Oxidative stress contributes to neurodegeneration in Huntington's disease (HD). However, the origins of oxidative stress in HD remain unclear. Studies in HD transgenic models suggest involvement of mitochondrial dysfunction, which would lead to overproduction of reactive oxygen species (ROS). Impaired mitochondria complexes occur in late stages of HD but not in presymptomatic or early-stage HD patients. Thus, other mechanisms may account for the earliest source of oxidative stress caused by endogenous mutant huntingtin. Here, we report that decreased levels of a major intracellular antioxidant glutathione coincide with accumulation of ROS in primary HD neurons prepared from embryos of HD knock-in mice (HD140Q/140Q), which have human huntingtin exon 1 with 140 CAG repeats inserted into the endogenous mouse huntingtin gene. Uptake of extracellular cysteine through the glutamate/cysteine transporter EAAC1 is required for de novo synthesis of glutathione in neurons. We found that, compared with wild-type neurons, HD neurons had lower cell surface levels of EAAC1 and were deficient in taking up cysteine. Constitutive trafficking of EAAC1 from recycling endosomes relies on Rab11 activity, which is defective in the brain of HD140Q/140Q mice. Enhancement of Rab11 activity by expression of a dominant-active Rab11 mutant in primary HD neurons ameliorated the deficit in cysteine uptake, increased levels of intracellular glutathione, normalized clearance of ROS, and improved neuronal survival. Our data support a novel mechanism for oxidative stress in HD: Rab11 dysfunction slows trafficking of EAAC1 to the cell surface and impairs cysteine uptake, thereby leading to deficient synthesis of glutathione.
Compelling data support a critical role for oxidative stress in the pathogenesis of Huntington's disease (HD), a disorder caused by polyglutamine expansion in huntingtin (Htt). However, mechanisms for origins of oxidative stress in HD are unclear. Oxidative stress occurs with overproduction of reactive oxygen species (ROS), or reduction in the antioxidant capacity, or both. Impaired functions of mitochondrial complexes, which would cause overproduction of ROS, occur in late stages but not in presymptomatic or grade I HD patients (Guidetti et al., 2001), suggesting that sources of oxidative stress independent of mitochondria exist in early HD.
Glutathione (GSH) is a major antioxidant in the brain (Dringen, 2000) and essential for protecting cellular constituents against ROS-induced damage by reacting with ROS through its free thiol group (Schulz et al., 2000; Ballatori et al., 2009). After reaction with ROS, GSH is oxidized into GSSG (glutathione disulfide), which can be converted back to GSH for reuse by GSH reductase, a flavoprotein that uses NADPH as the electron source. A rise in ROS usually stimulates a compensatory increase in GSH synthesis to maintain the normal redox balance. Maintenance of normal GSH levels is vital for neuronal survival (Li et al., 1997; Nicole et al., 1998; Wüllner et al., 1999). In the brain, glial cells store high levels of GSH and release GSH into the extracellular space (Dringen and Hamprecht, 1998). However, neurons cannot take up extracellular GSH (Aoyama et al., 2008) and need de novo synthesis that requires uptake of the rate-limiting precursor cysteine from the extracellular space (Dringen, 2000; Aoyama et al., 2008) through the neuronal Na+-dependent glutamate transporter EAAC1 (EAAT3) (Shanker et al., 2001; Aoyama et al., 2006). Knock-out of EAAC1 causes oxidative stress in neurons and age-dependent neurodegeneration, which can be rescued by administration of membrane permeable GSH precursor N-acetylcysteine (Aoyama et al., 2006). EAAC1 locates mainly in the cytoplasm (Rothstein et al., 1994; Conti et al., 1998; He et al., 2000; Nieoullon et al., 2006) and undergoes constitutive recycling that requires normal Rab11 activity (González et al., 2007).
Decrease of GSH levels has been reported to occur in peripheral plasma (Klepac et al., 2007) and erythrocytes (Zanella et al., 1980) of HD patients. Postmortem brains of HD patients have levels of GSH similar to those of non-HD controls (Sian et al., 1994). However, study of brain tissues does not allow discrimination between neurons and glial cells because in the HD brain astrocytic storage of high levels of GSH (Dringen and Hamprecht, 1998) can mask changes in neurons (Selkoe et al., 1982; Myers et al., 1991). Therefore, it is important to know whether mutant Htt disturbs GSH levels selectively in neurons. In this study, we examined levels of GSH in primary cortical neurons prepared from embryos of a knock-in mouse model of HD (HD140Q/140Q) and wild-type (WT) mice. We found that decreased levels of GSH and increased levels of ROS simultaneously occurred in HD140Q/140Q cortical neurons. We propose that glutathione dysregulation in HD140Q/140Q neurons is linked to a deficit in Rab11-dependent trafficking of EAAC1.
Materials and Methods
The cDNA encoding Rab11 (BC010722) was amplified by PCR, purified from agarose gel using QIAEX II kit (QIAGEN), and digested with BamHI/XhoI. After purification from agarose gel, the digested Rab11 cDNA fragment was cloned into the BamHI/XhoI sites of the HA-pcDNA3 plasmid. Rab11 mutant Rab11Q70L was generated by PCR-based site-specific mutagenesis. All Rab11 sequences were verified by DNA sequencing. For cloning of Rab11Q70L into the CSCW2-pgk lentivirus vector, HA-Rab11Q70L-pcDNA3 was digested with BamHI and XhoI to obtain Rab11Q70L, which was subcloned into the BamHI/XhoI sites of the pENTR1A plasmid. Rab11Q70L was then released from Rab11Q70L-pENTR1A after treatment with NheI and XhoI. This NheI/XhoI Rab11Q70L fragment was cloned into the NheI/XhoI sites of the CSCW2-pgk vector that encoded enhanced green fluorescent protein (EGFP). The two genes are independently expressed via a bicistronic regulation.
Generation of lentivirus and viral titration.
