Ubiquitination Regulates the Neuroprotective Function of the Deubiquitinase Ataxin-3 in Vivo*

Background: Ubiquitination of ataxin-3 enhances its DUB activity in vitro, but it is unknown whether it controls its function in intact organisms. Results: Ubiquitination at lysine 117 regulates the neuroprotective function of ataxin-3 in Drosophila. Conclusion: Ataxin-3 ubiquitination is a critical regulator of its in vivo functions. Significance: Ubiquitination may constitute a general regulatory mechanism of DUB activities in vivo. Deubiquitinases (DUBs) are proteases that regulate various cellular processes by controlling protein ubiquitination. Cell-based studies indicate that the regulation of the activity of DUBs is important for homeostasis and is achieved by multiple mechanisms, including through their own ubiquitination. However, the physiological significance of the ubiquitination of DUBs to their functions in vivo is unclear. Here, we report that ubiquitination of the DUB ataxin-3 at lysine residue 117, which markedly enhances its protease activity in vitro, is critical for its ability to suppress toxic protein-dependent degeneration in Drosophila melanogaster. Compared with ataxin-3 with only Lys-117 present, ataxin-3 that does not become ubiquitinated performs significantly less efficiently in suppressing or delaying the onset of toxic protein-dependent degeneration in flies. According to further studies, the C terminus of Hsc70-interacting protein (CHIP), an E3 ubiquitin ligase that ubiquitinates ataxin-3 in vitro, is dispensable for its ubiquitination in vivo and is not required for the neuroprotective function of this DUB in Drosophila. Our work also suggests that ataxin-3 suppresses degeneration by regulating toxic protein aggregation rather than stability.

Diverse cellular pathways are regulated by the post-translational modification of proteins with ubiquitin (Ub), 2 which can control substrate protein localization, interaction with protein partners, or turnover. Conjugation of Ub to specific substrates depends on the coordinated action of three proteins: the Ubactivating enzyme (E1) transfers Ub to an E2 Ub-conjugating enzyme which then, in the presence of an E3 (Ub-ligase), com-pletes the transfer of Ub to a lysine residue of a substrate protein through an isopeptide bond. A substrate can be mono-or polyubiquitinated, and different types of Ub-Ub linkages can be formed on proteins, with various outcomes. For example, mono-Ub of a membrane receptor can signal its internalization, whereas certain poly-Ub chains signal the degradation of a protein by the proteasome (for review, see Ref. 1).
Protein ubiquitination can be reversed by deubiquitinating enzymes (DUBs). DUBs are key players in various cellular processes, and several of them are linked to malignancies and neurological diseases (1)(2)(3)(4)(5). In fact, the regulation of DUB activity is being considered as an entry point into therapeutics for several human diseases (2, 4, 6 -8). Despite progress in understanding the functions of DUBs at the structural and cellular level, general understanding of DUB physiology, particularly in vivo, is limited.
Although in vitro biochemical and cell biological studies provide compelling evidence for the control of DUB activities by their own ubiquitination (9), the biological relevance of this type of regulation in vivo in intact organisms is presently uncertain. To begin to understand the physiological importance of DUB ubiquitination to their in vivo roles, we set a defined goal: to determine whether ubiquitination-dependent regulation of the catalytic activity of the DUB ataxin-3 is important for its protective function in vivo.
Ataxin-3 (see Fig. 1A) has been implicated in regulating protein stability in quality control pathways and has also been reported to regulate transcriptional events (16); still, its exact physiological functions are not well understood. An important feature of ataxin-3 is its polyglutamine (polyQ) region that, when abnormally expanded, causes age-related neurodegeneration in the polyQ-repeat disorder spinocerebellar ataxia type 3 (SCA3, also known as Machado-Joseph disease) through a mechanism that remains to be elucidated (16).
Elegant studies conducted by the Bonini laboratory showed that ataxin-3 serves a neuroprotective role in Drosophila melanogaster by suppressing degeneration caused by several polyQ disease proteins, including its own self (SCA3), huntingtin (Huntington disease) and ataxin-1 (SCA1) (17). PolyQ-repeat diseases are neurodegenerative disorders caused by abnormal expansion of the polyQ tract in otherwise unrelated proteins (18,19). Understanding how ataxin-3 operates and is regulated in vivo is therefore important for both neurodegenerative and neuroprotective processes. Here, we present evidence that ubiquitination of this DUB is critical for its neuroprotective role in vivo.

