Phosphorylation of Arfaptin 2 at Ser 260 by Akt Inhibits PolyQ-huntingtin-induced Toxicity by Rescuing Proteasome Impairment*

Huntington disease (HD) is caused by an abnormal expanded polyglutamine repeat in the huntingtin protein. Insulin-like growth factor-1 is of particular interest in HD because it strongly inhibits polyQ-hunting-tin-induced neurotoxicity. This neuroprotective effect involves the phosphorylation of huntingtin at Ser 421 by the prosurvival kinase Akt (Humbert, S., Bryson, E. A., Cordelie`res, F. P., Connors, N. C., Datta, S. R., Finkbeiner, S., Greenberg, M. E., and Saudou, F. (2002) Dev. Cell 2, 831–837). Here, we report that Akt inhibits polyQ-huntingtin-induced toxicity in the absence of phosphorylation of huntingtin at Ser 421 , suggesting that Akt also acts on other downstream effector(s) to prevent neuronal death in HD. We show that this survival effect involves the ADP-ribosylation factor-interacting protein arfaptin 2, the levels of which are increased in HD patients. Akt phosphorylated arfaptin 2 at Ser 260 . Lack of phosphorylation of arfaptin 2 ( htt ) at Ser 421 . This phosphorylation blocks the polyQ-huntingtin toxic effect. In addi-tion, Akt phosphorylates arfaptin 2 at Ser 260 . When phosphorylated, arfaptin 2 inhibits the polyglutamine-induced impairment of the proteasome and therefore protects striatal neurons from the toxic effects of polyQ-huntingtin. PKB , protein kinase B; PDK1/2 , 3-phosphoinositide-dependent kinase-1/2; PI-3,4,5-P3 , phosphatidylinositol 3,4,5-trisphos- phate; PI3K , phosphatidylinositol 3-kinase.

Huntington disease (HD) 1 is a fatal neurodegenerative disorder characterized by involuntary movements, personality changes, and dementia (1). The dominantly inherited causal gene encodes the huntingtin protein, which contains an abnormal polyglutamine expansion (polyQ) in HD patients. HD develops when this expansion exceeds 35 glutamine residues. There is a strong inverse correlation between the number of glutamines and the age at onset of the disease. HD leads to the selective dysfunction and death of striatal neurons in the brain. The presence of neuritic and intranuclear inclusions in neurons is also characteristic of the disease.
There is currently no effective treatment to prevent or to delay disease progression, and death usually occurs 10 -20 years after the appearance of the first clinical symptoms. We recently reported that the insulin-like growth factor-1 (IGF-1)/ Akt signaling pathway could have a beneficial effect in HD (2). In striatal neurons, the most vulnerable neurons in HD, IGF-1 completely blocks polyQ-huntingtin-induced toxicity (2). This effect is mediated by the serine/threonine kinase Akt, which directly phosphorylates polyQ-huntingtin at Ser 421 , inhibiting the toxic properties of polyQ-huntingtin. These findings reveal that the IGF-1/Akt signaling pathway is of particular interest in HD, as it directly affects the protein that causes the disease. We observed that Akt is altered in the brains of HD patients and particularly in the striatum, showing that this kinase plays an important role in HD (2).
Akt is a potent prosurvival kinase that exerts its effects by phosphorylating key substrates such as components of the cell death machinery (3). Akt could be particularly important in HD, as it abrogates the toxicity of polyQ-huntingtin by phosphorylating Ser 421 . It also acts on other substrates that promote survival by general mechanisms or by mechanisms relevant to HD. The phosphorylation by Akt of substrates other than huntingtin may be important in the latter stages of the disease. During the pathological process in HD, huntingtin is cleaved by several proteases, including caspases, calpains, and other unidentified proteases (4 -8). This cleavage generates short N-terminal amino acid fragments that contain the pathological polyglutamine expansion and that are more toxic to neurons than full-length huntingtin (4,9). Some of these short N-terminal fragments do not contain the Ser 421 phosphorylation site, suggesting that the direct neuroprotective effect of Akt on huntingtin is lost during proteolysis (10).
