Machado–Joseph disease, also known as spinocerebellar ataxia type 3, is the most common of the dominantly inherited ataxias worldwide and is characterized by mutant ataxin-3 misfolding, intracellular accumulation of aggregates and neuronal degeneration. Here we investigated the implication of autophagy, the major pathway for organelle and protein turnover, in the accumulation of mutant ataxin-3 aggregates and neurodegeneration found in Machado–Joseph disease and we assessed whether specific stimulation of this pathway could mitigate the disease. Using tissue from patients with Machado–Joseph disease, transgenic mice and a lentiviral-based rat model, we found an abnormal expression of endogenous autophagic markers, accumulation of autophagosomes and decreased levels of beclin-1, a crucial protein in the early nucleation step of autophagy. Lentiviral vector-mediated overexpression of beclin-1 led to stimulation of autophagic flux, mutant ataxin-3 clearance and overall neuroprotective effects in neuronal cultures and in a lentiviral-based rat model of Machado–Joseph disease. These data demonstrate that autophagy is a key degradation pathway, with beclin-1 playing a significant role in alleviating Machado–Joseph disease pathogenesis.
Machado–Joseph disease, also known as spinocerebellar ataxia type 3, is an autosomal dominant neurodegenerative disorder caused by a CAG trinucleotide repeat expansion within the coding region of the MJD1 gene (Kawaguchi et al., 1994; Zoghbi, 2000). This gene encodes ataxin-3, a polyubiquitin-binding protein whose physiological function has been linked to ubiquitin-mediated proteolysis (Doss-Pepe et al., 2003; Chai et al., 2004). The polyglutamine tract confers a toxic gain-of-function to the mutant protein leading to neuronal dysfunction and cell death (Zoghbi, 2000; Duenas et al., 2006). The mutant protein accumulates and forms neuronal intranuclear inclusions. If these inclusions are hallmarks of the disease (Schmidt et al., 1998), their exact role in the polyglutamine pathology is still unclear.
There is strong evidence that proteins with a mutant polyglutamine tract are inefficiently degraded by the ubiquitin-proteosome system but may be degraded by macroautophagy (hereafter called autophagy), a mechanism with a crucial role in degradation of insoluble aggregate-prone proteins and essential for neuronal survival (Cuervo, 2004; Williams et al., 2006). Mice lacking either one of the autophagic genes 5 or 7 (ATG5/ATG7) develop neurodegeneration and cytoplasmic ubiquitinated aggregates, similarly to what happens in neurodegenerative disorders (Hara et al., 2006; Komatsu et al., 2006). Furthermore, the activity of this pathway decreases with ageing (Cuervo, 2004; Vellai, 2009) and is affected in several neurodegenerative diseases (Shibata et al., 2006; Pickford et al., 2008; Crews et al., 2010). Although autophagy was initially considered as a non-specific ‘in-bulk’ clearance mechanism, recent evidence supports selectivity of this process, with specific molecules recognizing polyubiquitination as a signal for cargo sequestration and autophagic clearance. The sequestosome 1/p62 protein (p62) was the first cargo-recognizing molecule identified, with a domain that binds both cargo and autophagic machinery (Bjorkoy et al., 2005; Kirkin et al., 2009; Ichimura and Komatsu, 2010).
Neuronal activation of autophagy can result from upregulation of the autophagic protein 6/beclin-1 (Atg6/beclin-1) or inhibition of the mammalian target of rapamycin. Beclin-1 is an evolutionarily conserved protein that is essential for the nucleation step of autophagy, via its interaction with the class III phosphatidylinositol-3-kinase/Vps34 complex (Zeng et al., 2006). Beclin-1 is described to be a crucial and limiting protein for the autophagy pathway. Inactivation of the Caenorhabditis elegans beclin-1 gene causes apoptosis (Takacs-Vellai et al., 2005) and increases aggregation and neurodegeneration upon mutant huntingtin expression (Jia et al., 2007). Furthermore, deletion of one of the beclin-1 alleles in mice is sufficient to cause impaired autophagy and neurodegeneration (Pickford et al., 2008). A decrease in beclin-1 levels has been observed in aged brains and in Alzheimer’s and Huntington’s diseases (Shibata et al., 2006; Pickford et al., 2008) while its overexpression has been reported to reduce aggregation and improve neuronal function (Pickford et al., 2008; Spencer et al., 2009). No data are available regarding beclin-1 levels and modelling in Machado–Joseph’s disease. Therefore, it is crucial to investigate if the autophagy pathway is impaired in Machado–Joseph disease and if soluble and aggregated forms of mutant ataxin-3 can be selectively degraded by this pathway.
In the present study, we evaluated alterations in the molecular components of autophagy in tissue from patients with Machado–Joseph disease and in vitro and in vivo models of the disease. We found accumulation of proteins related with the autophagy pathway and autophagosomes in brains of patients with Machado–Joseph disease, as well as decreased beclin-1 levels in tissue from patients with Machado–Joseph disease and rodent models. Furthermore, mutant ataxin-3 aggregates were shown to be recognized by the autophagic receptor p62 while beclin-1 overexpression improved autophagosomal flux, mutant ataxin-3 clearance and neuronal function. Our data provide evidence that the autophagy pathway is altered in the brain of patients with Machado–Joseph disease and that beclin-1 overexpression is able to stimulate the autophagic flux, leading to a selective clearance of mutant ataxin-3.
