Calcium Ions Promote Superoxide Dismutase 1 (SOD1) Aggregation into Non-fibrillar Amyloid

Background: SOD1-enriched protein inclusions and Ca2+ overload are hallmarks in ALS-affected motor neurons. Ca2+ burden correlates with SOD1 aggregation in cellular models. Results: Ca2+ induces conformational changes that enhance and shift SOD1 aggregation from fibrils toward amorphous aggregates. Conclusion: SOD1 aggregation is enhanced and modulated by Ca2+. Significance: Ca2+ can behave as a pathogenic effector in the formation of ALS proteinaceous inclusions. Imbalance in metal ion homeostasis is a hallmark in neurodegenerative conditions involving protein deposition, and amyotrophic lateral sclerosis (ALS) is no exception. In particular, Ca2+ dysregulation has been shown to correlate with superoxide dismutase-1 (SOD1) aggregation in a cellular model of ALS. Here we present evidence that SOD1 aggregation is enhanced and modulated by Ca2+. We show that at physiological pH, Ca2+ induces conformational changes that increase SOD1 β-sheet content, as probed by far UV CD and attenuated total reflectance-FTIR, and enhances SOD1 hydrophobicity, as probed by ANS fluorescence emission. Moreover, dynamic light scattering analysis showed that Ca2+ boosts the onset of SOD1 aggregation. In agreement, Ca2+ decreases SOD1 critical concentration and nucleation time during aggregation kinetics, as evidenced by thioflavin T fluorescence emission. Attenuated total reflectance FTIR analysis showed that Ca2+ induced aggregates consisting preferentially of antiparallel β-sheets, thus suggesting a modulation effect on the aggregation pathway. Transmission electron microscopy and analysis with conformational anti-fibril and anti-oligomer antibodies showed that oligomers and amyloidogenic aggregates constitute the prevalent morphology of Ca2+-induced aggregates, thus indicating that Ca2+ diverts SOD1 aggregation from fibrils toward amorphous aggregates. Interestingly, the same heterogeneity of conformations is found in ALS-derived protein inclusions. We thus hypothesize that transient variations and dysregulation of cellular Ca2+ levels contribute to the formation of SOD1 aggregates in ALS patients. In this scenario, Ca2+ may be considered as a pathogenic effector in the formation of ALS proteinaceous inclusions.


Imbalance in metal ion homeostasis is a hallmark in neurodegenerative conditions involving protein deposition, and amyotrophic lateral sclerosis (ALS) is no exception. In particular, Ca 2؉ dysregulation has been shown to correlate with superoxide dismutase-1 (SOD1) aggregation in a cellular model of ALS.
Here we present evidence that SOD1 aggregation is enhanced and modulated by Ca 2؉ . We show that at physiological pH, Ca 2؉ induces conformational changes that increase SOD1 ␤-sheet content, as probed by far UV CD and attenuated total reflectance-FTIR, and enhances SOD1 hydrophobicity, as probed by ANS fluorescence emission. Moreover, dynamic light scattering analysis showed that Ca 2؉ boosts the onset of SOD1 aggregation. In agreement, Ca 2؉ decreases SOD1 critical concentration and nucleation time during aggregation kinetics, as evidenced by thioflavin T fluorescence emission. Attenuated total reflectance FTIR analysis showed that Ca 2؉ induced aggregates consisting preferentially of antiparallel ␤-sheets, thus suggesting a modulation effect on the aggregation pathway. Transmission electron microscopy and analysis with conformational anti-fibril and anti-oligomer antibodies showed that oligomers and amyloidogenic aggregates constitute the prevalent morphology of Ca 2؉ -induced aggregates, thus indicating that Ca 2؉ diverts SOD1 aggregation from fibrils toward amorphous aggregates. Interestingly, the same heterogeneity of conformations is found in ALS-derived protein inclusions. We thus hypothesize that transient variations and dysregulation of cellular Ca 2؉ levels contribute to the formation of SOD1 aggregates in ALS patients. In this scenario, Ca 2؉ may be considered as a pathogenic effector in the formation of ALS proteinaceous inclusions.