CSCW2-pgk with cDNA encoding EGFP or Rab11Q70L-CSCW2-pgk-EGFP was packaged into lentivirus. 293T producer cells were maintained in DMEM high-glucose media supplemented with 10% fetal calf serum (Atlanta Biologicals), l-glutamine (Invitrogen), and sodium pyruvate (Invitrogen) without antibiotics in a 37°C cell culture incubator (Thermo Fisher Scientific). The day before transfection, 7 × 106 of 293T cells were plated in a 15 cm dish and cultured as above. Transfection was performed according to the manufacturer's instructions (Invitrogen). For each transfection, 6.75 μg of CSCW2-pgk or Rab11Q70L-CSCW2-pgk plasmid DNA, 6 μg of pCMV-dR8.2 dvpr, 0.75 μg of pCMV-VSV-G, and 243 μl of FuGene 6 (QIAGEN) were used to prepare packaging mixtures. After applying packaging mixtures, 293T cells were cultured at 37°C as above for 24 h and changed into fresh culture media. Forty-eight hours later, culture media were collected and filtrated through a 0.45 μm filter to remove cell debris. Filtrated culture media were centrifuged at 4°C, 24,000 rpm in a T50.2 rotor (Beckman Coulter) for 1.5 h. The viral pellet was resuspended in 0.2 ml of PBS containing 0.5% BSA. Titers of virus were determined using a HIV-1 p24 antigen ELISA kit (ZeptoMetrix). Viral titers were 1.83 × 108 pg ml−1 for lenti-Rab11Q70L/EGFP and 1.1 × 108 pg ml−1 for lenti-EGFP.
Preparation of primary cortical neurons and infection of lentivirus.
Cortices were dissected from the brains of embryos of embryonic day 16 (E16) mice (WT and HD140Q/140Q), dissected free of meninges and other tissue, and incubated in PBS containing antibiotics (penicillin, streptomycin, and neomycin) and 0.25% trypsin for 10 min at 37°C (all reagents were purchased from Invitrogen, except where indicated). After incubation, cells were washed twice in PBS supplemented with calcium and magnesium, and then dissociated in Neurobasal medium containing B27 supplement, N2 supplement, 25 μm mercaptoethanol and l-glutamine (NBM), seeded in plates (BD Biosciences) with or without glass coverslips (Warner Instruments), and cultured in serum-free NBM media. Plates and glass coverslips were precoated with poly-l-lysine. Twenty-four to 48 h after plating, primary cultures were treated with cytosine arabinoside (AraC) for 24 h to inhibit proliferation of glial cells. Then, fresh NBM media was added to the cultures. Quantitative analysis of the primary cultures was performed 4 d after AraC treatment to determine the presence of astrocytes (GFAP labeling), microglia (CD68 labeling), and neurons (β-III-tubulin labeling). Results showed that 99.5% of the cells were neurons.
Primary neurons were infected with 30 multiplicities of infection of lentivirus expressing EGFP or Rab11Q70L/EGFP or not for indicated times as previously reported (Li et al., 2009a). Experiments were conducted when EGFP expression in >90% of neurons was confirmed with a fluorescence microscope.
Measurement of glutathione.
Measurement of GSH was conducted according to supplier instructions (BioVision Research Products). In brief, primary neuronal cultures were lysed in assay buffer (BioVision Research Products) at 8 d in vitro (DIV8) and collected into a 1.5 ml tube. After removing nuclei by centrifugation at 4°C, 14,000 rpm for 5 min in a table centrifuge (Eppendorf), 20 μl of perchloric acid (PCA) was immediately added to 60 μl of postnuclear supernatant (S1), mixed, and incubated on ice for 5 min. After a centrifugation at 4°C, 14,000 rpm for 5 min, 40 μl of PCA-preserved supernatants were mixed with 20 μl of 3N KOH for neutralization of the pH, incubated on ice for 5 min, and centrifuged at 4°C, 14,000 rpm for 5 min. Ten microliters of the neutralized sample was diluted with assay buffer to a final volume of 90 μl, mixed with 10 μl of o-phthalaldehyde (OPA) probe, and incubated at room temperature for 40 min. Fluorescence intensities of samples and standards were read on a fluorescence plate reader equipped with excitation/emission wavelength of 340/450 nm. Data were represented as nanomoles per milligram protein.
Measurements of reactive oxygen species in living neurons.
Primary WT and HD140Q/140Q neurons from E16 embryonic days were plated at 400,000 cells/ml on 24-well plates and cultured as above. At DIV8, neurons with or without virus were used for measuring reactive oxygen species as described previously (Valencia and Morán, 2001). In brief, neurons were changed into fresh complete medium and incubated with 2 μg ml−1 5-(and-6)-carboxy-2′,7′-dichlorofluorescein diacetate (carboxy-DCFDA) (Invitrogen) at 37°C for 30 min. After incubation, neurons were washed in prewarmed PBS and imaged using a Bio-Rad Radiance 2100 confocal laser-scanning microscope with krypton–argon laser. Images were acquired through a 60× Nikon Plan Fluor Apo objective (numerical aperture, 1.4) on an inverted Nikon Eclipse TE300 fluorescence microscope using 488 nm excitation and 515 nm emission wavelengths. Each field with at least two neurons was imaged. The focus was adjusted by visualizing neurites. After adjustment of focus, images were immediately collected. At least three different fields of neurons were obtained per culture dish and used for quantification.
Images were analyzed using NIH ImageJ. The soma and neural processes were manually tracked by tracing the edge of a cell body or a selected neurite. The pixels in the equivalent area adjacent to the cell body or process were also measured. The fluorescence intensity in the area outside the soma or neurites was subtracted from the measurement of fluorescence intensity in the cell body or neurite. Three criteria were used for selecting a neurite: the neurite was in focus, had a clear point of origin from the soma, and was well separated from other processes. The average area of all examined neurons for each condition was used for normalizing the fluorescence intensity per area of the soma region for each neuron. Data are represented as the mean average intensity per neuron with SD. The average length of all examined processes for each condition was used to normalize the fluorescence intensity for each process. Data are represented as the mean total intensity per length of a process with SD.
