Abstract
Accumulation of misfolded proteins and protein assemblies is associated with neuronal dysfunction and death in several neurodegenerative diseases such as Alzheimer's, Parkinson's, and Huntington's disease (HD). It is therefore critical to understand the molecular mechanisms of drugs that act on pathways that modulate misfolding and/or aggregation. It is noteworthy that the mammalian target of rapamycin inhibitor rapamycin or its analogs have been proposed as promising therapeutic compounds clearing toxic protein assemblies in these diseases via activation of autophagy. However, using a cellular model of HD, we found that rapamycin significantly decreased aggregation-prone polyglutamine (polyQ) and expanded huntingtin and its inclusion bodies (IB) in both autophagy-proficient and autophagy-deficient cells (by genetic knockout of the atg5 gene in mouse embryonic fibroblasts). This result suggests that rapamycin modulates the levels of misfolded polyQ proteins via pathways other than autophagy. We show that rapamycin reduces the amount of soluble polyQ protein via a modest inhibition of protein synthesis that in turn significantly reduces the formation of insoluble polyQ protein and IB formation. Hence, a modest reduction in huntingtin synthesis by rapamycin may lead to a substantial decrease in the probability of reaching the critical concentration required for a nucleation event and subsequent toxic polyQ aggregation. Thus, in addition to its beneficial effect proposed previously of reducing polyQ aggregation/toxicity via autophagic pathways, rapamycin may alleviate polyQ disease pathology via its effect on global protein synthesis. This finding may have important therapeutic implications.
The polyglutamine/CAG disorders comprise a group of neurodegenerative diseases that are associated with polyglutamine (polyQ) expansion mutations in the respective disease genes that are otherwise unrelated (Cummings and Zoghbi, 2000). Abnormally long polyQ stretches cause proteins to misfold and produce intracellular protein aggregates. It is believed that polyQ aggregation follows a stochastic nucleation-dependent process that initiates oligomerization, amyloid-like fibril formation, and the production of structures called inclusion bodies (IBs) (Perutz and Windle, 2001). Because polyQ misfolding/aggregation is associated with cellular toxicity (Ross and Poirier, 2004), it is crucial to understand the cellular mechanisms that control misfolding/aggregation with a view to the development of drugs that modify these pathways and alleviate disease.
The accumulation of intracellular IBs points to the inability of cells to dispose of mutant polyQ proteins using chaperone-assisted refolding (Muchowski and Wacker, 2005) and proteasome-mediated degradation (Jana and Nukina, 2003). Deciphering the mechanisms of degradation and clearance of polyQ-expanded proteins and how such mechanisms might be targeted using drugs is a major focus of current research. Macroautophagy (here referred to as autophagy) is a process alternative to that of proteasomal degradation by which some long-lived proteins and organelles are cleared (Shintani and Klionsky, 2004). Autophagy may be responsible for clearing polyQ-expanded proteins and their assemblies (Rubinsztein, 2006). Clearance by autophagy occurs by sequestration of the target organelle/protein into double-membrane structures called autophagosomes that fuse with endo/lysosomes and discharge their contents, which are subsequently degraded. The mammalian homolog of Atg8 MAP-LC3 (LC3) is a key mediator of autophagy: after LC3 is C-terminally cleaved (LC3 I), phosphatidylethanolamine is added to the C-terminal glycine by the Atg5/12 complex, generating LC3 II bound to the nascent autophagosomal membrane (Tanida et al., 2004). Because the Atg5/12 complex catalytically activates the lipidation of LC3, trace amounts of Atg5 can support substantial autophagy, whereas Atg5 knockout cells are totally deficient in autophagy (Hosokawa et al., 2006). The hallmark of autophagic activation is the formation of autophagosome puncta containing LC3 II, whereas the biochemical measurement of autophagic activity is expressed as the amount of LC3 II that accumulates in the absence or presence of lysosomal activity.
Autophagy was first implicated in the regulation of IB formation and clearance of aggregate-prone proteins based on the use of chemical activators/inhibitors, including the proautophagic drug rapamycin and knockdown of different autophagic genes (Rubinsztein, 2006). The finding that rapamycin and its analog CCI-779 protect against neurodegeneration in animal models of misfolding diseases (Ravikumar et al., 2004; Berger et al., 2006) opens up immense hopes for treating debilitating diseases such as the polyQ disorders. Rapamycin, a macrolytic lactone produced by Streptomyces hygroscopicus, has immunosuppressive, antimicrobial, and antitumor properties. It binds intracellularly to FK506 binding protein 12 and targets the protein kinase mammalian target of rapamycin (mTOR). Inhibition of phosphorylation of mTOR by rapamycin activates autophagy, and it has been suggested that rapamycin (or analogs) ameliorates neurodegenerative proteinopathies via activation of autophagy (Rubinsztein, 2006). However, mTOR has an impact on various downstream targets not necessarily involved in autophagy, including the control of protein synthesis (Dann and Thomas, 2006; Wullschleger et al., 2006), and because of these effects, it is currently being evaluated in several phase II clinical trials for cancer (Sabatini, 2006). It is therefore unclear whether rapamycin mediates its protective effects solely via autophagy.