Amyotrophic lateral sclerosis (ALS) 2 is a fatal neurodegenerative disease characterized by the selective degeneration of motor neurons in the spinal cord, brainstem, and cerebral cortex (1). Most cases of ALS are sporadic with no known genetic linkage, whereas ϳ10% are associated with familial causes (fALS). Both forms share a similar neurodegeneration pattern and are clinically and pathologically indistinguishable (2), thus suggesting a common molecular mechanism. One of the common features in all ALS patients is the presence of cytoplasmic proteinaceous aggregates in motor neurons, suggested to play a determinant role in neuron toxicity, degeneration, and cell death (3)(4)(5)(6)(7). In this respect, inclusions enriched in mutant variants of superoxide dismutase-1 (SOD1) are a well established hallmark of fALS-SOD1-associated forms of the disease (7)(8)(9)(10)(11). However, sporadic ALS patients also contain wild type SOD1 in proteinaceous inclusions (12)(13)(14)(15), which share aberrant conformations of mutant SOD1 clinical variants that are known to cause aggregation (16 -18). In fact, expression of recombinant peptide fragments of wild type SOD1 in cultured cells results in insoluble fALS-SOD1-like conformers, suggesting that SOD1 is inherently prone to aggregate (19). Therefore, fALS mutations and environmental adverse conditions are probably triggers of SOD1 toxic aggregation, which seems to be an intrinsic property of this protein. Thus, wild type SOD1 is suggested to be involved in ALS pathology (20,21), inducing toxicity that degenerates motor neurons (22). SOD1 is a highly stable homodimeric protein that holds in each monomer a disulfide bridge as well as a copper/zinc binuclear site. Interestingly, SOD1 aggregates from a transgenic fALS mouse model tend to be metal-deficient and/or lack the disulfide bond (23,24) immature conformers. In agreement, the folding pathway of SOD1 clinical mutants has been suggested to favor the accumulation of metal-depleted (apo) monomeric intermediates (25), which are prone to aggregate (26 -28). In fact, post-translational modifications are major structural determinants for SOD1 stability (29,30). Each SOD1 chain folds into a ␤-barrel that is flanked by two major loops, the zinc and the electrostatic loops, which together shape the active site. In the absence of metal coordination, the ␤-barrel and dimer interface remain intact, but the associated loops become highly disordered (31)(32)(33). Further, reduction of the native disulfide bridge promotes dimer dissociation, yielding monomers with a high level of conformational flexibility (34,35). Immature SOD1 states alone are, however, insufficient to explain SOD1-linked ALS pathogenesis; not all ALS-related SOD1 variants evidence increased destabilization of the apo state (36) and reduced apo state (37) with respect to the wild type form. Thus, additional chemical and biological factors must account for the toxicity of SOD1 in ALS. One lead arises from the fact that, although ubiquitously expressed in all tissues (38,39), SOD1 only aggregates within specific motor neurons in ALS. This preferential vulnerability suggests that unique properties within the affected neurons, besides SOD1 mutations, are likely to be mandatory for the onset of SOD1 aggregation. One such characteristic within the affected motor neurons is a vulnerability to calcium overload; indeed, these cells express highly calcium-permeable AMPA receptors (40 -44) with concurrent low calcium buffering capacity due to a lack of the Ca 2ϩ -buffering proteins parvalbumin and calbindin (45,46). In agreement, calcium ions accumulate in the spinal and brain stem motor neurons of ALS patients as well as in animal and cell models of ALS-SOD1 (47)(48)(49)(50)(51)(52)(53). Moreover, studies on these models have shown that Ca 2ϩ overload promotes and correlates with SOD1 aggregation (54,55). Interestingly, oculo-motor neurons, which are not affected in ALS, have a 5-6-fold higher Ca 2ϩ buffering capacity than the spinal and brain stem ALS-vulnerable motor neurons (56). This particular observation suggests that Ca 2ϩ may be a key factor in modulating SOD1 toxicity in ALS; indeed, endogenous Ca 2ϩ levels are systematically elevated in the specific motor neurons in which SOD1 proteinaceous aggregates are found. In this study, we have thus addressed in vitro the effect of this metal ion on the aggregation mechanism of SOD1, and the results obtained suggest a link between elevated Ca 2ϩ levels and SOD1 aggregation in ALS.