EXPERIMENTAL PROCEDURES
Constructs-All of the ataxin-3 constructs used in this report are the full-length human version of the protein and contain a normal polyQ stretch of 22 repeats. Ataxin-3 used in recombinant assays is in pGEX6P1 and was cleaved from GST by using PreScission Protease (GE Healthcare). Ataxin-3 that was cloned into pUASt and was used to generate transgenic fly lines has a His 6 tag appended at the C terminus of the protein, immediately succeeding the last native amino acid. Where indicated, lysine residues were mutated into the similar, but nonubiquitinatable amino acid arginine by using the QuikChange Mutagenesis kit (Stratagene) (13). CHIP and Ubch5C constructs have been described before (11,13,20).
Drosophila Lines and Procedures-Drosophila maintenance and husbandry were conducted in diurnally controlled environmental chambers at 25°C (24 -26). Drosophila transgenics were created using the Gal4-UAS system. Ataxin-3 constructs were cloned into the pUASt vector (27). Injection of pUASt constructs into the parental w 1118 line was performed by Duke University Model Systems. For histology, whole flies with removed proboscises were fixed in 2% glutaraldehyde/2% paraformaldehyde in Tris-buffered saline overnight. Fixed flies were then dehydrated in a series of 30, 50, 75, and 100% ethanol and propylene oxide, embedded in Poly/Bed812 (Polysciences), and sectioned at 5 m. Sections were stained with toluidine blue, as described previously (17,28). UAS-PolyQ78 Severe (originally MJD.tr-Q78-c211.  Table 1. For Western blotting, whole flies or dissected fly heads that had been frozen were treated in a Dounce homogenizer in boiling 2% SDS lysis buffer (50 mM Tris, pH 6.8, 2% SDS, 10% glycerol, 100 mM DTT), sonicated briefly, boiled for 10 min, centrifuged (or not) at 13,000 ϫ g for 30 min at room temperature, and supernatants were loaded into SDS-polyacrylamide gels (26). We used 50 l of lysis buffer per whole fly or 15 l of buffer per dissected head. For extraction using Triton X-100, the procedure was based on Ref. 29. Triton X-100 buffer was 50 mM Tris, pH 7.5, 1% Triton X-100, 1 mM EDTA, and complete protease inhibitor mixture (Sigma-Aldrich). Ten dissected fly heads per sample were homogenized, sonicated, and then centrifuged at 75,000 ϫ g (4°C) for 30 min. Supernatant and pellet were separated. Pellet was dissolved in 2% SDS/ 100 mM DTT and boiled for 10 min. Urea/SDS extraction procedure was based on Ref. 30. Extraction buffer was 8 M urea, 200 mM Tris, pH 7.5, 1 mM EDTA, 5% SDS. Ten dissected fly heads per sample were homogenized, sonicated, and heated at 65°C for 10 min. One sample was centrifuged (13,000 ϫ g at room temperature for 30 min), the other sample was not; supernatants were loaded into SDS-polyacrylamide gels. Photos of external fly eyes were taken with an Olympus SZ61 microscope equipped with a DP21 digital camera. Photos of histological sections were acquired with an Olympus BX53 microscope equipped with a DP72 digital camera, and retinal depth was quantified using cellSens Dimension software (Olympus).
Mouse Brain Lysate Preparation-Flash-frozen brains from CHIP knock-out mice or littermate controls (20) were homogenized in radioimmunoprecipitation assay buffer with protease inhibitors (Sigma-Aldrich), sonicated, and centrifuged at 75,000 ϫ g (4°C for 30 min). Supernatants were supplemented with 1% final SDS and 100 mM DTT, boiled for 10 min, and loaded into SDS-polyacrylamide gels.

Ubiquitination of Ataxin-3 Is Important for Its Ability to Suppress Severe Toxic Protein-dependent Degeneration in Vivo-
We recently reported that the catalytic activity of the DUB ataxin-3 ( Fig. 1A) is controlled by its ubiquitination at lysine 117 (13). This modification markedly enhances the DUB activity of ataxin-3 in vitro (11,13). Normal ataxin-3 (WT) and a version that contains only lysine at position 117, with all of the other lysines mutated into the similar but nonubiquitinatable amino acid arginine, are equally potentiated by their ubiquitination (Fig. 1B). However, it is unclear whether ubiquitination of ataxin-3 is important for its function in intact organisms.
