|Home | About | Journals | Submit | Contact Us | Français|
Polyglutamine (polyQ) expansion within the ataxin-7 protein, a member of the STAGA and TFTC chromatin remodeling complexes, causes the neurodegenerative disease spinocerebellar ataxia type 7 (SCA7). Proteolytic processing of ataxin-7 by caspase-7 generates N-terminal toxic polyQ-containing fragments that accumulate with disease progression and play an important role in SCA7 pathogenesis. To elucidate the basis for the toxicity of these fragments, we evaluated which post-translational modifications of the N-terminal fragment of ataxin-7 modulate turnover and toxicity. Here, we show that mutating lysine-257 (K257), an amino acid adjacent to the caspase-7 cleavage site of ataxin-7 regulates turnover of the truncation product in a repeat-dependent fashion. Modification of ataxin-7 K257 by acetylation promotes accumulation of the fragment, while unmodified ataxin-7 is degraded. The degradation of the caspase-7 cleavage product is mediated by macroautophagy in cell culture and primary neuron models of SCA7. Consistent with this, the fragment co-localizes with autophagic vesicle markers and enhanced fragment accumulation increases in these lysosomal structures. We suggest that the levels of fragment accumulation within the cell is a key event in SCA7 neurodegeneration, and enhancing clearance of polyQ-containing fragments may be an effective target to reduce neurotoxicity in SCA7.
Spinocerebellar ataxia type 7 (SCA7) is a dominantly inherited neurodegenerative disease characterized by latejonset degeneration of the cerebellum, brain stem, and retina. SCA7 is caused by expansion of the polyglutamine tract within the ataxin-7 protein (Lindblad et al., 1996; David et al., 1997; Del-Favero et al., 1998). SCA7 shares common properties of other polyglutamine expansion diseases, including Huntington’s disease (HD), dentatorubral–pallidoluysian atrophy (DRPLA), and spinal bulbar muscular atrophy (SBMA), but a distinguishing clinical feature is blindness correlating with the transcriptional dysregulation of specific retinal genes (La Spada et al., 2001; Helmlinger et al., 2006). Ataxin-7 is a homolog of the yeast Sgf73p, a component of the SAGA (Spt/Ada/Gcn5/acetyltransferase) chromatin remodeling complex, and is now known to function in the human STAGA (SPT3-TAF(II)31-GCN5L acetylase) and TFTC (GCN5 and TRRAP) complexes (Sanders et al., 2002; McMahon et al., 2005), although its exact role remains unclear. Mutant polyglutamine expansion disrupts the normal function of ataxin-7 in these complexes (Palhan et al., 2005; Helmlinger et al., 2006) and confers additional gain-of-function properties, such as aggregation by polyQ-containing fragments (Holmberg et al., 1998).
Protein turnover is a crucial mechanism for maintaining cellular viability over time. Accordingly, disruption of proteolytic pathways occurs in polyglutamine expansion diseases, Alzheimer’s disease and Parkinson’s disease (reviewed in (Olzmann et al., 2008)). The ubiquitin proteosome system (UPS), is one of two major proteolytic pathways. In this pathway, ubiquitin modification targets misfolded and short-lived proteins to the cytosolic 26S proteasome for degradation. PolyQ aggregates are resistant to proteasomal degradation, and disrupt global UPS activity (Bowman et al., 2005; Bennett et al., 2007). In other studies it was found that global UPS activity in vivo was not affected in the brains of polyQ disease transgenic mice (Tydlacka et al., 2008; Bett et al., 2009; Jeong et al., 2009). Ataxin-7 interacts with a portion of the 19S subunit (Matilla et al., 2001) and the polyQ-expanded form of ataxin-7 can inhibit proteasomal function (Wang et al., 2007). The second proteolytic pathway autophagy, degrades whole organelles and cytoplasmic material. Three types of autophagy have been identified: macroautophagy, microautophagy and chaperone-mediated autophagy (CMA). Macroautophagy (hereafter referred to as autophagy) can degrade polyQ-expanded fragments (Qin et al., 2003; Young et al., 2008). Upregulating autophagy can ameliorate polyQ-dependent toxicity in models of HD and SBMA (Yamamoto et al., 2006; Pandey et al., 2007; Sarkar et al., 2009).
The toxic effects of polyglutamine expansion are protein context-dependent, and can be modulated by post-translational modification. Proteolytic cleavage of these proteins by caspases generates short, polyglutamine-containing fragments with increased cellular toxicity (Wellington et al., 1998; Ellerby et al., 1999a; Ellerby et al., 1999b). Proteolytic cleavage products are frequently found in aggregated inclusions observed in both in vitro and in vivo polyglutamine disease models (Ross, 1997; Zhou et al., 2003) and in post-mortem tissue from patients (DiFiglia et al., 1997; Wellington et al., 2002). We have previously shown that ataxin-7 is cleaved by caspase-7 at amino acid positions D266 and D344 (Young et al., 2007). In SCA7 and HD mouse models, substitution of aspartate to alanine or asparagine at these sites blocks the formation of N-terminal truncation fragments and ameliorates disease symptoms ((Graham et al., 2006); Guyenet, et al., unpublished work).
