Implications of fALS Mutations on Sod1 Function and Oligomerization in Cell Models

Among the familial forms of amyotrophic lateral sclerosis (fALS), 20% are associated with the Cu,Zn-superoxide dismutase (Sod1). fALS is characterized by the accumulation of aggregated proteins and the increase in oxidative stress markers. Here, we used the non-invasive bimolecular fluorescence complementation (BiFC) assay in human H4 cells to investigate the kinetics of aggregation and subcellular localization of Sod1 mutants. We also studied the effect of the different Sod1 mutants to respond against oxidative stress by following the levels of reactive oxygen species (ROS) after treatment with hydrogen peroxide. Our results showed that only 30% of cells transfected with A4VSod1 showed no inclusions while for the other Sod1 mutants tested (L38V, G93A and G93C), this percentage was at least 70%. In addition, we found that 10% of cells transfected with A4VSod1 displayed more than five inclusions per cell and that A4V and G93A Sod1 formed inclusions more rapidly than L38V and G93C Sod1. Expression of WTSod1 significantly decreased the intracellular oxidation levels in comparison with expression of fALS Sod1 mutants, suggesting the mutations induce a functional impairment. All fALS mutations impaired nuclear localization of Sod1, which is important for maintaining genomic stability. Consistently, expression of WTSod1, but not of fALS Sod1 mutants, reduced DNA damage, as measured by the comet assay. Altogether, our study sheds light into the effects of fALS Sod1 mutations on inclusion formation, dynamics, and localization as well as on antioxidant response, opening novel avenues for investigating the role of fALS Sod1 mutations in pathogenesis.


Introduction
Amyotrophic lateral sclerosis (ALS) is an age-associated neurodegenerative disease characterized by progressive muscle paralysis and death [1]. Approximately 10% of ALS cases are associated with autosomal dominant mutations (fALS) [2]; among them, 20% are caused by Cu, Zn-superoxide dismutase (Sod1) mutations [3]. Sod1 is an abundant and important protein in eukaryotic cells. Its physiological function is mainly related to the reduction of oxidative stress by converting superoxide anions to hydrogen peroxide, which is in turn broken down to water and oxygen by peroxidases [4]. Recently, Sod1 was reported to translocate into the nucleus of yeast cells in response to oxidative stress, where it regulates the expression of antioxidant and repair genes [5].
More than 150 individual mutations have been identified in human Sod1 (hSod1), affecting in different ways the onset and the prognosis of fALS [6,7]. Nevertheless, despite intense studies, the molecular mechanisms connecting mutation Aline A. Brasil and Rayne S. S. Magalhães have equal contribution.
* Tiago F. Outeiro touteir@gwdg.de hSod1 to fALS etiology and pathology are still elusive. The toxicity of fALS hSod1 has been related to structural instability [8], misfolding, aberrant enzymatic activity (lost and gain of function) [9], and disturbance of redox homeostasis [10] as well as to the presence of Sod1-containing aggregates in juxtanuclear quality control compartment (JUNQ) [11]. In this context, while transgenic mice expressing the fALS-associated hSod1 mutants G93A and A4V exhibit ALS-like phenotypes, animal knockouts for endogenous Sod1 do not [9,12].
In vitro studies have shown that metal coordination and disulfide bond formation are relevant for Sod1 structural stability [13][14][15], suggesting that an unfolded Sod1 conformation is responsible for the increased aggregation propensity. In addition, it has been shown that aggregation propensity of recombinant Sod1 does not always predict aggregation in a cellular context. It is not surprising due to obvious differences in protein concentration and surrounding environment between in vitro and in cell [16]. Therefore, in this study, we set out to assess the contribution of fALS hSod1 (A4V, L38V, G93A and G93C) towards protein aggregation, oxidative stress, and DNA damage. Using the bimolecular fluorescence complementation (BiFC) assay, which allows to observe protein interactions in vivo, we found that these Sod1 mutations induce defective dimerization and, consequently, affect protein aggregation [17]. Importantly, we investigated, for the first time, the dynamics of fALS Sod1 aggregation in living cells and the subcellular partitioning of Sod1 inclusions in respect to the JUNQ compartment. We found that the fALSassociated hSod1 mutants tested are unable to overcome oxidative stress and DNA damage.
Altogether, our results clearly define a common mechanism of action of mutant Sod1, resulting in cytotoxicity, impaired function, aggregation, and in alterations in intracellular distribution, opening novel territory for the identification of targets for therapeutic intervention in ALS.