MATERIALS AND METHODS
Chemicals and Sample Preparation-All reagents were of the highest grade commercially available. SOD1 was expressed in E. coli BL21(DE3) strain and grown and purified as described (57). All SOD1 experiments were performed using the demetallated form (apo-SOD1). The preparation of apo-SOD1 was obtained following published procedures (58). Metal content of apo-SOD1 was confirmed using the colorimetric reagent Zincon (59). A Chelex resin (Bio-Rad) was used to remove contaminant trace metals from all buffer solutions and to maintain apo-SOD1 in the demetallated form. Concentration of SOD1 was determined using the extinction coefficient 10,800 cm Ϫ1 M Ϫ1 at 280 nm. The broad range of concentrations used throughout biophysical exper-iments relates to the specific requirements and limitations of each of the different techniques used.
Circular Dichroism (CD)-Far UV CD analyses were performed using a Jasco J-815 spectropolarimeter equipped with a Peltier-controlled thermostated cell support. CD spectra were the average of eight scans obtained by collecting data at 0.1 nm intervals from 260 to 190 nm. The results were expressed as mean residue molar ellipticity [] with units of degrees cm 2 / dmol, as calculated from the equation, where [] obs is the ellipticity measured in millidegrees, MRW is the mean residue molecular weight, c is the protein concentration in mg/ml, and l is the optical path length of the cell in cm. Spectra were recorded with 30 M apo-SOD1 samples in 50 mM Tris, pH 7.5, that were previously incubated overnight with increasing concentrations of CaCl 2 at 37°C and 600 rpm.
ATR-FTIR-Infrared spectra were performed on a Bruker IFS 66/S spectrometer equipped with a mercury/cadmium/telluride (MCT) infrared detector and a thermostatized Harrick BioATR II cell. All measurements were obtained in an ATR cell with 150 M apo-SOD1 and 300 M SOD1 aggregates at pH 7.5 formed in the absence and presence of 300 and 600 M CaCl 2 , respectively. Each spectrum comprises the mean of 150 scans taken at a resolution of 2 cm Ϫ1 . Spectra were corrected for the buffer and water vapor. Difference absorption spectra are the average of independent subtractions between three data sets of Ca 2ϩ -incubated and control samples. The 1750 -1700 cm Ϫ1 region is shown to demonstrate the absence of major contributions from noise or water vapor artifacts, thus validating the data. Region assignments were based on typical absorption regions for specific secondary structure elements (60).
ANS Binding Assay-ANS fluorescence emission enhancement was evaluated in a BMG Fluostar Optima fluorescence plate reader using a 370-nm excitation filter and a 480-nm emission filter. Samples of 15 M apo-SOD1 were prepared as triplicates in 50 mM Tris, pH 7.5, that were previously incubated overnight with increasing concentrations of CaCl 2 at 37°C and 600rpm in black 96-well plates (Nunc, catalog no. 732-2701). The ANS fluorescence emission spectrum was recorded in a Cary Varian Eclipse instrument possessing a Peltier-thermostated cell support.
Dynamic Light Scattering (DLS)-DLS measurements were carried out in a Malvern Zetasizer Nano ZS instrument equipped with a 4-megawatt helium-neon laser (632 nm). 60 M apo-SOD1 samples in 50 mM Tris, pH 7.5, were incubated with different concentration ratios of CaCl 2 /SOD1 , filtered through a 0.45-m filter, and incubated overnight at 37°C and 600 rpm before DLS analyses. A CaCl 2 /SOD1 molar ratio of 2 was used to compare the mean light scattering intensity variation over time with control. After 24, 48, 68, and 90 h of incubation, 50-l aliquots were collected and measured in a 45-l quartz cuvette (Hellma) at 37°C. The operating procedure was set to 17 runs, each being averaged for eight measurements. The resulting data were analyzed using DTS (version 6.32) software (Malvern).