To probe the actions of rapamycin on the formation and clearance of expanded polyQ proteins and IBs, we have taken advantage of clonal cell lines of autophagy-proficient (Atg5+/+) and -deficient (Atg5-/-) mouse embryonic fibroblasts (MEFs) that are easily amenable to biochemical and genetic rescue experiments. Using exon 1 of human Htt containing 97 glutamines and fused to enhanced green fluorescent protein (EGFP) (Ex1HttQ97-EGFP) as an aggregation prone model polypeptide, we show that autophagy-deficient cells accumulate insoluble Ex1HttQ97-EGFP more rapidly and form greater numbers of IBs compared with autophagy-proficient cells. Reexpression of Atg5 in Atg5-deficient cells reversed this phenotype. Most strikingly, rapamycin reduced the amount of insoluble Ex1HttQ97-EGFP and IBs to a similar degree in both Atg5+/+ and Atg5-/- cells. The formation of SDS-insoluble polyQ assemblies is a cooperative process that is highly dependent on the accumulation of a critical mass of the protein (Scherzinger et al., 1999; Colby et al., 2006). We suggest that a major effect of rapamycin is the reduction in protein synthesis required for polyQ aggregation and IB formation to occur.
Materials and Methods
Expression Vectors. Mammalian expression vectors encoding exon 1 of the HD gene with 25 or 97 glutamines fused at the C terminus to an EGFP tag were a gift from Erich Schweitzer and Alan Tobin (Brain Research Institute, University of California, Los Angeles, CA). The mouse Atg5 expression vector and adenovirus mRFP-LC3 have been described previously (Mizushima et al., 2001; Bampton et al., 2005).
Cell Culture, Transfection, Inclusion Load Measurement, and Microscopy. SV-40-transformed MEF from Atg5+/+ and Atg5-/- mice (Kuma et al., 2004) were cultured in Dulbecco's modified Eagle's medium (Invitrogen, Carlsbad, CA) containing 10% fetal bovine serum (Sigma, St. Louis, MO), 4.5 g/l glucose, 2 mM l-glutamine, 1 mM sodium pyruvate, 100 U/ml penicillin, and 100 μg/ml streptomycin (Sigma) in 5% CO2 at 37°C. Cells were propagated in 75-cm2 flasks and seeded on 12-mm poly(l-lysine)-coated glass coverslips in 24-well plates for fluorescence analysis or directly onto six-well plates for immunoblot/filtertrap analysis. Cells were trypsinized, counted, and seeded at a density of 4 × 104 cells/well in 24-well plates and 3 × 105 cells/well in 6-well plates. After an overnight culture, cells reached 60 to 80% confluence and were transiently transfected using Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer's instructions. The transfection medium was replaced with fresh medium after 4 h, and cultures were incubated for a further 20 h in the presence or absence of the following inhibitors: 200 nM rapamycin (Rap; Sigma), 50 nM bafilomycin A1 (BafA1; Sigma), and 0.3 to 0.01 μg/ml cycloheximide (CHX; Sigma). Cultures requiring longer time courses were split 24 h after transfection and reseeded at lower densities for harvesting after 48 to 96 h. Cells were fixed in 4% paraformaldehyde, washed in phosphate-buffered saline, and analyzed using epifluorescence microscopy with an Olympus X-170 microscope (Olympus, Tokyo, Japan). Images were collected using an AstraCam camera and Ultra-View software (PerkinElmer Life and Analytical Sciences, Waltham, MA). Inclusion load was calculated as the proportion of EGFP-expressing cells that contained IBs. At least 200 cells were counted per condition.
Protein Synthesis and Cell Counting. Cells were briefly washed free of methionine to avoid long-term methionine deprivation and labeled for 1 h in methionine-free RPMI medium (Sigma) containing 10% fetal bovine serum and 1.85 MBq of [35S]methionine (GE Healthcare, Chalfont St. Giles, Buckinghamshire, UK) and the appropriate additives. Cells were washed three times in methionine-containing Dulbecco's modified Eagle's medium, protein was precipitated in ice-cold 20% trichloroacetic acid, and after three washes with 5% trichloroacetic acid, the precipitate was dissolved in 15% SDS and radioactivity was measured by scintillation counting. Little tRNA was found in these pellets. Cells were counted after trypsinization using a hemocytometer.
Immunoblotting, Immunocytochemistry, Filter Trap Assay, and Resolubilization with Formic Acid. Cells were either collected with a cell scraper or trypsinized and counted using a hemocytometer before being pelleted and washed in phosphate-buffered saline. Material was prepared for immunoblotting and filter-trap detection according to Wanker et al. (1999). In brief, cells were lysed on ice for 30 min in filter-trap lysis buffer [50 mM Tris-HCl, pH 8.8, 100 mM NaCl, 5 mM MgCl2 0.5% (w/v) Nonidet P-40, and 1 mM EDTA] in the presence of Complete protease inhibitors (Roche, Indianapolis, IN). Insoluble material was pelleted by centrifugation at 16,000g for 10 min and resuspended in 100 μl of DNaseI buffer (20 mM Tris-HCl, 15 mM MgCl2, 0.5 mg/ml DNase I; Sigma) for 2 h at 37°C. Protein concentrations of soluble (supernatant fraction) and insoluble fractions (pellet) were determined using the Bicinchoninic acid kit (Sigma) and bovine serum albumin standards. Between 5 and 30 μg of insoluble material was diluted into 200 μl of 2% SDS, boiled for 5 min, and applied to a 96-well dot-blot apparatus (Bio-Rad Laboratories, Hercules, CA) containing a cellulose acetate membrane with 0.2-μm pore-size (Macherey-Nagel, Bethlehem, PA). Resolubilization of pellets with formic acid was