Fluorescence Microscopy
Forty-eight hours after transfection, H4 cells were washed with Dulbecco's phosphate-buffered saline (DPBS) and fixed with 4% paraformaldehyde (PFA) for 10 min at room temperature (RT). Followed by three washing steps with DPBS, cells were stained with Hoechst 33258 (Life Technologies-Invitrogen, Carlsbad, CA, USA) (1∶5000 in DPBS) for 5 min and maintained in DPBS for fluorescence microscopy. Fluorescence images were acquired with a Leica DMI 6000B microscope (Leica, Germany), with a ×40 objective. Scale bars were calculated by using ImageJ software and were included in the figure legends together with the actual magnification.

Quantification of Nuclear and Cytoplasmic Fluorescence Intensities
Nuclear and cytoplasmic fluorescence intensities were quantified using ImageJ software (http://rsbweb.nih.gov/ij/). Using the freehand tool, the nucleus and cytosol were selected and the respective intensities were measured. The results reflect the counting of at least 50 cells per condition.

Quantification of hSod1 Inclusions
Transfected cells were detected and scored based on the hSod1 inclusions pattern and classified into three groups: cells without inclusions, five or less inclusions (≤ 5 inclusions), and more than five inclusions (≥ 5 inclusions). Results reflect the counting of at least 50 cells per condition.

Live Cell Imaging
Images of H4 cells expressing BiFC-tagged hSod1 were recorded by using the Olympus IX81-ZDC microscope system (Olympus, Germany).Cells were maintained in DMEM,(Life Technologies-Invitrogen, CA,USA),supplementedwith 10% (v/v)FBSand1% (v/v) penicillin-streptomycin at 5% atmospheric CO 2 , with the incubation unit set to a temperature of 37°C. Live cells were imaged every 30 min over a time course of 20 h overnight.

Immunoblotting
Forty-eight hours post-transfection cells were washed with room temperature DPBS and then harvested. H4 cells were lysed with radioimmunoprecipitation assay (RIPA) lysis buffer (50 mM Tris pH 8.0, 0.15 M NaCl, 0.1% (w/v) SDS, 1% NP40 (v/v), 0.5% (w/v) Na-deoxycholate), 2 mM EDTA, and a protease inhibitor cocktail (one tablet/10 mL) (Roche Diagnostics, Mannheim, Germany). Protein concentration was determined using the Bradford assay (BioRad Laboratories, Hercules, CA, USA), and the gels were loaded with 30 μg protein after denaturation for 5 min at 95°C in a protein sample buffer (125 mM of 1 M Tris HCl pH 6.8, 4% in 4% (w/v) BSA/TBS-Tween overnight at 4°C. After washing three times in TBS-Tween for 5 min, the membranes were incubated for 1 h with anti-mouse IgG and anti-goat IgG horseradish peroxidase labeled secondary antibody (GE Healthcare, Bucks, UK) at 1∶6000 in 4% (w/v) BSA/TBS-Tween. Detection was carried out using luminol reagent and peroxide solution (Millipore, Billerica, MA, USA) and applied to the membrane 1 min before scanning with in Fusion FX (Vilber Lourmat, Collégien, France). The band intensity was estimated using the ImageJ software (NIH, Bethesda, MD, USA) and normalized against γ-tubulin.