We report here that Akt blocks the polyglutamine-dependent * This work was supported in part by Association pour la Recherche sur le Cancer Grant 4807, Fondation pour la Recherche Médicale, Fondation Banque Nationale de Paris Paribas, European Communities Concerted Action Early Pathogenic Markers of Slow Neurogenerative Diseases Grant QLK6-CT-2000-0384, and the Provital/P. Chevalier and Hereditary Disease Foundation Cure Huntington Disease Initiative (to F. S.). The Harvard Brain Tissue Resource Center was supported in part by United States Public Health Service Grant MH/NS 31862. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 1 The abbreviations used are: HD, Huntington disease; IGF-1, insulin-like growth factor-1; GST, glutathione S-transferase; Akt-ca, constitutively activated Akt; SGK, serum-and glucocorticoid-induced kinase; GFP, green fluorescent protein; NIIs, neuronal intranuclear inclusions; MT, microtubule; UPS, ubiquitin/proteasome system; ANOVA, analysis of variance. neuronal death and inclusion formation that are induced by an N-terminal fragment of huntingtin that does not contain Ser 421 . We demonstrate that this Ser 421 huntingtin-independent mechanism involves the ADP-ribosylation factor-interacting protein arfaptin 2. Phosphorylation of arfaptin 2 at Ser 260 by Akt promoted neuronal survival and decreased intranuclear inclusion formation in a neuronal model of HD. This demonstrates that arfaptin 2 is neuroprotective in HD. Finally, we show that phosphorylated arfaptin 2 inhibits the polyQ-huntingtin-induced blockade of the proteasome.
Cell Culture and Transfection-Primary cultures of striatal neurons were prepared from embryonic day 17 Sprague-Dawley rats and transfected at 4 days in vitro by a modified calcium phosphate technique (12). The mouse neuroblastoma NG108-15, human embryonic kidney 293, and the monkey fibroblast-like kidney COS-7 cell lines were cultured in Dulbecco's modified Eagle's medium supplemented with 10% bovine calf serum (Invitrogen) (or serum-starved when indicated) and antibiotics (50 units/ml penicillin and 50 g/ml streptomycin) and transfected by the calcium phosphate technique.
Kinase Assays-Kinase assays were performed as described (2,14). 293T cells were transfected with hemagglutinin-Akt vectors; soluble protein extracts were recovered; and hemagglutinin-Akt was immunoprecipitated. Recombinant SGK (Upstate Biotechnology, Inc.) was used. GST-fused proteins were produced in BL21 competent strains and purified with glutathione-Sepharose-coupled beads. Kinases and substrates were incubated for 30 min at 30°C with 5 Ci of [␥-32 P]ATP. The reaction products were resolved by 12% SDS-PAGE and stained with Coomassie Blue, and 32 P-labeled proteins were visualized by autoradiography.
Two-dimensional Phosphopeptide Mapping-After in vitro phosphorylation by Akt in the presence of [␥-32 P]ATP, two-dimensional phosphopeptide mapping of wild-type or S260A arfaptin 2 was performed as described previously (15). GST-fused wild-type and S620A arfaptin 2 were incubated with thrombin protease (Amersham Biosciences) following the manufacturer's instructions to remove the GST tag. After in vitro phosphorylation by Akt, two-dimensional phosphopeptide mapping of wild-type or S260A arfaptin 2 was performed (15). Briefly, gel pieces containing phosphorylated proteins were excised, destained, and dried. Arfaptin 2 was then digested with tosylphenylalanyl chloromethyl ketone-treated trypsin (75 g/ml; Worthington) in 25 mM ammonium carbonate buffer (pH 8) at 37°C for 12-16 h. Samples were dried, and peptides were collected in 10 l of electrophoresis buffer (10% glacial acetic acid and 1% pyridine (pH 3.5)). Samples were spotted 4 cm from the bottom and 5 cm from the right side of a thin-layer chromatography sheet (Analtech Inc.) together with basic fuschin dye. Firstdimension electrophoresis was carried out at pH 3.5 for 3 h at 350 V, and second-dimension chromatography was performed on dried sheets in chromatography buffer (37.5% pyridine, 25% butanol, and 7.5% acetic acid) for 4 h. Phosphopeptides were visualized with a PhosphorImager (Storm 820, Amersham Biosciences).
Antibody against Arfaptin 2 Phosphorylated at Ser 260 -Phosphopeptide CRDAGTRGRLEpSAQA was synthesized, coupled to keyhole limpet hemocyanin (Neosystem), and injected into rabbits. Polyclonal antibody to the keyhole limpet hemocyanin-coupled peptide was obtained and affinity-purified by two chromatography steps (Neosystem). Briefly, the serum was filtered (0.22-m filter), and following an addition of 1 M Tris (pH 8.0) up to a final concentration of 100 mM, it was applied to a SulfoLink column (Pierce) coupled to the non-phosphorylated peptide. The flow-through was collected, concentrated, and applied to the phosphopeptide column. Elution was performed with 100 mM glycine buffer (pH 2.7).