Materials and methods
Human brain tissue
Post-mortem putamen tissue from three patients with clinically and genetically confirmed Machado–Joseph disease (Patient 1: 62 years, 26/70 CAG repeats; Patient 2: 53 years, 23/69 CAG repeats; Patient 3: 62 years, 22/74 CAG repeats) and midfrontal cortical tissue (69 years, onset at 56) were obtained from the ‘Tissue Donation Programme of the National Ataxia Foundation, Minneapolis, MN, USA’ (VA Medical Centre, Albany, NY, USA). Controls with no evidence of neurological disease (Control 1: 55 years; Control 2: 56 years; Control 3: 67 years) were obtained from Neurology and Pathology Services of University Hospital of Coimbra, Portugal. The tissue was fresh when dissected, placed in 10% neutral buffered formalin and kept at 4°C.
Transgenic mice tissue
Brains of homozygous transgenic mice (n = 6), 10–13 weeks old (late stage of disease), expressing human full-length mutant ataxin-3 with 71 glutamines (Q71-C mice) (Goti et al., 2004) and wild-type C57BL/6 mice age matched (n = 6) were dissected fresh and kept at –80°C. Small punches of the striatum were collected with a Harris Uni-Core pen, with 2.0 mm diameter (Ted Pella, Inc).
The complementary DNAs encoding for human beclin-1 (Liang et al., 1999) and rat light chain 3 (Kabeya et al., 2000) fused to the enhanced green fluorescent protein were inserted downstream from the mouse phosphoglycerate kinase 1 promoter in a self-inactivating lentiviral transfer vector (Deglon et al., 2000). Viral vectors encoding for the human beclin-1, enhanced green fluorescent protein-light chain 3, human full-length wild-type ataxin-3 with 27 glutamines (Alves et al., 2008b) and human full length mutant ataxin-3 with 72 glutamines (Alves et al., 2008b), were produced in human embryonic kidney 293 T cells using a four-plasmid system described previously (de Almeida et al., 2002).
In vivo experiments
Adult male Wistar rats (Charles River Laboratories, Inc.), weighing 200–220 g were used. The animals were housed in a temperature-controlled room and maintained on a 12 h light/dark cycle. Food and water were available ad libitum. The experiments were carried out in accordance with the European Community Council directive (86/609/EEC) for the care and use of laboratory animals. For the stereotaxic injection of lentiviral vectors, concentrated viral stocks were thawed on ice and resuspended by vortexing. The rats were anaesthetized with ketamine (75 mg/kg intraperitoneally) and xylazine (10 mg/kg intraperitoneally). Particle content of lentiviral vectors was matched to 200 μg of p24/ml. The surgical procedure was performed as described previously (Alves et al., 2008b). The animals received a single injection of lentivirus in each hemisphere at the following coordinates: 0.7 mm rostral to bregma, 3.0 mm lateral to midline and 5.5 mm ventral from the skull surface, with the mouth bar set at 0.0 mm. In all experiments ipsilateral was compared to contralateral hemisphere and only experiments performed with the same batches of virus were compared.
Human fibroblasts culture
Human fibroblast cells were obtained from Coriell cell repositories and skin biopsy (control: 38 years, not affected; Machado–Joseph disease 1: 38 years, 74 CAG repeats; Machado–Joseph disease 2: 28 years, 82 CAG repeats). Cells were kept in culture in Dulbecco’s modified Eagle’s medium supplemented with 10% bovine serum, 1% non-essential amino acids, 2 mM L-glutamine, 100 U/ml penicillin and 100 µg/ml streptomycin (Gibco) at 37°C in 5% CO2/air atmosphere.
Striatal primary cells culture
Timed-pregnant Sprague–Dawley rats (Charles Rivers Laboratories) were killed by CO2 inhalation, and embryos (Embryonic Day 15) were collected in a Petri dish and placed on ice. Dissections were performed under a stereomicroscope and ganglionic eminences were isolated. Culture and infection protocol was performed as described elsewhere (Zala et al., 2005). Analysis was performed at 2 weeks post-infection (18 days in vitro).
Neuroblastoma cell culture
Mouse neuroblastoma cell line (Neuro-2 a cells) obtained from the American Type Culture Collection cell biology bank (CCL-131) were incubated in Dulbecco’s modified Eagle’s medium supplemented with 10% foetal bovine serum, 100 U/ml penicillin and 100 µg/ml streptomycin (Gibco) (complete medium) at 37°C in 5%CO2/air atmosphere. Cells were plated and lipofectamine transfected or infected with lentiviral particles expressing enhanced green fluorescent protein-light chain 3 or ataxin-3 and beclin-1/control, at the ratio of 10 ng of p24 antigen/105 cells 24 and 48 h after plating. At 2 weeks post-infection, cells were either lysed for western blot processing or lipofectamine transfected for the flow cytometry experiments.
Antibodies for immunohistochemistry and western blotting
The following primary antibodies were used: mouse monoclonal anti-ataxin-3 (1H9, Chemicon; 1:5000; overnight, 4°C); rabbit polyclonal anti-p62 antibody (Abgent; 1:50; overnight-48 h, 4°C); rabbit polyclonal anti-Atg16L antibody (Abgent; 1:50; overnight-48 h, 4°C); rabbit polyclonal anti-light chain 3 antibody (Abgent; 1:50; overnight-48 h, 4°C); rabbit polyclonal anti-ubiquitin (Dakocytomation; 1:1000; overnight 4°C); mouse monoclonal anti-beclin-1 (BD Biosciences; 1:1000; overnight, 4°C); mouse monoclonal anti-β-actin antibody (1:5000, Sigma); rabbit polyclonal anti-dopamine-and-cyclic Adenosine 5′-monophosphate-regulated phosphoprotein of 32 kDa (DARPP-32) (Chemicon; 1:5000, overnight 4°C); mouse monoclonal anti-neuronal nuclei protein antibody (Chemicon; 1:1000; overnight 4°C).