performed according to Hazeki et al. (2000). Pellets were treated in 100 μl of 100% formic acid for 1 h at 37°C, vacuum-centrifuged, and solubilized in 1× SDS-PAGE sample buffer (see below). Soluble material was supplemented with 4× SDS-PAGE sample buffer (1 M Tris-HCl, pH 6.8, 400 mM dithiothreitol, 8% SDS, and 40% glycerol), and 30 μg was used for analysis by SDS-PAGE (8-12.5%). Membranes were blocked in 5% milk for 1 h and probed with the following primary antibodies: mouse monoclonal anti-GFP (8371-1; BD Biosciences, San Jose, CA) at 1:4000, rabbit polyclonal anti-phospho-S6 (2211; Cell Signaling Technology, Danvers, MA) at 1:1000, mouse monoclonal anti-ERK (M12320; Transduction Laboratories, Lexington, KY) at 1:5000, rabbit polyclonal anti-actin (A2066; Sigma) at 1:1000, and mouse antivimentin (V6630; Sigma, 1:40). Immunocytochemical analysis was performed as in Bampton et al. (2005). Rabbit polyclonal anti-LC3 antibodies were gifts from Yasuo Uchiyama (Osaka University Graduate School of Medicine, Osaka, Japan) and Eiki Kominami (Juntendo University School of Medicine, Tokyo, Japan). Rabbit polyclonal anti-Atg5 antibody was described previously (Mizushima et al., 2001). Blots were subsequently probed with horseradish peroxidase-conjugated anti-mouse or anti-rabbit IgG (Jackson Immunoresearch Laboratories, West Grove, PA) at 1:5000 and visualized with enhanced chemiluminescence detection reagents (Amersham). Immunoblots and dot-blot signals were scanned with a flatbed scanner (HP Scanjet 5470c; Hewlett Packard, Palo Alto, CA), and densitometry was performed using ImageJ software (National Institutes of Health, Bethesda, MD). For ratiometric values, the integrated pixel intensity of each signal was calculated and divided by the signal intensity obtained under control conditions. For each sample, dot plots were repeated at two to three dilutions to ensure that the signal was not saturated, giving rise to a single value used for statistical analysis. Values from several experiments were then used to determine the mean (fold) difference in signal intensity. Input was normalized either according to cell number or protein content. For SDS-insoluble material, in some experiments, a parallel analysis of protein loading was conducted by immunoblotting for histone expression (M.A.K., PhD Thesis; data not shown).
Statistical Analysis. The mean value of replicates within an experiment (duplicates to quadruplicates) was taken as a single value when calculating standard deviations from multiple experiments. Multiple comparisons were made using ANOVA followed by Tukey's honestly significant difference post hoc test, pairwise comparisons were conducted using two-tailed Student's t test, and one sample t test or 95% confidence intervals were used for calculating the significance of ratiometric values. These values and the number of experiments performed for each result are indicated in the text and figure legends.
Results
Genetic Ablation of Atg5 Increased Ex1HttQ97-EGFP Accumulation and Inclusion Body Formation. We first tested whether the complete genetic ablation of autophagy (Atg5) in cells (MEFs) modulated the accumulation of polyQ-expanded huntingtin (Htt), because previous experiments have been performed with RNAi approaches or not under conditions of Htt synthesis (Iwata et al., 2005b; Shibata et al., 2006). Atg5+/+ and Atg-/- MEFs were transfected with cDNA encoding exon 1 of human Htt containing either 25 glutamines (Ex1HttQ25) or 97 glutamines (Ex1HttQ97) fused to EGFP. No discernible IBs in cells were found in MEFs of either type transfected with Ex1HttQ25-EGFP (Fig. 1). However, cytoplasmic or nuclear IBs were readily formed in Ex1HttQ97-EGFP-expressing cells of both types (Fig. 1). Evidence that Atg5+/+ MEFs were proficient to undergo autophagy whereas Atg5-/- MEFs were not was obtained by expression of mRFP-LC3. Figure 1 shows that Atg5+/+ MEFs contained several mRFP-LC3 puncta, whereas mRFP-LC3 expression in Atg5-/- MEFs was evenly diffuse, as shown previously (Bampton et al., 2005). We did not observe colocalization of mRFP-LC3 with IBs in wild-type cells at this time point (24 h).
The proportion of EGFP-expressing cells containing IBs increased over time in both cell types (Fig. 2A). Approximately twice as many Ex1HttQ97-EGFP-positive Atg5-/- MEFs contained IBs compared with Atg5+/+ MEFs after 1 to 2 days (Fig. 2A). Nuclear inclusions in approximately 10 to 15% of both types of cells were evident from the fact that nuclear DNA was “vacated” from spots in which the IBs had deposited (Fig. 1, arrows) (for quantification, see Fig. 4). There was no difference in the transfection rate between Atg5-/- and Atg5+/+ cells (quantified in Fig. 4), transfection efficiency varying between 50 and 60% in both types of MEFs (see Supplementary Fig. S1 for low-power fluorescent images of cells). It is important to note that at 24 h after transfection, we did not detect any differences in toxicity caused by either Ex1HttQ25-EGFP or Ex1HttQ97-EGFP expression in Atg5+/+ and Atg5-/- cells, as assessed by inspection of nuclear abnormalities (approximately 5% of EGFP+ve cells showed baseline toxicity as measured by nuclear fragmentation; see Fig. 4 for a quantitative comparison). Therefore the increase in IB formation of Ex1HttQ97-EFP in Atg5-/- cells was due neither to unequal transfection nor to any differential toxicity as a result of IB formation in our experiments.
