|Home | About | Journals | Submit | Contact Us | Français|
NDJ generated the Akt-DN mice and generated the data presented in Fig. 3. CEB generated the data in Fig. 1, LAD generated the data in Fig 2(c, d) and Fig 5(a), HBC generated the data in Fig. 4. SS initiated development of the cerebellar extract phosphorylation assay and generated the data in Fig. 6 except for Fig. 6c contributed by SL data in Fig. 7 were obtained by JMA and BA, in Fig. 8 by BA, and in Fig. 9 by SL. PJN contributed the data in Fig. S1. Data analyses and interpretation were performed by NDJ, JMA, HYZ, HBC, and HTO. NDJ and HTO wrote the paper. NDJ, JMA, HYZ, HBC, and HTO all edited the paper.
Spinocerebellar ataxia type 1 (SCA1) is one of nine inherited neurodegenerative disorders caused by a mutant protein with an expanded polyglutamine tract. Phosphorylation of ataxin-1 (ATXN1) at serine 776 is implicated in SCA1 pathogenesis. Previous studies, utilizing transfected cell lines and a Drosophila photoreceptor model of SCA1, suggest that phosphorylating ATXN1 at S776 renders it less susceptible to degradation. This work also indicated that oncogene from AKR mouse thymoma (Akt) promotes the phosphorylation of ATXN1 at S776 and severity of neurodegeneration. Here, we examined the phosphorylation of ATXN1 at S776 in cerebellar Purkinje cells, a prominent site of pathology in SCA1. We found that while phosphorylation of S776 is associated with a stabilization of ATXN1 in Purkinje cells, inhibition of Akt either in vivo or in a cerebellar extract-based phosphorylation assay did not decrease the phosphorylation of ATXN1-S776. In contrast, immunodepletion and inhibition of cyclic AMP-dependent protein kinase decreased phosphorylation of ATXN1-S776. These results argue against Akt as the in vivo kinase that phosphorylates S776 of ATXN1 and suggest that cyclic AMP-dependent protein kinase is the active ATXN1-S776 kinase in the cerebellum.
Spinocerebellar ataxia type 1 (SCA1) is caused by the expansion of a cytosine-adenosine-guanosine trinucleotide repeat that encodes a polyglutamine tract in the ataxin-1 (ATXN1) protein. Other inherited neurodegenerative diseases resulting from this pathogenic mechanism include Huntington’s disease, dentatorubral pallidoluysian atrophy, spinobulbar muscular atrophy, and SCAs 2, 3, 6, 7, and 17 (Orr and Zoghbi 2007).
While the ATXN1 (previously designated as the SCA1 gene) encoded protein ATXN1 is widely expressed in the CNS, the most frequent and severe pathological alterations are in Purkinje cells of the cerebellar cortex (Robitaille et al. 1995). Several studies indicate that the normal function of ATXN1 is linked to pathogenesis. For example, ATXN1 is normally located in the nucleus of Purkinje cells and mutant ATXN1 must enter the nucleus of Purkinje cells for it to cause disease (Klement et al. 1998). Subsequent studies revealed that wild-type ATXN1 has properties consistent with a role in the regulation of gene expression. These normal activities of ATXN1 include an bility to bind RNA (Yue et al. 2001), to shuttle between the nucleus and cytoplasm (Irwin et al. 2005), and to interact with several transcription factors (Okazawa et al. 2002; Tsai et al. 2004; Tsuda et al. 2005; Lam et al. 2006; Serra et al. 2006).
One transcription factor to which ATXN1 binds is the repressor Capicua (Lam et al. 2006). In the cerebellum, the majority of ATXN1 assembles into a soluble complex that includes Capicua. Moreover, SUMOylation is a type of post-translation modification often associated with regulating gene transcription (Seeler and Dejean 2003), and ATXN1 is SUMOylated on at least five lysine residues (Riley et al. 2005). ATXN1 interacts with the transcription factors Retinoic acid receptor-related Orphan Receptor (ROR) α and the RORα coactivator Tip60, and expression of mutant ATXN1 decreases mRNA transcripts regulated by the RORα complex (Serra et al. 2006). RORα is known to regulate the expression of several genes whose products are important for Purkinje cell development and function (Gold et al. 2003).