Additionally, lysine modification by small ubiquitin-like protein (SUMO1) or acetylation has been shown to modulate protein accumulation and toxicity in HD, SCA1, DRPLA and SBMA (Chan et al., 2002; Terashima et al., 2002; Steffan et al., 2004; Riley et al., 2005). A ubiquitin competitor, SUMO1, appears to stabilize proteins and decrease their degradation (Steffan et al., 2004). For SBMA, acetylation of polyQ-expanded androgen receptor impacts its aggregation and receptor trafficking (Thomas et al., 2004). In ataxin-7, we identified lysine-257 (K257), a highly conserved residue near the D266 caspase cleavage site, as an important modulator of fragment accumulation in vitro and in vivo. Through chemical disruption, we determined that autophagy mediates the turnover of the ataxin-7 fragment. Furthermore, the acetylase, CBP, and the deacetylase, HDAC7, regulated ataxin-7 turnover through acetylation of the fragment. The work presented here suggests that pathways that enhance the clearance of toxic fragments may effectively mitigate SCA7 pathogenesis.
Ataxin-7 cDNA containing either 10 or 92 CAG repeats in pcDNA3.1 (Invitrogen) were utilized in these studies as previously published (La Spada et al., 2001; Young et al., 2007). Site directed mutagenesis was performed to generate K→R substitutions at amino acid position K-223 using the primers 5′-GCGCATCCTCATCAAGTTCCAAGTTGTTGAGATCACCCAAAGAGA-AACTGCAGCTCAGGGGG3′ and 5′CCCCCTGAGCTGCAGTTTCTCTTTGGGTGATC-TCAACAACTTGGAACTTGATGAGGATGCGC3′; and at position K-257 using the primers 5′CATGGTAGAATCATGACACCCTCTGTGAGAGTGGAAAAGATTCATCCGAGAATGT AAGGC3′ and 5′GCCTTACATCCATTCTCGGATGAATCTTTTCCACTCTCACAGAGG-GTGTCATGATTCTACCATG3′. Mutated nucleotides are underlined in bold. Mutagenesis was performed using the Quikchange Site-Directed Mutagenesis kit (Stratagene) according to the manufacturer’s instructions. Primers were generated by Integrated DNA Technologies (IDT). All constructs were sequenced to confirm the appropriate mutation was introduced and CAG repeat length was not altered. For co-expression studies, plasmids encoding GCN5 and HA-ubiquitin were purchased from Addgene (GCN5 cat# 14424; Ub cat# 18712). The CBP-MT construct was a kind gift from Dr. Aleksey Kazantsev. HA-SUMO1 construct was from Dr. Zhao.
HEK293T cells were cultured in DMEM (Mediatech,) containing 1% penicillin/streptomycin (25 units/mL) and 10% heat-inactivated fetal bovine serum (DMEM complete) unless otherwise specified. Transient transfections were performed using Lipofectamine 2000 (Invitrogen) according to manufacturer’s instructions. Cells were seeded in 6-well dishes at approximately 2 × 106 cells/well. Each well was transfected with 4 μg total DNA and 20 μL Lipofectamine 2000 in 0.5 mL DMEM lacking serum and pen/strep. Following a 48 h incubation, cells were lysed directly in wells with 400 μL 1X Nupage LDS sample buffer (Invitrogen) diluted in M-PER (Thermo Scientific) containing protease inhibitors (1 tablet/10 mL, Complete, Mini, Roche), phosphatase inhibitor cocktail II (100 μL/10 mL, Calbiochem), and 25 mM N-ethylmaleimide (Sigma). HDAC inhibitors were used as indicated: final concentrations of Trichostatin A (Sigma) 50 μM, Na-butyrate (Sigma) 30 μM, and nicotinamide (Sigma) 30 mM were added to cell media 10 min prior to harvesting in lysis buffer described above.
Dharmacon On Target Plus SMARTpools (Thermo Scientific) were used for siRNA disruption of HDAC2 (M-003495), HDAC3 (M003496), HDAC4 (M-003496), HDAC5 (M-003498), HDAC6 (M-003499), HDAC7a (M-009330) and HDAC8 (M00350). HEK293T cells were seeded in 6-well dishes at a density of 1 × 106 cells/well in 2.5 mL DMEM complete medium. Cells were co-transfected with 1 μg DNA and 40 μmol RNA using 10 μL Lipofectamine 2000 in 0.5 mL DMEM according to manufacturer’s instructions. After 72 h incubation, cells were harvested as described above and analyzed by Western blotting.
Cerebellums from 5–10 6-day postnatal CD1 mouse pups were dissected into ice-cold HBSS (Gibco) and meningeal tissue removed. The cerebellums were then transferred to a clean dish, mechanically disaggregated, and trypsinized for 20 min at 37°C in 0.08% trypsin made by diluting 1 mL 0.25% trypsin into 2 mL warmed HBSS. Cells were collected by centrifugation and resuspended in 5 mL HBSS containing 12 mM MgSO4 and 5 U/uL DNAse 1 (Invitrogen). Cells were triturated with three increasingly smaller-bore fire-polished glass pipets, approximately 1 mm, 0.5 mm, and 0.25 mm in diameter. Cells were then collected by centrifugation once more and resuspended in DMEM complete. Cells (4.5 × 106) were transfected with 2 μg DNA using program G-013 on the Amaxa Nucleofector II with Amaxa Mouse Neuron Nucleofector kit. For immunocytochemistry, cells (5 × 105) were plated per well on 8-well poly-L-lysine coated glass slide (BD Biosciences, San Jose, CA) and cultured for 2–4 days in Neurobasal A medium (Gibco) containing 2% B27 (Gibco) and 1% penicillin/streptomycin unless otherwise indicated. For Western blot analysis, 2 × 106 cells per well were plated in 12-well poly-L-lysine (Sigma) coated polystyrene plates (Nunc). Following culturing, cells were either fixed in 2% PFA for imaging or harvested for Western blot analysis. All experiments and animal care were performed in accordance with the Buck Institute for Age Research IACUC guidelines.