Human tissues were from the Harvard Brain Tissue Resource Center (Belmont, MA). Brain samples were homogenized in Nonidet P-40 lysis buffer and cleared by centrifugation at 6000 ϫ g for 15 min at 4°C. 50 g of homogenates were subjected to Western blot analysis. Samples 1-5 correspond to brain numbers 4741, 4744, 4751, 4797, and 4740, respectively, as numbered by the Harvard Brain Tissue Resource Center. Quantification of Western blots was performed and is expressed relative to actin levels. Human biopsies were procured following the guidelines recommended by the National Institutes of Health. Proteins were loaded onto 10% SDS-PAGE; transferred to polyvinylidene difluoride membrane; and immunoblotted with anti-arfaptin 2 (1:1000) (13), anti-arfaptin 2/POR1 (1:100; N19, Santa Cruz Biotechnology, Inc.), and anti-␤-actin (1:5000; AC15, Sigma) antibodies.
Immunofluorescence-Transfected cells were grown on lamininand/or poly-D-lysine-coated glass coverslip, fixed with 4% paraformaldehyde for 20 min, and incubated with the following primary antibodies: anti-arfaptin 2 (1:200) (13), anti-His tag (1:500; Cell Signaling Technology), anti-GM130 (1:100, BD Biosciences), and fluorescein isothiocyanate-conjugated anti-␣-tubulin (1:100, Sigma). Pictures of fixed cells were captured with a three-dimensional deconvolution imaging system. Measurement of Neuronal Survival and Intranuclear Inclusions-Four days post-plating, primary cultures of striatal neurons were transfected with wild-type huntingtin or polyQ-huntingtin and GFP to identify the transfected cells. To be certain that each neuron synthesizing GFP also expressed the huntingtin construct, transfections were performed by the modified calcium phosphate technique with a high ratio of DNA to GFP DNA (10:1) (2). Under these circumstances, Ͼ95% of the GFP-positive neurons also expressed the huntingtin construct (data not shown). GFP-positive neurons were scored by fluorescence microscopy in a blinded manner 16 and 36 h post-transfection. Cell death occurring within the GFP-positive cells was determined as the difference in the number of surviving neurons between the two time points and is expressed as a percentage of cell survival. For intranuclear inclusion scoring, striatal neurons were transfected with vectors of interest and a plasmid encoding ␤-galactosidase (10:1 ratio). Neurons were fixed 5 days post-transfection, immunostained, and analyzed for the presence of ubiquitin-positive intranuclear inclusions (anti-␤-galactosidase antibody (1:300), 5 Prime 3 3 Prime, Inc.; and anti-ubiquitin antibody (1:100), Dako Corp.). Each graph represents at least two to three independent experiments performed in duplicate or triplicate. Each bar in a given graph corresponds to the scoring of ϳ2000 neurons in neuronal survival experiments and to 500 neurons for inclusion scoring. Data were submitted to complete statistical analyses (see figure legends).

Akt Exerts a Neuroprotective Effect Independently of Ser 421
Phosphorylation in Huntingtin-We tested whether Akt could promote neuroprotection independently of Ser 421 phosphorylation. We used an in vitro neuronal model of HD to determine whether Akt could block the neuronal toxicity induced by an N-terminal fragment of huntingtin not containing Ser 421 . In primary cultures of striatal neurons, fragments of huntingtin containing the abnormal polyglutamine expansion induce neuronal death and the formation of neuronal intranuclear inclu-sions (12). We transfected striatal neurons with constructs encoding the first 171 amino acids of huntingtin containing 17 glutamine residues (wild-type, construct 171-17) or 68 glutamine residues (mutant, construct 171-68) and with a construct encoding Akt-ca and analyzed neuronal survival (Fig. 1A). As expected, construct 171-68 induced a statistically significant decrease in neuronal survival compared with construct 171-17. However, Akt completely inhibited neuronal death induced by huntingtin fragment 171-68 (Fig. 1A). We then analyzed the effect of Akt on the formation of neuronal intranuclear inclusions (NIIs). We found that Akt significantly decreased the percentage of neurons containing NIIs (Fig. 1B). In vitro phosphorylation assays showed that neither fragment 171-17 nor fragment 171-68 was phosphorylated by Akt-ca (data not shown). This suggests that the protective effects of Akt are probably not due to phosphorylation between amino acids 1 and 171 of huntingtin.
