrf2 was overcome by 30 µM
H2O2.
View larger version (15K):
[in this window]
[in a new window]
|
Figure 5.
Nrf2 overexpression in mixed immature
cortical cultures protects neurons from
H2O2-mediated toxicity, but not
staurosporine-induced apoptosis. Ad-GFP- and ad-Nrf2-infected cultures
were allowed to express transgenes for 48 hr before exposure to 0.3-30
µM H2O2 (A;
n = 3) or 0.1-10 µM staurosporine
(B; n = 4) for a further 24 hr.
Cells were stained for NSE to evaluate neuronal viability. Data
represent the mean ± SEM from the indicated number of experiments
performed in quadruplicate; *p < 0.05, compared
with ad-GFP-infected control.
|
|
Neuronal injury from oxidative stress can exhibit features of an
apoptotic pathway (Whittemore et al., 1995; Tan et al., 1998). To test
whether Nrf2 upregulation could protect neurons from apoptosis induced
by a different pathway than direct exposure to ROS, we treated infected
cultures with 0.1-10 µM staurosporine, a potent inhibitor of phospholipid/Ca2+-dependent
protein kinase (Tamaoki et al., 1986). Interestingly, overexpression of
Nrf2 did not confer any protection from staurosporine-induced toxicity
under these conditions (Fig. 5B). In both ad-GFP- and ad-Nrf2-infected groups cell death was observed with 0.3 µM first, whereas maximal
toxicity was achieved with 10 µM.
Glutathione released from ad-Nrf2-infected glial is necessary and
sufficient for protecting neurons from oxidative glutamate toxicity
Given that a small number of ad-Nrf2-infected glia are present in
infected mixed cortical cultures (~2% of the total cell number
according to manual cell-counting data), it is possible that these
enhanced glial cells contributed to the observed protection of neurons.
To address this possibility, we tested whether ad-Nrf2-infected glia
transplanted into naive (uninfected) mixed cultures could protect
resident neurons. Ad-Nrf2-infected glia were plated directly into naive
cultures at 0.1-2% of the total cell number by using the coculture
setup, as detailed in Materials and Methods (Fig. 1A). Amazingly, up to 83.3 ± 22.3% of the
naive neurons were protected from 3 mM glutamate when the glia were transplanted at
only 0.5% (Fig.
6A-C). Complete
neuronal protection was achieved when glia were plated at 1.5% of the
total cell number. This protection was not observed with ad-GFP-,
ad-Nrf2DN-infected, or uninfected glia. It should be noted, however,
that the plated glial densities slightly underestimate the number of
glia during exposure to toxins 24 hr after plating due to glial
proliferation.
View larger version (56K):
[in this window]
[in a new window]
|
Figure 6.
A small fraction of infected glial cells is
sufficient to protect neurons from oxidative glutamate toxicity.
A, Representative images from glial-neuron coculture
setup (see Fig. 1A). Within a typical coculture
infected glia (green) and uninfected neurons (red) can be observed.
Uninf, Uninfected glia transplanted. B, Group data
obtained from plate scanning for NSE (red) fluorescence.
C, Decreased neuronal viability is demonstrated by a
loss of red fluorescence. The addition of Nrf2-overexpressing glia
restores NSE expression to levels found in an ad-GFP-infected group
that was not exposed to glutamate. Data represent the mean ± SEM
from three separate experiments performed in quadruplicate;
*p < 0.05. Scale bar, 80 µm.
|
|
Glia are known to release a number of protective factors that
enhance neuronal survival via a variety of mechanisms. Of particular importance is the release of sulfhydryl species such as GSH,
cysteine-glycine (CysGly), or cysteine, which can contribute to
neuronal GSH synthesis (Sagara et al., 1993a; Dringen et al., 1999;
Wang and Cynader, 2000). When enriched glial cultures were infected
with ad-Nrf2, both intracellular and released GSH were increased (Fig.
7A,B). When glia were infected with ad-Nrf2 and plated at 2 × 104 cells/ml (equivalent to 2% of the
total cell number), the GSH accumulated in the medium over 24 hr was
increased significantly to 11.5 ± 1.9 µM
as compared with 4.6 ± 0.8 µM with ad-GFP
or 2.8 ± 1.2 µM with ad-Nrf2DN. Filtering
the medium (<3 kDa size restriction) did not reduce the GSH measured
in the medium significantly, suggesting that the released GSH was not
protein-bound and was most likely in its free form. Indeed, neuronal
toxicity was prevented completely when naive cultures were exposed to 3 mM glutamate in the presence of 10 µM exogenous GSH for 24 hr (Fig.
7C). This suggests that the concentration of GSH released
into the medium by a small ad-Nrf2-infected glial component was
sufficient for neuronal protection from oxidative glutamate
toxicity.
View larger version (27K):
[in this window]
[in a new window]
|
Figure 7.