Phosphorylation of regulatory proteins is one mechanism that links neuronal function with changes in gene expression (e.g. Zhou et al. 2006). Interestingly, phosphorylation of ATXN1 at serine 776 is implicated in regulating its function as well as SCA1 pathogenesis (Chen et al. 2003; Emamian et al. 2003). The phosphorylation-resistant serine to alanine (S776A) substitution mitigates the ability of ATXN1[82Q] to induce neurodegeneration in transgenic mice even when the mutant ATXN1 harbored an expanded polyglutamine tract (Emamian et al. 2003). This amino acid change also affects the SUMOylation of ATXN1 in tissue culture (Riley et al. 2005) and the formation of large soluble complexes between ATXN1 and other cellular proteins (Lam et al. 2006). In addition, phosphorylation at S776 and length of the polyglutamine tract, regulate the interaction of ATXN1 with the RNA-binding motif protein RBM17 (Lim et al. 2008). Moreover, it is the large ATXN1/RBM17 containing complex that likely contributes to disease. Thus, insight into the cellular pathways that regulate the phosphorylation of ATXN1 at S776 is critical for understanding the function and pathogenesis induced by mutant ATXN1.
Previous work (Chen et al. 2003), using a Drosophila model of SCA1 and transfected tissue culture cells, suggested two important aspects of phosphorylating ATXN1 at S776. First, these studies showed that phospho-S776-ATXN1 is less susceptible to degradation than the phosphorylation-resistant A776-ATXN1. This work also suggested that oncogene from AKR mouse thymoma (Akt) promoted the phosphorylation of ATXN1 at S776 thereby increasing severity of disease. More recent work indicates that the effect phosphorylation has on ATXN1 is cell-type dependent (Jorgensen et al. 2007). In this study, we utilized transgenic mice and a cerebellar extract-based phosphorylation assay to investigate the effect of S776 on stability of ATXN1 in vivo and the regulation ATXN1-S776 phosphorylation in the mammalian cerebellum.
ATXN1[30Q]-S776 mice were from the A02 line and ATXN1[82Q]-S776 mice from line B05 (Burright et al. 1995). The ATXN1[30Q]-A776 transgene was generated by adding a point mutation in the original construct used to make the ATXN1[30Q]-S776 mice. The Akt-dominant negative mutation (DN) transgene was made by inserting K179M-Akt into the L7/Pcp2 gene (Oberdick et al. 1990). The Akt transgene was linearized with BamHI, gel isolated, purified by phenol/chloroform extraction, chloroform extraction and ethanol precipitation, and suspended in injection buffer (10 mM Tris–Cl, pH 8.0; 0.1 mM EDTA) at a concentration of 4 ng/mL. Embryo injections were performed by the Mouse Genetics Laboratory, University of Minnesota. PCR and Southern blot analyses were used to identify transgene positive animals.
The generation and genotyping of Sca1154Q/+ mice has been described (Watase et al. 2002). Previously described Akt1+/− mice (Cho et al. 2001) were generous gift from Dr M. J. Birnbaum (University of Pennsylvania). All mice were on the C57Bl/6 background. We crossed Sca1154Q/+ to Akt1+/−. The first generation consisted of wild-type, Akt1+/−, Sca1154Q/+, and Sca1154Q/+; Akt1+/− were subjected to rotarod analysis as previously described (Watase et al. 2002), using 7-week-old naïve animals. Rotarod performance was analyzed by repeated-measures anova.
Immunofluorescence staining of mouse cerebellum was performed using a rabbit monoclonal antibody that recognizes phosphoglycogen synthase kinase-3 (GSK-3) α/β, Ser21/9 (#9327; Cell Signaling Technology, Beverly, MA, USA). Fifteen-week-old FVB/n and AKT-DN mice were perfused and fixed as previously described (Emamian et al. 2003). Vibratome sections were stained with the anti-phospho-GSK-3 antibody at a dilution of 1 : 200. Rabbit anti-Alexa 488 (Invitrogen, Carlsbad, CA, USA) was used at a 1 : 500 dilution. Fluorescent images were scanned using a Fluoview 1000 IX2 inverted microscope (Olympus, Minneapolis, MN, USA). FVB/N and AKT-DN mice were imaged using identical confocal parameters.
Immunohistochemical staining was performed as previously described (Klement et al. 1998). Immunohistochemistry was performed on 5 mm microtome sections from paraffin-embedded brains. After rehydration epitopes were unmasked by steaming for 10 min in pH 6.0 citrate buffer. Staining was carried out using the ABC Elite kit (Vector, Burlingame, CA, USA). Sections were blocked for 20 min in normal serum; incubated overnight at 4°C (11750, 1 : 320 000 or PN1168, 1 : 1.3 × 106), washed briefly, and incubated for 30 min at 25°C with biotinylated anti-rabbit secondary antibody. The sections were washed, incubated with ABC reagent, washed again, exposed for several min to 3′, 3′ diaminobenzidine (DAB) substrate, washed, counter-stained with hematoxyline, dehydrated in graded alcohols and xylol, and mounted.
