Glycogen synthase kinase 3β suppresses polyglutamine aggregation by inhibiting Vaccinia-related kinase 2 activity

Huntington’s disease (HD) is a neurodegenerative disorder caused by an abnormal expansion of polyglutamine repeats in the N-terminal of huntingtin. The amount of aggregate-prone protein is controlled by various mechanisms, including molecular chaperones. Vaccinia-related kinase 2 (VRK2) is known to negatively regulate chaperonin TRiC, and VRK2-facilitated degradation of TRiC increases polyQ protein aggregation, which is involved in HD. We found that VRK2 activity was negatively controlled by glycogen synthase kinase 3β (GSK3β). GSK3β directly bound to VRK2 and inhibited the catalytic activity of VRK2 in a kinase activity-independent manner. Furthermore, GSK3β increased the stability of TRiC and decreased the formation of HttQ103-GFP aggregates by inhibiting VRK2. These results indicate that GSK3β signaling may be a regulatory mechanism of HD progression and suggest targets for further therapeutic trials for HD.

Consistent with its involvement in a variety of signaling pathways, GSK3β is associated with many diseases such as Alzheimer's disease 21 , cancer 22,23 , bipolar disorder 24 , diabetes 22 , and HD. Decreased GSK3β levels and activity are observed in the brains of R6/1 mice, an animal model of HD 25 , and reduced GSK3β levels are also found in the brains of human patients with HD 26 . However, the precise role of GSK3β in HD has not been elucidated.
In this study, we show that GSK3β directly interacted with VRK2 and inhibited VRK2 catalytic activity in vitro. Subsequently, GSK3β reduced VRK2-mediated degradation of TRiC, thereby suppressing polyQ-expanded Htt aggregation.

Results
GSK3β directly interacts with VRK2. We previously found that VRK2 indirectly affects the folding of polyQ proteins involved in HD 9,14 . To examine whether other proteins are involved in the regulation of VRK2mediated polyQ protein aggregation, we carried out immunoprecipitation analysis. We found that GSK3β was a VRK2 binding partner. To confirm the interaction between VRK2 and GSK3β , HEK293T cells were transfected with Flag-tagged human VRK2 and HA-tagged GSK3β followed by immunoprecipitation with α -Flag antibody. We found that Flag-VRK2 interacted with HA-GSK3β (Fig. 1a). To verify that GSK3β directly binds to VRK2 in vitro, recombinant GSK3β and VRK2 proteins were purified as GST (GST-GSK3β )-and Flag (Flag-VRK2)fused proteins in E. coli, respectively, and immunoprecipitation assays were performed with α -Flag antibody or control IgG. We found that VRK2 directly formed a complex with GSK3β (Fig. 1b). Furthermore, immunocytochemical analysis showed that EGFP-VRK2 colocalized with HA-GSK3β in the same subcellular compartments in SH-SY5Y (Fig. 1c) and HEK293A ( Supplementary Fig. S1a) cells.
To study the structural relationship between VRK2 and GSK3β , we performed in silico docking analysis. VRK2 and GSK3β structures found in the RCSB Protein Data Bank (PDB) were computationally docked into a 3D model using PatchDock. We found three binding interfaces with high scores in the protein-protein docking model (Fig. 1d). According to the crystal structure of VRK2 (PDB entry: 2V62) and GSK3β (PDB entry: 1I09), residues D269, L271, Q290, H296, K311, and H316 of VRK2 and D49, R50, D90, Y117, S119, G120, and K122 of GSK3β (Interface 1); residues Y191, K200, N205, R207, and D256 of VRK2 and R96, G202, Q206, E211, and E279 of GSK3β (Interface 2); and residues E66, G113, and S115 of VRK2 and D233, E279, and R282 of GSK3β (Interface 3) are crucial for the interaction (Fig. 1e). Furthermore, the electrostatic interaction model of Interface 1 showed that interactions occurred between acidic residues on GSK3β and basic residues on VRK2 ( Supplementary Fig. S1b). To understand the influence of surface charges on binding, we conducted point mutations of VRK2. K311, D256 and E66 were selected as key amino acid residues for binding to GSK3β because they have high affinity binding by hydrogen bond pairing. We observed that K311A point mutant (Interface 1), D256A point mutant (Interface 2), E66A point mutant (Interface 3) and K311/D256A/E66A triple mutant had weaker binding affinity for GSK3β compared with wild-type (WT) VRK2 ( Supplementary Fig. S1c). Taken together, these results indicate that VRK2 interacts with GSK3β in vitro, in vivo, and in silico.