Thioflavin T Fluorescence Binding-Real-time ThT fluorescence (480 nm) was recorded using a BMG Fluostar Optima fluorescence plate reader upon excitation at 440 nm. Readings were taken every 7 min, and the well plates were subjected to 5 min of agitation at 600 rpm prior to each fluorescence measurement. Assays were performed in black 96-well plates (Nunc, catalog no. 732-2701). Samples were prepared as triplicates at a wide range of concentrations of apo-SOD1 (20 -120 M) with and without adding CaCl 2 up to a molar ratio of Ca 2ϩ / SOD1 ϭ 2 and mixed with a 2ϫ molar ratio of ThT in 50 mM Tris, pH 7.5, at 37°C. Final sample volume in each lane was 200 l, and all wells contained one Teflon bead ( 1 ⁄ 8-inch diameter). Aggregation curve analyses were performed by fitting the data to the equation, where Y corresponds to fluorescence intensity, x is the time, and X 0 corresponds to the time of half-height of fluorescence intensity (t 50 ). The lag phase is calculated by t lag ϭ X 0 Ϫ 2, and the apparent rate velocity (k app ) ϭ 1/. Kinetic parameters have been determined from six independent determinations. Dot Blotting Using Amyloid Conformational Antibodies-10 l of SOD1 aggregates obtained at the plateau phase of each aggregation kinetic curve were dotted in triplicates onto PVDF membranes and probed with a 1:500 dilution of the anti-amyloid oligomer A11 antibody (AB9234, Merck Millipore) and 1:1000 for the anti-amyloid fibril OC antibody (AB2286, Merck Millipore) according to the manufacturer's instructions. Dots were visualized using a horseradish peroxidase-conjugated IgG secondary antibody with a chemiluminescence detection system (GE Healthcare). Images were recorded and analyzed using Quantity One analysis software from Bio-Rad.
Transmission Electron Microscopy-For visualization by TEM, 5 l of 300 M apo-SOD1 aggregates formed with and without adding CaCl 2 up to a molar ratio of Ca 2ϩ /SOD1 ϭ 2 were absorbed to carbon-coated collodion film supported on 400-mesh copper grids and negatively stained with 1% uranyl acetate. The grids were exhaustively visualized with a Jeol microscope (JEM-1400), operated at 80 kV.

RESULTS
Calcium Increases Apo-SOD1 ␤-Sheet Content and Hydrophobicity-We have started by investigating if Ca 2ϩ has an effect on SOD1 native structure. This follows the evidence that protein aggregation does not necessarily require the formation of an unfolded conformer but may rather result from dynamic fluctuations around the native state, due to environmental variations, disruption of post-translation interactions, or mutations (61). In particular, we focused on effects on ␤-sheet structures and hydrophobicity changes because these properties are often altered in aggregation-prone conformers (62). The effects of increasing Ca 2ϩ concentration on SOD1 secondary structure were investigated using far UV circular dichroism (Fig. 1A). We observed that incubation of the protein with CaCl 2 at up to a molar ratio of Ca 2ϩ /SOD1 of 3 results in a roughly linear change of the signal of the negative band centered at 218 nm (Fig. 1A, inset), thus suggesting a direct correlation between the presence of Ca 2ϩ and an increased content of SOD1 ␤-sheets. In a control experiment using NaCl, no change was observed. The impact of Ca 2ϩ on SOD1 secondary structure was further analyzed by ATR-FTIR analysis. Indeed, the difference FTIR absorption spectrum in the amide I region, obtained from subtracting the Ca 2ϩ incubated from the control SOD1 sample, also reveals secondary structure changes. An increase in ␤-sheet content (positive amplitude bands in the 1620 -1630 cm Ϫ1 region) is observed alongside a decrease in randomness (negative amplitude bands in the 1642-1652 cm Ϫ1 region) (Fig. 1B). The observed variations are thus suggestive of conformational rearrangements taking place in the presence of Ca 2ϩ .
To verify if Ca 2ϩ also promotes alterations on SOD1 hydrophobicity, we have used ANS, a fluorophore that interacts with exposed hydrophobic patches within ␤-sheeted structures (63). ANS is scarcely fluorescent when free in an aqueous neutral solution, emitting at 520 nm (Fig. 1C, trace a). However, a blue shift to around 480 nm as well as an increase in intensity of fluorescence emission are observed upon ANS binding to hydrophobic surfaces in proteins (64). In agreement with previous reports, apo-SOD1 binds ANS (Fig. 1C, trace b) (65); however, the presence of Ca 2ϩ results in a ϳ15% increase in ANS fluorescence emission (Fig. 1C, trace c). The intensity of ANS fluorescence emission was found to increase up to an excess molar ratio of Ca 2ϩ /SOD1 of 2. For higher molar ratios of Ca 2ϩ , the ANS emission tends to decrease, suggesting that fewer hydrophobic moieties are accessible (not shown). This could be an indication that intermolecular hydrophobic interactions are being formed at higher Ca 2ϩ concentrations, thus decreasing the exposed hydrophobic regions available for ANS binding. This would be compatible with an oligomerization process through structural reorganization involving hydrophobic burial (66). Overall, the results indicate that in the presence of Ca 2ϩ , SOD1 adopts distinct conformational states that are ␤-sheet-enriched.