TreadScan 2.0 (Clever Systems Inc., Reston, VA, USA) was used for gait analysis as described by Wooley et al. (2005). Four groups were tested, at a speed of 25 cm/s: non-transgenic FVB/n (n = 5–8); ATXN1[82Q] B05 (n = 6–10); Akt-DN (n = 7); Akt-DNx ATXN1[82Q] B05 (n = 4 or 5). Trials were begun with naïve mice at 4 weeks of age and repeated at 5, 7, 9, and 11 weeks of age. Twenty second videos (2000 frames) were recorded for each mouse per trial and a three sec section of 300 consecutive frames (11–14 steps) were selected for analysis. Rear stride time is the time between two sequential paw contacts with the treadmill for the same paw. Rear stride time values were an average of right and left paws. Stride time was measured in millisec by the TreadScan software. Parameters were calculated for consecutive strides of each foot over three sec (11–14 steps). The median values for the right and left rear feet were averaged. For statistical analysis groups were compared by t-test analysis.
Five micrograms of purified Glutathione-S-transferase (GST)-ATXN1 and 25 µg of mouse (FVB/n) cerebellar lysates were combined with 0.1 mM ATP in lysis buffer (0.25 M Tris–Cl, pH 7.5 containing 1× protease inhibitors (Roche Biochemicals, Indianapolis, IN, USA) and 1× phosphatase inhibitor cocktails 1 and 2 (Sigma, St Louis, MO, USA). The reaction was incubated for 1 h at 30°C and loaded on a 4–12% Bis-Tris polyacrylamide gel (Invitrogen). After electrophoresis, proteins were transferred to a nitrocellulose membrane (Protran, Piscataway, NJ, USA) and blocked with 5% milk in 1× phosphate-buffered saline (PBS) with 0.1% Tween-20. The membrane was probed with pS776 ATXN1 polyclonal antibody PN1168 (Emamian et al. 2003) at 1 : 1000 for 2 h in blocking solution followed by three washes with 1× PBS with 0.1% Tween-20. The membrane was probed with an enhanced chemiluminescence-anti-rabbit IgG-horseradish peroxidase secondary antibody (GE Healthcare, Piscataway, NJ, USA) at 1 : 2500 western blots were developed using the SuperSignal West Pico Chemiluminescent Kit (Thermo-Scientific, Rockford, IL, USA). Cerebellar nuclear and cytoplasmic fractions were prepared using the NE-PER Nuclear and Cytoplasmic Extraction Reagents (Pierce Biotechnology, Rockford, IL, USA) according to the manufacture’s protocol. Briefly, cerebella were homogenized in Cytoplasmic Extraction Reagent I, vortexed for 15 s, and incubated on ice for 10 min in Cytoplasmic Extraction Reagent II. Extract was centrifuged for 5 min at 16 000 g and supernatant (cytoplasmic extract) collected. Pellet was suspended in 200 µL ice-cold 1× PBS, spun for 5 min (13 000 g), pellet suspended in ice-cold Nuclear Extraction Reagent, vortexed extensively and centrifuged for 10 min (16 000 g) and supernatant, nuclear extract, collected. Fraction purity was assessed by western blotting using an anti-IkB antibody (Cell Signaling) as a marker for a cytoplasmic protein and anti-histone-3 (Cell Signaling) as a nuclear marker.
Cerebella were dissected from FVB/n mice and flash frozen in liquid nitrogen. Thawed cerebella were suspended in 0.3 mL lysis buffer [120 mM sodium chloride, 20 mM Tris-chloride pH 7.5, 1 mM phenylmethylsulphonyl fluoride, 1 mM dithiothreitol (DTT), 1× Roche protease inhibitors, and 1× Sigma phosphatase inhibitor cocktails 1 and 2] per cerebellum. Homogenates were generated on ice in a 7.5- or 15-mL Dounce homogenizer, typically with 18 strokes of the ‘A’ pestle and 36 strokes of the ‘B’ pestle. Homogenates were centrifuged for 5 min at 16 000 g to pellet nuclei and large membranes. A saturated solution of ammonium sulfate with 10 mM EDTA was used to generate ammonium sulfate cuts typically starting with a 30% cut and ending with a 90% cut. An appropriate volume of saturated ammonium sulfate was added to the cerebella lysate and incubated with rotation at 4°C for 15–30 min. Precipitated proteins were centrifuged at 16 000 g for 5–15 min. Pellets were suspended in a small volume of lysis buffer and supernatants were subjected to further ammonium sulfate fractionation. Fractions were dialyzed (10 000 molecular weight cutoff) twice in 4 L of dialysis buffer (120 mM sodium chloride, 20 mM Tris-chloride pH 7.5, and 1 mM DTT) and cleared by centrifuging at 16 000 g for 5 min. Supernatants were used for kinase assays.