We have previously demonstrated that part of the neuroprotective effect of IGF-1 is mediated through the phosphorylation of huntingtin at Ser 421 (2). We show here that Akt protected striatal neurons from polyQ-huntingtin-induced toxicity not only by directly phosphorylating huntingtin, but also by phosphorylating other substrates that in turn modulate polyQ-huntingtin-induced toxicity. This indirect effect of Akt led to inhibition of both the death of striatal neurons and the formation of  .n.) and GST-fused proteins as substrates. The reaction products were resolved by SDS-PAGE; the gel was stained with Coomassie blue (left panel); and the 32 P-labeled proteins were visualized by autoradiography (right panel). Akt-ca phosphorylated full-length arfaptin 2 and its C-terminal portion. Phosphorylation was abolished when Ser 260 was mutated to Ala in the C-terminal portion of arfaptin 2 (S260A). C, after in vitro phosphorylation by Akt and trypsin digestion, the resulting phosphopeptides of wild-type and S260A arfaptin 2 were resolved by two-dimensional migration. The major dot observed with wild-type arfaptin 2 (see arrows) was lost when Ser 260 could not be phosphorylated. D, human embryonic kidney 293 cells were transfected with wild-type or S260A arfaptin 2 and with Akt-ca or the corresponding empty vector and serum-starved for 24 h. Protein extracts were analyzed with antibody against arfaptin 2 phosphorylated at Ser 260 (P-Arfa-S260; upper panel) and anti-arfaptin 2/POR1 antibody (lower panel). E, kinase assays were performed using recombinant SGK and GST-fused proteins as substrates and analyzed as described for B. GST-fused human huntingtin-(384 -467) (GST-hu-htt) was used as a positive control for SGK phosphorylation. SGK phosphorylated GSTfused human huntingtin-(384 -467), but did not phosphorylate GSTfused arfaptin 2. CB, Coomassie Blue.
intranuclear inclusions (Fig. 1). Therefore, Akt may act on a substrate that directly controls polyQ-huntingtin-induced toxicity. To identify such an Akt substrate, we screened data bases for proteins containing an Akt consensus phosphorylation site (17). We found, among many others, arfaptin 2, which has been shown to regulate polyQ-huntingtin aggregation (13).
Arfaptin 2 Is Up-regulated in HD-The striatum is the most affected region in HD. To determine whether arfaptin 2 is modified in the pathological situation, we analyzed the levels of arfaptin 2 in human striatal samples from control (CT), grade 3 HD (HD3), and grade 4 HD (HD4) patients ( Fig. 2A). Brain extracts were analyzed using anti-arfaptin 2/POR1 antibody (upper panel) and anti-␤-actin antibody as a control for protein levels (lower panel). Arfaptin 2 levels were higher in brain extracts from HD patients than in those from control individuals (Fig. 2B). This agrees with previous work on HD transgenic mouse brains showing up-regulation of arfaptin 2 (13) and further suggests its involvement in HD.
Arfaptin 2 Is a Substrate of Akt-Arfaptin 2 has a putative Akt phosphorylation site at Ser 260 (Fig. 3A). To test whether Akt phosphorylates arfaptin 2 in vitro, we incubated GST-fused arfaptin 2 with Akt-ca or inactivated Akt in the presence of [␥-32 P]ATP. Akt-ca phosphorylated arfaptin 2, whereas inactivated Akt did not (Fig. 3B). To demonstrate that Ser 260 is phosphorylated by Akt-ca, we generated a GST-fused C-terminal fragment of arfaptin 2 (amino acids 249 -341) with either Ser 260 or a Ser-to-Ala mutation at this position (S260A). The C-terminal fragment of arfaptin 2 was no longer phosphorylated by Akt-ca when Ser 260 was replaced with Ala (Fig. 3B), even though the amount of the GST-fused protein was similar. We carried out two-dimensional phosphopeptide mapping of full-length wild-type and S260A arfaptin 2 to confirm Ser 260 as a site of phosphorylation of arfaptin 2 (Fig. 3C). The major phosphopeptide observed in wild-type arfaptin 2 (see arrows) was not seen when Ser 260 was mutated to Ala, suggesting Ser 260 as the main site of phosphorylation of arfaptin 2 by Akt.