Calcium Promotes Oligomerization and Conformational Heterogeneity of Apo-SOD1-In order to investigate a possible link between the ANS results and the formation of intermolecular interactions at increasing Ca 2ϩ concentrations, we have used DLS, a very sensitive technique for assessing changes in the oligomerization state of protein solutions (67). After 24 h of incubation at 37°C and constant agitation, a monomodal size distribution profile by volume with a mean peak averaging at a hydrodynamic diameter of ϳ5.4 Ϯ 0.4 nm was observed for all samples, irrespective of the presence of Ca 2ϩ up to a molar ratio of 4. However, the absence of a multimodal distribution by volume in DLS does not necessarily imply the absence of aggregates if the population of formed oligomeric species is very low relative to that of dimeric SOD1 (68,69). We have thus probed for oligomeric species by comparing the mean hydrodynamic diameter (Z-average size), a parameter extremely sensitive to the presence of minor populations of larger oligomers, as well as the mean count rate parameter for evaluation of the scattering intensity. From this combined analysis, we determined an increase in the mean hydrodynamic diameter for the Ca 2ϩincubated SOD1 samples, from 6.5 Ϯ 0.3 nm for the control to 7.9 Ϯ 0.9, 10.1 Ϯ 1.5, and 10.1 Ϯ 2.5 nm for samples at Ca 2ϩ / SOD1 molar ratios of 1, 2, and 4, respectively. An increase in mean light scattering intensity was also observed upon increasing concentrations of added Ca 2ϩ , from 356 Ϯ 18 nm for the control, to 377 Ϯ 10, 439 Ϯ 26, and 543 Ϯ 31 nm for samples at Ca 2ϩ /SOD1 ratios of 1, 2, and 4, respectively. The results suggest a correlation between increasing Ca 2ϩ /SOD1 ratios and enhanced oligomerization (Fig. 2), in agreement with Ca 2ϩ inducing aggregation and conformational heterogeneity on SOD1. The used Ca 2ϩ concentrations (60 -240 M) bear some resemblance to physiological ones because intracellular micromolar transients of free Ca 2ϩ are known to occur in vivo upon neuron stimulation (70 -72).
Calcium Enhances Apo-SOD1 Aggregation-In order to further characterize the effect of Ca 2ϩ on the aggregation propensity of SOD1, we have analyzed the variation of the mean light scattering intensity over time (Fig. 3A). We determined that the presence of Ca 2ϩ promotes an earlier formation of an aggregation nucleus (t Ͻ 48 h), whereas the onset of aggregation in the control (without Ca 2ϩ ) occurs substantially later (t Ͼ 70 h). This suggests that the conformational changes induced by the presence of Ca 2ϩ give rise to self-association-prone conformers that promote SOD1 oligomerization under physiological pH. The effect of Ca 2ϩ on SOD1 aggregation kinetics was further investigated monitoring binding of thioflavin T, a fluorescent probe that recognizes diverse amyloid-like fibrils and amorphous aggregates (73)(74)(75)(76) as well as oligomers and protofibers (77)(78)(79). In both cases a sigmoidal-type transition was observed, suggesting a nucleation-dependent aggregation mechanism (Fig. 3B). However, Ca 2ϩ was shown to impact the aggregation profile of SOD1, namely the lag time and fibrillation rate ( Table 1). This was evidenced by a decrease in the time required for the formation of the aggregation nucleus (which was decreased by ϳ26 h in the presence of Ca 2ϩ ) and by a decrease in the apparent elongation rate constants.   Table 1 for kinetic parameters and "Materials and Methods" for further details. Error bars, S.D.