GSK3β inhibits VRK kinase activity in vitro. As GSK3β and VRK2 directly interact and are both serine/threonine-protein kinases, we tested whether they phosphorylate each other in an in vitro kinase assay. To determine whether VRK2 is a substrate of GSK3β , recombinant His-VRK2 kinase-dead (KD) mutant protein was incubated with recombinant GST-GSK3β protein. We found that GSK3β did not phosphorylate VRK2 (Fig. 2a). Therefore, we repeated in vitro kinase assay to determine whether VRK2 phosphorylates GSK3β . We found that VRK2 did not phosphorylate GSK3β or affect its kinase activity (Fig. 2b). However, VRK2 autophosphorylation was decreased in the presence of GSK3β regardless of GSK3β kinase activity (Fig. 2b,c). These results suggest that VRK2 kinase activity is inhibited by GSK3β without protein phosphorylation and that GSK3β could act upstream of VRK2. Our previous study demonstrates that USP25 is phosphorylated and inhibited by VRK2. Among three fragments of UPS25, the fragment containing the regulatory domain between amino acids 655 to 780 from the N-terminal was phosphorylated by VRK2 9 . To clarify whether GSK3β inhibits the effect of VRK2 kinase activity on USP25 phosphorylation, we performed in vitro kinase assays with His-VRK2 protein, USP25 fragment, and GST-GSK3β . We found that VRK2-mediated phosphorylation of the USP25 fragment was reduced in the presence of WT or KD GSK3β (Fig. 2d). Furthermore, we confirmed that GSK3β decreased VRK2-mediated histone H3 phosphorylation ( Supplementary Fig. S2a). These results suggest that VRK2 kinase activity is inhibited by the presence of GSK3β protein regardless of its kinase activity.
GSK3β suppresses VRK2-mediated degradation of TRiC. The chaperonin TRiC is composed of eight homologous subunits, from CCT1 to CCT8. We previously found that VRK2 facilitates CCT4 polyubiquitination, which downregulates CCT4 14 . Based on the interaction between GSK3β and VRK2 and the observation that GSK3β is an upstream regulator of VRK2, we tested whether GSK3β regulates VRK2-mediated CCT4 downregulation. First, we examined the localization of GSK3β , VRK2, and CCT4. Fluorescence imaging showed that EGFP-VRK2 colocalized with Flag-GSK3β and HA-CCT4 in the same subcellular compartments of HEK293A cells (Fig. 3a). Next, we examined changes in the endogenous level of CCT4 with GSK3β overexpression. For this experiment, we used constitutively VRK2-overexpressing U2OS cells for visualizing prominent polyQ protein aggregates when polyQ-containing protein was overexpressed ( Supplementary Fig. S3a). We found that the level of CCT4 protein was significantly increased when WT or KD GSK3β was overexpressed (Fig. 3b,c). Next, we further examined CCT4 protein stability in two different cell lines with overexpression of Flag-VRK2 with or without HA-GSK3β . We found that GSK3β rescued CCT4 protein levels, whereas VRK2 overexpression decreased CCT4 protein levels (Fig. 3d,e). Taken together, these results suggest that GSK3β acts as a positive regulator of CCT4 by inhibiting VRK2.
Scientific RepoRts | 6:29097 | DOI: 10.1038/srep29097 GSK3β reduces VRK2-mediated polyQ aggregation. We previously found that VRK2 reduces CCT4 protein levels, resulting in the accumulation of HttQ103 aggregation 14 . To investigate the effect of GSK3β on VRK2-mediated aggregation of polyQ-containing protein, we evaluated polyQ aggregates using a GFP-fused polyQ-expanded Htt fragment. We coexpressed HttQ103-GFP and HA-GSK3β in HEK293A cells with or without overexpression of Flag-VRK2. Interestingly, HttQ103 insoluble aggregates were decreased in HA-GSK3β -overexpressing cells (Fig. 4a). Because GSK3β inhibits degradation of TRiC protein levels by    inhibiting VRK2 kinase activity, this suggests that functional TRiC plays a role in inhibiting pathogenic HttQ103 aggregation. In addition to the stabilization of TRiC, the amount of insoluble HttQ103 aggregates was also reduced by overexpression of GSK3β WT or KD (Supplementary Fig. S4a). We further confirmed that GSK3β reduced HttQ103 aggregation by filter-trap assay (Fig. 4b,c, Supplementary Fig. S4b). Consistent with the results of Western blotting and filter-trap assay, HttQ103 aggregates did not increase significantly in GSK3β overexpression with co-overexpression of VRK2 condition while VRK2 alone overexpression increased aggregates (Fig. 4d,e). Collectively, these results suggest that GSK3β -mediated VRK2 inhibition reduces HttQ103 aggregation through the stabilization of chaperonin.