The human brain tissue was fresh when dissected and placed in cold 10% neutral buffered formalin fixative solution. Rats were killed by sodium pentobarbital overdose, transcardially perfused with 0.1 M phosphate buffer solution and a 4% paraformaldehyde fixative solution (Fluka, Sigma) followed by brain removal. Both tissues were cryoprotected in 25% sucrose-0.1 M phosphate buffer solution solution for 48 h, dry ice-frozen and cut on a cryostat-microtome (Leica CM3050S) in 25–30 µm coronal sections. Slices were collected and stored in 48-well trays (Corning Inc.), free-floating in 0.1 M phosphate buffer solution supplemented with 0.12 µmol/l sodium azide. The plates were stored at 4°C until immunohistochemical processing.
The immunohistochemical procedure was initiated by incubating brain sections in phenylhydrazine diluted in phosphate buffer solution (1:1000; 15 min, 37°C; light imaging only), followed by a Tris-buffered saline pH 9 antigen retrieval method (30 min, 95°C, human tissue only), blocking in 10% normal goat serum in 0.1% Triton X-phosphate buffered solution (1 h, room temperature) and incubation with the respective primary and secondary antibodies diluted in blocking solution. For light imaging, the secondary antibody used was biotinylated and followed a reaction with the Vectastain elite avidin-biotin-peroxidase kit and by 3,3′-diaminobenzidine substrate (both from Vector Laboratories). Premounted sections were then counterstained with cresyl violet (2 min in cresyl violet solution followed by differentiation in ethanol 70%) or directly dehydrated in toluene and coverslipped with Eukitt mounting medium (O. Kindler GmbH and Co.). Light images were acquired with a Zeiss Axiovert 200 imaging microscope. For the fluorescent imaging, the secondary antibody used was coupled to a fluorophore (Alexa Fluor, Invitrogen) and followed a nuclei staining reaction with 4′,6-diamino-2-phenylindole (5 min, room temperature). Images were acquired with a confocal LSM Zeiss. All immunohistochemical analyses were performed in triplicate and using at least three sections of each experimental set.
Immunohistochemistry quantitative analysis
The human brain tissue analysis for the p62-, Atg16L- and light chain 3-positive cells were carried out counting for each staining, 10 fields (×100 magnification) of three different sections. Only cytoplasmic puncta-like immunoreactivity was scored positive. For each field and staining, normalization was relative to total number of nuclei (cresyl staining).
The quantitative analysis of enhanced green fluorescent protein-light chain 3 positive puncta was carried out counting eight fields (×100 magnification) of three different brain sections from four animals sacrificed at 12 weeks post-injection (n = 4). Hemispheres injected with mutant ataxin-3 and wild-type ataxin-3 were counted and only cytoplasmic puncta-like immunoreactivity was scored positive. For each field, normalization was relative to total number of nuclei (4′,6-diamino-2-phenylindole staining).
The quantification of ataxin-3 and ubiquitin-positive inclusions was performed by scanning 8–10 coronal sections spread over the anterior–posterior extent of the striatum (inter-section distance: 200 µm), using a ×10 objective on a Zeiss Axioplan2 Imaging microscope motorized for x, y and z displacements, and an image acquisition and analysis software (Mercator, Explora Nova). For each animal, the absolute number of inclusions in the striatum was calculated as described previously (Alves et al., 2008a) . For the quantification of DARPP-32 and neuronal nuclei protein depleted volumes, scanning of 8–10 coronal sections spread over the anterior–posterior extent of the striatum (inter-section distance: 200 µm), using a MCID™ acquisition and software analysis apparatus was performed. The volume was estimated as described elsewhere (Alves et al., 2008a) and data were expressed as the evaluated DARPP-32 or neuronal nuclei protein depleted volume (mm3).
Protein extraction and western blotting
For the protein extraction protocol, brain tissue and cells were lysed in radioimmunoprecipitation assay-buffer solution (50 mM Tris HCl, pH 8, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulphate) containing proteases inhibitors (Roche diagnostics GmbH) followed by a 4 s ultra-sound chase (1 pulse/s). Total protein lysates were stored at –80°C and protein concentration was determined with the Bradford protein assay (BioRad). Depending on the analysis, 20 (Neuro-2 a), 30 (transgenic mice brains and striatal primary cells) and 60 (human fibroblasts) µg of protein extract were resolved in sodium dodecyl sulphate-polyacrylamide gels (4% stacking, 12% running; 20–80%) with exception of striatal primary cells analysis that were resolved in different sodium dodecyl sulphate-polyacrylamide gels (4% stacking, 6% running; 50–50%). The proteins were transferred onto polyvinylidene difluoride membranes (GE Healthcare) according to standard protocols. The immunobloting procedure was performed as described previously (Alves et al., 2008a) with the respective primary antibodies, followed by incubation with the respective alkaline phosphatase-linked secondary antibody. Bands were visualized with Enhanced Chemifluorescence substrate (ECF) (GE Healthcare) and chemifluorescence imaging (VersaDoc Imaging System Model 3000, Bio-Rad). The western blot analyses were performed at least in triplicate for each experimental set. Membranes were stripped using a 0.1 M glycine pH 2.3 (30 min, room temperature) and reprobed with mouse monoclonal anti-β-actin antibody (1:5000, Sigma). Densitometric analysis was carried out using ImageJ software (NIH).