One function of ataxin-3 in vivo is its neuroprotective role in D. melanogaster. In the fruit fly, ataxin-3 suppresses degeneration caused by various polyQ proteins in a manner that depends on its catalytic activity (17). To test the hypothesis that ubiquitination of ataxin-3 is important for this in vivo function, we generated new transgenic Drosophila lines that express one of the following versions of ataxin-3 through the Gal4-UAS system: wild type (WT), catalytically inactive (C14A), lysine-less nonubiquitinatable (Lys-null), or containing only Lys-117 (Fig.  1C). All of the lysine residues in Lys-null are mutated into arginine.
The rationale to generate flies that express these specific forms of ataxin-3 was informed by our earlier studies, where we found that ataxin-3 is preferentially ubiquitinated at Lys-117 in vitro and in mammalian cell culture (13). Through in vitro reconstituted assays, we previously demonstrated that ubiquitination of ataxin-3 at other single lysine residues was modest  (21). K-117, lysine residue that is primarily ubiquitinated in cells (13). Triangles, relative positions of lysine residues other than Lys-117. UIMs, Ub-interacting motifs that bind poly-Ub chains at least 4 residues long (21). QQQ, polyQ region that, when abnormally expanded, causes SCA3 (16). . p values are from Student's t tests using data from the 6-h time point. C, new transgenic fly lines that express ataxin-3 through the Gal4-UAS system. Expression of UAS-ataxin-3 for these blots was driven by the ubiquitous sqh-Gal4 driver (24,43). Western blots are from 20 whole flies per genotype, probed with antiataxin-3 or anti-tubulin antibodies, as indicated. Flies were heterozygous for the Gal4 driver and ataxin-3. D, photos and sections of fly eyes expressing ataxin-3. None, flies contained only gmr-Gal4, in the absence of ataxin-3. The other groups were heterozygous for driver and UAS-ataxin-3. Flies were 1 day old. Genotypes for the flies shown in C and D are listed in Table 1. and did not enhance its DUB activity (13). Also, the DUB activity of Lys-null and nonubiquitinated WT ataxin-3 toward poly-Ub chains was similar in vitro (13). We reasoned that flies that express WT, Lys-null, or Lys-117 ataxin-3 are a suitable platform to investigate whether ubiquitination of ataxin-3 is important for its protective role in Drosophila.
Western blotting shows several ataxin-3-positive bands above unmodified ataxin-3, which, based on our previous work (11,13,22), are ubiquitinated forms of this protein. These slower migrating bands are absent in the Lys-null version of ataxin-3 (Fig. 1C). Because the rest of our studies in this report focus on the Drosophila compound eye, we confirmed that expression of none of the ataxin-3 forms used causes detectable anomalies to the external or internal structures of this organ ( Fig. 1D and data not shown).
We consistently observe several ubiquitinated ataxin-3 bands with WT, C14A, and Lys-117 in purified systems ( Fig. 1B and Refs. 11,13), in mammalian cell preparations and mouse brain (11,13), and in Drosophila (Fig. 1C). It is unclear whether the DUB activity of these different ubiquitinated species of ataxin-3 is enhanced to different extents. However, based on our previously published work monoubiquitination is sufficient for up-regulation (11,13).
Expression of an expanded polyQ repeat peptide (polyQ78 Severe ) selectively in fly eyes causes massive degeneration of external and internal structures ( Fig. 2A, left panel; this line was origi- Lys-null and Lys-117 ataxin-3. No statistically significant differences were found between WT and Lys-117. All p values from Student's t tests comparing Lys-null with C14A were Ͻ0.01. Degeneration scale: 0, no signs of depigmentation; 1, minimal depigmentation; 2, substantial depigmentation, but Ͼ50% of the external eye appears normal; 3, Ͼ50% of the eye shows depigmentation; 4, necrotic spots; 5, pervasive necrosis. Genotypes for flies in all panels are listed in Table 1. nally described in Ref. 28). PolyQ78 Severe is the isolated polyQ region of pathogenic ataxin-3, comprising a stretch of 78 glutamines and immediately adjoining amino acids (28). This peptide is devoid of any of the other domains of ataxin-3 (Fig. 1A) and probably serves best as a generic model of polyQ-dependent degeneration rather than solely of SCA3 (31,32).