Determination of cell viability.
Cell viability was determined by the ability of neurons to convert 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) into DMSO-soluble formazan blue crystals. Primary WT and HD140Q/140Q neurons from E16 embryonic days were plated at 800,000 cells/ml on 24-well plates and cultured as above. At DIV4, primary neurons were changed into fresh complete NBM and infected with lentivirus expressing EGFP alone or Rab11Q70L and EGFP for 7 d, respectively. At DIV11, neurons were changed into fresh complete NBM and incubated with 0.5 mg ml−1 of MTT in complete NBM for 15 min at 37°C. After MTT incubation, the medium was carefully removed. Formazan blue was dissolved in 100% DMSO. Absorbance was read at 540 nm on a multiwell spectrophotometer. The mean absorbance unit of WT neurons (no infection) or WT neurons expressing EGFP alone (viral infection) was used for calculating the percentage of MTT conversion by neurons under the corresponding condition (no infection or viral infection). The mean percentage of WT or WT-EGFP for each condition with SD was graphed.
Cysteine uptake assay.
Primary neurons in six-well plates were infected with virus or not after AraC treatment and routinely cultured. Five days after infection of virus, primary neurons were washed twice in prewarmed PBS and cultured in PBS for 1 h at 37°C. l-[14C]Cysteine uptake was performed at room temperature for 30 s in PBS containing 1% BSA. The final concentrations of l-[14C]cysteine were 0.1, 1, 5, 10, and 50 μm, respectively, and the final concentration of DTT was 25 mm. l-[14C]Cystine (20 μCi/ml; PerkinElmer) was stored at −80°C after adding DTT to a final concentration of 50 mm DTT. Without addition of DTT to l-[14C]cystine, radioactivity was hardly detectable in the cell lysates after uptake was performed.
To determine that primary neuronal cultures take up cysteine through amino acid transporters instead of the glutamate–cystine exchanger system, cysteine uptake was conducted at room temperature for 60 s in 50 mm HEPES-K, pH 7.4, 146 mm NaCl, 5 mm MgCl2, and 0.2 m sucrose in the presence or the absence of 1 mm dl-threo-β-benzyloxyaspartate (TBOA) (Tocris Bioscience), and in 50 mm HEPES-K, pH 7.4, 100 mm KCl, 5 mm MgCl2, and 0.2 m sucrose. To determine the contribution of EAAT2/GLT-1 to neuronal uptake of cysteine, cysteine uptake was performed in the absence or the presence of 100 μm dihydrokainate (DHK) (Tocris Bioscience). Before cysteine uptake, primary neurons were cultured in PBS for 1 h in a cell culture incubator and washed twice in 50 mm HEPES-K, pH 7.4, 5 mm MgCl2, 1 mm DTT, and 0.2 m sucrose. The concentration of l-[14C]cysteine was 2 μm, and the concentration of DTT was 10 mm. After cysteine uptake for indicated times, primary neurons were washed three times in cold PBS and lysed in 0.2 ml of cold 50 mm HEPES-K, pH 7.4, 100 mm KCl, 2 mm EDTA, and 1 mm DTT containing 1% Triton X-100 (TX-100) on ice for 15 min. Twenty microliters of lysates were used for scintillation count in triplicate. Data were represented as cpm per milligram of protein per second.
Biotinylation of cell surface proteins.
Biotinylation of proteins at cell surfaces was conducted according to the manufacturer's instructions. Briefly, primary WT and HD140Q/140Q cortical neurons at DIV8 with or without viral infection at DIV4 were extensively washed in cold PBS, pH 8.0. After incubation in cold PBS on ice for 15 min, cells were changed into fresh cold PBS containing 1 mm sulfo-NHS-SS-biotin (Pierce) and incubated on ice for 4 h. After extensive washes in cold PBS, cells were incubated in 0.1 mm Tris-Cl, pH 8.0, to quench nonreacted sulfo-NHS-SS-biotin on ice for 30 min and lysed in lysis buffer (50 mm HEPES, pH 7.4, 200 mm NaCl, 10 mm MgCl2, 1% TX-100, and protease inhibitors). Crude lysates were cleared by centrifugation at 4°C, 14,000 rpm for 10 min. The resulting supernatants were incubated with preequilibrated NeutrAvidin agarose (Pierce) at 4°C with rotation for at least 2 h. After incubation, samples were centrifuged at 4°C, 1000 rpm for 1 min. The resulting supernatants (intracellular proteins: unbiotinylated) were precipitated with chloroform/methanol and resuspended in sample buffer for SDS-PAGE and Western blot analysis. NeutrAvidin agarose bound with cell surface proteins (biotinylated) were washed three times in lysis buffer and boiled in sample buffer for SDS-PAGE and Western blot analysis.
The band density of biotinylated EAAC1 (biotinylated) from each set (WT vs HD) of experiments was measured using SigmaScan Pro or NIH ImageJ. At the exposure necessary to detect the signal of biotinylated EAAC1 (biotinylated), the signal for intracellular EAAC1 (unbiotinylated) was saturated. Therefore, biotinylated EAAC1 values were normalized against intracellular actin to obtain equal input for each set of experiments. Normalized values of biotinylated EAAC1 were used to calculate percentage of biotinylated EAAC1 in HD140Q/140Q neurons relative to WT neurons.
SDS-PAGE and Western blot.
SDS-PAGE and Western blot analyses were performed as described previously (Li et al., 2009a). In brief, samples were boiled in SDS-PAGE sample buffer for 5 min and loaded on a Tris-glycine gel. After electrophoresis, proteins were transferred onto nitrocellulose blots. The membranes were blocked with 5% nonfat milk in TBS containing 0.05% Tween 20 and incubated with antibodies. Concentrations of the primary antisera used for Western blot were anti-EAAC1 (1:500; Millipore Bioscience Research Reagents) and anti-actin (1:500; Sigma-Aldrich). Peroxidase-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories) were diluted at 1:5000. All antisera were diluted in TBS containing 1% BSA and 0.05% Tween 20. Blots were developed using enhanced ECL (Pierce).