To test whether the increase in IBs in the Atg5-/- cells relative to Atg+/+ cells correlated with an increase in the accumulation of SDS-insoluble Ex1HttQ97-EGFP, we used the filter-trap assay to measure the amount of SDS-insoluble Ex1HttQ97 protein formed in each cell type. Pellets remaining after protein extraction in 1% Nonidet P-40 were treated with DNase I, boiled in 2% SDS, and filtered onto a cellulose acetate filter using a dot-blot apparatus (Wanker et al., 1999), whereas respective supernatant proteins were separated by SDS-PAGE. Figure 2B shows that Ex1HttQ97 formed SDS-insoluble material in both types of MEFs, whereas Ex1HttQ25 did not. Quantification showed that there was a 2-fold increase of insoluble Ex1HttQ97-EGFP in Atg5-/- cells compared with Atg5+/+ cells, hence correlating with increased IB formation in autophagy-deficient versus autophagy-proficient cells (Fig. 2D, ▪, n = 4, p < 0.001).
To test for autophagic activity, we probed for LC3 by immunoblotting. Consistent with the lack of mRFP-LC3 puncta in Atg5-/- MEFs (Fig. 1B), no LC3 II was detected in Atg5-/- extracts from Atg5-/- cells immunoblotted for LC3, but extracts from Atg5+/+ MEFs expressed LC3 II, the latter being the autophagosome-associated form of LC3. A similar amount of LC3 I was expressed in both cell types. Equal input of soluble protein was confirmed with an antibody against ERK1 and ERK2 (tERKs). Further evidence for ongoing autophagy in Atg5+/+ cells was obtained by treatment with BafA1, which prevents LC3 II degradation in lysosomes and thus causes LC3 II to accumulate in autophagically proficient cells (Kabeya et al., 2000; Bampton et al., 2005). Figure 2C shows that BafA1 significantly increased the amount of LC3 II in Atg5+/+ cells compared with untreated cells, whereas no changes in LC3 occurred in Atg5-/- cells, consistent with the complete absence of autophagy in these cells. Blocking autophagy using BafA1 also significantly increased SDS-insoluble Ex1HttQ97-EGFP in the autophagy-proficient but not -deficient cells (Fig. 2C, quantified in Fig. 2D). The ratio of LC3 II/I varied between experiments (Fig. 2, B and C), but BafA1 always increased the amount of LC3 II by at least 2-fold (Fig. 2D). No SDS-insoluble material was detected in extracts of Ex1HttQ25-EGFP transfected cells of either genotype (Fig. 2B). We also did not detect EGFP signals on filters when filtrating the supernatant of Ex1HttQ97-EGFP-expressing cells after spinning at 16,000g (data not shown) but without boiling, suggesting that no SDS-insoluble oligomeric Ex1HttQ97-EGFP species of more than 200 nm (pore size of filter) were generated.
We analyzed whether the IBs are ubiquitinylated in both cell types, because this is a hallmark of all polyQ diseases in vivo, including HD. We found colocalization of ubiquitin with IBs in ∼5% of both Atg5+/+ and Atg5-/- cells (Fig. 3 and data not shown). We also found that lysosome-associated membrane protein-1 was associated with IBs in both cell types. Because cytoplasmic IBs are surrounded by intermediate filaments that form an “aggresome” (Waelter et al., 2001), we further probed for the intermediate filament protein vimentin. In both Atg5-/- and Atg5+/+ cells, IBs were surrounded by vimentin immunoreactivity. These results show that IBs are qualitatively similar in both cell types.
Together, these data show that Atg5-dependent degradation via autophagy plays an important role in determining the amount of insoluble polyQ-expanded Ex1Htt protein. The decrease in the propensity of cells to form IBs and insoluble Ex1HttQ97-EGFP correlates with their ability to perform autophagy.
Rapamycin Reduced the Amount of Insoluble Ex1HttQ97-EGFP and Inclusion Body Formation in Both Autophagy-Proficient and -Deficient Cells. We next investigated whether rapamycin requires an Atg5-dependent mechanism to modulate the amount of insoluble Ex1HttQ97-EGFP and IB formation. MEFs were treated with 200 nM rapamycin either 12 h before transfection, to instill high autophagic activity before onset of polyQ expression and IB formation, or were treated with rapamycin simultaneously with transfection. After 18 to 24 h, the percentage of EGFP-positive cells with IBs was determined, whereas insoluble Ex1HttQ97-EGFP was measured using the filter-trap assay as described above. To determine that rapamycin was active, we measured S6 phosphorylation. S6 is a ribosomal protein whose phosphorylation is regulated by S6 kinase in an mTOR-dependent manner (Nobukini and Thomas, 2004). Figure 4A shows that rapamycin added 12 h before transfection inhibited S6 phosphorylation in both cell types (transfected with Ex1HttQ25-EGFP or Ex1HttQ97-EGFP), indicating that rapamycin prevented mTOR activity independently of Atg5 activity. There was no difference in the amount of Ex1HttQ25-EGFP or Ex1HttQ97-EGFP expressed in either cell type treated with rapamycin compared with untreated cells when the total amount of soluble protein input was equalized between treatments, thus indicating that there is no differential destruction of the transfected proteins per se (Fig. 4A).