Whole-cell lysates were sonicated at 40 mA in 5×5 sec pulses. DTT was added to a final concentration of 10 mM and samples were heated to 95°C for 10 min. 10 or 25 μg protein per sample were subject to SDS-PAGE using 4–12% NuPage Bis-Tris gels under reducing conditions in MES running buffer. Gels were run at constant 200V for 1 h, transferred to 40 μM nitrocellulose membranes in 1X NuPage transfer buffer with 10% methanol at a constant 350 mA for 1 h. Membranes were incubated in TBS with 0.1% Tween 20 (TBS-T) and 5% milk for 1 h. Primary antibodies were diluted in TBS-T/5% milk at 1:10,000 for GAPDH (RDI-TRK5G4-6C5, Research Diagnostics, Inc), 1:1000 for ataxin-7 (PA1-749, Affinity Bioreagents, directed against N-terminal 17 amino acids of mouse and human ataxin-7), for HDAC7 (12174, Abcam), for tubulin antibody (Sigma), 1:500 for SUMO1 (40120, ActiveMotif), and 1:50 for lysine-acetylation (9441, Cell Signaling). Blots were developed using Pierce ECL or SuperSignal West Femto ECL reagents (Thermo Scientific). Band intensity was quantified using NIH ImageQuant TL v2005 or NIH ImageJ as indicated.
For immunoprecipitation, 500 μg total 293T lysates diluted in M-PER were applied to G-Sepharose beads complexed to ataxin-7 antibody (Affinity Bioreagents PA1-749) or c-myc beads (Pierce). Following overnight incubation at 4°C, beads were washed three times in TBST and three times in 0.1M Tris pH 7.4. Proteins were eluted into loading buffer by heating to 95°C for 10 min followed by centrifugation to collect the eluted fraction. Western blotting was carried out as described above.
Cells were fixed by adding 8% paraformaldehyde directly to culture medium to a final concentration of 2%. Cells were incubated for 30 min with gentle agitation, washed 2 × 5 min with room temperature PBS, and stored in PBS at 4°C prior to staining. Cells were permeabilized in 0.25% Triton in 1X TBS for 15 min, washed 2×5 min in TBS, and blocked in 5% NDS in TBS for 30 min. Primary antibody was diluted in 2.5% NDS in 1X TBS at 1:200 for ataxin-7 (La Spada et al., 2001; Garden et al., 2002; Young et al., 2007), 1:50 for LAMP-2 (clone M3/84, sc19991, Santa Cruz Biotechnology) and LC3 (M152-3, MBL) and 1:25 for ubiquitin (clone P4D1-A11, 05-944, Upstate) and incubated with cells overnight at 4°C. Cells were then washed 3×10 min in TBS and secondary antibody diluted in 2.5% NDS in 1X TBS (Alexa 488 anti-rabbit 1:250; Alexa 555 anti-mouse 1:500) was applied for 1 h at RT, washed, and cells were mounted with ProlongGold containing DAPI. Slides were cured overnight. Epifluorescence images were captured on a Nikon Eclipse E800. Filter cubes with the following specs: Red: Excitation 515–560 Dichroic 565 Emission 572–642, Green: Ex 460–500 Dichroic 505 Em 510–560 and Blue: Ex 340–380 Dichroic 400 Em 435485 were utilized. Confocal images were obtained using a Zeiss LSM 510 NLO microscope and Zeiss LSM Image Browser software. Laser settings used were 488 Ar, 543 He/Ne, and Chameleon Ultra tuned to 780nm. Following capture, all images were background adjusted in Adobe Photoshop using identical parameters for each channel.
The MoPrP-Flag-(1X)-SCA7(10/92Q)-myc(1X) expression vectors were derived as follows. A 0.9 kb chloramphenicol resistance cassette was PCR-cloned into the NotI, XhoI restriction sites in an SCA7 cDNA. This cloning step fused the N-terminal 143 bp and C-terminal 355 bp of SCA7 coding sequence to the ends of the selection cassette (deleting the intervening 2.4 kb of SCA7 sequence). This modified SCA7 cDNA was PCR-amplified using primers that were engineered to insert an N-terminal FLAG and a C-terminal myc epitope tag (5′-GCGGAGGTCGACGCCACCATGGACTACAAAGACGATGACGACAAGCTTATGTCGGA GCGGGCCGCGGATG-3′ and 5′-GAGGTGGTCGACTCAGCTATTCAGATCCTCTTCTGA-GATGAGTTTTTGTTCGGGACGTGCCTTTGGCTGATGAAG-3′). This 1.5 kb PCR product was digested with SalI and cloned into the XhoI site of MoPrP.Xho (Borchelt et al., 1996) yielding a new vector designated PrP-SCA7-Not/Xba (#435-3). The chloramphenicol resistance cassette was liberated from this vector using the flanking SCA7 restriction sites (AatII and XhoI) and the intervening AatII-XhoI SCA7 sequence was cloned into this vector restoring the full-length SCA7 coding sequence. SCA7 cDNA sequences containing: a wild-type 10Q and a mutant 92Q yielded a set of vectors designated: 410-3 and 411-3 respectively.