To unequivocally identify Ser 260 as an Akt phosphorylation site, we raised a polyclonal antibody that specifically recognizes arfaptin 2 phosphorylated at Ser 260 . As shown in Fig. 3D, the antibody against arfaptin 2 phosphorylated at Ser 260 recognized arfaptin 2 only when coexpressed with Akt-ca, but failed to recognize the same protein with the S260A mutation. These results show that Akt phosphorylates arfaptin 2 at Ser 260 in cells and that arfaptin 2 is a substrate of Akt.
Phosphorylation of Arfaptin 2 Modifies Its Cellular Distribution-Arfaptin 2 is located mainly in the perinuclear region of Chinese hamster ovary cells, in particular around the microtubule (MT)-organizing center and in cytoplasmic vesicular structures (13). To investigate whether phosphorylation alters the distribution of arfaptin 2, we transfected NG108-15 neuroblastoma cells with His-tagged wild-type or S260A arfaptin 2. We used immunofluorescence to compare the distribution of arfap- tin 2 and the Golgi marker protein GM130 (Fig. 4A). The transfected wild-type and endogenous arfaptin 2 proteins appeared in the cytoplasm as diffuse punctate staining and partially co-localized with GM130 in the perinuclear region. However, the distribution of S260A arfaptin 2 was very different. The protein was redistributed to bundle structures in the cytoplasm and was no longer detected in Golgi bodies or vesicles. To ensure that this effect was due to loss of phosphorylation at Ser 260 , we assessed the effect of constitutive phosphorylation of Ser 260 by replacing it with Asp (S260D). S260D arfaptin 2 had the same subcellular localization as arfaptin 2 in neuroblastoma cells (Fig. 4A) and in striatal neurons (data not shown). S260A arfaptin 2 also re-localized into bundle structures in neuritic extensions of striatal neurons (Fig. 4B). Quantification of bundle formation in neuroblastoma cells and neurons demonstrated that it was specifically associated with loss of phosphorylation at Ser 260 , as wild-type and S260D arfaptin 2 never formed bundles. In contrast, bundles were observed in 35-45% of NG108-15 neuroblastoma cells and neurons expressing S260A arfaptin 2 (Fig. 4C).
Arfaptin 2 Subcellular Localization Depends on the MT Network-Proteins that associate with MTs often induce bundles when overexpressed in cells (18). Bundle formation induced by loss of phosphorylation of arfaptin 2 at Ser 260 together with the localization of arfaptin 2 around the MT-organizing center led us to investigate the effect of phosphorylation on the distribution of arfaptin 2 in relation to the MT network (Fig. 5A). Unfortunately, the paraformaldehyde fixation step, which is essential for the detection of the anti-histidine antibody that localizes transfected arfaptin 2, slightly disrupted the MT network. Nevertheless, wild-type arfaptin 2 was located mainly around the MT-organizing center in a punctate staining pattern as described previously (13), whereas in cells expressing S260A arfaptin 2, the MT network was disrupted (Fig. 5A). To ensure that the localization of arfaptin 2 and its regulation by phosphorylation depend on MTs, we treated cells with nocodazole, an MT-depolymerizing agent. We found that both wildtype and S260A arfaptin 2 distributions were disrupted, with the proteins being concentrated in dense clusters throughout the cytoplasm (Fig. 5B). These results show that arfaptin 2 localization depends on the integrity of the MT network and that loss of phosphorylation of arfaptin 2 leads to an alteration of the MT organization.
We next studied the effect of phosphorylation on arfaptin 2 distribution in cells. We analyzed arfaptin 2 in whole cell extracts and in soluble and insoluble Nonidet P-40 fractions from COS-7 cells transfected with various arfaptin 2 constructs (Fig. 5C). Wild-type and S260D arfaptin 2 were concentrated in the soluble Nonidet P-40 protein fraction, whereas S260A arfaptin 2 was strongly enriched in the insoluble Nonidet P-40 protein fraction.
Thus, we have demonstrated by immunostaining and biochemical experiments that loss of phosphorylation of arfaptin 2 at Ser 260 dramatically alters the cellular distribution of arfaptin 2. This results in its accumulation into insoluble bundle structures and to the partial disruption of the MT network.