TABLE 1 Effect of Ca 2؉ on SOD1 aggregation kinetics under physiological pH
Apo-SOD1 aggregation (80 M) was monitored by ThT fluorescence emission in 50 mM Tris, pH 7.5, with or without Ca 2ϩ under 600 rpm agitation with a Teflon bead at 37°C. (n ϭ 6).

Kinetic parameters
In order to obtain insight into SOD1 oligomers formed at the onset of the exponential phase (t ϭ 48 h), we have compared the particle size distributions (Fig. 4A). In the control without calcium, although poorly populated, it is already possible to observe a broad distribution of oligomers (Fig. 4A, bottom). However, Ca 2ϩ populates a pool of larger oligomers (Fig. 4A, top). Indeed, Ca 2ϩ promotes the formation of an increased fraction of larger aggregates (Ͼ500 nm), whereas in the control, smaller aggregates (Ͻ500 nm) predominate (Fig. 4B).
The effect of Ca 2ϩ on SOD1 aggregation was also investigated under reducing conditions, a well established factor known to hasten SOD1 aggregation (28,80) (Fig. 5). A similar effect is observed with respect to non-reducing conditions; further, with a reductant, Ca 2ϩ also lowers SOD1 critical concentration for aggregation (Table 2). In summary, the presence of Ca 2ϩ potentiates SOD1 aggregation, and this effect is observed irrespective of the redox status of the intramolecular disulfide bond (Cys 57 -Cys 146 ).
Calcium Influences the Secondary Structure of SOD1 Aggregates-We have also examined the difference ATR-FTIR spectra to probe for conformational changes on SOD1 aggregates formed in the presence of Ca 2ϩ (Fig. 6). Ca 2ϩ aggregates display prominent ␤-sheet components at 1630 and 1695 cm Ϫ1 , likely to be associated with intermolecular antiparallel arrangements of the ␤-strands, because antiparallel ␤-sheets exhibit a strong band near ϳ1630 cm Ϫ1 and a weaker band near ϳ1990 cm Ϫ1 (81)(82)(83). On the other hand, in the control, a main ␤-sheet contribution at 1640 cm Ϫ1 is observed that can be attributed either to parallel ␤-sheets or to twisted antiparallel ␤-sheet structures, because both types of structure have absorption frequencies shifted to higher wave numbers than the corresponding antiparallel ␤-sheets and often exhibit a reduced/ negligible high wave number side band (84). Moreover, Ca 2ϩ aggregates have a high content of intermolecular ␤-sheets (band at 1612 cm Ϫ1 ) (85) as well as ␤-turns (bands at 1678 and 1664 cm Ϫ1 ); on the other hand, the control denotes a higher content of random structures, as evidenced by the 1655 cm Ϫ1 negative band.
Calcium Influences the Morphology of SOD1 Aggregate-Having established that Ca 2ϩ modulates the secondary structure of SOD1 aggregates, we moved to investigate their morphological differences. For this purpose, we have used conformational antibodies in combination with TEM. The conformational antibodies A11 and OC recognize generic epitopes of soluble oligomers and fibrils, within amyloid proteins, respectively (86,87). Dot blot analysis of SOD1 aggregates obtained after the plateau phase of the aggregation kinetics was reached (t ϳ200 h) showed that these are immunoreactive toward both antibodies, indicating that SOD1 aggregates comprise amyloid epitopes (Fig. 7). Interestingly, incubation with Ca 2ϩ results in different binding patterns; the presence of Ca 2ϩ increases the reactivity toward oligomers (A11) while simultaneously decreasing proportionally the immune response toward fibrils (OC). This result implies that Ca 2ϩ decreases the extent of fibril formation, suggesting that the aggregation pathway is diverted toward amorphous aggregates.