Translational competence of the transgenic constructs was verified by transient transfection into HEK293T cells using Superfect (Qiagen). The vector backbone was removed by Pf1F I digestion and then microinjected into oocyte pronuclei. Mice were in a C57BL/6J background. Founder mice were identified by PCR analysis of tail DNA. SCA7 genotying primers used (5′-CCGAGAATGGATGGCACACT-3′ and 5′-TGCGGTGGTTGCTGAGAGT-3′) amplify a 440 bp fragment of human SCA7. Southern and Western blot analysis were carried out as previously described (La Spada et al., 2001) and expression for PrP-mycFlag-SCA7-10Q-myc line A305 and PrP-Flag-SCA7-92Q-myc line B306 utilized in these studies were similar. Similar to our previous characterized SCA7-92Q-6076 transgenic line (La Spada et al., 2001), PrP-mycFlag-SCA7-92Q-line 306 over-expresses ataxin-7 at ~2.0-fold endogenous levels, displays retinal pathology by 12-weeks of age, and has a shortened lifespan (9 months).
SCA7 mice were perfused with PBS and then 4% PFA in PBS (20 ml each). Paraffin-embedded sections were washed with xylene for 2 X 5 min to deparaffinize the tissue. Sections were re-hydrated in consecutive ethanol washes (100% to 70%) before resuspension in 1X Tris-buffered saline for 15 min. All washes were performed at room temperature. For antigen retrieval, sections were microwaved in 10 mM citrate buffer, pH 6.0, for 5 min at 40% power in a 1100 W microwave oven (Sanyo). Sections were allowed to cool in the same buffer for 20 min at RT, then transferred to 1X TBS for 10 min. For immunostaining, sections were blocked in 10% normal donkey serum in 1X TBS for 1h at RT with chicken anti-ms IgG at 1:500. Primary antibody was diluted in 1% BSA in 1X TBS as described above and incubated overnight at 4°C. All subsequent steps were performed at RT. Slices were washed 3×10 min in 1X TBS. Secondary antibody was diluted 1:500 in 1% BSA in 1X TBS and applied to slices for 1 h. Ataxin-7 antibody K (1:200) was labeled with Alexa 488 donkey anti-rabbit; and LC3 (1:50) was labeled with Alexa 555 donkey anti-mouse. Following incubation in secondary antibody, slices were washed 3×20 min in 1X TBS and mounted in Prolong Gold with DAPI.
We have previously found that caspase-7 cleavage of ataxin-7 at D266 is an important determinant of cellular toxicity (Young et al., 2007) and in the progression of SCA7 in transgenic mice (Guyenet et. al., unpublished results). In other polyglutamine diseases, the interplay between ubiquitination and sumoylation of target lysines can strongly influence intracellular protein accumulation and toxicity (Steffan et al., 2004). To determine whether post-translational modification plays a role in ataxin-7 turnover, we analyzed the ataxin-7 protein for modification consensus sequences (Fig. 1). For sumoylation, we used both SUMOplot (Abgent) and SUMOsp (http://sumosp.biocuckoo.org/). SUMOplot is a strong predictor of consensus sequences, while SUMOpre incorporates mammalian phylogenetic similarity into its scoring algorithm to weight the consensus sites that are of particular evolutionary significance in mammals. Of the initial predictions, the two most highly predicted sites, K223 and K257, were both proximal to the D266 caspase-7 cleavage site (Fig. 1A,B). These residues are also adjacent to predicted phosphorylation sites at S224 and S255 (Fig. 1C), suggesting the potential for functional interaction between them. In addition, since acetylation/deacetylation-SUMOylation switch has been shown to regulate transcription factors/repressors such as MEF2 or HIC (Stankovic-Valentin et al., 2007), ataxin-7 lysines could be targeted by both sumoylation and acetylation.
To demonstrate that ataxin-7 is modified by SUMO, c-myc tagged ataxin-7 was immunopurified and incubated with the E1 and E2 ligases required for sumoylation in addition to SUMO-1 (+) or a non-conjugatable SUMO-1 mutant (−). Consistent with our predictions, ataxin-7 is modified by sumoylation (Supplementary Fig. 1). The major band migrated with an apparent molecular weight of 100 kDa (120 kDa for ataxin-7-92Q), whereas the slower migrating band showed an ~14-kDa increase in molecular mass, consistent with the addition of a single SUMO moiety. Indeed, the same bands were observed in 293T cells co-transfected with myc-ataxin-7 and SUMO-1 expression vectors (data not shown). Using a similar immunoprecipation strategy, we also found that ataxin-7 and fragments expressed in 293T cells or in are immunoreactive to anti-acetylation antibodies (Fig. 2A). In addition, we found that ataxin-7 immunoprecipitated cellular lysates from MoPrP-flag-SCA7-10Q-myc or MoPrP-flag-SCA7-92Q-myc transgenic mice are immunoreactive to anti-acetylation antibodies (Fig. 2B) suggesting the modification occurs in vivo.