For hydroxyapatite fractionation, A 50–90% ammonium sulfate cut of mouse cerebellar lysate was loaded onto a 10-mL hydroxy-apatite (Sigma) column in dialysis buffer and eluted with a 0–720 mM gradient of sodium phosphate pH 7.5 in dialysis buffer. The fifteen fractions with the most eluted protein were collected and analyzed by western blot or in a kinase assay. Hydroxyapatite fractions were not dialyzed, as 720 mM sodium phosphate only minimally inhibits the kinase reaction (data not shown).
Kinase reactions were set up with 10 µg bacterially purified fusion protein between glutathione-S-transferase and fragment Vof human ATXN1 and a variable amount (typically 1–20 µg) of cerebellar lysate (or fractionated cerebellar lysate). These were incubated for 60 min at 30°C in 30 µL reactions containing 5 µCi 32P-adenosine triphosphate (3000 Ci/mmol) in kinase buffer (45 mM Tris-chloride pH 7.5, 33 µM cold ATP, 40 mM glycerol phosphate, 10 mM magnesium chloride, 1 mM ethylene glycol tetraacetic acid, and 1 mM DTT). Glutathione agarose resin (Invitrogen) was washed twice in wash buffer (120 mM sodium chloride, 20 mM Tris-chloride pH 7.5, 50 mM EDTA, 0.05% NP-40, and 1 mM DTT). After kinase reaction incubation, 15 µL packed beads in 1 mL wash buffer was added to each tube. Tubes were incubated with shaking at 25°C for 15 min. Beads were washed 3 times in wash buffer, spinning briefly in between at 16 000 g to pellet the agarose beads. After the final wash, fusion protein was removed from the beads with addition of 2× sample buffer (125 mM Tris-chloride pH 6.8, 2% sodium dodecyl sulfate, 20% glycerol, 0.01% bromphenol blue, 10% beta-mercaptoethanol, and 2 mM EDTA). Samples were run on 4–12% bis-Tris acrylamide gels (Invitrogen), dried and exposed to film.
Previous work demonstrated the importance of the serine at position 776, S776, in the ability of ATXN1[82Q] to induced neurodegeneration in transgenic mice (Emamian et al. 2003). Further studies found the effect of phosphorylating S776 to vary depending on cell type (Jorgensen et al. 2007). To examine whether S776 affects the accumulation of ATXN1 in a SCA1 susceptible mammalian neuron, transgenic mice over-expressing one of either two forms of human ATXN1-30Q, with either a serine at position 776 (S776) or an alanine at this residue (A776), in their Purkinje cells were crossed to Atxn1−/− mice to allow the measurement of ATXN1 and ATXN1 mRNA levels without interference from endogenous murine Atxn1 or Atxn1 mRNA. To avoid problems associated with the extraction of insoluble ATXN1[82Q], ATXN1[30Q]-expressing ATXN1[30Q] mice were examined. Determining the ratio of protein to mRNA in cerebella extracts was used as a measure of the relative stabilities of ATXN1[30Q]-S776 and ATXN1[30Q]-A776. As shown in Fig. 1(a), cerebella from hemizygous ATXN1[30Q]-S776 mice had half the amount of ATXN1 mRNA as detected in extracts prepared from ATXN1[30Q]-A776 mice. Yet, the amount of ATXN1 present in ATXN1[30Q]-S776 cerebellar extracts was several fold higher than the amount of ATXN1[30Q]-A776 protein detected (Fig. 1b). When calculated as an ATXN1 protein/ATXN1 mRNA ratio, ATXN1[30Q]-S776 mice had a ten-fold greater amount of protein for a given amount of ATXN1 mRNA as did ATXN1[30Q]-A776 mice (Fig. 1c). These results are consistent with the amino acid at position 776 of ATXN1 having a role in modulating the level of ATXN1 in Purkinje cells.