Detection of DNA Damage
The comet assay, a single cell gel electrophoresis-based method, was used to the detect DNA single-and double-strand breaks, according to the protocol previously described in [21]. Agarose coated slides were prepared by dipping the slides into a 1% (w/v) low-gelling temperature agarose (PeqLab) and allowed to air-dry. Approximately 5 × 10 4 cells/mL were harvested in DPBS, mixed with 1% agarose and placed on a precoated slide. After agarose has gelled, alkaline lysis was performed by submerging the slides in alkaline buffer (1.2 M NaCl, 100 mM Na 2 EDTA, 0.1% (w/v) sodium lauryl sarcosinate, 0.26 M NaOH, pH > 13) and stored overnight at 4°C. After overnight lysis, the alkaline rinse solution (0.03 M NaOH, 2 mM Na 2 EDTA, pH = 12.3) was used to wash the slides three times and to conduct the Slides were neutralized with distilled water and stained with 2.5 μg/mL of propidium iodide for 20 min at RT. DNA damage quantification was carried out by examining at least 50 comet images from each slide. The images were acquired with an epifluorescence microscope (Leica DMI 6000B microscope, Leica, Germany). CometScore software (TriTek Corp) was used to determine the tail moment (product of the tail length and the percentage of total DNA in the tail) of each individual comet image.

Analysis of Intracellular Oxidation
The generation of intracellular reactive oxygen species (ROS) was assessed in H4 cells expressing BiFC-tagged hSod1 or untagged hSod1 (WT and mutants). Briefly, the cells were seeded in 48-well plates and washed once with DPBS. For ROS production measurement, cells were incubated with 25 μM 2′,7′-dichlorofluorescin diacetate (DCFDA, Sigma) at 37°C for 30 min. Following the incubation period, two washing steps with DPBS were performed to remove excess probe and fluorescence intensity (excitation 485 nM; emission 535 nM) was measured using the microplate reader Infinite M2000 PRO, Tecan. After three basal measurements, cells were challenged with 5% (v/v) H 2 O 2. The fluorescence values were recorded up to 45 min and normalized to those obtained from non-H 2 O 2 -induced cells.

Statistical Analyses
Data were analyzed using Graph Pad Prism 5 (San Diego, California, USA) software and were expressed as the mean ± SD of at least three replicates. Statistical differences from WT Sod1 were calculated using one-way ANOVA and two-way ANOVA with Bonferroni correction. Significance was assessed for, where an asterisk corresponds to p < 0.05, double asterisks corresponds to p < 0.01, and triple asterisks corresponds to p < 0.001.

Results fALS Mutations on hSod1 Promote Oligomerization and Impair Nuclear Localization
In order to assess the effect of hSod1 mutations on oligomerization and subcellular localization, we used the BiFC-based assay which involves the fusion of Sod1 with non-fluorescent fragments of the Venus protein [22]. This assay is based on the reconstitution of a functional Venus fluorescent protein upon dimerization of hSod1, enabling the direct visualization of the formation of hSod1 oligomers and inclusions in living cells (Fig.1a). WT hSod1 formed only dimers/oligomers in H4 cells, but not inclusions, unlike all other variants tested (Fig. 1b, c). The inclusions formed by the hSod1 mutants tested exhibited different sizes and shapes (Fig. 1b). In addition, the number of inclusion per cell was also variable (Fig. 1c). For hSod1 A4V, 80% of the cells displayed small round inclusions. For the other mutants, around 30% of the cells displayed larger and elongated inclusions. A4V and L38V inclusions were found scattered over the cytoplasm, and G93A and G93C formed a larger number of inclusions per cell.
In addition, we also found that all fALS hSod1 mutations reduced the nuclear localization of Sod1 to half of that observed with WT hSod1 (Fig. 1b, d). The levels of A4V, G93C, and L38V hSod1 variants were identical, except those of G93A, suggesting this mutant may be more unstable (Fig.  1e, f).
Next, we asked whether the inclusions formed by Sod1 were localized in stress granules (SG), and costained cells with an antibody against G3BP, an established marker of SGs [23,24]. We found that none of the inclusions formed by the hSod1 mutants tested in this study colocalized with G3BP (Fig. 2).
In order to investigate the site of accumulation of hSod1 in H4 cells, we monitored the localization of the inclusions using von Hippel-Lindau (VHL) fused to mCherry as a marker for the JUNQ compartment. VHL is a heterologous protein that misfolds and forms soluble aggregates that are targeted to the JUNQ compartment, where they colocalize with the quality control machinery [25]. To improve the visualization of VHL-mCherry proteins into JUNQ, H4 cells were incubated with the proteasome inhibitor MG132 (10 μM) for 7 h, 48 h after transfection. The inhibition of the proteasome did not change the visualization of the hSod1 variants into JUNQ, suggesting that the levels of fALS inclusions were sufficiently high under normal conditions (Fig. 3). Magnification ×1000. c Quantification of the number of inclusion per cell. At least 50 cells were counted per condition and classified in three different groups: blue, purple, and yellow bars represent the percentage of cells without any inclusion, with five or less inclusions and cells with more than five inclusions, respectively. Data was combined from at least three independent experiments. One-way ANOVA, with Bonferroni correction, was used for statistical analysis to compare differences between WT vs mutant cells (*** p < 0.001). d Subcellular localization of BiFC-hSod1. BiFC fluorescence intensity was quantified using ImageJ. For each condition, 50 cells were analyzed. e Representative immunoblot confirming expression of VN-Sod1 and Sod1-VC fragments. f Quantification of the immunoblots. Data is expressed as mean ± SD of at least three replicates. One-way ANOVA, with Bonferroni correction, was used for statistical analysis with significance level of p < 0.001 represented by three asterisks To investigate the dynamics of the formation of Sod1 inclusions in vivo, live-cell imaging was performed during 20 h post-transfection. The time of appearance of the first inclusions was recorded and compared among the variants of hSod1 tested. Upon careful analysis of the data, we verified that A4V and G93A hSod1 inclusions appeared about 3 h earlier than L38V and G93C hSod1 inclusions (Fig. 4) and persisted in the cells for more than 5 h. This clearly suggests a distinct aggregation dynamics between A4V/G93A and L38V/G93C fALS mutations.