Flow cytometry analysis
For the transient transfection experiments, Neuro-2 a cells were plated and transfected with lipofectamine (Invitrogen) 24 h after plating. Lipofectamine was used according to the manufacturer’s protocol, and cells transfected with enhanced green fluorescent protein-light chain 3 with wild-type or mutant ataxin-3 plus beclin-1 or cytoplasmic LacZ as control (1:1:1 DNA ratio). After 24, 48, 72 and 96 h post-transfection, cells were harvested with trypsin/ethylenediaminetetraacetic acid (Gibco), washed and resuspended in 0.5–1.0 ml of complete medium without phenol red for fluorescence activated cell sorter analysis. For the stable enhanced green fluorescent protein-light chain 3 expression analysis, cells were plated, infected with enhanced green fluorescent protein-light chain 3 lentiviral particles and transfected with lipofectamine for the ataxin-3 and beclin-1/control constructs (1 : 1 DNA ratio) at 2 weeks post-infection. The analysis was performed 48 h post-transfection. For the starvation conditions, cells were incubated for 3 h with Hanks buffer saline solution modified to have 1.8 g/l of glucose (Sigma). For all conditions, cells were harvested with trypsin/ethylenediaminetetraacetic acid (Gibco), washed and resuspended in 0.5–1.0 ml of Hanks buffer saline solution for fluorescence activated cell sorter analysis. Samples were kept on ice until use and analysis of 50 000 viable cells per sample was performed on a fluorescence activated cell sorting flow cytometer (Becton Dickinson), where the data of viable cell counts was plotted as enhanced green fluorescent protein fluorescence intensity (Fl1 channel, 530 ± 30 nm). The flow cytometry analysis was performed in triplicate for each experimental set.
Molecular components of the autophagy pathway abnormally accumulate in Machado–Joseph disease brain tissue
An abnormal number of autophagosomes within cells has been reported as a hallmark of altered autophagy in the brain (Kegel et al., 2000; Nixon et al., 2005). For this reason, we investigated if components of the autophagy pathway were abnormally accumulating in post-mortem brain tissue of patients with Machado–Joseph disease. The immunoreactivity for autophagy components in an affected brain area—the putamen—of three autopsy-confirmed patients with Machado–Joseph disease (Patients 1–3, Fig. 1) was compared to three age-matched controls (Controls 1–3, Fig. 1) and an unaffected brain area, midfrontal cortex, of an autopsy-confirmed patient with Machado–Joseph disease (Fig. 1).
We found high immunoreactivity for mutant ataxin-3 aggregates (Fig. 1A) and for components involved in sequential steps of the autophagy pathway (Fig. 1B–E) in the putamen of patients with Machado–Joseph disease but not in controls. A strong nuclear and cytoplasmic puncta-like immunoreactivity for the sequestosome 1/p62 protein (Fig. 1B) was observed in the putamen of patients with Machado–Joseph disease but not in controls (n = 3, Fig. 1E; 38.51 ± 0.21 versus 10.20 ± 3.85 in controls; P = 0.0029). This protein binds ubiquitinated proteins, redirects them to autophagosomes and interacts with the light chain 3-II protein, before being degraded with other cellular components (Bjorkoy et al., 2005; Kirkin et al., 2009; Ichimura and Komatsu, 2010). A robust cytoplasmic puncta-like immunoreactivity was also observed for the autophagic protein 16, (Atg16L) protein (Fig. 1C), which integrates the Atg12-Atg5-Atg16L complex, in the putamen of patients with Machado–Joseph disease but not in controls (n = 3, Fig. 1E; 44.79 ± 1.42 versus 8.75 ± 3.05 in controls; P = 0.0004). This complex is important for the lipidation of light chain 3 and is present in the autophagosomal membrane during the elongation phase, being a marker of immature autophagosomes (Ichimura and Komatsu, 2010). Finally, we analysed the microtubule-associated protein 1 light chain 3 (LC3) (Fig. 1D). This protein is an essential component of the autophagosomal membrane and a marker of all autophagosomal structures, immature and mature autophagosomes, as well as autophagolysosomes. A clear cytoplasmic-puncta-like immunoreactivity was observed in the putamen of patients with Machado–Joseph disease but not in controls (n = 3, Fig. 1E; 41.01 ± 5.69 versus 6.25 ± 0.72 in controls; P = 0.003).
These data suggest impairment of the autophagic pathway, since autophagosomes are efficiently cleared in the central nervous system (Boland et al., 2008) and accumulation of p62 and light chain 3-positive autophagosomes has been associated with impairment in the trafficking to lysosomes and protein degradation (Bjorkoy et al., 2005; Komatsu et al., 2007; Pankiv et al., 2007).
Early activation of the p62 autophagic receptor and late accumulation of light chain 3-positive autophagosomes
Autophagy has been described as an essential cellular mechanism for the clearance of ubiquitin-positive inclusions (Hara et al., 2006; Komatsu et al., 2006). Recently, p62/SQSTM1 was identified as an important constituent of ubiquitinated protein aggregates, allowing its selective degradation through the autophagy pathway (Bjorkoy et al., 2005; Pankiv et al., 2007). Therefore, to investigate whether mutant ataxin-3 inclusions were targeted for autophagy-mediated degradation, ubiquitin (Fig. 2A, upper panel) and p62 (Fig. 2A, lower panel), immunostaining was performed in rats injected with mutant ataxin-3 or wild-type ataxin-3 at 4 (n = 4, Fig. 2A) and 8 weeks (n = 4, data not shown) post-injection. At 4 weeks, mutant ataxin-3 inclusions had already co-localized with ubiquitin in the nucleus. Importantly, the cytoplasmic autophagic receptor p62 was found translocated to the nucleus and co-localizing with ubiquitinated mutant ataxin-3 inclusions (Fig. 2A, lower panel).