To determine whether modification of the predicted ataxin-7 sumoylaton sites K223 or K257 are important for regulating the stability of an ataxin-7 cleavage fragment, we made single lysine to arginine substitutions at these residues in a truncated ataxin-7-10Q or 92Q construct expressing amino acids 1–266. This protein is identical to the fragment generated by caspase-7 cleavage at D266 of ataxin-7. Substitution of arginine (R) for lysine (K) maintains size and charge characteristics, but is not a substrate for post-translational modification. Constructs expressing the ataxin-7-10Q (1–266), ataxin-7-10Q (1–266) K223R, ataxin-7-10Q (1–266) K257R, ataxin-7-92Q (1–266), ataxin-7-92Q (1–266) K223R, and ataxin-7-92Q (1–266) K257R fragments were transiently transfected into HEK293T cells and protein was analyzed by Western blot analysis (Fig. 3A,B).
As shown in Fig. 3A, these mutations result in loss of the shifted bands (14-kDa increments) that represent modified forms of ataxin-7-10Q or -92Q proteins (denoted with asterisks), as well as the high molecular weight material in the lanes with the ataxin-7-92Q proteins consistent with aggregation. In addition, protein levels of the ataxin-7 K257R mutants are significantly decreased, indicating that modification at this site regulates the stability or turnover of the ataxin-7 (1–266) fragment. Moreover, this effect is polyQ repeat-dependent, conferring a much larger (~6-fold) decrease in the context of 10Q than 92Q (<2-fold) (Fig. 3B). These results were also observed in primary cerebellar neurons transiently transfected with the ataxin-7 fragment (Fig. 3C), indicating that the molecular consequences of the K257R mutation are consistent across cell types and validating the use of HEK293T cells for further molecular characterization. Of note, in primary cerebellar neurons, the clearance of the ataxin-7-92Q fragment is impaired when compared to ataxin-7-10Q fragment and mutation at K257 promotes clearance of both proteins.
To confirm that the reduction in ataxin-7 (1–266) K257R protein reflected loss of total protein and not the accumulation of insoluble aggregates, we performed filter trapping of HEK293T lysate from cells expressing ataxin-7-10Q and ataxin-7-92Q, with either no lysine substitutions, K223R, or K257R, onto cellulose acetate membranes. As shown in Fig. 3D, we found that the 10Q-containing fragments do not aggregate, consistent with our findings from Western blot analysis, and the 92Q-containing fragments aggregate proportionately to the amount of soluble protein for each mutant, indicating that K257R mutation confers loss of total protein and not simply reduction of soluble protein. Finally, this modulation appears to be specific to the 1–266 fragment, as the K257R mutation does not measurably alter the turnover of full-length ataxin-7 (data not shown).
Since the K257R mutation appears to substantially decrease the stability of the 1–266aa ataxin-7 fragment, we were interested in identifying whether degradation was mediated by the ubiquitin proteasome system (UPS) or through autophagy. To identify the mechanism of degradation for the K257R mutant fragments, we chemically disrupted these pathways using epoxomicin and 3-methyladenine (3-MA), respectively (Fig. 4). Epoxomicin binds to subunits of the 26S proteasome and selectively disrupts its activity, while 3-MA blocks autophagy by inhibiting class III PI3K activity, preventing autophagosome formation. HEK293T cells were transiently transfected with the N-terminal 1–266 fragments of the ataxin-7, ataxin-7-10Q (1–266), ataxin-7-10Q (1–266) K223R, ataxin-7-10Q (1–266) K257R, ataxin-7-92Q (1–266), ataxin-7-92Q (1–266) K223R or ataxin-7-92Q (1–266) K257R and treated with either epoxomicin (25 μM) or 3-MA (10 mM) 48 hours after transfection. As shown in Fig. 4A,B, Western blot analysis demonstrated that UPS inhibition had a negligible effect on ataxin-7-10Q or -92Q protein levels, while disruption of macroautophagy resulted in an increase in ataxin-7. Addition of 3-MA resulted in a significant increase in the steady-state levels of ataxin-7-10Q (1–266), ataxin-7-10Q (1–266) K223R, ataxin-7-10Q (1–266) K257R, ataxin-7-92Q (1–266) K223R or ataxin-7-92Q (1–266) K257R. In particular, the levels of ataxin-7 (1–266) K257R or K223R proteins were similar to the unmodified ataxin-7 fragment. Interestingly ataxin-7-92Q (1–266) levels were not altered by either 3-MA or epoxomicin.
We also found that blocking macroautophagy in primary neurons expressing ataxin-7 resulted in similar increases of ataxin-7 protein levels (data not shown). These findings support a primary role for autophagy, but not proteasomal degradation, in the turnover of the ataxin-7 fragments. Relevant to the gain-of-function properties for polyglutamine expansions, the ataxin-7-92Q (1–266) was resistant to degradation by autophagy. Since mutation at either K223 or K257 resulted in a protein susceptible to degradation, it is likely that post-translational modification at these sites plays a critical role in blocking degradation of the mutant fragment.