After two washes in PBS, primary neurons on coverslips were fixed in 4% paraformaldehyde/PBS at room temperature for 15 min, quenched in 50 mm NH4Cl containing 0.1% Triton X-100, washed twice in PBS containing 0.1% Triton X-100, blocked with 1% BSA in PBS containing 0.05% Tween 20, and labeled with primary antibodies. After the primary antibody step, targeted proteins were visualized by incubation of cells on coverslips with secondary antibodies conjugated with BODIPY FL (Invitrogen) or Cy-3 (Jackson ImmunoResearch Laboratories). Primary antibodies against Rab11 (1:500; ABR–Affinity BioReagents) and EAAC1 (1:100; Millipore Bioscience Research Reagents) were used. Secondary antibodies including Hoechst 33258 (Invitrogen) were diluted at 1:1000. All antisera were diluted in PBS containing 1% BSA and 0.05% Tween 20. Microscopy was performed using 60× or 100× oil Nikon Plan Apo objective mounted on an inverted Nikon Eclipse TE300 fluorescent microscope. Individual images for each wavelength (405, 488, and 568 nm) were collected separately using a Bio-Rad 2100 LaserSharp confocal system equipped with krypton–argon and blue diode lasers. The scanned images from each fluorescence channel were merged using PhotoShop.
Fluorescence intensity of immunoreactive EAAC1 signals in soma and neurites of neurons and the corresponding background signals were captured with confocal microscopy and measured using NIH ImageJ. The average size of soma of WT (n = 30) and HD140Q/140Q neurons (n = 30) in the samples was not different. The average area of all examined neurons (WT plus HD, 60 neurons) was used for normalizing the EAAC1 fluorescence intensity in the soma and neurites of neurons. The mean intensity of EAAC1 was graphed.
Two-tailed Student's t test was performed to determine statistical significance between two research groups. One-way ANOVA and post hoc analysis were used for determination of statistical significance for multiple comparisons.
A decline in levels of GSH in HD neurons precedes neuronal death
We examined GSH levels in primary WT and HD140Q/140Q mouse cortical neurons after culture for DIV8 using a GSH assay. This assay takes advantage of the selective reaction of OPA with GSH. OPA itself has a very low fluorescence but generates strong fluorescence signal after reaction with GSH. We found that HD140Q/140Q neurons had significantly lower levels of GSH compared with WT neurons (n = 3, mean ± SD, in nanomoles per milligram protein per minute, WT vs HD140Q/140Q: 10 ± 1.1 vs 7.2 ± 0.7; p < 0.01).
Since loss of intracellular GSH can cause neurodegeneration in vitro (Li et al., 1997; Nicole et al., 1998; Wüllner et al., 1999), we looked for clues of cell death of primary HD140Q/140Q neurons. MTT conversion assay showed that at DIV8 when GSH levels have already declined, HD140Q/140Q cortical neurons did not manifest significant death relative to WT neurons (mean ± SD of percentage of WT, WT vs HD140Q/140Q: 100 ± 3.5 vs 102.8 ± 4.4%; Student's t test, no significance) (Fig. 1). However, the percentage of MTT conversion by HD140Q/140Q neurons relative to WT neurons dropped to 66.7% at DIV11 (mean ± SD of percentage of WT, WT vs HD140Q/140Q: 100 ± 4.6 vs 66.7 ± 4.3%; Student's t test, p < 0.00001) (Fig. 1), indicating significant death of HD neurons at DIV11. Thus, a decline in GSH levels precedes neuronal death and may be a pathogenic mechanism in HD140Q/140Q mice.
HD neurons have elevated levels of reactive oxygen species
ROS are normal by-products of oxidative phosphorylation and quickly removed by antioxidants, including GSH. Having shown that HD neurons had decreased levels of GSH, we next examined whether levels of ROS are different in HD140Q/140Q neurons compared with WT neurons at DIV8. Carboxy-DCFDA is a cell-permeable nonfluorescent dye that is oxidized into highly green fluorescent DCF in the presence of oxidants, including ROS. The intensity of fluorescence of DCF generated from carboxy-DCFDA is proportional to the amount of ROS in cells (Pérez-Severiano et al., 2004; Kirkland et al., 2007; Gerstner et al., 2008; Vaughn and Deshmukh, 2008). We found that both WT and HD140Q/140Q neurons exhibited DCF signal in the somata and neurites. The DCF signal in somata was much greater than in neurites (Fig. 2). Quantification of the fluorescence intensity revealed significantly higher levels of DCF signal in the somata and neurites of HD140Q/140Q neurons compared with those of WT neurons (n = 28 WT and 24 HD neurons, mean ± SD in average fluorescence intensity in the soma per neuron, WT vs HD140Q/140Q: 147 ± 39.7 vs 192 ± 27.0; p < 0.001; n = 33 WT and 21 HD neurons, mean ± SD in average fluorescence intensity in the process per length of a process, WT vs HD140Q/140Q: 0.068 ± 0.049 vs 0.208 ± 0.123; p < 0.001). The increase in DCF fluorescence intensity in combination with lower levels of GSH in HD140Q/140Q neurons (Fig. 1) indicate that the presence of mutant Htt caused a redox imbalance—reduction in GSH and accumulation of ROS—and oxidative stress.