Pretreatment with rapamycin significantly decreased the proportion of EGFP-positive cells containing Ex1HttQ97-EGFP IBs by 40 to 45% in both Atg5+/+ and Atg5-/- MEFs (Fig. 4B). To ensure that rapamycin treatment did not affect the transfection efficiency or toxicity of the Ex1Htt transgenes in either Atg5 cell type, we monitored both. We and others have shown previously that analysis of nuclear morphology as measured by nuclear fragmentation and condensation using DNA stains is a reliable marker of cell toxicity under these conditions and strongly correlates with other markers of cell death (Wyttenbach et al., 2001). Figure 4C shows that the toxicity associated with the expression of Ex1HttQ25- or Q97-EGFP was at a baseline level (5%) under our experimental conditions (after 18-24 h after transfection) and not different in the two Atg5 cell types (□). Furthermore, rapamycin treatment did not modulate cell survival compared with control conditions (▪), demonstrating that the reduction of IBs in both Atg5 cell lines induced by rapamycin (Fig. 4B) was not due to a differential toxicity (mean ± S.D., n = 3, ANOVA, p = 0.6). To make sure that the differential increase in IBs between Atg5+/+ and Atg5-/- cells (Figs. 2B and 4B) and the decrease in IB formation by rapamycin was not due to unequal transfection rates, we measured the transfection under the various conditions by counting the number of EGFP-positive cells in the total cell population after each experiment in parallel with the analysis of toxicity and IB. As shown in Fig. 4D, we obtained transfection efficiencies of 50 to 60%. It is noteworthy that neither the cell type nor rapamycin affected the rate of transfection (mean ± S.D., n = 3, ANOVA, p = 0.3). Because we observed a minor proportion of IBs in the nuclear compartment (Fig. 1), we also quantified the proportion of Atg5+/+ and Atg5-/- cells containing IB located in the nucleus versus cytoplasmic localization and whether this distribution is modulated by rapamycin. Figure 4D shows that 10 to 15% of nuclear IBs in both cell types was not changed under rapamycin treatment (mean ± S.D., n = 3, ANOVA, p = 0.9).
Having shown that rapamycin treatment reduced IBs in both cell types, we next investigated whether this reduction was also observed in the amount of insoluble material. IBs are cellular structures (or aggresomes) that may not provide an adequate estimation of the amount of polyQ aggregation. However, we measured a similar reduction in SDS-insoluble Ex1HttQ97-EGFP induced by rapamycin when this was assayed by filter trap, or after solubilizing the SDS-insoluble pellet with formic acid, thereby controlling for equal protein input and loading between the different conditions (Fig. 5A, quantified in B). There was no significant change in the amount of insoluble Ex1HttQ97-EGFP formed in cells that had been treated with rapamycin at the time of transfection compared with control (see supplementary Fig. S2 for raw data). Thus, rapamycin can decrease the amount of insoluble Ex1HttQ97-EGFP and IB load, but this effect occurs in an Atg5-independent manner and with a considerable delay after its addition.
Ravikumar et al. (2004) found that mTOR was inactivated in polyQ-expressing cells, was bound to a polyQ-expanded N-terminal portion of Htt, and sequestered into IBs and thus suggested that autophagy is endemically activated in HD (and maybe other polyQ diseases) as a protective response. In MEFs expressing Ex1HttQ97-EGFP, we failed to find an increase in mTOR immunoreactivity after solubilization of IBs using formic acid. mTOR was also not trapped on the stacking gel in conjunction with Ex1HttQ97-EGFP (data not shown), so the degree of sequestration of mTOR may be cell-specific and time-dependent.
Our finding that rapamycin reduced IBs and SDS-insoluble material in autophagy-deficient cells was unexpected. Hence, we next investigated through which mechanism rapamycin reduced IBs and SDS-insoluble Ex1HttQ97-EGFP.
Rapamycin Reduced the Formation of Insoluble Ex1HttQ97-EGFP by Lowering the Amount of Soluble Protein Input. To investigate how rapamycin may be reducing the load of insoluble Ex1HttQ97-EGFP and IBs in Atg5-/- MEFs, the kinetics of cell cycle and protein expression were investigated, because rapamycin is well documented to be a cell cycle suppressant and an inhibitor of protein synthesis (Dann and Thomas, 2006; Sabatini, 2006). To measure this, we counted the number of cells and calculated the amount of decrease due to rapamycin treatment compared with untreated cells set as the control value (presented as percentage of change). Indeed, approximately 25% fewer cells were generated in rapamycin-treated MEFs over the experimental period (36 h) irrespective of Atg5 genetic background or expression of Ex1HttQ25-EGFP or Ex1HttQ97-EGFP (Fig. 6A, left; 0.001 < p < 0.05 for every condition relative to untreated control; see Supplementary Fig. S3 for raw data). There was no statistical difference in rapamycin-induced decrease in cell number between untransfected cells or cells transfected with Ex1HttQ25-EGFP or Ex1HttQ25-EGFP. The reduction in total cell protein in rapamycin-treated cultures compared with untreated cultures was approximately 37%, irrespective of Atg5 genetic status or Ex1HttQ25-EGFP or Ex1HttQ97-EGFP expression (Fig. 6A, middle; 0.001 < p < 0.05 for every condition relative to untreated control; see Supplementary Fig. S3 for raw data), and no difference in the amount of this reduction between untransfected or transfected cells with either construct under rapamycin treatment was observed. When the total amount of protein per cell was calculated, the amount of protein was diminished on average by 17 ± 1.8% in rapamycin-treated cells compared with that in untreated controls, irrespective of genetic background or expression of Ex1HttQ25-EGFP or Ex1HttQ97-EGFP proteins (Fig. 6A, right). Thus, we conclude that rapamycin caused a significant reduction in protein per cell (p < 0.001, t test on pooled results, n = 10 for each genotype).
To test whether the reduction in total protein per cell due to rapamycin treatment included protein translated from Ex1Htt-expressing plasmid, we examined the amount of Ex1HttQ25-EGFP produced to avoid the uncertainty associated with the insolubility of Ex1HttQ97-EGFP (Ex1HttQ25-EGFP is soluble under all conditions examined, see above). Figure 6B shows that the average reduction in expression of Ex1HttQ25-EGFP per cell as a result of rapamycin was approximately 14% in both Atg5+/+ and Atg5-/- cells, similar to the reduction found in total protein per cell. There was no statistical difference in the reduction of total protein per cell and the reduction of Ex1HttQ25-EGFP per cell as a result of rapamycin treatment. This finding also suggests that Atg5-null cells have no increased protein expression from the plasmids compared with Atg5 wild-type cells because of the lack of the autophagic protein degradation system.