Discussion
HD is a neurodegenerative disorder caused by an expanded CAG repeat sequence (encoding a poly-glutamine stretch) in the Htt gene 27 , which causes protein aggregation in brain cells and progressive cell death, especially in the striatum and cortex 1,2 . Accumulation of mutant Htt protein is toxic and affects important cellular functions such as transcription, axonal transport, synaptic transmission, and mitochondrial function 4 . Mutant Htt interacts with several cellular proteins and sequesters into nuclear aggregates or forms cytoplasmic inclusions that cause progressive neuronal degeneration in HD mouse models. The number of glutamine residues correlates with the severity of HD symptoms, and the pathological hallmark of HD is the formation of polyQ-expanded aggregates 3,4 . Therefore, it is important to prevent this process. Therapeutic strategies for HD and other neurodegenerative diseases have focused on elevating molecular chaperones 5 or activating various cleaning mechanisms, including proteolytic systems 28 and autophagy 29 . Chaperones promote refolding or degradation of misfolded proteins and prevent protein aggregation. A previous study reports that eukaryotic chaperonin TRiC attenuates the accumulation of Htt-polyQ protein aggregates and their cytotoxicity 6 . Therefore, functional TRiC is involved in decreasing pathogenic HttQ103 aggregate formation.
Recent studies indicate that TRiC directly interacts with and regulates the transcriptional activity of HSF1 30 . Also, GSK3β phosphorylates HSF1 on residue serine 303 31 . Thus, we performed co-immunoprecipitation to test whether GSK3β binds to HSF1 in the presence of VRK2. Binding between GSK3β and HSF1 was not significantly changed by VRK2 overexpression (Supplementary Fig. S5a), suggesting that HSF1 constitutively binds to GSK3β in the presence or absence of VRK2. Next, we examined whether GSK3β -mediated inhibition of VRK2-induced TRiC destabilization affects HSF1 activity by examining transcription levels of the HSF1 target gene hsp70 using real-time quantitative PCR. We found no change in HSF1 transcriptional activity under our experimental conditions ( Supplementary Fig. S5b).
We previously found that VRK2 specifically binds to TRiC/CCT subunits, causing chaperonin downregulation through the ubiquitin-proteasome system 14 . This results in a lack of sufficient chaperonin action to mediate proper protein folding in VRK2-overexpressing cells. The level of active VRK2 enzymes may be an important indicator of HD onset and progression. To prove this, alternatively, we investigated VRK2 protein levels in a Huntington's disease model, R6/2 transgenic mouse, which is a well-studied transgenic animal and most widely used as HD mouse model. The amount of VRK2 protein was relatively increased in the R6/2 transgenic mouse brain lysates compared to wild-type mouse brain lysates at 4 weeks ( Supplementary Fig. S6a). In support of this concept, recently, many genome-wide association studies suggest that VRK2 is a risk factor for neurological disorders including schizophrenia 12,13 and epilepsy 11 . VRK2 mRNA levels are upregulated in schizophrenic patient brains compared with normal control brains 12 , and VRK2 gene rs3732136 polymorphism is associated with schizophrenia in the worldwide population 32 . Therefore, genetic variations could affect expression of the VRK2 gene, and malfunctions in VRK2 kinase activity might contribute to susceptibility to neurodegenerative diseases.