HD neurons are deficient in taking up cysteine
Next, we looked for a mechanism of deficient GSH in HD cortical neurons. We examined uptake of cysteine, which is required for de novo synthesis of GSH in neurons. Uptake of radiolabeled l-cysteine was conducted at room temperature for 30 s. In the presence of low concentrations of l-cysteine (0.1, 1, or 5 μm), both WT and HD140Q/140Q neuron-enriched cultures took up cysteine with a similar rate. However, HD140Q/140Q neuron-enriched cultures took up significantly less cysteine than WT neuronal cultures did with the concentration of l-cysteine at 10 and 50 μm (mean ± SD of cpm per milligram protein per second, WT vs HD140Q/140Q: 10 μm, 775 ± 86.6 vs 296 ± 43.3; 50 μm, 945 ± 88.6 vs 511 ± 44.6; p < 0.01) (Fig. 3A). Thus, insufficient supply of cysteine can contribute to low GSH levels in HD neurons.
The inhibitor TBOA markedly inhibits l-cysteine uptake in cortical neurons (Chen and Swanson, 2003). Addition of TBOA to the uptake buffer reduced uptake of l-cysteine by >90% in WT and HD cultures (supplemental Fig. S1A, available at www.jneurosci.org as supplemental material). As previously shown for primary cortical neurons, cysteine uptake through glutamate transporters is Na-dependent (Chen and Swanson, 2003). The glial glutamate–cystine exchanger is a sodium-independent transporter that exchanges intracellular glutamate for extracellular cystine and is absent in neurons (Aoyama et al., 2008). Replacing sodium with potassium in the uptake buffer reduced l-cysteine uptake by >90% in both WT and HD neuron-enriched cultures (supplemental Fig. S1A, available at www.jneurosci.org as supplemental material). Glial cells were sparse (<0.5%) in our neuron-enriched cultures. The requirement for sodium to take up cysteine suggests that the glutamate–cystine exchanger is not contributing to the deficient uptake of cysteine in HD neurons.
Cultured neurons have been reported to express glial glutamate transporter EAAT2/GLT-1 (Chen et al., 2002, 2004; Chen and Swanson, 2003). To address the contribution of EAAT2/GLT-1 to cysteine uptake in our neuronal cultures, we treated neurons with EAAT2/GLT-1 selective inhibitor, DHK, at a concentration of 100 μm as previously described (Chen and Swanson, 2003). DHK treatment caused a similar and small reduction in uptake of cysteine in WT and HD neurons compared with untreated cultures (n = 3; mean ± SD of percentage of untreatment, WT vs HD140Q/140Q: 90 ± 16 vs 88 ± 7.9%; one-tailed Student's t test, p = 0.44), suggesting that in our cultures EAAT2/GLT1 transporter does not contribute significantly to cysteine uptake. Our results with DHK differ from those of Chen and Swanson (2003) who reported that 100 μm DHK reduced cysteine uptake to 60–65% of untreated primary neuronal cultures. However, they indicated that their neuronal cultures have ∼4% glial cells, whereas our cultures have only 0.5% cells positive for glial markers. In cultures with more glia, cysteine uptake would occur through glial localized EAAT2/GLT1 transporter and be more inhibited by the presence of DHK.
Specific reduction of EAAC1 at the cell surface in HD neurons
The neuronal transporter EAAC1 was originally identified as a glutamate transporter. It was later determined that EAAC1 is the primary route for cysteine uptake (Aoyama et al., 2006), with 10- to 20-fold more affinity to cysteine than GLT-1 or GLAST (glutamate–aspartate transporter) (Zerangue and Kavanaugh, 1996). Since HD140Q/140Q cortical neurons had impaired uptake of cysteine and cortical neurons express EAAC1 (Conti et al., 1998), we investigated whether the distribution of EAAC1 at cell surfaces is altered in HD140Q/140Q neurons by an in vitro biotinylation assay, which is widely used to study cell surface levels of proteins including overexpressed EAAC1 (González et al., 2007). To inhibit endocytosis of biotinylated EAAC1, the biotinylation assay was performed on ice so that biotinylated EAAC1 maximally represented the steady-state levels of EAAC1 at the cell surface. Western blot analysis showed that the signal of intracellularly localized actin was detected mainly in the unbiotinylated samples (Fig. 3B, bottom panel, unbiotinylated), suggesting that the biotinylation reagent did not enter the cytoplasm. The levels of actin signal were comparable in WT and HD samples, indicating that similar numbers of WT and HD neurons were used in the assay. Consistent with a predominantly intracellular localization of EAAC1 (Rothstein et al., 1994; Conti et al., 1998; He et al., 2000; Nieoullon et al., 2006), we found that the majority of EAAC1 was unbiotinylated (Fig. 3B, top and middle panels). Immunoreactive signals of unbiotinylated EAAC1 in HD140Q/140Q neurons appeared comparable with those in WT neurons (Fig. 3B, top panel), suggesting that mutant Htt does not change the protein level of EAAC1. At optimal exposure of Western blot films for detecting intracellular EAAC1 (Fig. 3B, top panel), the signal for biotinylated (cell surface) EAAC1 was not detected in either WT or HD neurons. However, when the exposure time was prolonged (Fig. 3B, middle panel), levels of biotinylated (cell surface) EAAC1 were detected in both WT and HD samples but were significantly lower in HD140Q/140Q neurons than those in WT neurons. These data suggest that HD neurons have less EAAC1 distributed at the cell surface.