We next measured whether the reduction of 37% of total protein per cell as a result of rapamycin treatment (Fig. 6A) was correlating with a similar reduction in global protein synthesis as measured by direct [35S]methionine incorporation. Thirty hours after the addition of rapamycin, we added [35S]methionine to Atg5+/+ cells for 1 h. We observed a significant reduction in [35S]methionine incorporation (Fig. 6C; mean ± S.D., n = 3; t test, p < 0.001). We then quantified the reduction as a percentage change under rapamycin treatment compared with untreated cells and found that there was a 39 ± 3% reduction in [35S]methionine incorporation (Fig. 6D, left bar). This 39% reduction was almost identical with the reduction of 37 ± 3% found for total soluble protein (Fig. 6D, middle bar) that we calculated in parallel in each of these experiments. Calculating the ratio of change in [35S]methionine incorporation over the reduction of total protein content, this value was no different from 0, indicating that the reduction in the global amount of protein (as shown in Fig. 6, A and D) is probably due to an inhibition of protein synthesis.
Inhibition of Protein Synthesis by Cycloheximide Reduced the Level of Insoluble Ex1HttQ97 and Inclusion Body Formation Similar to Rapamycin. To investigate how a decrease in protein synthesis per cell affects the propensity of cells to form insoluble Ex1HttQ97-EGFP and IBs, a concentration of CHX, an inhibitor of protein synthesis, was sought that matches the decrease in the amount of soluble protein obtained with rapamycin. Cycloheximide was used because it binds to ribosomes reversibly, thus ensuring a response that is proportional to ribosome occupancy over extended periods of treatment. In dose-response experiments using rapamycin, we did not observe a significant difference in the amount of reduction in global protein synthesis between 50 and 200 nM rapamycin (data not shown), and hence, we aimed at a CHX concentration that would match that induced by 200 nM rapamycin.
As expected, the amounts of insoluble Ex1HttQ97-EGFP protein (Fig. 7A) and IB formation (Fig. 7B) were highly dependent on the concentration of CHX used in both cell types. Between 0.03 and 0.1 μg/ml CHX, a similar amount of insoluble Ex1HttQ97-EGFP was detected in the filter trap assay compared with the insoluble material obtained from rapamycin-treated cells (200 nM) irrespective of Atg5 genetic status (Fig. 7A). The amount of IBs formed with rapamycin also closely matched that observed with this concentration range of CHX (Fig. 7B), the amount of insoluble Ex1HttQ97-EGFP and IB formation being equally reduced. Moreover, steady-state protein levels were reduced similarly by the same range of CHX and rapamycin (see Supplementary Fig. S5). Finally, when we compared the amount of [35S]methionine incorporation over 1 h in cells pretreated with rapamycin (200 nM) and cells preincubated with different concentrations of CHX for 24 h, a similar inhibition of [35S]methionine incorporation was found with rapamycin and approximately 0.02 μg/ml CHX (rapamycin, 41.1 ± 7.2% compared with 49 ± 4% with 0.02 μg/ml CHX; mean ± range, two independent experiments). This result suggests that a concentration of CHX that similarly inhibits protein synthesis compared with 200 nM rapamycin also produces equivalent reduction in insoluble Ex1HttQ97-EGFP and IBs obtained by 200 nM rapamycin. Hence, a small reduction in the input of Ex1HttQ97-EGFP has major effects on the kinetics of formation of insoluble Ex1HttQ97-EGFP and IBs. It is interesting that at higher concentrations (0.1-0.3 μg/ml), CHX also reduced the amounts of LC3 I/II in Atg+/+ and LC3 I in Atg5-/- cells to an equivalent extent by the end of the treatment (Fig. 7A), suggesting that LC3 protein is being turned over quite quickly during this time by nonautophagic mechanisms.
Atg5 Overexpression Partially Restored Autophagic Activity in Atg5-/-Cells and Reduced Insoluble polyQ-Expanded Ex1Htt. The characteristics of cell lines can diverge rapidly, even when they derive from a common origin. To further investigate the role of autophagy in the control of insoluble Ex1HttQ97-EGFP and IB load, Atg5 was expressed in Atg5-/- and Atg5+/+ cells by transfection. Atg5 expression in Atg5-/- cells induced the formation of the Atg5/Atg12 conjugate and restored the ability of the cells to produce LC3 II to approximately 20% of the levels measured in Atg5+/+ cells (Fig. 8A). It is interesting that we found colocalization of overexpressed Atg5 with IBs (see Supplementary Fig. S4A) and a significant amount of Atg5 (but not Atg5/12 conjugate) accumulated in the pellets of both types of MEF cells expressing Ex1HttQ97-EGFP (Supplementary Fig. S4B shows results for Atg5-/- cells). Whether this Atg5 is disabled from performing its function (because no Atg12 accumulated in this fraction) remains to be resolved. Together, these results suggest that some of the overexpressed Atg5 protein in Atg5-/- cells is functional, consistent with the recent finding of Hosokawa et al. (2006) using the same cell clone and DNA plasmids.