Release of GSH from glia is both sufficient
and necessary for conferring neuronal protection. A,
B, Infection with ad-Nrf2 increases total intracellular
GSH/GSSG as well as GSH released into the medium (MEM/5% FCS, no
phenol red). C, Exogenous addition of reduced GSH
concurrently with glutamate treatment protects neurons from oxidative
glutamate toxicity. D, Glial GSH release is necessary
for Nrf2-dependent neuronal protection. A membrane-delimited coculture
(see Fig. 1B) was used, allowing enriched glial
cultures to be pretreated separately with the GSH synthesis inhibitor
BSO and then to be washed and added to wells containing neurons. BSO
pretreatment (200 µM) for 24 hr produces long-term
reduction of intracellular GSH/GSSG and release of GSH from glia and
abolishes glial-mediated neuronal protection. Data represent the
mean ± SEM of at least three independent experiments performed in
triplicate; #*p < 0.05.
|
|
To test whether GSH release from ad-Nrf2-infected glia was
necessary for the observed neuronal protection, we used BSO, an irreversible blocker of -GCS, to inhibit glial GSH synthesis (Griffith and Meister, 1979). To avoid potential effects of BSO on
neuronal GSH, we plated infected glia on culture plate inserts and
treated them with 200 µM BSO for 24 hr before transfer
into neuronal cultures by using the membrane-delimited coculture setup (Fig. 1B). For this setup the glial cells were
situated in an upper layer 1-2 mm above naive neurons in the bottom of
a well. BSO pretreatment of ad-Nrf2-infected glia suppressed levels of released GSH to ~3 µM for >24 hr after
washout (data not shown) and completely blocked the Nrf2-mediated
protection of neurons (Fig. 7D).
Small molecule inducers of Nrf2 increase neuronal survival during
oxidative glutamate toxicity
Compounds such as tBHQ and dimethyl fumarate, known inducers of
Nrf2 activation and phase II gene induction, have been shown to protect
various cell lines against oxidative stress caused by
H2O2 and dopamine exposure
(Duffy et al., 1998; Li et al., 2002). In this study we have found that
direct pretreatment of immature cortical culture with 10-20
µM tBHQ for 24 hr efficiently blocks oxidative glutamate
toxicity (Fig. 8A). To
compare the protective potential of tBHQ-treated glia with
ad-Nrf2-overexpressing glia, we treated enriched glial cultures
with 5-20 µM tBHQ for 24 hr to allow for
upregulation of an antioxidant response before they were transplanted
into naive cortical cultures via the coculture setup (Fig.
1A). Glia pretreated with 20 µM tBHQ provided neuronal protection from 3 mM glutamate when they were transplanted at a
plating density of 5% of the total cell number (Fig.
8B). However, protection was not seen when
tBHQ-treated glia were transplanted at 2%, a density sufficient for
neuronal protection when ad-Nrf2-infected glia were used. Treating
glial with tBHQ increased intracellular GSH content, as previously
described (Eftekharpour et al., 2000), and also increased released GSH
in a dose-dependent manner (data not shown).
View larger version (20K):
[in this window]
[in a new window]
|
Figure 8.
Neuronal protection can be achieved by
activation of endogenous Nrf2 with the use of a small molecule inducer.
A, Immature cortical cultures pretreated for 24 hr with
10 and 20 µM tBHQ are protected from 1-3 mM
glutamate exposure. Partial protection is conferred by tBHQ treatment
at 3 mM glutamate. B, Selective induction of
endogenous Nrf2 in glia led to partial neuronal protection from
oxidative glutamate toxicity. Glia pretreated with a range of tBHQ
concentrations (0-20 µM) for 24 hr were transplanted
into naive neuronal cultures at a plating density of 5% of the total
cell number, using a coculture setup (see Fig.
1A). Data represent the mean ± SEM from
three independent experiments performed in quadruplicate;
*p < 0.05.
|
|
| |
Discussion |
Nrf2 overexpression boosts antioxidant potential of glia:
protection of neurons from oxidative glutamate toxicity
Overexpression of the Cap`n'Collar transcription factor Nrf2
upregulates a strong antioxidant response in both neurons and glia
dissociated from rat cortex. We show that protection of neurons can be
achieved by upregulation of the glial antioxidant response alone.
Central to this protection was the augmented synthesis and release of
glial GSH, which can be used toward strengthening the neuronal GSH
pool. This conclusion was based on several lines of evidence. First,
ad-Nrf2 infection of glia significantly increased intracellular and
released GSH. Second, microarray analyses indicated the coordinated
induction of GSH biosynthesis, use, and export enzymes. Third,
exogenous GSH addition mimicked neuroprotection offered by
ad-Nrf2-infected glia, and selective depletion of glial GSH with the
use of BSO abolished this neuroprotection. Thus Nrf2-dependent enhancement of glial GSH release appears to be necessary and sufficient for neuronal protection. However, it is important to note that suppression of glial GSH synthesis may have deleterious effects on
glial health and possibly could alter the expression of other glial-derived protective factors. Although GSH plays an important role
for protection in the oxidative glutamate toxicity paradigm, we cannot
rule out completely the involvement of other glial-derived factors.
