Site-Specific Perturbations of Alpha-Synuclein Fibril Structure by the Parkinson's Disease Associated Mutations A53T and E46K

Parkinson's disease (PD) is pathologically characterized by the presence of Lewy bodies (LBs) in dopaminergic neurons of the substantia nigra. These intracellular inclusions are largely composed of misfolded α-synuclein (AS), a neuronal protein that is abundant in the vertebrate brain. Point mutations in AS are associated with rare, early-onset forms of PD, although aggregation of the wild-type (WT) protein is observed in the more common sporadic forms of the disease. Here, we employed multidimensional solid-state NMR experiments to assess A53T and E46K mutant fibrils, in comparison to our recent description of WT AS fibrils. We made de novo chemical shift assignments for the mutants, and used these chemical shifts to empirically determine secondary structures. We observe significant perturbations in secondary structure throughout the fibril core for the E46K fibril, while the A53T fibril exhibits more localized perturbations near the mutation site. Overall, these results demonstrate that the secondary structure of A53T has some small differences from the WT and the secondary structure of E46K has significant differences, which may alter the overall structural arrangement of the fibrils.


Introduction
Aggregation of the neuronal protein alpha-synuclein (AS) is strongly implicated in both sporadic and familial forms of Parkinson's disease (PD), and with a variety of other neurodegenerative diseases collectively termed synucleinopathies. AS is the main protein component of Lewy bodies, intraneuronal cytoplasmic inclusions that are the defining neuropathological hallmark of PD [1]. Missense mutations in AS, including A53T [2], A30P [3], and E46K [4], are associated with early-onset disease, suggesting that the mutant proteins differ either in their aggregation propensity or in the stability or toxicity of the resultant aggregated species.
Like wild-type (WT) AS, the mutant proteins A30P, E46K and A53T are largely unstructured in solution [5,6,7]. However, the mutant proteins differ in their fibrillation kinetics in vitro; A53T and E46K fibrillize more rapidly than WT [7,8], while A30P fibrillizes more slowly, preferentially adopting a protofibrillar intermediate state [9]. To better resolve the potential similarities and differences among fibrils comprised of mutant and WT AS, we have employed magic angle spinning (MAS) solid-state nuclear magnetic resonance (SSNMR) spectroscopy. This technique is uniquely suited to structural studies of amyloid proteins, such as fibrillar AS [10,11,12,13], which are insoluble and do not form Xray diffraction-quality crystals. Several SSNMR-derived structures or structural models of amyloid fibrils now have been deposited in the protein data bank (PDB) [14,15,16,17,18]. Recently, we used MAS SSNMR to demonstrate that the A30P mutation does not affect the secondary structure of AS fibrils when compared sitespecifically with WT AS fibrils [19].
The current study focuses on the A53T and E46K mutant fibrils. While previous reports indicated significant structural differences between WT and A53T fibrils [20], those studies were complicated by batch-to-batch variability in fibril polymorphs for the WT protein [10]. We have established carefully controlled sample preparation conditions that allow reproducible measurements of AS chemical shifts [21] and have subsequently made extensive assignments of the fibril core of WT protein [12]. The same protocol for fibril preparation that we used for WT protein is also applicable to the mutant proteins [19], allowing a more direct and complete comparison of WT and mutant fibrils. Here we show that, in contrast to a prior report [10], A53T fibrils have a highly similar secondary structure to WT fibrils, with small, localized perturbations near the mutation site, as well as for some residues numbered in the range 60-70 and 80-90. E46K fibrils present larger perturbations in secondary structure, which we attribute to the large electrostatic alteration at position 46.

Protein sample preparation
Uniformly (U)-15 N, 13 C labeled WT, A53T and E46K AS were prepared following previously published protocols [21]. In summary, recombinant protein expression was performed in E. coli BL21(DE3) using minimal media supplemented with BioExpress (Cambridge Isotopes). Purification was performed by chemical lysis and two chromatographic steps (hydrophobic interaction and gel exclusion), resulting in ,60 mg of pure protein from each L of culture. Sample purity was confirmed by SDS-PAGE gel electrophoresis and mass spectrometry.