Arfaptin 2 Modulates PolyQ-huntingtin-induced Toxicity: Effect of Phosphorylation by Akt-We studied the physiological consequences of arfaptin 2 expression and phosphorylation of Ser 260 on polyQ-huntingtin-induced toxicity in primary cultures of striatal neurons (Fig. 6A). We cotransfected striatal neurons with construct 171-17 or 171-68 and various constructs of arfaptin 2 and analyzed cell survival. Wild-type and S260D arfaptin 2 partially rescued neurons from construct 171-68-induced neuronal death, whereas S260A arfaptin 2 enhanced neuronal death. We then analyzed striatal neurons transfected with the same constructs for the presence of NIIs (Fig. 6B). S260D arfaptin 2 decreased the percentage of neurons containing nuclear inclusions, whereas wild-type arfaptin 2 did not. In contrast, S260A arfaptin 2 significantly increased this percentage. These results show that arfaptin 2 phosphorylated at Ser 260 has a neuroprotective effect on polyQ-huntingtin-induced toxicity by reducing both neuronal death and nuclear inclusions.
We cotransfected primary cultures of striatal neurons with huntingtin, arfaptin 2, and Akt-ca to see whether the Ser 260dependent toxicity of arfaptin 2 is related to phosphorylation by Akt at this site (Fig. 6, C and D). As we have shown (Fig. 1A), Akt-ca inhibited neuronal death induced by construct 171-68.
In the presence of wild-type arfaptin 2, Akt still exerted a neuroprotective effect and blocked polyQ-huntingtin-induced neuronal death (Fig. 6C). However, when S260A arfaptin 2 was expressed in neurons, Akt-ca only partially rescued polyQhuntingtin-induced neuronal death (Fig. 6C). Thus, the neuroprotective effect of Akt is at least partly mediated through phosphorylation of arfaptin 2. When construct 171-68, Akt, and wild-type arfaptin 2 were coexpressed in striatal neurons, a decrease in formation of NIIs was observed (Fig. 6D). This effect is related to phosphorylation of Ser 260 by Akt-ca, as this was not observed in the absence of Akt (Fig. 6B) and was abolished when Ser 260 was mutated to Ala (Fig. 6D). Therefore, phosphorylated arfaptin 2 exerts a protective effect in HD that is mediated via phosphorylation of Ser 260 by Akt.
Arfaptin 2 Phosphorylation Rescues PolyQ-huntingtin-induced Proteasome Impairment-The accumulation of S260A arfaptin 2 in cells and the toxicity of this protein with increased formation of NIIs led us to test whether phosphorylation of arfaptin 2 could modulate proteasome function (Fig. 7A). Although the function of arfaptin is not fully understood, arfaptin 2 has been shown to regulate proteasome activity (13). We used GFPu-1 cells stably producing a short degron (CL1) fused to GFP such that the fusion protein is targeted for proteasomal degradation (16). Proteasome activity was monitored by the fluorescence in GFPu-1 cells: an increase in fluorescence indicates inhibition of protein degradation. We cotransfected GFPu-1 cells with wild-type or expanded polyglutamine N-terminal fragments corresponding to exon 1 of huntingtin (exon 1-17 or 1-68, respectively) and wild-type or S260A arfaptin 2. The levels of GFP fluorescence were monitored by fluorescence-activated cell sorting analysis. Cells expressing exon 1-68 were more fluorescent than cells expressing exon 1-17 ( Fig. 7A) (16,19), indicating that abnormal polyglutamine expansion in huntingtin impairs the proteasome. The fluorescence intensity was significantly lower in cells transfected with wild-type arfaptin 2 and exon 1-68. This shows that arfaptin 2 inhibits the proteasome blockade induced by polyQ-huntingtin. In contrast, non-phosphorylatable arfaptin 2 was unable to prevent polyQhuntingtin-induced inhibition of protein degradation by the proteasome. Thus, arfaptin 2 inhibits the polyglutamine-induced proteasome blockade, thereby facilitating protein degradation. This beneficial property is lost in the absence of Ser 260 phosphorylation. DISCUSSION We have shown that arfaptin 2 is neuroprotective in a neuronal model that has characteristics that are observed in HD patients, such as intranuclear inclusions of polyQ-huntingtin Cell survival was significantly decreased by construct 171-68, but was significantly improved by cotransfection with wild-type or S260D arfaptin 2. In contrast, cotransfection with S260A arfaptin 2 significantly reduced