When the amount of insoluble material was examined using the filter-trap assay (Fig. 8B), overexpression of Atg5 in Atg5-/- cells reduced the amount of insoluble material that accumulated each day. Already after 1 day, the amount of insoluble Ex1HttQ97-EGFP in Atg5-/- cells expressing Atg5 decreased by 27 ± 8% compared with 2 ± 1% change in Atg5+/+ cells (mean ± S.E.M., n = 4, p < 0.02, t test; Fig. 8C). To compare the rate of reduction of insoluble Ex1HttQ97-EGFP under conditions of Atg5 overexpression over several days, we split and replated the transfected cells after 2 days of transfection and measured the rate of decrease of Ex1HttQ97-EGFP in both cell types (Fig. 8D). Atg5 overexpression accelerated the reduction of Ex1HttQ97-EGFP in Atg5-/- cells but not in Atg5+/+ cells. It is important to note that overexpression of Atg5 did not reduce the amount of total protein harvested from the Atg5-/- or Atg5+/+ cells (unlike rapamycin), suggesting that the decrease in Ex1HttQ97-EGFP caused by Atg5 is independent of protein synthesis (data not shown). Moreover, the number of cells harvested at each time point was similar between the four conditions, demonstrating that Atg5 overexpression had no effect on the cell cycle. It should be noted that insoluble Ex1HttQ97-EGFP also decreases over time in cells not overexpressing Atg5 from days 2 to 4 (Fig. 8, B and D). This is probably due to continuous cell division, which reduces the number of plasmids present in each cell. A differential loss of cells containing insoluble Ex1HttQ97-EGFP between days 2 and 4 may also contribute to this finding. Together, these data show that reexpression of Atg5 in autophagy-deficient cells achieves a reduction in the amount of insoluble Ex1HttQ97-EGFP.
Discussion
In the present study, we showed that autophagy-deficient cells lacking Atg5 expression accumulate more misfolded insoluble polyQ protein (Ex1HttQ97-EGFP) and form more IBs than control cells and that this effect can be partially reversed by re-expressing Atg5. These data suggest that the lack of autophagy increases polyQ aggregation and/or reduces the clearance of aggregation-prone polyQ proteins. Our results obtained through a genetic approach (complete genetic knockout of Atg5) are consistent with those of recent studies (Iwata et al., 2005a,b; Kouroku et al., 2007)
Rapamycin is a well characterized activator of autophagy and has been reported previously to alleviate toxicity of different aggregate-prone proteins (Rubinsztein, 2006). In these reports, the authors suggested that the beneficial effects of rapamycin resulted from its ability to reduce aggregation-prone toxic proteins via autophagy. Hence, one would predict that rapamycin would reduce polyQ aggregation in autophagy-proficient cells but not in autophagy-deficient cells. We tested this idea by treating Atg5+/+ and Atg5-/- cells with rapamycin and found that rapamycin reduced soluble and insoluble aggregate-prone Ex1Htt fragments independent of autophagic activity. The evidence for this effect of rapamycin is as follows: 1) rapamycin reduced insoluble Ex1HttQ97 and IB load in Atg5-null cells; 2) inhibition by rapamycin, which rapidly inhibited phosphorylation of S6 kinase within 1 h, was only apparent if the cells were preincubated with rapamycin for approximately 12 h before expression of Ex1HttQ97 for 24 h, a time frame that suggests long-term rather than short-term actions of the drug are required for the effect to occur; 3) rapamycin reduced the amount of soluble protein per cell within this time frame by approximately 17% in both Atg5+/+ and Atg5-/- cells, and this amount of inhibition was correlated with the amount of [35S]methionine incorporated over 1 h at the end of the incubation period, indicating that the reduction in global protein per cell occurred because of a reduction in protein synthesis. The similar extent of rapamycin-induced reduction in the amount of soluble Ex1HttQ25-EGFP per cell indicates that protein synthesis from the plasmid was similarly affected; and 4) the reduction in SDS-insoluble polyQ aggregation and IB formation induced by rapamycin could be mimicked by the use of low concentrations of CHX, which also caused a similar partial reduction in the extent of [35S]methionine incorporation. Thus, it seems that treatment with rapamycin can have a critical impact on the mass of soluble protein required for polyQ aggregation and IB formation independently of autophagy through a relatively modest reduction in protein input.
It is likely that autophagy too can reduce polyQ aggregation not only through clearance of formed polyQ aggregates but also by reducing the amount of polyQ protein submitted to these processes. These two mechanisms could be acting in an additive way. It is conceivable that some of the effects of rapamycin observed in previous studies may be due to a reduction in protein synthesis. In two studies in which rapamycin was used during the period of Htt synthesis (Drosophila melanogaster, transgenic mice) (Berger et al., 2006), it was not reported whether there were changes in expression levels of the transgenes, and hence, it is possible that some of the protective effects were due to the inhibitory action of rapamycin on translation. However, a subtle reduction in the synthesis of an Htt fragment due to rapamycin may be difficult to detect in animal models. Rapamycin treatment has been proposed to inhibit translation of specific mRNAs rather than resulting in global inhibition of translation (Grolleau et al., 2002). However, we found similar amounts of reduction in total protein per cell and in the amount of Ex1HttQ25-EGFP synthesized per cell, suggesting that there is no selectivity in nuclear versus plasmid-based CMV promoter-mediated expression. At present, we cannot distinguish between a possible effect of rapamycin on global translation versus a more specific effect on a subset of mRNAs, although it is clear that the function of some genes required for cell cycle are suppressed.