In the present study, we demonstrated that GSK3β and VRK2 directly interact and co-localize in cells using in vitro and in vivo binding assays. Structure-based docking analysis showed the presence of many hydrogen bonds and high-affinity interactions between GSKβ and VRK2. This docking analysis suggests that binding Interfaces 2 and 3 of GSK3β are interacting sites that bind to VRK2 outside of the active site. GSK3β is positioned near the catalytic domain of VRK2, and the binding of GSK3β to VRK2 altered protein flexibility in regions involved in substrate binding and turnover, suggesting their importance to kinase activity. Thus, GSK3β may inhibit VRK2 catalytic activity by disrupting its flexibility. The inhibition of VRK2 catalytic activity by GSK3β may also inhibit VRK2-induced degradation of TRiC, which could suppress polyQ-expanded Htt aggregation. However, a previous study demonstrates that GSK3β inhibitor protects polyglutamine protein-induced cell death but had inconsistent effects on the number of Htt aggregates in cells 33 . GSK3β inhibitor also has variable effects in a transgenic mouse model of HD 34 and has beneficial effects when combined with other drugs 35,36 . However, the main reason why GSK3β inhibitors do not have reliable effects may be due to markedly reduced levels and activity of GSK3β in the striatum and other brain regions in R6/1 HD mouse models and the frontal cortex of postmortem HD patients 25 . Our findings support previous observations of decreased levels and reduced activity of GSK3β in HD mouse models and postmortem human brains. Therefore, the level of GSK3β in the brain may be critical for controlling HD progression.
In conclusion, we propose that GSK3β has positive effects on HD by regulating VRK2 and the stability of chaperonin TRiC. GSK3β appears to inhibit VRK2 kinase activity via protein-protein interaction in a kinase activity-independent manner, thereby reducing ubiquitination-proteosomal degradation of TRiC and suppressing polyQ protein aggregation (Fig. 5). Our results could aid in the development of efficient ways for preventing or treating HD by regulating GSK3β levels.

Antibodies. Antibodies used in this study included: anti
Filter-trap assay. Cells were lysed in RIPA buffer, and sonicated lysates were diluted in 1% SDS-PBS and boiled at 95 °C for 5 min. The membrane was pre-equilibrated with 1% SDS in PBS. Proteins were loaded onto a 0.22-μ m nitrocellulose membrane. A dot blotter apparatus (Bio-Rad, Hercules, CA) was used for the application of samples, and anti-GFP antibody (Santa Cruz, CA) was used for detection.
Fluorescence microscopy. Transfected cells were maintained for 24-48 h, fixed with 4% paraformaldehyde for 20 min, and immunostained with appropriate primary and fluorescence-conjugated secondary antibodies. Coverslips were mounted onto slides using Fluoromount (Sigma). All images were obtained using a laser scanning confocal microscope (model FV1000; OLYMPUS), and FV10-ASW2.0 fluoviewer software was used for image analysis.
Structural modeling and analysis. PatchDock was used to study the structural relationship between GSK3β and VRK2 37 . According to the crystal structure of GSK3β (PDB entry: 1I09) 38 and VRK2 (PDB entry: 2V62) 39 , the best structure for docking was selected and further analyzed using PyMOL.
Establishment of stable cell lines. Stable cell lines were generated by G418 selection. Briefly, pNTAP-mock or pNTAP-hVRK2 vector was transfected into the normal U2OS cell line using electroporation and incubated for 24 h. Cells were selected by 800 μ g/ml G418 with media changes every 2 days until single colonies formed. After single colony isolation, stable cells were maintained with 200 μ g/ml G418-containing media.
Quantitative real-time reverse transcription PCR. Total RNA was isolated using TRI reagent (Molecular Research Center, Cincinnati, OH) and reverse transcribed using ImProm-II (Promega) according to the manufacturer's instructions. For detection and quantification, the StepOnePlus real-time PCR system (Applied Biosystems) was used.
Ethics statement. Approval of the study protocol was obtained from the Pohang University of Science and Technology Institutional Animal Care and Use Committee (POSTECH IACUC) (Approval ID: 2013-03-004). All animal experiments were carried out according to the provisions of the Animal Welfare Act, PHS Animal Welfare Policy, and the principles of the NIH Guide for the Care and Use of Laboratory Animals. All mice were maintained under conventional conditions at the POSTECH animal facility under institutional guidelines.

Statistical analysis.
Quantitative data are shown as mean ± standard error of the mean (SEM). Statistical analyses were performed using Student's t-test and one-way analysis of variance (ANOVA), two-way ANOVA and post-hoc Tukey's multiple comparison tests using GraphPad software.