α-Syn transmembrane channels
Lysenin and Aerolysin are the only experimentally-determined protein structures that have concentric β-barrels [31,32]. Like Lysenin and Aerolysin, α-Syn can exist in a soluble form and interact with membranes to form channels. Atomic force microscopy images of α-Syn channel assemblies [21] are remarkably similar in size and shape to those obtained for Lysenin [66] and Aerolysin channels [67] (Figs. 20 & 23).
NAC β-hairpins comprise a transmembrane pore and the Nt and Ct domains comprise the soluble domains in our models. However, there are some ambiguities regarding the structures of these domains. The Nt domains could form amphipathic α-helices on the membrane surface or form Type 1P concentric β-barrels as illustrated in Fig. 20. We favor the latter model because it is more like the structures of Lysenin and Aerolysin and appears more consistent with the atomic force microscopy images of α-Syn channels. However, we do not exclude the possibility that the Nt domain forms amphipathic α-helices, possibly as the channels are forming, and that the Ct domain is disordered.
Fig. 20 shows two alternative models for the signature sequence end of Sy5 and the NAC β-hairpin. Both models of the pore resemble the transmembrane region of Lysenin and Aerolysin channels. Although the transmembrane pore-forming regions of Lysenin and Aerolysin are less hydrophobic than the NAC domain of α-Syn and these proteins probably are not strictly homologous, their sequences have some similarities as indicated by the following structural alignment; residues that are identical in two or three sequences are bold, residues with black backgrounds are exposed to lipids in the structures and model, and the underlined regions are the loops connecting the two pore-forming β strands.
Lysenin poreT RT VTA THSIGSTISTGDAFEIGS VE VS YS HSHEES QVS M
α-Syn NACTN VGGA VVTGVT AVAQK——T VE GAGSIAA ATGFV KK
Aerolysin poreTNT YG LSEKVTT KNKFK-WPLVGETE LS IEIAA NQS WAS Q
In the first model, the hydrophilic Q79 and K80 side-chains of the putative β-turn within Sy7 would extend into the cytoplasm were the positively charged K80 would likely interact with negative charged lipid head groups. The only other charged residue in this region, E83, extends into the water-filled pore in our model and may increase cation permeation through the pore. In the second model the signature region of Sy5 forms an amphipathic α-helix on the membrane surface and the first putative β-strand (Sy6-Sy7a) is shifted upwardly by two positions. K80 can interact with E83 of the adjacent subunit as well as with lipid head groups. Hydroxyl groups of T72 and S87 extend into the pore in both models. The only polar side-chain groups that would be exposed to lipid alkyl chains are T75 and T92; fewer polar side-chain groups than are exposed to lipids for Lysenin or Aerolysin channels. Although the 11-residue-long β-hairpin of Sy7 is absent in β-Syn, the basic structure could still form with the β-turn occurring at the GVGN residues linking Sy6 to Sy8. This truncation of the putative transmembrane region and removal of positively charged residue at the β-turn may prevent initial interaction with the membrane and/or formation of a channel. If so, this may help explain why β-Syn is not toxic.