H previous studies [40], we found the amount of aggregate formation cells in mutant-type ataxin-3 as much higher than that in normal control; demonstrating polyQ expansion could induce the formation of aggregates. Although there was no significantly difference in both aggregate cell counting and density quantification between ataxin-3-68Q and ataxin-3-68QK166R, we could found the tendency that aggregate density of ataxin-3-68Q was slightly higher than that of ataxin-3-68QK166R, which support the DOPS results of insoluble fraction detection and indicate that SUMOyla-tion of mutant-type ataxin-3 might partially increase its stability and probably promote aggregate formation. It has been reported that protein aggregates could sequester polyQ proteins which affects their normal biological function [39] and finally result in polyQ diseases. SUMOylation of the polyQ proteins might influences their aggregation and toxicity. For example, SUMOylation of the polyQ-expanded AR decreases the amount of the SDS-insoluble aggregates [41], and study on huntingtin proposed that SUMOylation may explain the intriguing cell death observed in polyQ disorders [42]. As what we show in Figure 5, SUMO-1 modification of mutant-type ataxin-3 increased the early apoptosis rate of the neurons, indicating that SUMOylation might enhance the stability of mutant-type ataxin3, thus increase its cytotoxicity, however the concrete mechanism still needs intensive study in future. In conclusion, our study demonstrated that SUMOylation on K166, the first described residue of SUMO-1 modification of ataxin-3, partially increased the stability of mutant-type ataxin-3, and the rate of apoptosis arisen from the cytotoxicity of the modified protein. Those support the hypothesis that SUMO-1 modification has a toxic effect on mutant-type ataxin-3 and participates in the pathogenesis of SCA3/MJD. Further studies in Drosophila models should be done to confirm these findings.The Effect of SUMOylation on Ataxin-Figure 4. SUMO-1 modification partially increased ataxin-3-68Q stability. HEK293 cells were transfected with GFP-ataxin-3 or GFP-ataxin3K166R. Immunoblotting analysis showed difference between the soluble (S) and insoluble (I) ataxin-3 in 20Q and 68Q with or without K166 (A). At 48 h after transfection, both aggregates formation cells and its immunoflurescence density were quantified. Plasmid groups: 1. GFP-ataxin-3-20Q; 2. GFP-ataxin-3-20QK166R; 3. GFP-ataxin-3-68Q; 4. GFP-ataxin-3-68QK166R. Statistical significance was assessed with a one-way ANOVA. The amount of aggregates formation cells: 1 and 3: P,0.05 (*); 1 and 2: P.0.05 (**); 3 and 4: P.0.05 (***) (B). Immunoflurescence density of aggregates: 1 and 3: P,0.05 (*); 1 and 2: P.0.05 (**); 23977191 3 and 4: P.0.05 (***) (C). At 24 h after transfection, cells were treated with CHX (100 mg/ml) to prevent protein synthesis. Cells were harvested at 0, 1, 3, 7, 15 h after CHX treatment, subject to 12 SDS-PAGE, and analyzed by immunoblotting with anti-GFP antibody (D). doi:10.1371/journal.pone.0054214.gMaterials and Methods Plasmid constructionPlasmids for myc-ataxin-3 and SUMO-1 in pcDNA3.1-mycHis(-)B (Invitrogen) have been described previously [32]. Ataxin3K8R, ataxin-3K166R, and ataxin-3K206R were all generated by sitedirected mutagenesis using long primers and Eliglustat web overlap methods with primers M1/M2, M3/M4, M5/M6, respectively. GFP-ataxin-3 and GFP-ataxin-3K166R were constructed by subcloning the PCR product amplified using primers M1/M2 with pc.H previous studies [40], we found the amount of aggregate formation cells in mutant-type ataxin-3 as much higher than that in normal control; demonstrating polyQ expansion could induce the formation of aggregates. Although there was no significantly difference in both aggregate cell counting and density quantification between ataxin-3-68Q and ataxin-3-68QK166R, we could found the tendency that aggregate density of ataxin-3-68Q was slightly higher than that of ataxin-3-68QK166R, which support the results of insoluble fraction detection and indicate that SUMOyla-tion of mutant-type ataxin-3 might partially increase its stability and probably promote aggregate formation. It has been reported that protein aggregates could sequester polyQ proteins which affects their normal biological function [39] and finally result in polyQ diseases. SUMOylation of the polyQ proteins might influences their aggregation and toxicity. For example, SUMOylation of the polyQ-expanded AR decreases the amount of the SDS-insoluble aggregates [41], and study on huntingtin proposed that SUMOylation may explain the intriguing cell death observed in polyQ disorders [42]. As what we show in Figure 5, SUMO-1 modification of mutant-type ataxin-3 increased the early apoptosis rate of the neurons, indicating that SUMOylation might enhance the stability of mutant-type ataxin3, thus increase its cytotoxicity, however the concrete mechanism still needs intensive study in future. In conclusion, our study demonstrated that SUMOylation on K166, the first described residue of SUMO-1 modification of ataxin-3, partially increased the stability of mutant-type ataxin-3, and the rate of apoptosis arisen from the cytotoxicity of the modified protein. Those support the hypothesis that SUMO-1 modification has a toxic effect on mutant-type ataxin-3 and participates in the pathogenesis of SCA3/MJD. Further studies in Drosophila models should be done to confirm these findings.The Effect of SUMOylation on Ataxin-Figure 4. SUMO-1 modification partially increased ataxin-3-68Q stability. HEK293 cells were transfected with GFP-ataxin-3 or GFP-ataxin3K166R. Immunoblotting analysis showed difference between the soluble (S) and insoluble (I) ataxin-3 in 20Q and 68Q with or without K166 (A). At 48 h after transfection, both aggregates formation cells and its immunoflurescence density were quantified. Plasmid groups: 1. GFP-ataxin-3-20Q; 2. GFP-ataxin-3-20QK166R; 3. GFP-ataxin-3-68Q; 4. GFP-ataxin-3-68QK166R. Statistical significance was assessed with a one-way ANOVA. The amount of aggregates formation cells: 1 and 3: P,0.05 (*); 1 and 2: P.0.05 (**); 3 and 4: P.0.05 (***) (B). Immunoflurescence density of aggregates: 1 and 3: P,0.05 (*); 1 and 2: P.0.05 (**); 23977191 3 and 4: P.0.05 (***) (C). At 24 h after transfection, cells were treated with CHX (100 mg/ml) to prevent protein synthesis. Cells were harvested at 0, 1, 3, 7, 15 h after CHX treatment, subject to 12 SDS-PAGE, and analyzed by immunoblotting with anti-GFP antibody (D). doi:10.1371/journal.pone.0054214.gMaterials and Methods Plasmid constructionPlasmids for myc-ataxin-3 and SUMO-1 in pcDNA3.1-mycHis(-)B (Invitrogen) have been described previously [32]. Ataxin3K8R, ataxin-3K166R, and ataxin-3K206R were all generated by sitedirected mutagenesis using long primers and overlap methods with primers M1/M2, M3/M4, M5/M6, respectively. GFP-ataxin-3 and GFP-ataxin-3K166R were constructed by subcloning the PCR product amplified using primers M1/M2 with pc.