Degeneration caused by polyQ78 Severe is marked by loss of pigmentation of the external eye, loss of the functional units (the ommatidia; internal eye sections), a dramatic reduction of retinal depth, and the presence of aggregated structures/inclusions ( Fig. 2A, left panel). The inclusions were previously shown to contain polyQ78 Severe and are reminiscent of neuronal inclusions found in human postmortem brains afflicted by various polyQ diseases (17-19, 28, 31).
Expression of exogenous WT ataxin-3 dramatically suppresses degeneration from polyQ78 Severe , recapitulating in our hands the effect demonstrated by Bonini and colleagues (17). In the presence of normal ataxin-3, depigmentation of the external eye is essentially undetectable, the ommatidia are intact, and there are no visible inclusions ( Fig. 2A, middle panel). Unlike WT ataxin-3, its catalytically inactive version does not detectably suppress degeneration ( Fig. 2A, right panel), demonstrating that the DUB activity of this protease is at the core of its protective role. Therefore, we asked whether ubiquitination of ataxin-3, which enhances its catalytic activity in vitro (Fig. 1B and Refs. 11, 13) is important for its neuroprotective function.
Nonubiquitinatable Lys-null ataxin-3 somewhat suppresses polyQ78 Severe -dependent degeneration of Drosophila eyes early in the life of adults (Fig. 2B). However, by 2 weeks of age the potency of Lys-null ataxin-3 to suppress degeneration is markedly lower than that of WT or Lys-117 ataxin-3 (Fig. 2B,  day 14; Fig. 2C). Compared with eyes that co-express ubiquitinatable ataxin-3 and the disease protein, those that express Lysnull ataxin-3 alongside polyQ78 Severe show more severe degeneration and reduced retinal depth (Fig. 2, B and C).
To further understand the time course of protection from polyQ78 Severe by the different forms of ataxin-3, we examined depigmentation and necrosis of the external eye for 30 days (Fig. 2D). We categorized eyes using a scale from 0 to 5, where 0 denotes no visible signs of depigmentation and 5 means pervasive depigmentation and necrosis of the external fly eye. From these observations, it is clear that Lys-null ataxin-3 does not suppress degeneration as well as Lys-117 ataxin-3 (statistical significance at p Ͻ 0.01), although Lys-null ataxin-3 performs better than the catalytically inactive version ( Fig. 2D; statistical significance at p Ͻ 0.01). We should note that eyes co-expressing Lys-117 and the toxic protein trend slightly higher in quantification of degeneration compared with eyes expressing WT ataxin-3 and polyQ78 Severe , but these differences do not reach statistical significance. Based on these collective results (Fig. 2), we conclude that ataxin-3 ubiquitination at Lys-117 is critical for its neuroprotective role in Drosophila against severe forms of polyQ-dependent degeneration. (17). A major histological hallmark of polyQ-dependent degeneration is the presence of aggregated bodies/inclusions in various models of polyQ-repeat diseases and in human postmortem brain tissue (17-19, 28, 31). PolyQ78 Severe , as highlighted by boxes in Fig. 2A, leads to numerous inclusions in fly eyes. These structures are completely or nearly completely absent in the presence of catalytically active ataxin-3 in young flies, although they do accumulate with age (Fig. 2B, compare day 1 with day 14 in histological preparations). Therefore, we used Western blotting to examine whether ataxin-3 suppresses polyQ78 Severe -dependent degeneration by decreasing its protein levels, which could explain the reduced presence of inclusions. We tested different buffers to extract soluble and aggregated species of polyQ78 Severe from dissected fly heads (Triton X-100, SDS, or urea and SDS; Fig. 3A). We proceeded with a 2% SDS buffer that provided the best differentiation between SDS-soluble and SDS-resistant forms of polyQ78 Severe .  Table 1.

Ataxin-3 Suppresses Degeneration in Drosophila without Affecting Levels of the Toxic PolyQ Protein-It is unclear how ataxin-3 performs its neuroprotective role in Drosophila
Western blots from 1-day-old adult flies indicate that toxic protein levels are not lowered by the presence of WT ataxin-3 compared with degenerating eyes that do not express any ataxin-3 (Fig. 3B). In fact, we sometimes observe less polyQ78 Severe protein in the absence of any ataxin-3 compared with when the active DUB is present (Fig. 3C, right panel). We detect two types of polyQ species in Western blots: SDS-soluble and SDS-resistant, as also described by previously published work from other investigators (e.g. in Refs. 33,34). Neither type of polyQ species is reduced in the presence of ataxin-3 (Fig. 3).