fALS hSod1 Mutants Do Not Efficiently Protect against ROS and DNA Damage
In order to assess whether BiFC-tagged Sod1 retains function, we measured intracellular ROS levels using the oxidative stress sensitive 2′,7′-dichlorofluorescein (DCFH) probe. Using this approach, we also analyzed whether A4V, L38V, G93A, and G93C fALS mutations affected hSod1 function. We found that the levels of ROS observed in cells carrying an empty vector were similar to those observed in cells transfected with hSod1 constructs. Thus, we then asked whether we could detect differences in the handling of ROS in cells transfected with the hSod1 constructs. After 30 min of exposure of the cells expressing the different hSod1 variants to hydrogen peroxide (H 2 O 2 ), we found that the WT, but not the mutant hSod1, reduced the levels of ROS (Figs. 5 and 6). This suggests that the fusions of WT hSod1 with the BiFC Venus protein fragments retain dismutase function, leading to reduced levels of ROS. In contrast, the Sod1 mutants lacked dismutase activity, suggesting the mutations affected the antioxidant activity of Sod1. This suggests the following: (i) the fusions of WT hSod1 with the BiFC Venus protein fragments do not impair the antioxidant role of the enzyme; (ii) A4V, L38V, G93A, and G93C fALS mutations cause hSod1 loss of function.
Next, to investigate whether the different hSod1 variants protected from DNA damage, we measured DNA single-and double-strand breaks using the comet assay. In this assay, the tail moment is expressed as the relation between the tail length  (the smallest detectable size of migrating DNA) and the fraction of total DNA in the tail (the number of relaxed/broken pieces) [26]. Interestingly, we found that WT hSod1 protected against DNA damage, confirming that the BiFC tags did not affect Sod1 activity. In contrast, expression of A4V, L38V, and G93C hSod1 mutants did not protect against DNA damage as  Chronological and comparative analyses of the dynamics of inclusions formation. Data is expressed as mean ± SD of at least three replicates. One-way ANOVA, with Bonferroni correction, was used for statistical analysis with significance level of *** p < 0.001, comparing hSod1A4V with L38V, G93C, and G93C hSod1 mutations WT hSod1, having tail moments about five times larger, suggesting that these fALS Sod1 mutants are not able to maintain genomic stability due to their reduced presence in the nucleus (Fig. 1d) Fig. 5 FALS mutants do not protect cells against ROS production. Intracellular ROS measurements were conducted in transfected H4 cells expressing the BiFC-tagged hSod1 by using the oxidant-sensing probe 2′,7′-dichlorofluorescein (DCFH). Thirty minutes after treatment with 5% H 2 O 2 , total cellular levels of ROS were significantly lower in H4 cells expressing BiFC-tagged WT Sod1 in comparison to fALS mutations A4V, L38V, G93A, and G93C levels. At basal levels, no differences in intracellular ROS between WT, and the four Sod1 mutations were observed. Data is expressed as mean ± SD of at least three replicates. Two-way ANOVA, with Bonferroni correction, was used for statistical analysis with significance level of * p < 0.05, comparing WT with Sod1 mutants exhibited the highest levels of DNA damage, which were about 50% higher than those observed for cells expressing WT Sod1.
Discussion hSod1 mutations associated with fALS are prone to misfolding and aggregation [27]. Various in vitro and in vivo studies have assessed the effects of fALS-associated mutations on hSod1, in an attempt to afford novel insight into the molecular basis of the disease [13-15, 17, 28-31]. Several studies suggested that the toxicity of hSod1 mutants is related with the respective amino acid substitution [32] and with alterations in the intracellular localization of the protein [11,25,33]. Moreover, the cytotoxic mechanisms exerted by Sod1 mutants have also been associated with a bulk saturation of clearance mechanisms (e.g., proteasomal or autophagic protein degradation), saturation of chaperone function and dysfunction of mitochondria, alterations in apoptotic pathways, and axonal disorganization and disrupted axonal transport [34][35][36][37]. However, the precise molecular underpinnings of fALS are still unclear.