Endogenous EAAC1 recycles through Rab11-positive endosomes and accumulates in HD neurons
Exogenously expressed EAAC1 recycles through Rab11-positive recycling endosomes (González et al., 2007). To determine whether endogenous EAAC1 also recycles through Rab11-positive endosomes, we double labeled WT and HD140Q/140Q neurons that were cultured on glass coverslips with anti-EAAC1 and anti-Rab11 antibodies. In agreement with previous observations in normal primary hippocampal and cortical neurons (Coco et al., 1997; González et al., 2007), we found that, in both WT and HD140Q/140Q cortical neurons, immunoreactivity of endogenous EAAC1 occurred mainly at punctate structures in the cytoplasm of soma and neurites. Immunoreactive endogenous EAAC1 and endogenous Rab11 colocalized in some structures (Fig. 4), suggesting that, like overexpressed mycEAAC1, endogenous EAAC1 traffics through Rab11-positive endosomes. Immunoreactive EAAC1 was more concentrated in somata of HD140Q/140Q cortical neurons than in somata of WT neurons (Fig. 4). Quantification of fluorescent signals revealed a significant increase in somata of HD140Q/140Q cortical neurons compared with WT neurons (n = 30 WT and 30 HD140Q/140Q neurons; mean ± SD in fluorescence intensity, WT vs HD: 52 ± 3.8 vs 62 ± 4.9; Student's t test, p < 0.0001). Together with biotinylation data in Figure 3, it is suggested that, compared with WT neurons, HD neurons have less EAAC1 on the cell surface and more in the cytoplasm.
Expression of dominant-active Rab11 increases cysteine levels and intracellular GSH in HD neurons
The cell surface abundance of mycEAAC1 in glioma cells relies on normal Rab11 activity (González et al., 2007). Rab11 activation is diminished in the brain of HD140Q/140Q knock-in mice (Li et al., 2009a) and in skin fibroblasts established from HD patients (Li et al., 2009b). Therefore, we asked whether increasing Rab11 activity by expressing dominant-active Rab11Q70L could increase cell surface targeting of EAAC1 and consequently rescue the deficit in cysteine uptake and elevate levels of GSH in HD neurons. Expression of Rab11Q70L/EGFP or EGFP alone did not alter the protein level of EAAC1 in WT and HD neurons (supplemental Fig. S2, available at www.jneurosci.org as supplemental material). However, when compared with expression of EGFP alone, viral expression of dominant-active Rab11Q70L significantly increased levels of biotinylated EAAC1 in HD neurons (mean ± SD, percentage of WT-EGFP, HD-EGFP vs HD-Rab11Q70L: 57 ± 13 vs 97 ± 11; p = 0.016) (Fig. 5A). We also observed a slight but not significant increase of biotinylated EAAC1 in WT neurons after expression of Rab11Q70L compared with expression of EGFP alone (mean ± SD, percentage of WT-EGFP, WT-Rab11Q70L: 116 ± 10.8; p = 0.1244) (Fig. 5A). In normal cells, there may be a mechanism for controlling the cell surface abundance of proteins within a certain range. Nonetheless, the biotinylation data suggest that enhancement of Rab11 activity can improve cell surface targeting of EAAC1 in neurons.
In accordance with elevation of EAAC1 at cell surfaces, expression of Rab11Q70L enhanced both WT and HD140Q/140Q cortical neurons to take up cysteine in the presence of 10 or 50 μm cysteine compared with neurons without viral infection or with viral expression of EGFP alone (mean ± SD, cpm per milligram protein per second, one-way ANOVA, WT neurons: p < 0.001 for both 10 and 50 μm cysteine; HD140Q/140Q: p < 0.00001 for both 10 and 50 μm cysteine) (Fig. 5B). Cysteine uptake was elevated to the level of WT in HD neurons after viral expression of Rab11Q70L (Fig. 5B). There were also significantly higher levels of intracellular GSH in the presence of Rab11Q70L/EGFP compared with expression of EGFP alone or no viral infection in both WT neurons (sixfold increase) and HD neurons (threefold increase) (mean ± SD, nanomoles per milligram protein, one-way ANOVA, p < 0.00001 for both WT-Rab11Q70L and HD-Rab11Q70L) (Fig. 5C). We noticed that there was no linear relationship between enhanced cysteine uptake and increased GSH levels on expression of Rab11Q70L in primary neurons. This is because constitutive expression of Rab11Q70L over several days might lead to cumulative buildup of GSH inside of neurons, which cannot be reflected by a transient measurement of cysteine uptake at certain time points. Collectively, increasing the activity of Rab11 accelerated targeting of EAAC1 to the cell surface normalized the uptake of cysteine and increased levels of GSH in HD neurons.
Expression of dominant-active Rab11 enhances clearance of reactive oxygen species
Next, we examined the effect of expressing dominant-active Rab11 (Rab11Q70L) on levels of ROS in WT and HD neurons. Primary neurons were infected at DIV4 with lentivirus expressing EGFP or Rab11Q70L/EGFP and stained with carboxy-DCFDA at DIV8. The signals for EGFP were low and comparable in WT and HD neurons (data not shown). After treatment with carboxy-DCFDA, fluorescence signals increased markedly in cell bodies of WT and HD neurons. The increase in fluorescence compared with that of EGFP alone was defined as the DCF fluorescence. In the presence of EGFP alone, HD cortical neurons had higher levels of ROS (DCF signal) than WT neurons expressing EGFP alone or Rab11Q70L/EGFP (mean ± SD, fluorescence intensity in soma per neuron; one-way ANOVA, p < 0.001) (Fig. 6A,B), indicating that lentiviral expression of EGFP alone did not change the clearance of ROS in HD neurons relative to WT neurons. In the presence of Rab11Q70L expression, levels of DCF fluorescence in the somata of HD neurons significantly declined compared with HD neurons expressing EGFP alone (mean ± SD, fluorescence intensity in soma per neuron, EGFP vs Rab11Q70L: 173 ± 35.9 vs 139 ± 34.6; p < 0.01) (Fig. 6B). The ROS levels in HD neurons expressing Rab11Q70L were not significantly different compared with those in WT neurons expressing Rab11Q70L/EGFP (Fig. 6B). These data suggest that increasing Rab11 activity helps clear ROS in HD neurons.
WT neurons expressing Rab11Q70L had increased levels of DCF fluorescence compared with WT neurons expressing EGFP alone (mean ± SD, fluorescence intensity in soma per neuron, EGFP vs Rab11Q70L: 120 ± 34.0 vs 138 ± 40.8; p < 0.01) (Fig. 6). Increasing Rab11 activity in neurons that have normal Rab11 activity (WT neurons) may have different effects compared with effects in neurons that have diminished Rab11 activity.