PolyQ78 Severe protein levels are similar among WT, Lys-null, and Lys-117 ataxin-3-expressing eyes (Fig. 3C). However, expression of C14A ataxin-3 sometimes is associated with lower levels of polyQ78 Severe compared with other versions of the DUB (Fig. 3C, compare polyQ78 Severe levels between the two different experiments).
By 2 weeks of age, there is consistently less toxic protein detectable in fly eyes expressing C14A or Lys-null ataxin-3 compared with those that express WT or Lys-117 versions (Fig.  3D). The lower levels of polyQ78 Severe are an artifact of eye degeneration. This artifact reflects a loss of the overall eye caused by polyQ78 Severe in the presence of C14A or Lys-null ataxin-3, but not so much in the presence of WT or Lys-117 ataxin-3. Retinal loss is highlighted by blotting for the eye pigment rhodopsin (Fig. 3). Note how rhodopsin is nearly undetectable in eyes whose degeneration is not suppressed. The loading control, tubulin, indicates the total amount of lysate from dissected whole heads, which also includes non-eye tissue such as the brain.
Based on these results, ataxin-3 does not seem to protect from degeneration by decreasing the protein levels of polyQ78 Severe . Instead, the data in Fig. 3 (and also in Fig. 4, discussed in the next section) suggest that the protective role of this DUB lies elsewhere. As we observe few to no inclusions in polyQ78 Severe -expressing fly eyes in the presence of catalytically active ataxin-3 (Fig. 2, day 1), perhaps its protective function stems from an ability to reduce inclusion formation without altering overall protein levels.

Ataxin-3 Ubiquitination Is Important for Its Ability to Sup-
press Mild Degeneration-To further understand the importance of ataxin-3 ubiquitination to its protective role in vivo, we examined whether this modification is also required for milder forms of polyQ-dependent degeneration. Perhaps ubiquitination of ataxin-3 is important to suppress severe degeneration but dispensable in milder cases of toxicity. This type of detail would provide a better understanding of the role of ataxin-3 and its ubiquitination-dependent regulation in polyQ toxicity in Drosophila.
To investigate this possibility, we used a fly line containing a different chromosomal insertion of polyQ78 that expresses the same construct at levels that cause milder degeneration (28), here denoted as polyQ78 Mild . As shown in Fig. 4A, expression of this toxic protein by itself in fly eyes leads to inclusions and mild ommatidial loss in newly eclosed adult flies (compare with the extensive degeneration caused by polyQ78 Severe in Fig. 2A). By 2 weeks of age, most of the ommatidia have disappeared.
Expression of WT, Lys-null, and Lys-117 ataxin-3 each has a similar suppressive effect in this milder form of polyQ-dependent toxicity for the first 2 weeks of life (Fig. 4A). By 1 month of age, however, it is again clear that ubiquitination of ataxin-3 is important for its suppressive role. Eyes that co-express Lys-null ataxin-3 alongside polyQ78 Mild show marked loss of internal eye structure, whereas WT and Lys-117 ataxin-3 continue to suppress toxicity from this polyQ protein (Fig. 4B). As with polyQ78 Severe (Fig. 3), none of the ataxin-3 versions lowers the levels of polyQ78 Mild protein (Fig. 4C), supporting the notion that ataxin-3 suppresses degeneration without decreasing levels of the toxic protein. Together with the data from above, these results indicate that ataxin-3 ubiquitination at Lys-117 regulates the function of this DUB in vivo and is especially important in later stages or more severe forms of polyQ-dependent degeneration.
CHIP Is Dispensable for Ataxin-3 Ubiquitination and Its Neuroprotective Function in Drosophila-Ataxin-3 is ubiquitinated by the E3 ligase CHIP in reconstituted systems in vitro (11,13,20) and cooperates with this ubiquitin ligase to regulate  Table 1.
the turnover of select CHIP substrates in cultured mammalian cells (20). CHIP functions in protein quality control pathways by targeting various misfolded proteins for proteasomal degradation through its interaction with heat shock proteins (35,36). Consequently, we investigated whether CHIP is important for the ubiquitination and neuroprotective function of ataxin-3 in vivo.