Electron microscopy
A53T, E46K and WT AS fibril samples were treated with Karnovsky's fixative. After negatively staining with 2% ammonium molybdate (w/v), samples were applied on formvar carbon coated grids (300 mesh). Samples were viewed with a Hitachi H600 Transmission Electron Microscope operating at 75 kV.

a-Synuclein fibril preparation for solid-state NMR experiments
After 3 weeks of fibrillation, samples were ultracentrifuged for 1 hr at 100,000 g. The resultant pellets were washed with distilled water and ultracentrifuged again for 1 hr at 100,000 g. The supernatant was removed and the fibril pellets were dried under a stream of N 2 (g) until the final mass was unchanged. The dry powders were packed into 3.2 mm (thin or standard wall) NMR rotors (Varian, Inc., Palo Alto and Walnut Creek, CA and Fort Collins, CO; now part of Agilent Technologies, Santa Clara, CA and Loveland, CO), rehydrated to 36% water by mass and kept hydrated by Kel-F and rubber spacers, as previously described [12].  15 N-13 C-13 C and 13 C-15 N-13 C correlation experiments, band-selective SPECIFIC cross-polarization (CP) [25] was employed for heteronuclear polarization transfer between 15 N and 13 C and DARR [26] mixing for 13 C homonuclear polarization transfer. All E46K experiments were acquired under 13.3 kHz MAS and at a variable temperature (VT) of 10uC. All A53T experiments were acquired under 12.5 kHz MAS and at a VT temperature of 0uC. 1D 13 C and 2D 13 C-13 C were also collected at a range of temperatures from 210uC to +20uC to confirm that the spectra did not depend to a significant extent on the exact sample temperature. Chemical shifts were externally referenced using adamantane, with the downfield 13 C peak at 40.48 ppm [27].

Solid-state NMR data collection and analysis
NMR datasets were processed in NMRPipe [28] by applying back linear prediction to the direct dimension, and in all dimensions Lorentzian-to-Gaussian apodization was applied prior to zero filling and Fourier transformation. Peak picking, assign- ments and peak heights were obtained with NMRViewJ software [29] using Gaussian peak integration methods. Chemical shifts were deposited in the Biological Magnetic Resonance Bank (BMRB codes #18208 and #18207 for E46K and A53T, respectively).

Sample characterization
Monomers of each AS sample were incubated at 37uC and 200 rpm shaking, and fibrillation kinetics were measured by ThT  Figure 1A), consistent with previous reports [7,8,9,30,31,32,33]. Previous studies indicate possible differences in morphology between the mutants and WT based on FE and AFM investigations [13,34], while similar morphologies were observed by electron microscopy [32,35] (Figure 1B). Under these conditions, the rehydrated SSNMR samples of both mutants yielded high resolution 13 C-13 C 2D spectra ( Figure 1C,D), with narrow linewidths (,0.2 ppm) indicating a high degree of homogeneity throughout the fibrils, as previously observed for the WT [12]. The spectra were therefore determined to have more than sufficient sensitivity and resolution to perform chemical shift assignments in combination with 3D experiments.