The aggregation of mutant Sod1 in living cells and the alteration in conformation and localization have been monitored by tagging the protein with fluorescent fluorophores [31,38]. Although such approaches may alter the biological function of proteins, they enable important studies that cannot be performed otherwise.
Given that hSod1 spontaneously dimerizes in normal conditions and forms a homodimer [39], we took advantage of the BiFC assay, based on the complementation of two nonfluorescent fragments of the fluorescent protein, in order to investigate the effect of fALS-associated mutants on hSod1 dimerization, oligomerization and aggregation, and on functional readouts of activity.
Initially, we observed that BiFC-tagged Sod1 A4V, L38V, G93A, and G93C dimerized and formed intracellular inclusions. This effect was particularly strong for the A4V mutant, which significantly increase in the percentage of cells with inclusions.
It is still unclear whether Sod1 inclusions are toxic, and additional studies using different model systems will be instrumental to address this. Recent studies have connected protein inclusions with SGs [24]. Using spinal cord motor neurons [40] or HeLa cells as models [23], it was shown that inclusions formed by G93A and A4V Sod1 mutants colocalize with G3BP, an established marker of SGs. In our study, we did not observe colocalization of the inclusions formed with G3BP, suggesting differences in the types of inclusions formed. Our observations are consistent with those reported recently, where inclusions of G93A Sod1 did not colocalize with SGs in HEK 293 cells [40]. Thus, our results suggest that, in normal conditions, the Sod1 mutants studied do not colocalize with SGs.
Sod1 aggregation is directly dependent of dimer dissociation, which is promoted by mutations that weaken the interactions at the dimer interface [41,42]. FRET analysis indicated that fALS mutations impaired Sod1 dimerization, consequently affecting protein aggregation [17]. In addition, it was reported that mutant Sod1 tagged with YFP formed cytoplasmic inclusions in cultured cells, in contrast to WT Sod1 [43]. Consistently, WT Sod1-GFP chimeric protein maintain its dismutase activity and does not really form inclusions when expressed in PC12 cultured cells [44]. Thus, considering only the number of inclusions counted per cell, the A4V mutation promoted the most harmful effect in comparison to the other mutations. Live-cell imaging revealed that A4V and G93A mutations formed inclusion faster than G93C and L38V, suggesting, for the first time, that Sod1 variants A4V and G93A need the same time to aggregate and have more propensities to form inclusions over a time course. It has been shown that A4V variant had the most considerable dimerization defect and the strongest Sod1 aggregation activity, which is in agreement with the earlier aggregate formation observed in Fig. 3 [17]. In addition, G93A accumulated at lower levels, indicating that this altered protein is more rapidly degraded in cells. In comparison to other mutations related to fALS, G93A showed early disease onset related to the worst prognostic [45]. Moreover, in vitro studies that investigated the correlation between the propensity for aggregation and conformational stability of Sod1, showed that G93A have the highest conformation instability compared to other mutations, including A4V [46]. These fALS-linked mutations are located throughout the threedimensional structure of the Sod1 protein. The G93A is located on turns of β-sheets of the β-barrel, and the A4V mutation is at the dimer