To assess whether altering the activity of endogenous Akt in Purkinje cells in vivo affects ATXN1 phosphorylation, a dominant-negative Akt (Akt-DN) transgene was constructed such that its expression would be directed to Purkinje cells using the Pcp2/L7 regulatory region (Oberdick et al. 1990; Burright et al. 1995). There are three highly homologous isoforms of Akt in mammals (Akt1, Akt2, and Akt3). As shown in the Allen Brain Atlas (Lein et al. 2007), at least two, Akt 1 and Akt 2, are expressed by Purkinje cells. Thus to down-regulate Akt activity in Purkinje cells regardless of isoform we used a dominant negative strategy. The open reading frame for a dominant negative form of Akt, in which the lysine residue at position 179 of Akt was replaced with a methionine (Coffer et al. 1998), was inserted into the fourth exon of the Pcp2 gene that had all potential translational start sites removed (Fig. 2a). This amino acid substitution is within the ATP-binding region of Akt and renders it inactive and able to function as a dominant negative inhibitor of Akt activity (Franke et al. 1995). Transgenic mice, designated Akt-DN, were generated with Purkinje cell expression of the dominant-negative Akt. Immunostaining cerebellar sections from an Akt-DN mouse with an anti-FLAG antibody confirmed that expression was restricted to Purkinje cells of the cerebellar cortex (Fig. 2b). Akt-DN expression was for the most part localized to the cytoplasm of Purkinje cells, the predominant subcellular location of Akt in Purkinje cells (Jorgensen et al. 2007).
To determine if expression of the Akt-DN transgene affected Akt activity in Purkinje cells, the phosphorylation status of a well-characterized Akt substrate GSK-3 was assessed by immunofluorescence. GSK-3 is inhibited by Akt-mediated phosphorylation at Ser21 of GSK-3α and Ser9 of GSK-3β (Coffer et al. 1998). Immunostaining cerebellar sections with an antibody that recognizes phospho-GSK-3α/β (Ser21/9) showed that the amount of phospho-GSK-3 detected in Purkinje cells was greater in a wild-type FVB/N cerebellum than in an Akt-DN cerebellum (Fig. 2c and d, respectively). This result is consistent with the Akt-DN protein having an inhibitory effect on the activity of Akt in Purkinje cells of Akt-DN mice. Somewhat surprisingly, given the importance of Akt for neuronal survival (Brunet et al. 2001), the Akt-DN mice showed no evidence of abnormal Purkinje cell morphology (Fig. 2e and f) or signs of neurological dysfunction (Fig. 5b).
To examine the effect of decreasing Akt activity in Purkinje cells on wild type ATXN1, homozygous ATXN1[30Q] mice from the A02 line (Burright et al. 1995) were crossed to hemizygous DN-Akt mice. Double transgenic mice were generated and cerebellar extracts were prepared from them and littermate hemizygous SCA1-30Q mice. Using an ATXN1-phospho-S776 polyclonal antibody (Emamian et al. 2003), western blots of extracts from mice expressing DN-Akt and ATXN1[30Q] showed that phosphorylation of ATXN1[30Q]-S776 was not inhibited. In fact, although variable, the amount of pS776-ATXN1 increased significantly in DN-Akt/ATXN1[30Q] mice compared to mice expressing ATXN1[30Q] (Fig. 3a). Consistent with the idea that ATXN1 is stabilized by phosphorylation of S776 (Chen et al. 2003), DN-Akt/ATXN1[30Q] cerebella extracts also showed a significant increase in the amount of total ATXN1 (Fig. 3b).
Whether decreasing Akt activity in Purkinje cells impacts mutant ATXN1 with an expanded polyglutamine tract was examined by crossing homozygous ATXN1[82Q] mice from the B05 line (Burright et al. 1995) with the Akt-DN transgenic mice. Phospho-S776 ATXN1[82Q] in Purkinje cells was assessed by an immunohistochemical approach using the ATXN1-phospho-S776-specific antibody PN1168 (Emamian et al. 2003). Cerebellar sections were stained with serial dilutions of the anti-ATXN1-phospho-S776 antibody PN1168 and anti-total ATXN1 antibody 11750. At an anti-ATXN1 11750 titer below the threshold for detecting ATXN1 in Purkinje cells in ATXN1[82Q] mice (Fig. 4a), the antibody was able to detect ATXN1 in nuclei of Purkinje cells of ATXN1[82Q] /Akt-DN double transgenic mice (Fig. 4b). In a similar fashion, at an anti-ATXN1-phospho-S776 antibody PN1168 titer below the threshold for detecting phospho-ATXN1 in Purkinje cells in ATXN1[82Q] mice (Fig. 4c), was able to robustly detect phospho-ATXN1 in nuclei of Purkinje cells of ATXN1[82Q]/Akt-DN double transgenic mice (Fig. 4d). As a control, no staining was detected in Purkinje cell nuclei from Atxn1−/− (Fig. 4e) or ATXN1[82Q]-A776 mice (Fig. 4f). These results indicate that expression of Akt-DN in Purkinje cells did not lead to an inhibition of ATXN1[82Q]-S776 phosphorylation and if anything led to an increase in the amount of total and S776-phosphorylated mutant ATXN1.