In addition, western blot analysis of striatal samples from late stage Machado–Joseph disease transgenic mice (Goti et al., 2004) (Q71-C mice, n = 6, Fig. 2B–D) and age-matched wild-type animals (wild-type mice, n = 6, Fig. 2B–D) revealed that p62-positive aggregates (stacking gel, molecular weight >250 kDa) were significantly increased in Q71-C mice (n = 6, Fig. 2C; 0.31 ± 0.009) when compared to controls (wild-type mice, n = 6, Fig. 2C; 0.28 ± 0.005; P = 0.019). This increase of p62 aggregated form was accompanied by a significant reduction in p62 soluble levels (running gel, molecular weight: 50 kDa; n = 6, Fig. 2C; 0.66 ± 0.020 versus 0.76 ± 0.014 in controls; P = 0.003) as well as an increase of light chain 3-II autophagosomal marker (n = 6, Fig. 2D; 1.99 ± 0.061 versus 1.59 ± 0.12 in controls; P = 0.014 relative to light chain 3-I; 0.93 ± 0.021 versus 0.74 ± 0.066 in controls; P = 0.017 relative to actin). To further assess autophagosomal trafficking in Machado–Joseph disease, we cloned the autophagosomal marker enhanced green fluorescent protein-light chain 3 in the lentiviral vector (Supplementary Fig. 1C, E). This vector was co-injected with the mutant or wild-type ataxin-3-expressing lentiviral vectors in the rat brain. Histochemical analyses were performed at 1 (n = 2, data not shown), 4 (n = 7, data not shown), 8 (n = 7, data not shown) and 12 weeks post-injection (n = 4, Fig. 2E and F) by staining for nuclei (4′,6-diamidino-2-phenylindole, blue), ataxin-3 (red) and direct visualization of enhanced green fluorescent protein (light chain 3, green). Cytoplasmic autophagosomal accumulation (arrows) was observed only at 12 weeks post-injection (Fig. 2E and F), a late stage of disease in this model, in a small but highly significant number of cells expressing mutant ataxin-3 (n = 4, Fig. 2F; 4.08 ± 0.13), while almost no accumulation was observed upon wild-type ataxin-3 expression (n = 4, Fig. 2F; 0.48 ± 0.044; P = 0.0007). A high co-localization of light chain 3 protein with perinuclear ataxin-3 aggregates was also found, starting at 4 weeks (data not shown) up to 12 weeks (Fig. 2E, arrow heads). Quantitative real-time polymerase chain reaction confirmed that wild-type and mutant ataxin-3 s were expressed at similar levels (Supplementary Fig 2C).
Altogether, these results suggest that ubiquitinated ataxin-3 inclusions are recognized by the autophagic receptor p62 and therefore by the autophagic machinery in an early stage of disease. Moreover, impairments in this pathway are apparent at late stages of disease, possibly due to saturation in the degradative capacity of the autophagic machinery to clear mutant proteins.
Levels of the autophagic protein beclin-1 are reduced in fibroblasts from patients with Machado–Joseph disease and rodent models
To further investigate the basis of the late stage-autophagy impairment, we evaluated beclin-1 endogenous levels in a late stage Machado–Joseph disease transgenic mouse model (n = 6, Fig. 3A and B), in fibroblasts derived from patients with Machado–Joseph disease (n = 2, Fig. 3C and D) and in the Machado–Joseph disease rat model at 1 (n = 2, data not shown), 4 (n = 4, data not shown) and 8 weeks post-injection (n = 4, Fig. 3E). Beclin-1 levels were significantly decreased in transgenic mutant ataxin-3 mice (Q71-C mice, n = 6, Fig. 3A and B; 0.89 ± 0.011) compared to control (wild-type mice, n = 6, Fig. 3A and B; 1.053 ± 0.026; P = 0.0002). In addition, beclin-1 levels were also significantly reduced in fibroblasts from patients with Machado–Joseph disease (Fig. 3C and D; control: 1.15 ± 0.038; Machado–Joseph disease 1: 0.86 ± 0.087; Machado–Joseph disease 2: 0.69 ± 0.05; *P < 0.05; **P < 0.01). In the lentiviral-based Machado–Joseph disease rat, beclin-1 was trapped in nuclear mutant ataxin-3 inclusions at 8 weeks post-injection (Fig. 3E), while in controls (wild-type ataxin-3 expression; Fig. 3E) beclin-1 had a widespread cytoplasmic distribution within the cell. This could indicate that along with the progression of the disease, beclin-1 becomes trapped in the insoluble mutant ataxin-3 inclusions, while its soluble and functional levels progressively decrease. These data suggest that reduced levels of beclin-1 protein are present in Machado–Joseph disease pathogenesis and may contribute to the late-stage associated impairment of autophagy and progression of the disease.