cell survival. *, p Ͻ 0.05 (post hoc Fisher's analysis); **, p Ͻ 0.01; ***, p Ͻ 0.0001. NS, not significant. B, data from six independent transfections (ANOVA; F (3,39) ϭ 11.16; p Ͻ 0.0001) revealed that wild-type arfaptin 2 had no effect on construct 171-68-induced formation of NIIs. However, cotransfection with S260A arfaptin 2 significantly increased the number of neurons containing inclusions, whereas cotransfection with S260D arfaptin 2 reduced it. *, p Ͻ 0.05 (post hoc Fisher's analysis); **, p Ͻ 0.01; ***, p Ͻ 0.0001. C, data from two independent experiments (ANOVA; F (4,72) ϭ 12.03; p Ͻ 0.0001) revealed a significant decrease in cell survival induced by construct 171-68 that was completely blocked by the presence of Akt-ca or of Akt-ca and wild-type arfaptin 2. In contrast, neuronal death in cells expressing construct 171-68 with Akt-ca and S260A arfaptin 2 (Arfa SA) was significantly different from all other cases. *, p Ͻ 0.05 (post hoc Fisher's analysis); ***, p Ͻ 0.001. D, data from two independent transfections (ANOVA; F (3,31) ϭ 3.97; p Ͻ 0.02) revealed significant differences in the percentage of cells containing NIIs in the presence of Akt-ca or of Akt-ca and wild-type arfaptin 2, but not in the presence of Akt-ca and S260A arfaptin 2. *, p Ͻ 0.05 (post hoc Fisher's analysis); **, p Ͻ 0.01. and neuronal death. We have demonstrated that arfaptin 2 is a substrate of Akt and that phosphorylation of arfaptin 2 by Akt at Ser 260 promotes survival of striatal neurons. This shows that arfaptin 2 is involved in the neuroprotective effect of Akt in HD (Fig. 7B). We have also demonstrated that arfaptin 2 phosphorylation restores proteasome activity that is inhibited by the presence of polyQ-huntingtin in cells.
Our results are consistent with the study of Peters et al. (13) demonstrating that arfaptin 2 regulates aggregate formation of mutant huntingtin. Here, we have shown that this regulation is dependent on the phosphorylation state of arfaptin 2. When arfaptin 2 was constitutively phosphorylated (S260D mutation), it reduced the formation of inclusions in primary cultures of striatal neurons. However, loss of phosphorylation of arfaptin 2 at Ser 260 enhanced aggregation of huntingtin. Interestingly, C-terminal deletion mutants of arfaptin 2 that lack the Ser 260 phosphorylation site enhance polyQ-huntingtin aggregation in PC12 cells (13). In contrast, we have found that wild-type arfaptin 2 does not modify aggregate formation. This may be due to differences described under "Experimental Procedures" or in the phosphorylation status of arfaptin 2 in each of the cell systems. In our study, it is likely that wild-type arfaptin 2 was phosphorylated, as wild-type arfaptin 2 and S260D arfaptin 2 had the same localization in neuroblastomas (Fig. 4) and similar protective functions (Fig. 6).
We have found that arfaptin 2 levels are increased in the brains of HD patients, which is consistent with studies in HD transgenic mice (13). We have shown that arfaptin 2 is neuroprotective and that this increase may reveal a protective mechanism in affected cells. In addition, we have previously demonstrated that Akt is altered in the latter stages of the disease (2). We did not detect phosphorylated forms of arfaptin 2 in the brain samples (data not shown) and therefore cannot determine whether the alteration of Akt regulates the phosphorylation of arfaptin 2 at Ser 260 in the disease. This lack of detection may be due to post-mortem intervals.
Our study identifies a novel substrate of Akt and points out a new neuroprotective mechanism in HD. Akt phosphorylates arfaptin 2, which subsequently rescues the polyglutamine-induced proteasome deficit. It has been previously shown that Akt regulates ubiquitination and degradation of proteins. For example, Akt stabilizes Mdm2 and XIAP (X-linked inhibitor of apoptosis protein) (20,21). This mechanism involves the phosphorylation of Mdm2 and XIAP by Akt, which leads to the inhibition of their ubiquitination and thus inhibits degradation by the proteasome. The prosurvival effect of Akt can also be mediated by the enhanced degradation of negative regulators of cell growth and survival. Akt also promotes degradation of tuberin and a FOXO family transcription factor, FOXO3a, by a phosphorylation-dependent ubiquitination mechanism (22).