interface. Studies using spectroscopic techniques [47,48], thermal stability measurement [49], and disulfide reduction [32] have demonstrated that G93A and A4V mutation contribute to a reduction in protein stability, which may be related to their toxic gain of function. It has been shown that accumulation of toxic G93A Sod1 in the JUNQ compartment interferes with the quality control function, inhibiting the degradation of other misfolded proteins by sequestering Hsp70, thereby blocking proteasomal function. It was previously demonstrated that Sod1 A4V is a JUNQ-like aggregate which colocalizes with ubiquitin [33]. Thus, we verified whether the BiFC-tagged G93A Sod1 was indeed directed to JUNQ. Remarkably, we observed that all mutations analyzed colocalized with VHL protein. Thus, the mutant Sod1 localization in the JUNQ was confirmed a wellestablished hallmark feature of fALS, which increases the harmful effects of the mutations.
In addition to the observation of Sod1 inclusions, we observed that about 40% of dimeric Sod1 WT is located in the nucleus, while only 20% of mutant Sod1 was found in this compartment, showing a significant increase in fluorescence intensity of mutSod1 in the cytoplasm. These results indicate that the presence of Sod1 mutations decreases the ability to this protein to move into the nucleus. It has been shown in yeast cells that exposure to oxidants is sufficient to promote Sod1 nuclear localization through Dun1-Sod1 interaction and regulation of Sod1 by phosphorylation at S60, 99 [5]. In the nucleus, Sod1 affects the expression of oxidative resistance and repair genes [5] and interacts with DNA [50].
In this study, we confirmed that BiFC-tagged Sod1 mutants respond differently to mild exposure to H 2 O 2 in comparison to WT Sod1. Sod1 mutants do not properly respond against intracellular ROS production. One possible explanation is that A4V, G93A, G93C, and L38V hSod1 as aggregates at cytoplasm are unable to enter the nucleus (Fig. 1d). Another possibility is that the fALS mutations impair the ROS signaling that mediates hSod1 nuclear translocation; by remaining at cytoplasm, fALS hSod1 mutants aggregate.
Although all fALS-associated hSod1 mutants showed a similar distribution between cytosol and nucleus (80 versus 20%), leading to an impaired ability to prevent DNA damage when compared to WT Sod1, G93A hSod1 exhibited the worst performance. The levels of damage were twofold higher in G93A hSod1 expressing cells than in cells which contain only the endogenous Sod1, empty vector, meaning that G93A hSod1 is non-active and harmful with respect to DNA protection.
Our study reveals an in vivo aggregation timeline between Sod1 mutants, which were probably responsible to induce DNA damage as well as a decreased nuclear localization in comparison to the WT form. Regarding the stages of ALS disease when Sod1 aggregates are formed, some aspects have been reported. Sod1 aggregate was found to accumulate at the highest levels as symptoms appear in mouse models of fALS [45,51]. Moreover, inclusion-like structures were seen in symptomatic mice that express the ALS mutant G85R-SOD1, especially in mice motor neuron at the end-stage paralysis [52]. More recently, a timing of mutant Sod1 aggregation was reported in prion-like transmission studies with G85R-Sod1 expressing mice [53]. ALS patients harboring the A4V or G93A mutation have a short mean survival time of only about 1.5 years post diagnosis [17,54]. On the contrary, G93C mutation patients present a long survival time [17]. The ephemeral survival time of fALS patients carrying G93A or A4V mutations correlates with the fastest aggregation observed in cells in the current study. Ultimately, our findings may impact on the development of therapeutic strategies aimed at modulating the aggregation of Sod1, by providing novel and detailed information about the behavior of WT and mutant Sod1 in living cells.