We used two different UAS-RNAi lines that target Drosophila CHIP. The extent of knockdown that we achieved with each line, according to quantitative real-time PCR (qRT-PCR) is shown in Fig. 5A. The qRT-PCR results underestimate the actual extent of CHIP knockdown in fly eyes. Whereas expression of UAS-CHIP-RNAi is restricted to the eyes, RNA is isolated from whole dissected heads, which include non-eye tissue. CHIP knockdown in fly eyes is therefore expected to be considerably higher than what qRT-PCR reports.
We examined the ubiquitination pattern of WT ataxin-3 in the presence or absence of CHIP RNAi through Western blotting. As shown in Fig. 5B, we do not observe differences in the ubiquitination pattern of ataxin-3 from dissected fly heads, and the overall levels of ataxin-3 protein are also not affected by CHIP knockdown.
We also examined ataxin-3 protein in CHIP knock-out mouse brains. We again did not notice a difference in the overall levels of ataxin-3 in the absence of CHIP (Fig. 5C). Importantly in the absence of CHIP, ataxin-3 is more heavily ubiquitinated than when this ligase is present (Fig. 5C). Together, these data from flies and mice argue against CHIP being required for ataxin-3 ubiquitination in vivo.
Finally, we examined whether CHIP plays a role in ataxin-3dependent protection. CHIP knockdown by itself exacerbates external eye degeneration that is caused by polyQ78 Severe , highlighted by the presence of necrotic spots (Fig. 5D). Expression of ataxin-3 suppresses polyQ78 Severe -dependent degeneration, as also seen in Fig. 2. However, this protective function does not appear to be perturbed by the knockdown of CHIP (Fig. 5D). According to these results, the neuroprotective function of ataxin-3 does not depend on Drosophila CHIP.

DISCUSSION
Various regulatory processes control the catalytic activities and functions of DUBs. Through in vitro reconstituted assays, we recently demonstrated a dramatic up-regulation of the catalytic activity of the DUBs ataxin-3 and JosD1 by their own conjugation to ubiquitin (10,11,13). Here, our main objective was to determine whether ubiquitination can regulate DUB roles in vivo.
Ubiquitination at Lys-117 dramatically enhances the ubiquitin protease activity of ataxin-3 in vitro (Fig. 1), but it does not alter the subcellular localization of this protein in cells (13). Now, we found that ubiquitination of ataxin-3 at Lys-117 is important for the ability of this DUB to suppress polyQ-dependent degeneration in Drosophila (Figs. 2 and 4). These findings lead us to conclude that direct up-regulation of enzymatic FIGURE 5. CHIP is dispensable for ataxin-3 ubiquitination and protective function. A, histograms show mRNA levels of Drosophila CHIP when it is knocked down in fly eyes using two different UAS-RNAi lines driven by gmr-gal4. At least 15 dissected fly heads were analyzed per group. Experiment was conducted twice. Note that this is an underestimate of actual CHIP knockdown in fly eyes: knockdown was restricted to the eyes, but quantitative real-time PCR assays were conducted using the entire fly head, which includes tissues where CHIP mRNA was not targeted. B, ubiquitination pattern of WT ataxin-3 when Drosophila CHIP is knocked down in fly eyes. Western blots from 15 fly heads per genotype are shown. Driver, gmr-Gal4. Arrows, ubiquitinated WT ataxin-3. Asterisk, nonspecific band that is sometimes observable with anti-ataxin-3 antibody. C, ataxin-3 ubiquitination in CHIP knock-out mouse brains. Brain lysates from mice of the indicated ages were probed in Western blots with an anti-ataxin-3 antibody. Arrows, ubiquitinated ataxin-3. D, left, external eye photos. D, right, quantification of degeneration from flies expressing the indicated UAS constructs through the gmr-Gal4 driver. All flies were heterozygous for all constructs. Images are representative of at least five flies per genotype, with similar results. Shown in histograms are means Ϯ S.D. (error bars) from at least five flies per genotype from two crosses per genotype, using the degeneration scale described in Fig. 2D. p values are from Student's t tests. Genotypes for flies are listed in Table 1. activity by ubiquitination refines the protective function of ataxin-3 in vivo.