Beclin-1 overexpression improves clearance of aggregated, oligomeric and soluble mutant ataxin-3
These data prompted us to investigate whether the upregulation of the autophagy pathway, through increase in the levels of beclin-1, could improve mutant ataxin-3 clearance and delay the development of neuropathology. For that purpose, we co-expressed human beclin-1 (Supplementary Fig. 1) and mutant ataxin-3 in primary striatal cells (Supplementary Fig. 2) and in the rat model. Western blot analysis of primary striatal cells infected with lentiviral vectors expressing wild-type ataxin-3 and mutant ataxin-3 alone and co-infected with human beclin-1 at 2 weeks post-infection showed three different species of ataxin-3: (i) insoluble aggregates; (ii) soluble oligomers; and (iii) soluble mutant and wild-type ataxin-3. Strikingly, radioimmunoprecipitation assay buffer-insoluble ataxin-3 aggregates (stacking gel, molecular weight >250 kDa) were only visible in samples containing mutant ataxin-3 (Fig. 4A) and not in samples co-expressing mutant ataxin-3 and beclin-1 (Fig. 4A). Optical densitometry analysis of the membrane (Fig. 4C) revealed a drastic reduction of aggregated ataxin-3 in beclin-1-treated (0.074 ± 0.0034) as compared to non-treated mutant ataxin-3 cells (0.75 ± 0.18; P = 0.0039). A similar decrease of ataxin-3 oligomers (stacking-running gel interface, #, molecular weight: 250 kDa) (Fig. 4A and C) in beclin-1 treated (0.079 ± 0.0050) compared to non-treated samples (0.86 ± 0.23; P = 0.0065) was observed. Finally, beclin-1 led to a significant decrease of radioimmunoprecipitation assay buffer soluble mutant ataxin-3 (running gel, molecular weight: 64 kDa; Fig. 4A and C; 0.85 ± 0.036 versus 2.85 ± 0.65 in control; P = 0.012). Non-overexposed membranes were used for quantification (Supplementary Fig. 2B). Probing of the membranes with an ubiquitin antibody (Fig. 4B) and quantification (Fig. 4D) revealed the presence of ubiquitin-positive aggregates in the stacking fraction of the gel (top to #; 2.68 ± 0.40) that were reduced in cells co-expressing beclin-1 (0.41 ± 0.032; P = 0.0002).
In the Machado–Joseph disease rat model (Fig. 4E–H), immunohistochemistry and quantitative analysis for ataxin-3 and ubiquitin-positive inclusions (Fig. 4F–H) revealed a significant beclin-1-mediated reduction at 4 weeks in the number of ataxin-3 inclusions (n = 8, Fig. 4F; Fig. 4G left; 121 683 ± 21 341 versus 185 866 ± 32 343 inclusions in controls; P = 0.039) as well as in the number of ubiquitin inclusions (n = 8, Fig. 4F; Fig. 4H left; 52 629 ± 7762 versus 77 077 ± 11 901 inclusions in controls; P = 0.042). This beclin-1-mediated clearance was even more pronounced at 8 weeks post-injection (n = 8, Fig. 4F and G; ataxin-3 inclusions 133 947 ± 24 883 versus 218 062 ± 21 594 inclusions in controls; P = 0.0033) (Fig. 4F and H; ubiquitin inclusions 62 066 ± 11 013 versus 99 789 ± 8891 inclusions in controls; P = 0.0006). Quantitative real-time polymerase chain reaction confirmed that mutant ataxin-3 and mutant ataxin-3 plus Beclin-1 striatal cells express ataxin-3 at similar levels (Supplementary Fig. 2D). Together these findings indicate that beclin-1 was able to clear soluble, oligomeric and aggregated toxic species of ataxin-3 over time.
Beclin-1 overexpression reduces neuronal dysfunction mediated by mutant ataxin-3
In view of these results, we next investigated if clearance of ataxin-3 different species would result in neuroprotective effects. Given that lentiviral expression of mutant ataxin-3 produces a depletion of neuronal DARPP-32 marker and neuronal nuclei protein that can be precisely quantified (Alves et al., 2008a), immunohistochemistry for DARPP-32 (Fig. 5A) and neuronal nuclei protein (Fig. 5A) was performed. Beclin-1 overexpression led to a significant reduction of neuronal dysfunction as revealed by the decrease at 4 weeks (Fig. 5B and C) of DARPP-32 (n = 8, Fig. 5B; 1.25 ± 0.17 versus 1.94 ± 0.29 mm3 in controls; P = 0.0019) and neuronal nuclei protein-depleted volumes (n = 8, Fig. 5C; 0.80 ± 0.11 versus 1.63 ± 0.30 mm3 in controls; P = 0.0092). Importantly, this protective effect was conserved at 8 weeks post-injection as revealed by DARPP-32 (n = 8, Fig. 5B; 1.82 ± 0.27 versus 3.14 ± 0.40 mm3 in controls; P = 0.0003) and neuronal nuclei protein immunoreactivity (n = 8, Fig. 5D; 1.26 ± 0.16 versus 2.38 ± 0.11 mm3 in controls; P = 0.000007). In addition, beclin-1 overexpression was associated with reduced microglial (Supplementary Fig. 4A; 4 weeks) and astroglial activations (Supplementary Fig. 4A; 8 weeks). Unexpectedly, beclin-1 silencing in the brain of adult wild-type rats expressing mutant ataxin-3 did not aggravate mutant ataxin-3-induced neuropathology (Supplementary Fig. 3). This might be due to an insufficient downregulation of protein levels (67% in primary striatal cultures; Supplementary Figs 2 and 3) which, within the short period of time (8 weeks), failed to produce significant exacerbation of Machado–Joseph disease pathology.
Overall, these results indicate that beclin-1 overexpression mitigates Machado–Joseph disease neuropathology.
