This study was aimed to address two issues. The first one refers to the possibility to speed up the aggregation process thought an increase of local concentration of aS by generating covalent dimers. The second issue explored is the topology of the individual aS molecules as building blocks of the aggregated fibers. aS dimers formation has been reported in the literature as product of redox chemistry, as described as an example for Tyr125 based dimerization upon nitrative stress 
. The formation of dimers was also reported in vitro
, cellular and animal models for Y39C and Y125C aS mutants. In this case the cysteine substitutions at the reported positions in the aS sequence were proposed to increase dimers formation and accelerate protein aggregation and cellular toxicity 
probably through the formation of oligomeric species.
In our study the procedure used to produce aS dimers relies on two strategies: the first was focused on the formation of dimers with a minimal number of constraints on the fibrillation process. aS molecules were linked with different reciprocal orientations through their N-terminals (V3C), the C-terminals (adding 141G and 142C) or directly cloned in tandem. The second strategy led to the production of a fourth dimer, in which the two fibrils core regions (residues 1–104/29–140) were juxtaposed. The dimerization of aS per se is neither sufficient to trigger the conversion to beta structure of the individual aS molecules nor to trap a transient potential beta formed in one of the two molecules of the dimers. NMR analysis of the structures of all produced dimers did not show any relevant difference with aS. The only modifications in the NMR spectra could be referred to the covalent link between the two monomers. Analogous results were obtained in the CD and IR experiments. Moreover, the intrinsic propensity of unfolded aS to acquire alpha-helix conformation in the presence of membrane-mimetic systems, such as SDS, is maintained in the dimer molecules.
The kinetic properties of the aggregation process, for the different dimer types, were first analyzed in terms of fibril formation. The ThT assay, that reports the growth of fibrils, showed only for the DC an aggregation behavior characterized by the presence of a shorter lag phase in comparison to length observed for aS and all other dimers, thus suggesting that the all the latter aggregate following a qualitatively similar nucleated polymerization process. Differences emerged in the fibril elongation phase starting immediately after the lag phase, as the CC dimer presented a steeper rise in the ThT fluorescence signal, which could be interpreted as a facilitated accretion of the dimer building blocks on the fibrils as a consequence of the topology of the individual aS molecules in the fibrils. This possibility will be discussed again later within the context of the proposed structural model for the fibrillar assemblies. The DC dimer presents a more aggregation prone behavior, but the end product seems to differ from the mature fibrils formed by aS, an observation that hinders the direct comparison of the kinetic data from these species.
To explore an earlier stage in the aggregation process, we analyzed the formation of oligomers as detected by fluorescence polarization assay. Using a protocol described by Luk et al. 
, we first monitored the formation of aS oligomeric species by following the changes in the value of fluorescence polarization of reporter labelled monomeric aS (1% of the dimers). The formation of oligomeric species was not accelerated by dimerization, the CC dimers presented the same lag phase and slope of the raising part of the fluorescence polarization intensity curve of the wt aS. Furthermore, the NN dimer, that in the ThT assay showed a slow rate of increase in fluorescence in comparison to aS, showed a significant increase in the length of the lag phase of oligomers formation. Taken together, these kinetic data suggest that the constraints posed by formation of covalent aS dimers act differently at the different stages of the aggregation process.
The quantitative morphological characterization of aS fibrillar products based on TEM and AFM images is in accord with the fibril structure model first proposed by Fink and coworkers 
, in which mature aS fibrils are formed by the entwining of two sub-components called protofibrils into a left-handed-twisted bundle. The striking coincidence between the diameters of aS protofibrils and NC fibrils (see Results
section and Figure S7
) allows an interesting observation on the aggregation process of aS mature fibrils. The fact that aS protofibrils are rarely observed, and are invariably either (i) branching out from mature fibrils, or (ii) very short if observed in isolation, suggest that during the aggregation of aS, the joining of protofibrils happens concurrently with their formation (and possibly elongation), and not afterwards as proposed in the so-called “hierarchical” aggregation models. In contrast, NC fibrils, although morphologically very similar to aS protofibrils, reach often lengths of microns, suggesting that their ability to side-joining is severely hindered with respect to their aS counterparts.
The structural characterization of the fibrillation products of the dimers may provide the rationale to contextualize the results presented here in the light of the recent SSNMR studies on aS fibrils. Several laboratories focused on the definition of the aS monomer structure within fibrils 
. The converging picture, with some inconsistency on the boundaries of the individual β-strands, identifies 5 consecutive strands going from residues ~36 to ~95. In this frame Vilar et al. 
, by evaluating the distance restraints between residues extracted from the 13
C proton driven spin diffusion spectra, proposed a relative topology for the individual strands that defines a fold for the monomer in which the five strands define a plane. This model of the fold of the monomer within the fibril has been recently confirmed by Pulsed EPR on a spin labeled aS, in which the distance between the residues 41/42 and 90/91 was estimated to range from 3,9 to 5,1 nm, according to relative orientation in the strands 
. The stacking along the main axis of the fibers was deduced from the diffraction patterns described to be around 1,1 nm 
. Starting from the structure deduced from the results described above, the constraints posed by the covalent dimerization in the different aS constructs to the building up of the fibers become informative. The NN and the CC present a long linker between the core regions of the two monomers of the dimers. The average diameters of their fibrils are very similar to aS and also the distribution of the diameter values is close to that of aS. In this case the linker connects the ends on the same side of the plane defined by the individual folded monomers. A relevant feature that is lost in both NN and CC dimers in comparison to aS is the diameter periodicity, previously ascribed to protofibrils twisting, as shown in Figure S6
. In the case of the NC dimer the linking region, residues 96–140 of the first molecule and 1–36 of the following molecule, has to connect two points at the opposite side of the plane. This is not an issue in the structure of the core region, the IR is almost indistinguishable from that of aS, but the connecting loop hinders the formation of the twisted fibrils. This is documented by the shift to lower values of the heights distribution detected for the NC dimers. This observation hints on a role of either the N or the C term in the formation of the twisted structure, however, the absence of the twisted mature fibers and the substantial decrease in height observed for the C-terminal deletion mutant 
strongly support the possibility that the latter is the determining feature.
The last evidence that provides further support to the model proposed by Vilar et al. 
, comes from the DC dimer. Again the IR of the DC based fibrils does not show any difference with aS suggesting that the core region secondary structure is conserved. However, the absence of a linker region between the end of the first folded monomer and the beginning of the sequence that should form the adjacent plane compromise the possibility to form the same type of fibrils that are formed by aS. Here, the engagement of one part of the dimer in the formation of a folded monomer in the fibril core leaves the second part of the DC molecule in wrong orientation to pile up along a growth axis of the fibril, the consequence is that the second part of the DC monomer may either be free or get involved in the formation of other parallel fibrils. This view is substantiated by the imaging data, which present a non-homogeneous distribution of heights for the DC base fibrils due to the superposition of growing fibrils.
In conclusion, this study allows to propose both a tentative scheme of the tertiary structure organization of the monomers in the fibrils, in which as previously proposed 
they define a stacking of planar elements and to pinpoint a role for the non core region in the quaternary structure of the mature fibrils.