Several covalent posttranslational modifications, such as ubiquination, sumoylation, methylation, and acetylation can target lysine residues. Post-translational modifiers of ataxin-7 stabilization could be mediated by small ubiquitin-like protein (SUMO-1). SUMO-1 is one of a family of small ubiquitin like proteins (SUMO-1-4) described to participate in a multitude of cellular functions including transcription, nuclear translocation, and inflammation. SUMO-1 directly modifies proteins using three steps similar to ubiquitination using the E1 equivalent activating complex (SAE I and II) and E2 also known as Ubc9. Another enzymatic pathway that modifies lysines is histone acetylases (HATs). Transcriptional regulatory proteins, including GCN5, PCAF, p300, CBP, TAFII250 possess intrinsic histone acetyltransferase HAT activity acetylase. The role of lysine acetylation in regulating nuclear proteins and transcriptional factors has been well established. Histone acetylation affects chromatin structure and gene expression; tumor suppressor protein p53 can be acetylated at multiple sites, and different acetylation modifications have distinct effects on p53 function. Relevant to our studies, ataxin-7 is in the acetytransferase complex, STAGA, which contains the acetyltransferase GCN5.
To characterize the contribution of possible lysine modification, including ubiquitination, sumoylation, and acetylation, to ataxin-7 stability and turnover, we co-transfected ataxin-7-10Q or -92Q (1–266) with constructs expressing ubiquitin, SUMO1, GCN5 or CBP (Fig. 5A). We found that only one of the lysine modifying enzymes, CBP, when co-expressed, led to significant increases in ataxin-7 accumulation, while GCN5 showed a modest decrease. This suggests that acetylation may play a role in stabilizing the ataxin-7 fragments. Unexpectedly, ubiquitin or sumoylation co-expression did not alter or increased the ataxin-7 fragment levels. Our initial prediction was that, like huntingtin, ubiquitin over-expression would speed degradation of ataxin-7, while SUMO1 over-expression might result in the stabilization of the protein (Bence et al., 2001; Steffan et al., 2004; Bennett et al., 2007).
Since CBP co-expression with ataxin-7 increased protein levels, we investigated the effect of CBP on the steady state levels of ataxin-7 (1–266) and the K257 mutant. As shown in Fig. 5B,C, we found that the ataxin-7-10Q K257R did not reach the same levels as the ataxin-7-10Q (1–266) with CBP co-expression suggesting that acetylation of ataxin-7 at this site contributes to ataxin-7 stabilization. The ataxin-7-92Q K257R gave a similar effect when both the soluble and insoluble fractions were accounted for (Fig. 5B).
Histone deacetylases (HDACs) maintain the acetylation levels of histone and non-histone proteins. Mammalian HDACs are classified into four classes: type I, type IIa,b, type III and type IV. Class I and II HDACs have an active site zinc coordinated by histidine and aspartate residues. The class I HDACs (HDAC1, HDAC2, HDAC3 and HDAC8) are widely expressed in most tissues and are localized in the nucleus (shuttle-contain import/export nuclear localization signals). Class IIa HDACs (HDAC4, HDAC5, HDAC7 and HDAC9) contain a bipartite structure with an N-terminal domain which interacts with transcription factors and class I HDAC3. Since we found lysine acetylation influences clearance of the ataxin-7 fragments, we investigated whether modulation of deacetylation activity by siRNA knockdown of HDACs altered ataxin-7 fragment levels. To assess the effect of HDAC activity on ataxin-7 stability and clearance, we screened siRNA to HDAC 2 through 8 cotransfection with ataxin-7-10Q (1–266) or ataxin-7-92Q (1–266) fragments (data not shown). Among the seven HDACs tested, we found HDAC7 modulates the turnover of ataxin-7. Upon knockdown of HDAC7, we observed an increase ataxin-7-10Q (1–266) or ataxin-7-92Q (1–266) fragments levels compared to controls (Fig. 6). Thus, co-transfection experiments with HDAC7 siRNA resulted in the accumulation of ataxin-7.
Next we performed immunofluorescence analysis on primary cerebellar neurons over-expressing the 10Q or 92Q ataxin-7 1–266 fragment with no mutation, or the K223R, or K257R mutation (Fig. 7A). One question relevant to our study is if the K257R mutant, which is rapidly degraded by autophagy, has altered localization. As reported previously, we find that ataxin-7 displays a primarily nuclear localization with some cytoplasmic staining. We found that while the K223R mutant appears to localize similarly to the non-mutant fragment, the K257R mutant displays primarily a cytoplasmic localization in addition to increased levels of endogenous LAMP-2, a lysosomal marker and protein involved in autophagy.
Additionally, we performed co-staining of cells over-expressing ataxin-7 and ubiquitin to determine if ubiquitin-positive staining was similarly altered in the presence of the ataxin-7 K257R mutant (Fig. 7B). Consistent with the localization observed in Fig. 6A, we observed altered ataxin-7-K257R fragment localization in the cytoplasm, coincident with cytoplasmic ubiquitin staining. Interestingly, in cells expressing the ataxin-7-K257R fragment, the ubiquitin staining appears more intense and dispersed than in the other ataxin-7-expressing cells.