One quantitative measure of Purkinje cell pathology in SCA1 mice is thickness of the molecular layer (Zu et al. 2004). As shown in Fig. 5(a) ATXN1[82Q] and Akt-DN/ATXN1[82Q] animals manifested the same degree of molecular layer thinning, i.e. atrophy of Purkinje cell dendrites that progressed with age (12–20 weeks). Thus, by this assessment of Purkinje cell pathology expression of Akt-DN failed to decrease severity and progression of ATXN1[82Q]-induced disease.
The neurological status of the ATXN1[82Q] and Akt-DN/ATXN1[82Q] mice was assessed by analyzing gait performance using the TreadScan system (Wooley et al. 2005). Rear stride time was the gait parameter evaluated as this was previously validated as a sensitive measure for detecting early signs of gait deficits (Wooley et al. 2005; He et al. 2006). At no age examined was the gait performance of Akt-DN mice found to differ significantly from that of wt FVB/n animals (Fig. 5b). Thus, expression of Akt-DN had no detectable effect on gait performance in FVB/n mice. By 5 weeks ATXN1[82Q], and Akt-DN/ATXN1[82Q] animals showed a significant increase in stride time compared to FVB/N with Akt-DN/ATXN1[82Q] rear stride times significantly increased over ATXN1[82Q] mice at two of five ages out to 13 weeks. From these data we concluded that the Akt-DN/ATXN1[82Q] double transgenic mice were as neurologically compromised as the ATXN1[82Q] mice.
To determine if Akt1 gene dosage affects SCA1 disease, we crossed Akt1+/− animals with the Sca1154Q/+ knock-in mice, a model that recapitulates the selective neuropathology of SCA1 patients and that has ataxia and motor deficits that can be quantified by accelerating rotarod analysis at 7 weeks of age (Watase et al. 2002). Both Sca1154Q/+ and Sca1154Q/+; Akt1+/− animals showed significantly impaired performance (p < 0.001) on the rotarod compared with Akt1+/− and their wild-type littermates (Fig. S1a). The performance of Sca1154Q/+; Akt1+/− mice on the rotarod was not significantly different from that of Sca1154Q/+. These data show that removing one copy of Akt1 does not affect impaired motor coordination of Sca1154Q/+ mice. Moreover, loss of one allele of Akt1 had no effect on survival and weight loss in Sca1154Q/+ mice (Fig. S1b and c). Altogether, our results show that Akt1 probably does not modify SCA1 phenotypes. We cannot exclude that other AKT isoforms might compensate for the partial loss of Akt1 given the significant homology between AKT isoforms.
As a further step towards dissecting the phosphorylation of S776 of ATXN1, we developed a phosphorylation assay using cerebellar extracts. The approach was to assess phosphorylation at S776 of exogenously added GST-ATXN1 using wild type FVB/N cerebellar extracts as a kinase source. Phosphorylation was assessed using the phospho-S776 specific PN1168 antibody. Cerebellar extracts were found to specifically phosphorylate GST-ATXN1-S776 to a much greater degree than GST-ATXN1-A776 (Fig. 6a). The phosphorylation of GST-ATXN1-S776 increased with increasing amount of added cerebellar extract (Fig. 6b), indicating that the in vivo kinase for this site is active in the extracts. To examine whether the in vivo kinase is Akt an immunodepletion study was performed. Figure 6(c) shows that immunodepletion of Akt did not significantly affect the phosphorylation of ATXN1-S776.
Fractionating the cerebellar extract into cytoplasmic and nuclear fractions revealed that the ATXN1-S776 kinase was enriched in the cytoplasmic fraction while the nuclear fraction contained an undetectable level of the kinase after an 1-h incubation (Fig. 6d). Interestingly, pS776-endogenous Atxn1 was enriched in the nuclear fraction (Fig. 6d). This raises the possibility that the subcellular site of S776 phosphorylation might be separate from the subcellular site of pS776-ATXN1 action.