Expression of dominant-active Rab11 promotes neuronal survival
We established that expression of Rab11Q70L rescues a deficit in cysteine uptake, increases GSH levels, and lowers ROS levels in HD neurons. We examined whether expression of dominant-active Rab11Q70L can promote survival of HD neurons at the steady state. The MTT assay revealed that viral expression of Rab11Q70L in neurons starting at DIV4 led to increased survival in HD neurons at DIV11 compared with neurons expressing EGFP alone (one-way ANOVA, p < 0.001) (Fig. 7). Notably, introducing Rab11Q70L into HD neurons restored the survival of HD neurons to levels of WT neurons (mean ± SD, percentage of WT-EGFP; one-way ANOVA, no significance) (Fig. 7). Thus, neuronal survival in primary HD neuronal cultures was improved by increasing Rab11 activity.
Indications of oxidative stress are apparent in HD, but the origins of oxidative stress are not established. GSH is a major antioxidant in the brain (Dringen, 2000). It is a tripeptide formed by glutamic acid, cysteine, and glycine. Depletion of intracellular GSH leads to death of neuronal cells in vitro (Li et al., 1997; Nicole et al., 1998; Wüllner et al., 1999), suggesting that maintenance of normal levels of GSH is essential for neuronal survival. Neurons require de novo synthesis of GSH, which involves two enzymatic reactions and the uptake of the rate-limiting precursor cysteine from the extracellular space (Dringen, 2000; Aoyama et al., 2008). Neurons take up cysteine via the neuronal Na+-dependent glutamate transporter EAAC1 (Shanker et al., 2001; Aoyama et al., 2006). In this study, we show that HD neurons had insufficient supply of the essential substrate cysteine for de novo synthesis of GSH because of aberrant trafficking of EAAC1 and accumulated intracellular ROS before the detection of cell death and/or mitochondrial dysfunction. These findings provide support for a novel mechanism of oxidative stress in HD neurons with normally expressed full-length mutant Htt protein and suggest that oxidative stress is an important pathogenic factor in HD.
Under normal circumstances, an overproduction of ROS usually elicits a compensatory increase in cysteine uptake for generation of more GSH, in which the cysteine residue provides the thiol group to inactivate ROS. In contrast, we found cysteine uptake to be reduced in HD140Q/140Q neurons compared with WT neurons, thereby impairing inactivation of ROS. Hence, our data suggest that deficient levels of GSH in HD neurons arise from an inadequate supply of cysteine for GSH synthesis, thereby causing intracellular accumulation of ROS in neurons. Of course, we cannot rule out the possibility that deficient GSH synthesis and ROS overproduction are independent events occurring contemporaneously in HD140Q/140Q neurons. Even so, deficient GSH would contribute to an increased ROS content.
In eukaryotic cells including neurons, the mitochondrion is one of the major consumers of GSH because of its role in generating ROS (Starkov, 2008). A moderate decline in GSH levels caused by a deficit in cysteine uptake could over time escalate oxidative stress in all subcellular organelles that rely on the antioxidant function of GSH, including mitochondria. We found here by MTT assays that mitochondria dysfunction occurred later than the decline in GSH levels in primary neuron-enriched cultures (Fig. 1). Mitochondria dysfunction occurs in advanced HD but not in the early stages of the disease (Guidetti et al., 2001). Rodents and primates treated with 3-nitropropionic acid (3-NP) to compromise the mitochondrial function show phenotypes reminiscent of advanced HD in humans (Beal et al., 1993; Brouillet et al., 1993, 1995). These observations suggest that impairment of mitochondrial function is important in the progression of the disease. The appearance of mitochondrial dysfunction later in the disease may be the result of a steady buildup of ROS and increase in oxidative stress from a chronic deficiency in GSH synthesis. It will be important to know in future studies what subcellular organelle(s) and molecules are affected by deficient synthesis of GSH in the early stages of HD as a way to identify therapeutic targets.
Wyttenbach et al. (2002) showed that expression of Htt exon-1 with elongated CAG repeats in cells induces an elevation in ROS levels and a decline in GSH levels. They proposed that the excess of ROS occurs because mutant Htt exon-1 causes mitochondria dysfunctions, which are major ROS generators (Benchoua et al., 2006; Solans et al., 2006; Fukui and Moraes, 2007; Oliveira et al., 2007; Starkov, 2008). They suggest that the depletion of GSH is a downstream effect of excess ROS. Based on our analysis of primary neurons that express endogenous full-length mutant Htt, we suggest that mutant Htt causes a deficiency in GSH synthesis because of insufficient supply of cysteine and thereby elevates ROS. The HD cell model used by Wyttenbach et al. (2002) may be more representative of a model of late-stage HD when mitochondria are known to be involved. In our model in primary neurons the changes observed in the presence of endogenous full-length mutant Htt might be more representative of early stages of HD.
EAAC1 is the main route for neurons to take up cysteine (Aoyama et al., 2006). Thus, any factor that causes a reduction in cell surface EAAC1 can lower cysteine uptake and subsequently affect GSH levels. It is unlikely that the decrease of cysteine uptake in HD140Q/140Q neurons results from a decrease in the overall protein level of EAAC1 because WT and HD140Q/140Q neurons had comparable levels of EAAC1 in the cytoplasm, where EAAC1 is mainly localized (Rothstein et al., 1994; Conti et al., 1998; He et al., 2000). Unlike glial glutamate transporters in glial cells, only a small proportion of EAAC1 in neurons is present at the cell surface (Rothstein et al., 1994; Conti et al., 1998; He et al., 2000). This distribution pattern of EAAC1 is similar to that of transferrin receptor, which constitutively recycles through Rab11-positive recycling endosomes. In agreement with findings using overexpressed EAAC1 (González et al., 2007), we show that the trafficking of endogenous EAAC1 is also regulated through Rab11-positive endosomes. In HD140Q/140Q cortical neurons, immunoreactive EAAC1 was accumulated in the soma compared with those of WT cortical neurons, suggesting that HD140Q/140Q neurons have a deficit in the regulation of EAAC1 trafficking.