Chemical shift comparison of the A53T and E46K mutants versus wild-type
Chemical shifts were assigned using the standard complement of multidimensional SSNMR experiments for both mutant fibrils, as listed in Table S1. The analysis proceeded by de novo backbone walks and did not rely upon the WT assignments. Common amide frequencies in the three 3D spectra-NCACX, NCOCX and CAN(co)CX-were used to establish sequential connectivities, and side-chain chemical shifts were confirmed in combination with the 2D 13 C-13 C spectra, resulting in site-specific assignments, as illustrated in Figure 2. Assignments were further confirmed by examining the strips through the CA frequencies in the NCACX and CAN(co)CX experiments and the carbonyl carbon (CO) frequency in the NCOCX and CAN(co)CX experiments. The residues assigned sequentially for E46K were K46-V49, T59-Q79, G84-F94 and G106-P108. The residues assigned sequentially for A53T were S42-V49, V52-V55 and T59-K97. Chemical shift assignments are included in Tables S2 and S3 for E46K and  A53T, respectively. Chemical shifts (especially for 13 C) depend strongly upon secondary structure [36,37,38,39,40] and therefore serve as highly reliable tool for evaluating structural perturbations arising from the AS mutations. As shown in Figure 3, E46K exhibits chemical shift perturbations throughout the amino acid sequence. In the WT secondary structure [12], residue E46 is located in one of the long b-sheet strands. Therefore, it is not surprising that E46K mutation, where an ionic residue is replaced with a cationic residue at pH 7.4, leads to major chemical shift perturbations (|Dd| avg ,1.6 ppm) versus the WT throughout the fibril core, due to the alternation in electrostatics. In contrast, the A53T mutant exhibits relatively minor chemical shift perturbations (|Dd| avg ,0.5 ppm), and the perturbations are very much localized located in certain regions of the fibril core. The 13 CA and 15 N chemical shift perturbations (Figure 4) in A53T AS fibrils versus the WT are found near the mutation site, as well as in the 60 s and 80 s.

Relative signal intensities and secondary structure comparison between the mutants and wild-type
The relative rigidity of regions within the fibril core was evaluated based on the signal intensities of a dipolar-based 3D CANCO experiment. Our recent investigations demonstrated that the region reported by Giasson et al to be responsible for fibrillation (residues 71-82 for the WT [41]) exhibited some of the highest signal intensities within the fibril core [12]. Together with several other site-specific measurements, we demonstrated that these residues are among the most rigid residues in the structure. As demonstrated in Figure 5, these regions are within the stronger regions of the core compared to the other signals in both mutants. The site-specific pattern of residues 71-82 and the relative signal intensity in the 90 s versus other regions in the sequence is somewhat different for E46K versus WT or A53T.
Because chemical shifts are highly sensitive to atomic environment [36,37,38,39,40], backbone dihedral angles (W and Y) and secondary structure can be predicted in a semi-empirical manner using the TALOS+ program [42] to obtain the backbone dihedral angles and secondary structure for E46K and A53T AS fibrils ( Figure 6). While the TALOS+ results demonstrate that A53T exhibits small changes in secondary structure compared to the WT, E46K exhibits substantial chemical shift changes relative to WT, indicating differences in secondary structure patterns. These results demonstrate that the secondary structure is somewhat similar for A53T and WT; whether this arises from a common three-dimensional fold remains to be determined. In contrast, for E46K, the chemical shift values indicate significant differences in secondary structure for the E46K mutant relative to WT; it is likely that the overall structural arrangement of the E46K mutant therefore is different from that of the WT protein.

Conclusion
In summary, we have site-specifically compared the structures of AS fibrils from the E46K and A53T early-onset PD mutants to the WT using MAS SSNMR. Our results demonstrate minor sitespecific perturbations near the mutation site, the 60 s and 80 s for A53T and large chemical shift perturbations within the entire core for E46K. In addition, they reveal that the residues described to be essential for the WT fibrils formation by Giasson et al are also within some of the most stable residues in the core for the mutants.
Taken all these results together, this study demonstrates the presence of site-specific perturbations in the fibril structure by the A53T and E46K mutants. We believe these results will assist future investigations to elucidate how these mutants affect the fibrillogenesis of AS compared to the WT in the presence of phospholipid vesicles [43,44]. These studies might be significantly relevant to understand the role of the early-onset PD mutants in the disease.

Supporting Information
Table S1 Description of the multidimensional experiments acquired to obtain de novo 13 C, 15 N chemical shift assignments. (DOC)