We have recently reported that apo WT SOD1 gives rise to soluble oligomers under aerobic conditions when the protein is kept at 37°C and at a concentration and pH close to physiological, i.e., 100 µM and pH 7. The resulting soluble oligomers are formed by intermolecular disulfide covalent bonds and by non-covalent interactions between beta strands, forming amyloid-like structures capable of binding ThT 
The SOD1 mutants (), which are reported to be linked to fALS disease, were selected following the criteria indicated in . We selected mutants spread over the entire protein: some of the mutations are located at the subunit-subunit interface, some add a positively charged residue, others substitute an hydrophobic residue with an hydrophilic one, or some produce a simple side chain size variation. Other mutations are on residues either located on secondary structural elements, namely β strands, or just outside them, with change to residues favoring a helical conformation. We also selected a mutation, on a residue in a loop, which introduces a negative charge in the hydrophobic core of the protein. Finally, we introduced two mutations not reported to be related to fALS. Each of these mutant proteins was analyzed with respect to its behavior toward oligomerization, and correlations were sought between the mutant behavior, the nature of the mutation, and its location on the sequence. We purposely introduced the mutations on the “real” WT SOD1 and not on the thermostable form where the two free cysteines (6 and 111) are changed to alanine and serine, respectively (so-called AS SOD1) 
. This is an essential condition as we have shown that the presence of these two cysteines is the key for SOD1 oligomerization 
Mutants studied in this work and their criteria of selection
Each of the mutants, in the zinc-reconstituted as well as in the apo form, retained the dimeric quaternary structure, as assessed by gel filtration chromatography, which also demonstrated the absence of any significant amount of high molecular weight species. Circular dichroism (CD) spectra on both metallated and apo proteins indicated that the secondary structure present in WT SOD1 was fully conserved in all of the mutants. It has been shown previously that reduction of the intramolecular disulfide bond of apo WT SOD1 causes complete monomerization 
. We therefore inferred, from the dimeric state of the apo form of all the mutant proteins, that the intrasubunit disulfide bond was intact. For some of the mutants (T54R, V97M and I113T), the folded state of the proteins and the intact disulfide bond were also experimentally confirmed from their 1
N HSQC NMR spectra since their cysteine residues 57 and 146 have shifts very close to those observed for oxidized WT SOD1, but far from those of reduced WT.
Optical and fluorescence spectroscopies, the latter with the use of the ThT, coupled with gel chromatography, showed that the zinc-bound proteins did not give rise to any oligomeric species when they are incubated at pH 7, 37°C, 100 µM concentration, for periods of time longer than a month. The absence of formation of any large molecular weight species was confirmed by gel filtration chromatography. Consistently, turbidity at 400 nm showed no insoluble precipitate. These data indicate that, similar to WT SOD1 
, the zinc-bound forms of any SOD1 mutant are stable even under prolonged incubation at 37°C.
By contrast, the behavior of the metal-free (apo) form was dramatically different. Upon incubation at 37°C in the air, a progressive increase in ThT-binding fluorescence was observed for the apo form of all the mutants. shows the ThT-binding behavior of the eleven SOD1 mutants, together with that of apo WT SOD1, over a period of more than one year. The temperature dependence of this process was tested for incubations ranging from 15 to 40°C. While at 15°C the oligomerization process is much slower starting after 5–6 days, in the range 36–40°C a difference of one degree Celsius almost doubles the detected rates (data not shown). When a reducing agent such as DTT was added to the solutions of the mutants (tested for T54R, V87M, D90A, G93A, V97M, I113T and L144F), the oligomeric species were destroyed, leading to monomeric species, thus showing that the oligomerization occurs through disulphide bonds. The soluble oligomers, which appear to have a similar amyloid-like structure, as judged by their ThT-binding behavior, are stabilized by H-bond interactions between beta strands of SOD1 subunits. To test further for the existence of these non-covalent interactions, GdnHCl, a cauthropic agent that breaks hydrogen bonds, was added to the oligomers. For each of the mutants tested (T54R, V87M, D90A, I113F, I113T and L144F), the ThT-binding fluorescence was quenched in few minutes, whereas gel filtration of the resulting solutions showed that high-molecular-weight species remained present. While the loss of ThT-binding ability is due to the disruption of the amyloid-like structure of the oligomeric assemblies, the persistence of the oligomeric state is due to the covalent disulfide bonds between the free cysteines of the monomeric subunits.