As a means towards identifying the ATXN1-S776 kinase active in the cerebellum, we examined whether a kinase could be found that co-fractionated with ATXN1-S776 phosphorylation activity. The first approach selected was to fractionate FVB cerebellar lysates by sequential ammonium sulfate precipitations using nine cuts starting at 0–30% and going to 80–90%. To assess the ability of each ammonium sulfate fraction to phosphorylate S776 of ATXN1 a more sensitive radioactive phosphorylation assay was developed that used a GST-fusion peptide containing the COOH-terminal portion of ATXN1 from the ataxin-1 and high mobility group box domain (Lam et al. 2006) to the C-terminus as the substrate (fragment V, Fig. 7a). Phosphorylation activity peaked in two ammonium sulfate fractions, fractions 3 and 7 (Fig. 7b). The phosphorylation in fraction 7 was on intact fragment V and was greatly reduced when blocked by substituting an Asp residue at position 776 of fragment V (Fig. 7c). Thus, we concluded that the cerebellar kinase activity for ATXN1-S776 peaked in ammonium sulfate fraction 7.
Western blot analysis of the cerebellar ammonium sulfate fractions revealed that Akt peaked in fraction 5 and not with the S776 phosphorylating activity in faction 7 (Fig. 7d). Cyclic AMP-dependent protein kinase (PKA), another kinase strongly predicted as a potential ATXN1-S776 kinase using the Scansite program (Obenauer et al. 2003), fractionated with the peak of S776 phosphorylating activity in fraction 7 (Fig. 7e). Thus, the ammonium sulfate fractionation analysis is consistent with the other data that Akt is not the cerebellar ATXN1-S776 kinase and raised PKA as a candidate S776 kinase.
The fractionation pattern of a variety of other kinases was also assessed by western blot analysis, including the top matches from Group-based Prediction System (Xue et al. 2008) and Scansite (PKA alpha, beta, and gamma; death-associated protein kinase 3; protein kinase C alpha, beta, gamma, and lambda; Cdc-like kinase 2) as well as additional kinases with some overlap in consensus sequence with the ATXN1 sequence surrounding S776 (calmodulin-dependent kinase 2 alpha; glycogen synthase kinase 3 alpha and beta; apoptosis signal-regulated kinase 1). Only PKA co-fractionated with the peak of S776 phosphorylation in fraction 7 of the ammonium sulfate fractionation (data not shown and Fig. 7e). Note that PKA fractionated in two peaks, which may represent the inactive PKA complex of catalytic and regulatory subunits in fractions 1–3 and the active PKA in fractions 6–8.
To confirm that PKA is the primary ATXN1 kinase from mouse cerebellar cytysol, we took the 50–90% ammonium sulfate fractions (equivalent to fractions 6–9 in Fig. 7b–e) and further fractionated it by binding to a hydroxyapatite column and eluting with a salt gradient of sodium phosphate. The S776 kinase activity peaked in fraction 11 (Fig. 7f), as did the PKA immunoreactivity (Fig. 7g). Thus, PKA co-fractionated with the S776 phosphorylation activity under both ammonium sulfate and hydroxyapatite fractionation conditions.
There are several small molecules that inhibit the kinase activity of Akt and PKA. When the PKA/Protein Kinase C inhibitor staurosporine (Bernsteel et al. 2008) and a 17-residue peptide Protein Kinase (cAMP-dependent, catalytic) Inhibitor-amide that corresponds to the active region of the heat-stable inhibitor of PKA (Glass et al. 1989) were added to the ATXN1-fragment V assay, both significantly reduced the S776 phosphorylation in a dose-dependent fashion. In contrast, an N10-substituted phenoxazine Akt inhibitor (Thimmaiah et al. 2005) had no significant affect on S776 phosphorylation using the fragment V assay (Fig. 8). Lastly, a 50% immunodepletion of PKA from cerebellar extracts (Fig. 9c) significantly reduced phosphorylation of ATXN1-S776 as detected using full-length GST-ATXN1 and a polyclonal phospho-S776 antibody (Fig. 9a and b).