The morphology of the aggregates was further investigated using TEM (Fig. 8). In the absence of Ca 2ϩ , SOD1 forms amy-   Table 2 for kinetic parameters and "Materials and Methods" for further details.
loid-like fibrils with characteristic ultrastructural properties with an approximate diameter of 10 nm and several m in length. In the presence of Ca 2ϩ , fibrils still formed, but they were shorter and less abundant (Fig. 8, bottom panels). Further-more, other species were present, namely amorphous aggregates (Fig. 8, bottom right panel, arrow) and oligomers of different sizes (Fig. 8, bottom panels, arrowheads), thus forming a more heterogeneous sample. These results fully agree with those obtained using conformational antibodies. TEM observations suggest that Ca 2ϩ interfered with SOD1 fibrillogenesis but did not avoid the generation of intermediate amyloidogenic species, such as oligomers and fibrillar aggregates. Thus, the preferential non-fibrillar aggregates and oligomers that are formed by the presence of Ca 2ϩ retain the ability to bind ThT and accounted for most of the intensity observed in this assay (Fig. 1C). Interestingly, electron microscopy of Lewy body-like SOD1-enriched inclusions from the spinal cord of an ALS mouse model revealed that its core consisted of a heterogeneous mass of tangled filamentous material covered with small granules (14,88,89). This suggests that SOD1 inclusions in vivo are composed of protein aggregates with heterogeneous mor- phologies, like the ones here found to be formed in vitro in the presence of Ca 2ϩ .

DISCUSSION
Ca 2ϩ is increased in ALS and is particularly abundant in the specific motor neurons affected in this disease, where SOD1enriched proteinaceous inclusions are found. As a ligand, Ca 2ϩ is particularly versatile because it binds sites with irregular geometry, thus facilitating its nonspecific association to proteins (90,91). In agreement, a bound Ca 2ϩ was described in a crystal structure of an SOD1 clinical mutant (92) although sequence analysis and bioinformatics failed to identify any canonical Ca 2ϩ binding motif within SOD1. Here we have assayed the effect of Ca 2ϩ on the SOD1 aggregation mechanism. Overall, the results obtained show that under physiological pH, Ca 2ϩ induces conformational changes on the SOD1 fold that increase its propensity to aggregate. Specifically, Ca 2ϩ increases SOD1 ␤-sheet content and hydrophobicity, which are features likely to result in favorable intramolecular interactions leading to the formation of aggregates (93,94). Indeed, the presence of Ca 2ϩ was shown to promote SOD1 aggregation, both under oxidizing and reducing conditions, diverting the aggregation pathway toward non-fibrillar aggregates and oligomers instead of fibrils. Recent literature suggests that amorphous rather than amyloid-type aggregates are found in SOD1-ALS patient tissue (95), and the reported effects of Ca 2ϩ would go along these lines, but this is still a matter of debate in the field (96,97). This modulation effect by a metal ion has also been demonstrated for other proteins involved in neurodegenerative disorders, highlighting the influence of the chemical neuronal environment with respect to metal ion homeostasis in pathologic aggregation processes (98). Interestingly, Ca 2ϩ can also induce ␣-synuclein aggregation, as shown by cell culture and in vitro studies (99 -101). This evidence is particularly relevant because pathologic ␣-synuclein accumulation in Parkinson disease selectively affects dopaminergic neurons that are particularly vulnerable to Ca 2ϩ dyshomeostasis, correlating calcium overload with the onset of the disease (102)(103)(104)(105). Also, in the case of SOD1, the observed effect of Ca 2ϩ can have a particularly meaningful physiological relevance because SOD1 only undergoes aggregation within specific neuron cells of ALS patients (38,39) that simultaneously show Ca 2ϩ overload (47)(48)(49)(50)(51)(52)(53). Moreover, it was recently reported that calcium overload in the cytoplasm of cultured motor neurons expressing an ALSassociated mutant promotes SOD1 aggregation into inclusions. Importantly, Ca 2ϩ increase is suggested not to be a consequence of SOD1 aggregation because the rise in Ca 2ϩ levels in these cells occurs prior to the formation of SOD1 inclusions (54,55). In any case, SOD1 aggregates promote a further increase in the levels of cytosolic Ca 2ϩ (55). In agreement, increasing the Ca 2ϩ load exacerbates the insolubility and toxicity of a SOD1 clinical variant (106 -109). On the other hand, an opposite effect is observed when Ca 2ϩ buffering in cellular and animal models of fALS is increased, which resulted in a protective effect toward aggregation (110 -112). Our study provides in vitro evidence supporting the hypothesis that Ca 2ϩ overload in ALS can actually play a role in SOD1 deposition. Nevertheless, there must be other factors involved in SOD1 aggregation because otherwise SOD1 would probably also aggregate on dopaminergic neurons, which are neurons that feature high levels of labile cytosolic calcium.