Previous characterization of SCA7 transgenic models has demonstrated the production of proteolytic cleavage products (Zander et al., 2001; Garden et al., 2002; Young et al., 2007). In order to evaluate if autophagy is relevant to disease pathogenesis, we characterized a newly generated MoPrP-flag-SCA7-10Q-myc and MoPrP-flag-SCA7-92Q-myc transgenic mouse model ((La Spada et al., 2001); Guyenet, et al., unpublished results) for LC3 expression and the presence of punctate LC3 positive structures. The appearance of LC3-positive puncta is indicative of the induction of autophagy. As previously noted, MoPrP-flag-SCA7-92Q-myc transgenic mice have a dramatic neurological phenotype, premature death and accumulate proteolytic fragments (Guyenet, et al., unpublished results). As expected, expression of the polyglutamine-expanded ataxin-7 causes severe Purkinje cell degeneration (Fig. 8A). Unlike our previous model, we detected robust expression of ataxin-7 in the Purkinje cells (Fig. 8A). Using an antibody to LC3, immunostaining of cerebellum from MoPrP-flag-SCA7-92Q-myc mice demonstrated an increase in the expression of LC3 (Fig. 8A) as well as increased LC3 positive structures (Fig. 8B) in the Purkinje cells when compared to MoPrP-flag-SCA7-10Q-myc transgenic mice or non-transgenic littermates. Quantification of LC3-positive puncta revealed the MoPrP-flag-SCA7-92Q-myc Purkinje cells had at least five puncta per cell while the MoPrP-flag-SCA7-10Q-myc did not (Fig. 8C). This suggests the autophagy is induced in the MoPrP-flag-SCA7-92Q-myc transgenic mice.
Identifying post-translational modifications of polyglutamine proteins and understanding their downstream effect has become increasingly important in determining the mechanisms of polyglutamine disease pathogenesis. There are multiple examples in HD and the polyglutamine–dependent SCAs demonstrating that phosphorylation, ubiquitination, SUMOylation, and acetylation affect key behaviors of these proteins that are often altered in disease, including: 1) localization into the nucleus (Riley et al., 2005), mediated by phosphorylation-dependent SUMOylation; and 2) protein stability and accumulation, mediated by competitive binding of ubiquitin and SUMO (Steffan et al., 2004). In this study, we were interested in the turnover of polyglutamine-containing ataxin-7 fragments following cleavage by caspase-7, specifically, whether their accumulation could be modulated by post-translational modification. The clearance of toxic protein forms is an important factor in ameliorating the so-called “proteinopathies” neurodegenerative diseases caused by protein misfolding and aggregation (for review, see (Williams et al., 2006)). In this study, we provide evidence that Lys-257, near the Asp-266 cleavage fragment C-terminus (Young et al., 2007), is a modification target that alters ataxin-7 fragment accumulation in a repeat-dependent manner.
One of the most striking findings of this study is that a proteolytic cleavage product of caspases, in this case the substrate is ataxin-7, is specifically regulated by autophagy. This mechanism of regulation may have broader implications in the regulation of cell death pathways by caspases. It is possible that caspase cleavage products are not indiscriminately degraded but in some cases the process is regulated by post-translational modification. This form of regulation may have implications for caspase cleavage products that act as amplifiers of the cell death process or as dominant-negatives to inactivate survival complexes. Since CBP, but not GCN5, had an impact on ataxin-7 fragment turnover, we suggest that acetylation/deaceylation of proteolytic fragments generated during cell death may represent an important level of regulation by specific HATs/HDACs for accelerating or attenuating apoptosis. Consistent with the role of acetylation in regulating cell death, Ku70 acetylation promotes the dissociation of Bax from Ku70 and promotes cell death (Li et al., 2007). Further caspase cleavage of HDAC3 results in cytoplasmic accumulation, acetylation of nuclear targets and concurrently enhanced cell death (Escaffit et al., 2007).
The role of autophagy in ameliorating disease-induced toxicity has been explored in a number of proteinopathy models, including AD and PD, as well as the polyglutamine models HD and SCA3 (Yamamoto et al., 2006; Bilen and Bonini, 2007). The general proteinopathy model posits that misfolded protein aggregates, polyglutamine-containing or otherwise, are refractory to UPS-mediated degradation and that the resulting intracellular accumulation triggers the numerous downstream effects that lead to cell dysfunction and death. Since autophagy is a large-scale, generally non-selective process (excluding chaperone-mediated autophagy), the misfolded aggregates that are no longer subject to UPS-mediated proteolysis are still available for autophagic degradation. Thus, increased activation of autophagy may compensate for proteasomal malfunction, clearing protein aggregates and attenuating downstream aggregation-dependent malfunction. In HD, autophagy has been shown to specifically mediate degradation of a polyQ-containing N-terminal fragments of huntingtin (Qin et al., 2003). Moreover, the induction of autophagy to ameliorate neurodegeneration in HD models has been extensively investigated (Sarkar et al., 2008). Our results extend these studies and suggest that caspase generated cleavage products from polyQ disease proteins are cleared by autophagy and post-translational modifications may regulate this process. In accord with a significant role for autophagy in neurodegeneration, we found that SCA7 mice have increased autophagy in Purkinje cells as assessed by LC3-positive autophagosome number and LAMP2 immunochemistry. The upregulation of autophagy is likely a compensatory response to the accumulating build-up of proteolytic ataxin-7-92Q cleavage products.