The data provided here show that a phospho-resistant alanine at residue 776 of ATXN1, replacing a serine, destabilizes the protein in Purkinje cells. This finding is consistent with previous studies using transfected cell lines and a Drosophila model of SCA1 indicating that phosphorylation of ATXN1 at S776 stabilized the protein (Chen et al. 2003). Chen colleagues also suggested that Akt is the ATXN1-S776 kinase. In contrast, we found that inhibiting Akt activity either in vivo or in a cerebellar extract failed to decrease the amount of phospho-S776-ATXN1 detected. However, several lines of evidence indicated that PKA is the S776 kinase, including co-fractionation with the peak of S776 phosphorylating activity in cerebellar cytosol (Fig. 7), and reduction of S776 phosphorylation by cerebellar lysates upon addition of two PKA inhibitors (Fig. 8) and following PKA immunodepletion (Fig. 9). Although Purkinje cells amply express both Akt and PKA (Allen Brain Atlas – Lein et al. 2007), the data reported here are consistent with the conclusion that Akt is not the S776 kinase with PKA being a stronger candidate for the ATXN1-S776 kinase active in the cerebellum.
A cerebellar signaling pathway linked to regulating PKA activity is the rebound potentiation, a form of synaptic plasticity at GABAergic synapses between inhibitory interneurons (basket and stellate neurons) and Purkinje cells. In a model described by Kawaguchi and Hirano 2002; it was assumed that there is a relatively high basal level of PKA activity in Purkinje cells that is reduced with the activation of GABABR during depolarization at inhibitory synapses. If this assumption were accurate, a high basal activity of PKA in Purkinje cells would require that the basal intracellular concentration of cAMP be relatively high, perhaps via the activation of the glutamate receptor mGluR1 that activates adenyl-cyclase and increases cAMP (Miyashita and Kubo 2000). Since phosphorylation of ATXN1 at S776 promotes neurodegeneration (Serra et al. 2004; Chen et al. 2003), it is tempting to speculate that perhaps a high basal activity of PKA in Purkinje cells contributes to their enhanced sensitivity to the toxic affects of mutant ATXN1 in SCA1.
Multiple mechanisms regulate the specificity by which proteins are phosphorylated. These include structure of the catalytic site on the kinase, activation state of the kinase, interactions between the kinase and substrate, as well as interactions involving scaffolding and adaptor proteins. Such regulatory mechanisms are required since the number of potential phosphorylation sites in the proteome is on the order of 700,000 for any one kinase (Ubersax and Ferrell 2007). The data reported by Chen et al. 2003 demonstrate that the peptide sequence in ATXN1 surrounding S776 can function as a recognition motif for Akt. Thus, it is likely that some higher level regulatory mechanism prevents Akt in the cerebellum from phosphorylating ATXN1 at S776 as effectively as it can in the systems used by Chen et al. 2003. At this time it is unclear why Akt is not effective in phosphorylating S776 of ATXN1 in the cerebellum.
Disparate effects of altering a kinase-mediated signaling pathway between Drosophila and rodent models of neurological disease are becoming increasingly common, perhaps reflecting that in mammals the kinome is more complex than in the fly (Manning et al. 2002; Caenepeel et al. 2004). For example, loss of function mutations in the putative mitochondrial protein PINK1 (PTEN-induced kinase 1) are linked to recessive forms of Parkinson’s disease (Cookson et al. 2007). In Drosophila loss of function of the PINK1 homolog leads to dramatic morphological abnormalities of mitochondria (Clark et al. 2006; Park et al. 2006; Yang et al. 2006). While in the mouse, deletion of PINK1 impairs mitochondrial functions in the absence of gross changes in the ultrastructure or the total number of mitochondria (Gautier et al. 2008). Parkinson’s disease is also associated with the deposition of phospho-S129-α-synuclein into Lewy bodies in brain tissue. In a Drosophila model of α-synuclein induced neurodegeneration, phosphorylation at S129 was associated with enhanced pathology (Chen and Feany 2005). However, in a rat model of Parkinson’s disease S129 phosphorylation decreased degeneration (Gorbatyuk et al. 2008). In closing, the studies reported here on the phosphorylation of ATXN1 in the cerebellum provide further support for the importance of examining findings from Drosophila and cell culture systems with those from an in vivo mammalian model of the human neurodegenerative disease.
This research was supported by NIH/NINDS grant NS45667 (HTO) and HHMI-CIA (HYZ and HTO) and NIH/NINDS grant NS27699 (HYZ). HYZ is a HHMI Investigator.
Additional Supporting Information may be found in the online version of this article:
Figure S1. Heterozygosity for a null Akt1 allele has no effect on Sca1154Q/+ phenotypes.
Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.