Figure 1. The number of α-Syn monomers/oligomer at different times during formation. Most oligomers have less than 12 monomers in studies by Cremades et al. [2] (copied with permission). We added dashed lines to indicate 6, 12, 18, and 30 monomers: the number of monomers in some of our models.
Even the secondary structures of the assemblies alter with time. Recently Zhou and Kurouski [3] used atomic force microscopy-infrared spectroscopy to analyze the structure of α-Syn oligomers at different stages of aggregation. The nanometer resolution of this approach allows different aggregate species to be analyzed simultaneously. They observed high heterogeneity. Initially most assemblies were dominated by random coil and α-helical structures, but some contained β-sheets (mostly antiparallel). With time the amount of random coil/α-helical and antiparallel β-structure decreased while that of parallel β-structures increased. By day 7 parallel β-fibrils dominated, but some spherical assemblies with significant antiparallel β-content persisted. Lashuel et al. [4] obtained electron microscopy (EM) images of α-Syn annular protofibrils that assemble to form tubular protofibrils that then morph into fibrils. Chen et al. [5] developed cryoEM images of toxic cylindrical oligomers composed of ~ 18 and 29 α-Syn monomers. Fibrils are the only assemblies for which atomic-scale structures have been determined, and then only for portions of α-Syn [6-11]. They all include at least some segments from residues 15 – 97 that form in-register parallel cross-beta sheets. However, the exact sequential locations of the β-strands, overall folding patterns of the monomers, and interactions between fibrils vary among these structures. Also, these fibrils may not be the same as those occurring in the human brain [12]. None appear consistent with previously reported EM images of ribbons [13,14].
But functional roles of synucleins and their pathologies may involve interactions with lipids and/or fatty acids absent in all thein-vitro studies mentioned above (reviewed in Mori et al. [15]). The first 95 residues of α-Syn form two antiparallel helices on the surface of negatively charged membranes [16]. These interactions are especially prevalent in presynaptic vesicles where α-Syn apparently interacts with other proteins. However, Synucleins also occur elsewhere and aggregate into larger assemblies. Lewy bodies commonly found in PD patients contain large quantities of synucleins and amyloid beta amyloids but are also chocked full of lipids, fatty acids and organelles. Meade et al. [1] observed a series of helical α-Syn polymorphs that occur in the presence of lipid vesicles. Giorgia De Franceschi et al. [17] observed similar fibrils when the ratio of docosahexaenoic acid to α-Syn is relatively low. These lipid-α-Syn fibrils have much larger diameters than those formed in the absence of such amphiphiles. The α-Syn conformation is dependent upon the ratio of lipid to α-Syn; when the ratio is high α-helical oligomers form on the surface of lipid droplets [17]. In contrast, van Diggelen et al. [18] identified two α-Syn/fatty acid oligomer assemblies that were both rich in antiparallel β-structure but had differing epitopes. Eichmann et al. [19] have published EM images of high-density lipoproteins formed by all three families of synucleins in the presence of sphingomyelin. Also, numerous studies have found that α-Syn permeabilizes membranes and forms channels in both artificial membranes [20-23], and neural cells [24-26].
Several familial mutations that contribute to PD have been identified, most of which are sequentially near each other. However, it is unclear how or whether these alter oligomer conformations or the binding of this segment into a cleft within the Cyclophilin A protein [27].
Findings that numerous α-Syn assemblies typically coexist in equilibrium, alter slowly with time, have only partially ordered and/or fractionally occupied secondary structures, and are affected by cofactors, complicates experimental determination of their structures and what makes some toxic. Also, computational modeling that relies on molecular dynamic simulations of randomized assemblies is problematic because oligomeric and fibril development often occurs over days or weeks and involves many monomers whereas most molecular dynamic simulations last less than a millisecond for a relatively small number of monomers. Furthermore, assembly conformations may depend upon factors not included in the simulations.
We utilize an alternative approach to modeling oligomers and transmembrane channels that concentrates on β-barrels, most of which are concentric. Previously we published concentric β-barrel models of amyloid-β42 (Aβ42) hexamers, dodecamers, annular protofibrils, and transmembrane channels [28-30]. Several aspects of our models including six-stranded antiparallel β-barrels and concentric β-barrels were unprecedented. Since then, however, determination of the structures of two channel-forming toxins, lysenin [31] and aerolysin [32], have demonstrated that assemblies composed of multiple identical subunits related by radial symmetry and that contain concentric β-barrels occur naturally. Also, Laganowsky et al. [33] found that a segment from an amyloid-forming protein, alphaB crystalline, can adopt a radially symmetric six-stranded antiparallel β-barrel hexamer structure (which they call cylindrin) similar to the core β-barrel of our Aβ42 hexamer model, Do et al. [34] found that the sequences of portions of Aβ can also form this motif, and Serra-Batiste et al. [35] found that Aβ42 channels have β-barrel structures with only two distinct monomer conformations.
Stimulated by these and other supportive experimental findings, we have renewed our modeling efforts. Although some details of our current generation of models have been altered and are still evolving, the basic concept of assemblies composed of concentric β-barrels [30] has been retained. Here we extend those concepts and techniques to develop models of a variety of non-fibril Synuclein assemblies. Given the difficulties described above, attempts to model these assemblies in the absence of high-resolution structural data may seem ludicrous. A saving grace is that numerous lower resolution microscopy studies have revealed circular and cylindrical assemblies composed primarily of β secondary structure. The simplest way to model circular and cylindirical β-structures is as β-barrels, which are highly constrained if all subunits are identical and they are symmetric and concentric.