It is well documented that astrocytes can protect neurons from
damage caused by various ROS (Desagher et al., 1996; Lucius and
Sievers, 1996; Tanaka et al., 1999; Kirchhoff et al., 2001). Previous
studies have shown that an adequate astrocyte GSH content is essential
because depletion of GSH with BSO abolishes astrocyte-mediated neuroprotection (Drukarch et al., 1997; Chen et al., 2001). One mechanism underlying this protection is the continuous glial delivery of GSH and/or GSH precursors to neurons for GSH synthesis (Yudkoff et
al., 1990; Sagara et al., 1993a,b; Dringen et al., 1999, 2000). Interestingly, GSH efflux becomes markedly increased when the astrocytes are exposed to oxidative stress (Sagara et al., 1996), a
response that may be initiated by Nrf2 activation.
Although the events necessary for the neuronal uptake of GSH and its
precursors remain uncertain, evidence suggests that a number of
sulfhydryl species can be used by neurons, including cysteine, cystine,
and the CysGly dipeptide (Murphy et al., 1990; Sagara et al., 1993a,b;
Kranich et al., 1996; Wang and Cynader, 2000). Two major paths can be
initiated with the release of GSH into the medium/CSF (Fig.
9). GSH can be metabolized into CysGly and Glu-X conjugates by the glial ectoenzyme -GT (Dringen et al.,
1999). CysGly can be taken up by neurons or can be hydrolyzed further
by the ectoenzyme aminopeptidase N to release cysteine (Dringen et al.,
2001). Alternatively, a constant release of GSH can potentiate an
extracellular thiol/disulfide exchange reaction with cystine to produce
cysteine and a GSH-cysteine conjugate (Wang and Cynader, 2000). There
is currently little direct evidence to support the uptake of GSH by
neurons. However, both sodium-dependent and -independent GSH
transport systems have been isolated from brain cells, including
astrocytes (Kannan et al., 2000), bovine brain capillary cells (Kannan
et al., 1996), and immortalized mouse brain endothelial cells (Kannan
et al., 1999). Furthermore, sodium-dependent GSH transport in rat
synaptosomal membrane vesicles exhibited high-affinity kinetics
(Km = 4.5 ± 0.8 µM) (Iantomasi et al., 1999) and may be
physiologically relevant for uptake considering the low GSH levels
found in CSF (5.87 ± 0.29 µM) (Wang and
Cynader, 2000).
View larger version (24K):
[in this window]
[in a new window]
|
Figure 9.
Schematic diagram of GSH biosynthesis and release
pathways that may be involved with Nrf2-dependent coupling of GSH
between astrocytes and neurons. Astrocyte GSH synthesis and release are
more robust with Nrf2 overexpression. Higher levels of xCT (system
xc) expression were detected, which
promotes cystine uptake and provides a precursor for GSH synthesis.
Microarray analyses indicate that all major enzymes involved in GSH
biosynthesis are upregulated also. GSH release from astrocytes is
increased, leading into several possible pathways of extracellular GSH
metabolism, including an initial breakdown by -glutamylcysteine
transpeptidase (GT) and possible further breakdown by aminopeptidase
(Apep) into the glutathione precursors Cys, Gly, and CysGly suitable
for neuronal uptake (Dringen et al., 1999, 2001). Alternatively,
extracellular GSH may be taken up by neurons directly or may
contribute to the reduction of cystine into cysteine, which may be a
source of sulfhydryl species for neuronal uptake (Sagara et al., 1993;
Wang and Cynader, 2000). Neurons may also uptake cystine for
glutathione synthesis (Murphy et al., 1990). Microarray analyses also
indicate the upregulation of additional factors involved in
detoxification, ROS scavenging, and NADPH production that may work
together with GSH to protect neurons. Genes that are upregulated
significantly by Nrf2 overexpression are underlined. Intracellular and
extracellular concentrations of glutamate are average values from a
combination of previous studies in the human brain and CSF (Siegel,
1981). CSF levels of cystine are typically very low at ~0.2
µM (Lakke and Teelken, 1976; Araki et al., 1988; Wang and
Cynader, 2000). The Invitrogen MEM used in this study is
formulated to contain 100 µM cystine.
|
|
For a toxicity paradigm we primarily have used oxidative glutamate
toxicity in which neurons are exposed to millimolar levels of
extracellular glutamate to inhibit cystine uptake completely via the
cystine/glutamate antiporter, system
xc, leading to GSH
depletion and oxidative stress (Murphy et al., 1989, 1990). In
vivo, oxidative stress resulting from glutamate toxicity
potentially may play a role in cell death during stroke, injury, and
neurodegenerative disease (Coyle and Puttfarcken, 1993). Under
TOP
ABSTRACT
Introduction