Our work provides new information on the neuroprotective role of ataxin-3, about which little is known. Because this DUB has been reported to regulate the stability of some misfolded proteins in cultured mammalian cells (16), one would hypothesize that ataxin-3 protects from polyQ-dependent degeneration by regulating the turnover of the toxic peptide. In Drosophila, however, ataxin-3 does not affect overall levels of the polyQ protein (Figs. 3 and 4). These results suggest that ataxin-3 suppresses polyQ-dependent degeneration by functioning at steps other than the overall stability of the toxic protein.
One of the described functions of ataxin-3 in cultured cells is its ability to enhance the presence of aggresomes/inclusions by aggregation-prone proteins (13,37). In fly eyes, ataxin-3 also affects inclusions that contain misfolded proteins, but in this instance by reducing, rather than enhancing, their presence, and without altering total protein levels of the toxic species (Figs. 2-4). These findings suggest that ataxin-3 regulates the equilibrium between inclusion-prone polyQ species and more soluble forms or conformations. The difference in aggregation "handling" by ataxin-3 in cells compared with flies might stem from the use of cultured cells versus intact organisms. Together, however, these findings strengthen the connection of ataxin-3 to protein quality control, particularly in protein solubility/aggregation. Future studies are needed to determine how precisely the DUB activity of ataxin-3 protects from degeneration.
Because ataxin-3 cooperates with the E3 ubiquitin ligase CHIP in vitro and in cultured cells to help target CHIP substrates for degradation (20), we tested whether CHIP is important for the neuroprotective role of this DUB in flies. We found that Drosophila CHIP is not required for the protective function of ataxin-3 in fly eyes (Fig. 5), suggesting that the functional cooperation between ataxin-3 and CHIP is restricted to specific substrates. As we also did not observe reduced ataxin-3 ubiquitination when CHIP was knocked down in flies or knocked out in mice, we further conclude that CHIP is likely not the only E3 ubiquitin ligase that ubiquitinates ataxin-3 in vivo.
We do not commonly observe robust bands consistent with ubiquitinated ataxin-3 in control mouse brains (Fig. 5, Ref. 11, and data not shown), but we consistently observe ubiquitinated ataxin-3 in flies (this paper). This variation could be due to inherent differences in ataxin-3 ubiquitination in mice compared with Drosophila. Another possibility might be the longer process of isolating protein from mouse brain compared with flies, which may lead to the rapid disappearance of small amounts of ubiquitinated proteins. Regardless of the exact reason for the difference in basal ataxin-3 ubiquitination between these two organisms, our results indicate that CHIP is not singularly required for ataxin-3 ubiquitination in vivo.
Finally, we comment on an apparent inconsistency with findings from Drosophila, mouse models of degeneration and cultured cells about the protective function of ataxin-3. As we demonstrated in this work and as had been published previously (17), catalytically active ataxin-3 protects against polyQdependent degeneration in flies. In cell culture, murine Atxn3 knock-out cells do not tolerate heat stress well compared with cells that have endogenous ataxin-3 (38). A study conducted in mice also suggested that co-expression of SCA3-causing and normal ataxin-3 transgenes led to a milder phenotype than expression of pathogenic ataxin-3 alone (39). That investigation, however, did not include all of the necessary controls to make a definitive claim about protection from normal ataxin-3, as it was not its primary focus. Overall, these results could explain why SCA3 patients with two copies of the disease allele have a more severe phenotype than those with one normal and one SCA3 copy of ATXN3 (40), and why age of onset for SCA3 patients is slightly improved by the presence of the normal ATXN3 allele (41). An inconsistency about the protective role of ataxin-3 arises, however, from another mouse study: co-expression of normal ataxin-3 alongside the SCA3 version did not seem to have a detectably beneficial effect in other murine models of SCA3 (42). This report did not include a catalytically inactive version of ataxin-3 as a control. Perhaps future mouse studies should include forms of this DUB that are enzymatically deficient and/or utilize Atxn3-null lines to control for the normal functions of this enzyme in vivo. The apparent discrepancy notwithstanding, data from various models collectively support a protective role for ataxin-3.
In summary, our findings establish a physiological importance for ubiquitination-dependent regulation of the catalytic activity of ataxin-3 in Drosophila. This work leads us to suggest that the ubiquitination of other DUBs also dictates their functions in vivo by modulating their protease activities in perhaps unexpected ways.