An interesting question is whether rapamycin mediates all of its effects through mTOR. To date, there are no known alternative targets of rapamycin other than FK506 binding protein 12, which inhibits the protein kinase mTOR (Wullschleger et al., 2006), suggesting that the inhibitory effects of rapamycin on the synthesis (this study) and the clearance of aggregate prone proteins via autophagy (Rubinsztein, 2006) occur via mTOR. However, not all of the effects of mTOR are mediated by autophagy. The serine/threonine kinase mTOR is a highly conserved integrator of mitogenic and nutrient inputs and controls cell proliferation, cell growth, and endocytosis. These activities are probably mediated through modulation of protein synthesis (Dann and Thomas, 2006; Wullschleger et al., 2006), consistent with our observation of a reduction in insoluble Ex1HttQ97-EGFP and IBs in the autophagy-deficient Atg5-/- MEFs. The extent of inhibition of protein synthesis of approximately 17% we observed is in keeping with the reported decrease of 15 to 50% in translation rates induced by mTOR inhibition in mammalian cells (whereas in yeast, inhibition is nearer to 100%) (Jefferies et al., 1994; Terada et al., 1994). Hence, it is likely that the effect of rapamycin on protein synthesis is cell type-dependent. It remains to be seen to what extent this drug affects protein synthesis in neurons.
The reduction in insoluble Ex1HttQ97-EGFP achieved with rapamycin was mimicked by the use of CHX. Several reports have demonstrated that CHX inhibits autophagy (Kovács and Seglen, 1981) by preventing fusion of autophagosome with endo/lysosomes (Lawrence and Brown, 1993). Consistent with this view, IBs were cleared more quickly in the presence of rapamycin in an inducible cell model after expanded Ex1Htt protein synthesis was turned off, but IBs seemed to accumulate when CHX (10 μg/ml) was present with rapamycin (Ravikumar et al., 2002). In the above studies, 2 to 10 μg/ml CHX was used, which inhibits protein synthesis and autophagy more than 95%, whereas in our study, the concentration that mimicked rapamycin's effect was between 0.01 and 0.03 μg/ml, which elicited only a partial block on protein synthesis and presumably on autophagy. However, our important observation is that the same reduction in insoluble material and IBs were elicited by CHX in Atg5-/- MEFs, in which no autophagy can take place. Hence, the phenomenon we observed here is unlikely to be related to autophagy.
Autophagic clearance of Htt fragments may be promoted independently of mTOR (and S6K/Akt) via a pathway dependent on insulin receptor substrate 2 and stimulated via insulin and IGF-1 (Yamamoto et al., 2006). The notion that there are mTOR-independent pathways of autophagy is suggested by the finding that rapamycin inhibited autophagy suppression by insulin but not by regulatory amino acids in liver cells (Kanazawa et al., 2004), and yeast susceptibility to osmotic stress differed when induction of autophagy was by starvation or rapamycin (Prick et al., 2006). Indeed, it has been shown recently that small-molecule enhancers of mTOR can act independently of rapamycin to enhance the clearance of mutant huntingtin fragments and A53T α-synuclein (Sarkar et al., 2007). These findings suggest that multiple mechanisms may regulate autophagic clearance of misfolded/aggregated proteins. However, irrespective of how the proteins are cleared, a reduction in the critical mass of aggregation-prone proteins such as that described in the present study will always be beneficial in reducing protein aggregation.
Reducing the expression of Htt is, in theory, a promising strategy to alleviate disease: a decrease in the expression of wild-type Htt up to 50% is not detrimental (based on studies in heterozygotes), whereas decreasing the expression of mutant Htt using a short interfering RNA approach has been found to significantly reduce HD pathology in mice (Wang et al., 2005; Machida et al., 2006). This idea is supported by the recent findings of Colby et al. (2006), who proposed that an efficient therapeutic strategy for HD lies in reducing the rate of mutant Ex1Htt aggregation by only modestly reducing Htt expression levels. From our study, we cannot conclude that rapamycin would have a beneficial effect by reducing Htt synthesis in all cases. Future in vivo studies are required to address this point.
It is important to note that in HD in vivo, mutant Htt is constantly produced; hence, the mechanisms by which rapamycin regulates the accumulation and turnover of mutant Htt should be studied under condition of continuous Htt expression as we have done in our cellular model. The stochastic model of Colby et al. (2006) was applied under conditions in which no further protein synthesis was taking place during the misfolding process. If rapamycin can reduce Htt concentrations in HD, it might be beneficial even under conditions in which autophagy is already working at its maximal rate.
Acknowledgments
We thank Dr. Yasuo Uchiyama (Osaka University Graduate School of Medicine, Osaka, Japan) and Dr. Eiki Kominami (Juntendo University School of Medicine, Tokyo, Japan) for the provision of anti-LC3 antibodies, and Drs. Eric Schweitzer and Alan Tobin for the original constructs of Ex1HttQ25-EGFP and Ex1HttQ97-EGFP. Most of this work was conducted in Aviva M. Tolkovsky's laboratory under joint supervision of Andreas Wyttenbach and Aviva M. Tolkovsky.
Footnotes
-
This study was supported by the Wellcome Trust before October 2006 (to M.A.K., A.M.T.), the Hereditary Disease Foundation (to F.H.), and the Medical Research Council (to A.W., S.H.).
-
ABBREVIATIONS: polyQ, polyglutamine; CHX, cycloheximide; GFP, green fluorescent protein; EGFP, enhanced green fluorescent protein; Htt, huntingtin protein; Ex1Htt, huntingtin exon 1; HD, Huntington's disease; IB, inclusion body; MEF, mouse embryonic fibroblast; mRFP, monomeric red fluorescent protein; mTOR, mammalian target of rapamycin; BafA1, bafilomycin A1; ANOVA, analysis of variance; PAGE, polyacrylamide gel electrophoresis; Rap, rapamycin; FK506, tacrolimus; CCI-779, temsirolimus; ERK, extracellular signal-regulated kinase; tERK, total extracellular signal-regulated kinase.
-
↵ The online version of this article (available at http://molpharm.aspetjournals.org) contains supplemental material.
- Received November 12, 2007.
- Accepted December 31, 2007.
- The American Society for Pharmacology and Experimental Therapeutics