Beclin-1 improves autophagosomal flux and ataxin-3 clearance in neuroblastoma cells expressing mutant ataxin-3
To investigate whether beclin-1 effects were mediated through activation of the autophagic pathway, we performed an autophagosomal flux analysis in neuroblastoma cells (Neuro-2 a) transiently (Fig. 6B–E) and stably expressing enhanced green fluorescent protein-light chain 3 (Fig. 6F). The enhanced green fluorescent protein-light chain 3 protein acts similarly to the endogenous light chain 3 and translocates from the cytoplasm (light chain 3-I) to the autophagosomal membrane (light chain 3-II) where it remains until complete fusion to lysosomes (Bampton et al., 2005; Fig. 6A, Step 1–2). Once inside the lysosomes, the acidic pH will block the emission of enhanced green fluorescent protein fluorescence (Fig. 6A, Step 3) (Mizushima et al., 2010). Therefore, an increase in the autophagosomal flux leads to an increased enhanced green fluorescent protein-light chain 3 degradation, and concomitant reduction of its fluorescence emission (Shvets et al., 2008).
For the transient enhanced green fluorescent protein-light chain expression flow cytometry analysis, Neuro2a cells were transiently transfected with enhanced green fluorescent protein-light chain 3 and ataxin-3 in the presence of beclin-1/control vectors. Flow cytometry was performed on live cells 24, 48, 72 and 96 h post-transfection. Enhanced green fluorescent protein-light chain fluorescence emission progressively increased reaching its maximum at 48 h post-transfection. A similar fluorescence pattern was observed between control-treated ataxin-3 wild-type (Fig. 6B, orange) and mutant ataxin-3 (Fig. 6B, blue). In cells co-expressing mutant ataxin-3 and beclin-1 (Fig. 6C, red), the fluorescence intensity of enhanced green fluorescent protein-light chain 3 was significantly lower compared to cells co-expressing mutant ataxin-3 and control cDNA (Fig. 6C, blue) (n = 3, Fig. 6C; 48 h: 10.25 ± 0.1 versus 11.65 ± 0.53 in control cDNA, P < 0.01; 72 h: 9.60 ± 0.56 versus 11.30 ± 0.28 in control cDNA, P < 0.001). Furthermore, the fluorescence intensity was lower in mutant ataxin-3/beclin-1 expressing cells (Fig. 6D, red) as compared to wild-type ataxin-3/beclin-1 cells (Fig. 6D, green) (n = 3, Fig. 6D; 48 h: 10.25 ± 0.1 versus 12.78 ± 0.63 in wild-type ataxin-3, P < 0.001; 72 h: 9.60 ± 0.11 versus 10.75 ± 0.05 in wild-type ataxin-3, P < 0.05), suggesting that the autophagosomal flux of mutant ataxin-3 is more affected by beclin-1 overexpression as compared to the wild-type ataxin-3. The fact that the results obtained with wild-type ataxin-3 /beclin-1 (Fig. 6E, green) or control-treated cells (Fig. 6E, orange) were not significantly different, confirms this interpretation.
In order to confirm these results, the same experiment was performed with cells stably expressing enhanced green fluorescent protein-light chain 3 (Fig. 6F). The flow cytometry analysis was performed 48 h post-transfection of ataxin-3 and beclin-1/control constructs in cells incubated for 3 h in complete medium (full bars) or under starvation conditions (vertical stripes bars). As expected, the fluorescence intensity of cells stably expressing enhanced green fluorescent protein-light chain 3 was higher than cells transiently expressing enhanced green fluorescent protein-light chain 3. In agreement with prior results, under complete medium conditions (Fig. 6F, full bars) beclin-1 led to a selective improvement in the autophagosomal flux of cells expressing mutant ataxin-3 (n = 3, Fig. 6F, red; 80.33 ± 0.88 versus 96.60 ± 1.14 in control-treated, blue; P < 0.001) but not in cells expressing wild-type ataxin-3 (n = 3, Fig. 6F, green; 96.13 ± 2.26 versus 93.80 ± 1.18 in control-treated, orange). The starvation conditions (Fig. 6F, striped bars) led to a robust and significant decrease in the enhanced green fluorescent protein-light chain 3 fluorescence intensity in all experimental sets analysed (decrease from 35.62 to 47.25%; P < 0.001). Furthermore, under starvation, autophagosomal degradation continued to be more relevant for the mutant ataxin-3 (n = 3, Fig. 6F, blue-starvation conditions; 50.96 ± 2.84 versus 60.39 ± 2.39 in wild-type ataxin-3, orange-starvation conditions, both control treated; P < 0.01). With concomitant beclin-1 expression, the degradation was even more prominent for the mutant compared to the wild-type ataxin-3 (n = 3, Fig. 6F, red-starvation conditions; 47.33 ± 1.45 versus 61.58 ± 0.61 for wild-type ataxin-3, green-starvation conditions; P < 0.001). Next, to further infer about autophagic preferential clearance of mutant versus wild-type ataxin-3, a western blot analysis of Neuro-2 a cells co-infected with ataxin-3 plus beclin-1/control vectors was performed 2 weeks post-infection (Fig. 6G and H). In the presence of beclin-1, mutant ataxin-3 was more extensively cleared than wild-type ataxin-3 (n = 3, Fig. 6G and H; 1.31 ± 0.08 versus 1.66 ± 0.009 for wild-type ataxin-3; P < 0.05). In conformity, a decrease in mutant ataxin-3 levels upon beclin-1-treatment was also observed when compared to control-treated (n = 3, Fig. 6G and H; 1.31 ± 0.08 versus 1.66 ± 0.07 in controls; P < 0.05).
Altogether, these data show that beclin-1 is able to stimulate autophagosomal flux and therefore promote protein clearance, with selective higher degradation efficiency for mutant, as compared to wild-type, ataxin-3.