Recent studies report that the glial transporter EAAT-2/GLT-1 is also present in neurons where it can mediate cysteine uptake (Chen et al., 2002, 2004; Chen and Swanson, 2003). We found that treatment with the EAAT2/GLT-1 selective inhibitor DHK (100 μm) had a similar and relatively modest inhibitory effect on cysteine uptake in WT and HD neurons. Thus, it is unlikely that EAAT-2/GLT-1 has a significant contribution to defective cysteine uptake in HD neurons. Furthermore, EAAT2/GLT-1 has 10- to 20-fold lower affinity for cysteine than does EAAC1 (Zerangue and Kavanaugh, 1996). Overall, our data suggest that the deficient uptake of cysteine results from improper targeting of EAAC1 to the cell surface of HD neurons.
That impaired endosomal recycling of EAAC1 compromises the capacity of HD neurons to synthesize GSH is a novel pathogenic mechanism to explain oxidative stress in HD. Interestingly, mutations in vps13A, a protein involved in endosomal recycling, causes autosomal recessive chorea acanthocytosis (Rampoldi et al., 2001; Ueno et al., 2001), a late-onset progressive neurodegenerative disorder closely resembling HD. EAAC1 trafficking requires normal Rab11 activity (González et al., 2007). We have observed defective Rab11 activation in HD (Li et al., 2009a,b). Accordingly, introducing dominant-active Rab11Q70L into HD cortical neurons normalized cysteine uptake, increased GSH synthesis, enhanced removal of ROS, and improved neuronal survival. Expression of Rab11Q70L in HD neurons could not restore cysteine uptake or GSH to the same levels in WT neurons expressing Rab11Q70L. It is possible that HD neurons may have deficits in other signaling pathways, which can regulate cell surface targeting of EAAC1 from endosomal compartments. Such signaling pathways include protein kinase A, protein kinase C, phosphatidylinositol 3-kinase and glial factors (Davis et al., 1998; Guillet et al., 2005; Akiduki and Ikemoto, 2008; Lortet et al., 2008). Future studies will be needed to know whether these signaling pathways are impaired in HD neurons.
How dominant-active Rab11Q70L provides the beneficial effect to HD neurons is not clear. One possibility is that enhancement of Rab11 activity accelerated the overall recycling rate of EAAC1 (Fig. 5A). It is known that Rab11 regulates trafficking of a variety of other critical proteins and cholesterol, so that other Rab11 effects might be involved. Brain proteins that are known Rab11-dependent cargos include transferrin receptor (iron uptake), brain-enriched Na+/H+ exchanger NHE5 (cell volume and pH maintenance) (Diering et al., 2009), glycine transporter GLYT2 (glycine uptake) (Núñez et al., 2009), glucose transporter 1 (glucose uptake) (Wieman et al., 2007), and AMPA receptor (long-term potentiation) (Park et al., 2004). The presence of Rab11Q70L in HD neurons may also improve functions related to these proteins.
We found that HD cortical neurons had a deficit in cysteine uptake. The involvement of insufficient cysteine supply in HD pathogenesis has indirect support. Treatment of R6/2 mice with cystamine significantly improves both motor performance and survival (Dedeoglu et al., 2002; Bailey and Johnson, 2006). Cystamine is an organic disulfide formed from the dimer of cysteine, cystine, through decarboxylation. Recent studies show that administration of cystamine raises cysteine levels in the brain of both R6/2 mice (Fox et al., 2004) and transgenic mice expressing full-length mutant Htt (YAC128 mice) (Pinto et al., 2005). Cystamine also protects 3-NP-induced mitochondrial dysfunction in immortal Hdh111Q striatal cells (Mao et al., 2006). Furthermore, treatment of rats with N-acetylcysteine, a membrane-permeable form of cysteine, before administration of 3-NP significantly reduces 3-NP-induced striatal injury (Fontaine et al., 2000). Regardless of the mechanisms of action, approaches that increase cysteine levels in neurons may be effective for HD.
In summary, our study supports the idea that the presence of mutant Htt causes a deficit in targeting of the neuronal glutamate/cysteine transporter EAAC1 to the cell surface and this in turn impairs uptake of extracellular cysteine and hampers generation of glutathione to handle ROS in neurons. Given that EAAC1 also mediates uptake of glutamate, we predict that HD GABAergic neurons, including medium spiny neurons, may be defective in synthesis of inhibitory neurotransmitter GABA with glutamate as the substrate. Our study suggests that impaired recycling of EAAC1 is attributable to deficient Rab11 activity. We have previously shown that mutant Htt inhibits guanine nucleotide exchange on Rab11 in brains of young HD knock-in mice (Li et al., 2009a). Manipulation of Rab11 activity may be beneficial for slowing the progression of HD. Future studies will be necessary to investigate ways to enhance Rab11 activity in HD neurons in vivo.
This work was supported by National Institutes of Health Grant NS038194 (N.A.) and the Huntington's Disease Society of America (M.D.). X.L. was supported by a John J. Wasmuth Postdoctoral Fellowship from the Hereditary Disease Foundation. A.V. was supported by fellowships from Fundacion Mexico en Harvard, A.C., and the Hereditary Disease Foundation. This publication was made possible by Grant 5P30 DK32520 from the National Institute of Diabetes and Digestive and Kidney Diseases. We thank Dr. Brian Healy of the Massachusetts General Hospital Biostatistics Center for his advice on the data analysis.
The authors declare no competing financial interests.
- Correspondence should be addressed to either Xueyi Li or Marian DiFiglia, Cellular Neurobiology Laboratory and Department of Neurology, Massachusetts General Hospital and Harvard Medical School, Charlestown, MA 02129. or