Formation of ThT-binding structures when apo SOD1 mutants and WT are incubated at 37°C.
The rates of oligomerization and consequently of fluorescence increase, is found to depend on the nature of the mutation, being for some mutants strikingly different from that of apoWT SOD1. In particular, three mutants (G93A, V97M and I113T) showed a very fast initial rate of oligomerization, more than twice that of WT SOD1. I35T, a mutant not currently linked to fALS, showed a significantly slower rate of oligomer formation than these three, but the rate was still higher relative to the WT SOD1 protein. The other mutants showed rates of oligomer formation very similar to apo WT SOD1 or, in one case, i.e., T54R SOD1, slightly lower than it. Thus the rates of aggregation for the fALS mutant proteins studied can be divided in two groups, some with rates of oligomerization very similar to that of apo WT SOD1 and three others much faster, with one, non-ALS mutant SOD1 oligomerizing at an intermediate rate. It is important to stress that in no case did a human SOD1 apoprotein fail to form soluble, high molecular weight oligomers. Repeated trials of each of the mutants also established that the kinetics of oligomer formation were highly reproducible as monitored by increases in ThT fluorescence. The data for T54R, D90A, I113T SOD1, and WT SOD1 apoproteins are shown in figure S1
It is not only the location of the mutation, but also the nature of the amino acid substitution that determines the oligomerization rate. For example, mutation of Ile113 induces a very fast rate of oligomerization, much faster than WT, when is substituted with Thr, but close to that of WT when Ile is replaced by Phe.
Similar oligomerization behavior is also observed for the two mutants, up to now not reported to be involved in ALS. The I35T mutant, located in β3 strand, showed a fast increase in ThT fluorescence (), and therefore in the rate of oligomerization, as also evidenced by the gel chromatographic analysis (data not shown), while the other mutant here investigated, L67V, ologomerizes with a slower rate, similarly to WT SOD1. The oligomerization of mutants I35T and L67V, in their apo forms, supports our mechanism suggesting that the process would take place for any mutation only when the protein is in the apo form, but not in the zinc-bound form, as we have extensively verified.
In any case, it is important to note that, despite the different aggregation rates, the fluorescence limits at very long times (about 1 year) are similar, indicating the formation of very large molecular weight oligomers for WT and all of the mutant SOD1 proteins studied ( and figure S1
The pattern of oligomerization, detected through fluorescence, was paralleled by gel filtration data. The data for the two mutants with the two extreme oligomerization rates, as observed in the fluorescence experiments, (I113T and T54R) are shown in and . After about 100 hours, while I113T was mostly in high molecular weight states, even if not yet as the final, highest molecular weight ones, mutant T54R was still essentially all in the dimeric state. Gel filtration data also indicated that the final status (after about one year) contains a distribution of high MW species, up to the column cut off (7×106
Da). The presence of intermediate MW species is also observed during the long period of the aggregation process; they eventually evolve towards the final very high MW species. Multi-angle light scattering analyses of the samples along the oligomerization process also showed that the average molecular weight was increasing as a function of time (, Table S1
), with a decrease of the fraction of the dimeric species and the increase of that of the oligomers. The increase in ThT-fluorescence and that in molecular weight from light scattering go in parallel, with a linear correlation between the fraction of aggregated specie (non-dimer) and the ThT-binding fluorescence at each period of incubation (). This correlation indicates that similar increase in fluorescence corresponds to similar increase in oligomeric species. Multi-angle light scattering data of samples after long periods of incubation, when they reach a steady state condition, provide very similar average molecular weights for all the mutants (of the order of 106
Da), indicating that the oligomers eventually have essentially the same size, independently of the rate of aggregation.
Formation of oligomeric structures when apo I113T SOD1 mutant is incubated at 37°C.
Formation of oligomeric structures when apo T54R SOD1 mutant is incubated at 37°C.
Correlation between percentage of aggregated species and ThT-binding fluorescence.