The ubiquitin/proteasome system (UPS) is involved in several neurodegenerative disorders, including HD and other polyglutamine disorders (23). In polyglutamine disorders, aggregates may allow the temporary storage of misfolded toxic proteins prior to their degradation by the proteasome (12). In agreement with this hypothesis, to clear the misfolded and aggregated proteins, the proteasome must show efficient ubiquitination and degradation activity (12, 24 -26).
Several studies have revealed that the UPS is impaired during disease progression. Misfolded and/or aggregated proteins inhibit the UPS and lead to the accumulation of ubiquitin conjugates in cells (16,19). In addition, polyglutamine-containing proteins are not efficiently degraded by the UPS. This results in the further accumulation of expanded polyglutamine-containing short peptides. These are then prone to aggregation and thereby compromise the function of the UPS (27). Finally, the age-dependent decrease in proteasome activity also contributes to the accumulation of short fragments of polyQ-huntingtin (28). Therefore, aggregates are probably both inhibitors of the UPS and the consequence of UPS inhibition (16,29). Thus, the identification of molecules or pathways that regulate the UPS is of potential therapeutic value.
Our study shows that phosphorylated arfaptin 2 leads to recovery from the proteasome impairment induced by polyQhuntingtin. How does arfaptin 2 regulate proteasome function? Although arfaptin 2 interacts with Arf6 and Rac1, its function remains unknown (30 -32). The nature of the proteins that FIG. 7. Phosphorylation of arfaptin 2 at Ser 260 rescues polyQhuntingtin-induced proteasome impairment. A, exon 1-17 (ex1-17) and exon 1-68 (ex1-68) were cotransfected with vectors encoding wildtype or S260A arfaptin 2 (Arfa) or the corresponding empty vector in GFPu-1 cells. Data are from three independent experiments (ANOVA; F (3,16) ϭ 13.90; p Ͻ 0.0001). Fluorescence intensity was significantly increased by exon 1-68, but not when arfaptin 2 was coexpressed. In contrast, in the presence of S260A arfaptin 2, fluorescence was not decreased. ***, p Ͻ 0.001 (post hoc Fisher's analysis). NS, not significant. B, IGF-1 is neuroprotective in HD through the activation of several pathways. The closely related kinases Akt and SGK are activated by IGF-1 and in turn phosphorylate huntingtin (htt) at Ser 421 . This phosphorylation blocks the polyQ-huntingtin toxic effect. In addition, Akt phosphorylates arfaptin 2 at Ser 260 . When phosphorylated, arfaptin 2 inhibits the polyglutamine-induced impairment of the proteasome and therefore protects striatal neurons from the toxic effects of polyQ-huntingtin. PKB, protein kinase B; PDK1/2, 3-phosphoinositidedependent kinase-1/2; PI-3,4,5-P3, phosphatidylinositol 3,4,5-trisphosphate; PI3K, phosphatidylinositol 3-kinase. interact with arfaptin 2 and its subcellular localization (13) suggest that arfaptin 2 may regulate cytoskeletal remodeling involving the MT network. This is supported by our finding that loss of phosphorylation of arfaptin 2 at Ser 260 disrupts the MT network in neuroblastoma cells and neurons. As for aggresomes (33), an intact MT network is required for aggregate formation. This indicates that active transport along MTs participates in the sequestration of aggregates and in their subsequent processing by the proteasome (34). Arfaptin 2 may be a positive regulator of this process. As a defective proteasome is involved in several neurodegenerative disorders, it will be of interest to address this mechanism in future works.
We have previously shown that, upon IGF-1 activation, SGK phosphorylates polyQ-huntingtin at Ser 421 and abrogates its toxicity (14). This kinase does not phosphorylate arfaptin 2. These data further emphasize the importance of Akt in HD because Akt targets different proteins to protect neurons against polyQ-huntingtin (Fig. 7B). Akt abolishes the toxicity of full-length polyQ-huntingtin, which contains Ser 421 . Akt also phosphorylates arfaptin 2 at Ser 260 , thereby facilitating protein degradation. This effect on proteasome function is of particular interest. During the course of the disease, polyQhuntingtin is cleaved by several proteases, generating smaller and more toxic N-terminal fragments that do not contain Ser 421 . The phosphorylation of specific substrates by Akt may be a complementary process during the progression of the disease, further suggesting that the Akt pathway is of therapeutic value in HD.