Like the “histone code” of histone modification, it is becoming increasingly clear that non-histone-associated proteins are subject to the same modifications, potentially with similar regulatory effects. For ataxin-7, we find that lysine modification has significant consequences regarding stability of the 1–266 fragment generated by caspase-7 cleavage. Since ataxin-7 is a member of the STAGA complex, which also contains the histone acetyltransferase GCN5, we hypothesized that ataxin-7 might be subject to GCN5-dependent acetylation. However, results from this study suggest that CBP, not GCN5, may be the relevant acetyltransferase targeting the proteolytic cleavage product of ataxin-7. CBP has long been implicated in polyglutamine-dependent neurodegeneration. Evidence that CBP is sequestered in polyglutamine aggregates, accompanied by histone hypoacetylation and transcriptional repression, has led to a model in which CBP depletion through sequestration in aggregates prevents its normal function in transcriptional activation of CREB-dependent gene expression (Rouaux et al., 2004). This model has been debated, although it was recently shown that CBP sequestration was not always accompanied by nuclear depletion, and that both were required for toxic effects (Jiang et al., 2006), perhaps explaining some of the previous conflicting results. In addition, over-expression of CBP has been shown to mitigate polyglutamine-dependent toxicity in neuronal models, supporting an important role for CBP function in neuronal stability. In this study, over-expression of CBP with ataxin-7 1–266 leads to a significant increase in the accumulation of this fragment, suggesting that it directly acetylates ataxin-7. This is consistent with Western blot analysis with acetylation antibodies in which the ataxin-7 1–266 is acetylated (data not shown). Of note, we found that GCN5 promoted the degradation of full-length ataxin-7 suggesting acetylation regulates normal turnover by a distinct mechanism (data not shown).
Additionally, we find that the modification-dependent regulation of these fragment levels is decreased in the presence of an expanded polyglutamine tract; in this case, 92Q vs. 10Q. This dampening effect may be one mechanism by which expanded polyQ fragments accumulate to aggregation-prone or toxic levels. That cleavage products of polyQ-containing proteins may be specifically regulated by post-translational modification is a previously unreported mechanism. Moreover, the idea that polyQ expansion disrupts this regulation, contributing to fragment accumulation and toxicity, is also a novel explanation for the deleterious effects of polyQ expression on the cell.
Ataxin-7 belongs to the STAGA multiprotein acetylation complex; however it remains unclear what the direct interactions are between this protein and the components of the acetylation/deacetylation machinery, which ultimately controls gene expression patterns. The critical role of deacetylation in ameliorating symptoms of SCA7 phenotype has been demonstrated by a number of research groups. Latouche and colleagues in a SCA7 Drosophila model demonstrated that administration of the HDAC inhibitor sodium butyrate was neuroprotective (Latouche et al., 2007). Interestingly, a different group suggested that treatment with the same HDAC inhibitor relieved the expression of CRE/CBP transcription co-activators in a normal and expanded ataxin-7 cellular model (Strom et al., 2005). This observation further strengthened the critical role of ataxin-7 as a modulator of the acetylation/deacetylation balance, which is altered by the presence of CAG repeats in the SCA7 phenotype. In the present study, we examined whether there is a functional interaction of ataxin-7 and a specific member of the HDAC family. Screening of seven HDACs (2–8) suggested that HDAC7 predominantly regulates ataxin-7 fragment turnover and stability. The regulatory role of HDAC7 on ataxin-7 levels was further confirmed by disruption of HDAC7 expression using siRNA knockdown technology, which resulted in a build up of ataxin-7 levels. Those findings may be particularly useful for future studies designing therapeutic agents associated with a specific class of HDACs to promote ataxin-7 clearance, and most importantly, restoring aberrant gene expression patterns caused by the expanded CAG repeats.
In summary, we demonstrate that N-terminal ataxin-7 fragments resulting from caspase cleavage are subject to regulated clearance. The regulatory mechanism appears to be post-translational modification, as disruption of modification sites through mutation alters the fragment stability. Importantly, polyQ expansion partially inhibits the regulatory effects of post-translational modification. We propose that the de-regulation of proteolytic product turnover is an important mediator of polyQ-induced toxicity and cell death. This work may suggest that deacetylation of the ataxin-7 fragment, for example, by promoting HDAC activity, may have an ameliorate SCA7 by promoting fragment clearance. Our results also emphasize the importance of protein context in understanding the same post-translational modification in polyglutamine expansion diseases. For ataxin-7 fragments, acetylation is found to stabilize the protein while a recent study for huntingtin demonstrates acetylation of mutant huntingtin increases the turnover of the protein (Jeong et al., 2009). Future work will evaluate the role of the proteases in the lysosome that are involved in the clearance of the ataxin-7 fragment.
SUMO1 modification of full-length ataxin-7. C-myc tagged ataxin-7 was immunopurified and incubated with the E1 and E2 ligases required for sumoylation in addition to SUMO-1 (+) or a non-conjugatable SUMO-1 mutant (−). p53 is shown as a positive control for sumoylation. Western blots were probed with c-myc antibody (top) and SUMO-1 antibody (bottom). Asterisks denote modified ataxin-7.
This work was supported by the NIH NS040251 (LME), NIH NS062413 (LME), NIH EY14061 (ARL) and the Nathan Shock P30AG025708. SM was supported by NIH training grant T32 AG000266. We would like to thank Caitlin Rugani for genotyping of SCA7 mice, Junli Zhang for carrying out injection of the SCA7 10Q transgene, Dr. Aleksey Kazantsev for the kind gift of CBP-MT and Yingming Zhao for HA-SUMO1. Thanks to Juliette Gafni for suggestions on the work and manuscript.