In this study, we provide evidence of an impairment of the autophagy pathway in advanced Machado–Joseph disease and demonstrate that an early stimulation of the beclin-1 autophagic pathway increases mutant ataxin-3 clearance and reduce its neurotoxicity.
In the first part of the study, we have shown that the disease modifies endogenous levels of autophagic proteins, in tissue of patients with Machado–Joseph disease, rats (Alves et al., 2008b) and transgenic mice (Goti et al., 2004). The accumulation of autophagosomes in brains of patients with Machado–Joseph disease, as well as in advanced stages of the rat and mouse model, suggests that autophagy is compromised in more advanced stages of the disease, in accordance with what has been described for Alzheimer’s (Nixon et al., 2005; Yu et al., 2005; Jaeger et al., 2010) and Parkinson’s diseases (Crews et al., 2010). In healthy neurons, the autophagy pathway is constitutively active and highly efficient and therefore the autophagic accumulation here observed most likely arises from insufficient clearance of autophagosomes (Boland et al., 2008). Additional analyses in the Machado–Joseph disease rodent models suggested that the autophagic cargo recognition step is functional, as ubiquitinated ataxin-3 inclusions were detected by the autophagosomal receptor p62 and found co-localized with light chain 3. This finding is of major importance, since p62 participates in the effective targeting of ubiquitinated proteins to autophagosomes (Bjorkoy et al., 2005; Kirkin et al., 2009). Despite the fact that the autophagy pathway is functional in early stages of disease, mutant proteins progressively accumulate along with the autophagosomes in a late stage, which means that the autophagy machinery probably reaches saturation when the amount of mutant protein synthesized is higher than the degradative capacity of the autophagy system (reviewed in Cuervo, 2004; Wong and Cuervo, 2010). In agreement with that hypothesis, levels of the autophagic protein beclin-1 were found to be decreased in fibroblasts from patients with Machado–Joseph disease and rodent models for the disease. This finding is in agreement with what previously described for Huntington’s (Shibata et al., 2006) and Alzheimer’s diseases (Pickford et al., 2008; Jaeger et al., 2010). A decrease in beclin-1 levels has been also described in ageing (Shibata et al., 2006) and there is evidence that the ageing-related decline of the autophagic function may be involved in the delayed onset of neurodegenerative diseases (Simonsen et al., 2008).
Beclin-1 and mammalian target of rapamycin are part of large protein complexes, which regulate the two signalling pathways controlling autophagy (Yorimitsu and Klionsky, 2005). Inhibition of the mammalian target of rapamycin complex activates autophagy while the beclin-1 complex directly activates autophagosome formation (Yorimitsu and Klionsky, 2005). In our study, beclin-1 overexpression proved to achieve a robust effect regarding the stimulation of the autophagosomal flux, mutant ataxin-3 clearance and neuroprotection. Beclin-1 mediated clearance of soluble and insoluble toxic forms of mutant ataxin-3, leading ultimately to neuroprotective effects. Furthermore, the fact that beclin-1-mediated clearance was more pronounced for the mutant relative to the wild-type ataxin-3 proves that this pathway can be selective for the degradation of mutant proteins, challenging the original idea of autophagy as a bulk degradation system (reviewed in Rubinsztein, 2006).
While completing the experiments for this article, beneficial effects of autophagy stimulation in a model of Machado–Joseph disease, through administration of a rapamycin analogue, temsirolimus, were reported (Menzies et al., 2009). Our study not only supports those results but also reinforces them with a more mechanistic analysis and an autophagy-specific approach. Since mammalian target of rapamycin is a protein kinase that regulates multiple important cellular functions (Wullschleger et al., 2006), potential non-specific effects may be obtained with an mammalian target of rapamycin inhibition strategy. Furthermore, rapamycin itself has been described to decrease polyQ protein synthesis (King et al., 2008; Li et al., 2008). On the other hand, beclin-1 interacts with the class III phosphatidylinositol 3-kinase/VPS-34 complex and activates autophagy through a different mechanism, which directly activates autophagic-related proteins and therefore should allow autophagy stimulation in a specific manner (Cao and Klionsky, 2007).
In conclusion, this study correlates the reduced levels of beclin-1 protein with a late impairment of the autophagy pathway in Machado–Joseph disease and shows that selective activation of the beclin-1-autophagic pathway promotes degradation of mutant ataxin-3, identifying this pathway as a molecular target of major importance for therapy in Machado–Joseph disease.
This work was supported by the Portuguese Foundation for Science and Technology (PTDC/SAU-FCF/70384/2006 and PTDC/SAU-NEU/099307/2008), the National Ataxia Foundation (NAF Research Award 2010), the Commissariat à L’Energie Atomique (CEA) and the Association Française pour les Myopathies (AFM; SB/NF/2010/2008 Number 15079CA). Isabel Nascimento-Ferreira, Lígia Sousa-Ferreira, Sandro Alves and Isabel Onofre were supported by the Portuguese Foundation for Science and Technology.
Supplementary material is available at Brain online.
We thank Dr Tamotsu Yoshimory for providing the enhanced green fluorescent protein-light chain 3 construct, Dr Beth Levine for the Beclin-1 construct, Dr Olinda Garcia for post-mortem human brain tissue; and the technical assistance of Dr Luísa Cortes (confocal microscopy), Sylvain Martineau (real-time PCR), Aurélie Delzor (in vivo experiments), Carole Malgorn (primary cell cultures), Clévio Nóbrega (Neuro-2 A cell culture) and Sara Trabulo/ Dr Isabel Nunes (flow cytometry).
autophagic related protein
dopamine-and-cyclic AMP-regulated phosphoprotein of 32 kDa
sequestosome 1/p62 protein