Cyril C.
Curtain
*ab,
Nigel M.
Kirby
c,
Haydyn D. T.
Mertens‡
c,
Kevin J.
Barnham
ab,
Robert B.
Knott
d,
Colin L.
Masters
b,
Roberto
Cappai
a,
Agata
Rekas
d,
Vijaya B.
Kenche
b and
Timothy
Ryan
b
aDepartment of Pathology and Bio21 Molecular Science and Technology Institute, The University of Melbourne, Victoria 3010, Australia. E-mail: ccurtain@unimelb.edu.au
bThe University of Melbourne, Florey Institute of Neuroscience and Mental Health, Victoria 3010, Australia
cSAXS/WAXS Beamline, The Australian Synchrotron, Clayton, Victoria 3168, Australia
dAustralian Nuclear Science and Technology Organisation (ANSTO), Kirrawee, NSW 2232, Australia
First published on 21st October 2014
The 140 residue intrinsically disordered protein α-synuclein (α-syn) self-associates to form fibrils that are the major constituent of the Lewy body intracellular protein inclusions, and neurotoxic oligomers. Both of these macromolecular structures are associated with a number of neurodegenerative diseases, including Parkinson's disease and dementia with Lewy bodies. Using ensemble optimisation modelling (EOM) and small angle X-ray scattering (SAXS) on a size-exclusion column equipped beamline, we studied how the distribution of structural conformers in α-syn may be influenced by the presence of the familial early-onset mutations A30P, E45K and A53T, by substituting the four methionine residues with alanines and by reaction with copper (Cu2+) or an anti-fibril organic platinum (Pt) complex. We found that the WT had two major conformer groups, representing ensembles of compact and extended structures. The population of the extended group was increased in the more rapidly fibril-forming E45K and A53T mutants, while the compact group was enlarged in the oligomer-forming A30P mutant. Addition of Cu2+ resulted in the formation of an ensemble of compact conformers, while the anti-fibril agent and alanine substitution substantially reduced the population of extended conformers. Since our observations with the mutants suggest that fibrils may be drawn from the extended conformer ensemble, we propose that the compact and extended ensembles represent the beginning of oligomer and fibril formation pathways respectively, both of which have been reported to lead to a toxic gain of function. Manipulating these pathways and monitoring the results by EOM and SAXS may be useful in the development of anti-Parkinson's disease therapies.
We have shown previously that SAXS and ensemble optimisation modelling (EOM) of native α-syn can be used to study the distribution of conformers based on their Rg and Pr distribution.14 Oxidation of the protein's four methionine residues markedly reduces the proportion of the most extended members of the ensemble of conformers. Since the native protein readily aggregates to form the fibrils associated with Parkinson's disease Lewy bodies, and the oxidised version is non-fibrillogenic,12 it is plausible that the extended conformer is a key stage on the fibril formation pathway. In this communication we present data from a size exclusion column (SEC) SAXS-EOM study of conformational changes in familial early PD onset mutants of α-syn and the influence of Cu2+ and an inhibitor of fibril formation on conformer distribution.
Copper/glycine (CuGly) 10 mM was prepared as described previously16 and the synthesis and characterisation of the anti-fibril agent ([PtII(N,N-dimethyl-2-[2-(quinolin-8-yl)-1H-benzimidazol-1-yl]ethanamine) Cl2] referred to as VK7) has also been described previously.17 All buffers and other reagents were of analytical grade or the highest possible purity.
The protein samples were subjected to in-line size-exclusion chromatography (SEC) using an automated HPLC apparatus (Shimadzu) fitted with a 10 mm by 300 mm Superdex 75 size exclusion column (GE Healthcare) with a bed volume of 22 mL equilibrated with 10 mM pH 7.4 phosphate buffer at a flow rate of 0.6 mL min−1. 100 μL of each sample at 10 mg mL−1 was injected from a 96 well plate auto-changer and the column's eluent was directed through a UV protein absorption detector set at 280 nm with a 10 mm path length to a 1.5 mm quartz capillary where it was exposed to the X-ray beam. At least 700 detector images of sequential 2 s exposures (2.1 s repeat time) were collected, corresponding to a total elution volume of 14.5 mL. When there was no further signal post the elution of the main α-syn absorbance peak, SAXS data collection was halted. After each sample the column and capillary were flushed with 100 μL of 6 M guanidine hydrochloride solution to remove any aggregated protein and the column was allowed to re-equilibrate in the phosphate buffer for a further 45 mL. Absolute intensities were calibrated using water in the 1.5 mm ID quartz capillary.
A small amount of aggregation, virtually undetectable by its absorption at 280 nm, occurred in the SEC column causing an increase in scattering that impacted on the quality of the scattering profiles represented by the leading edge of the SEC protein peak. This effect is illustrated in the Fig. S1 (ESI†). The aggregate formation could be triggered by contact with the column packing or could represent the beginnings of the aggregate nucleation that also seems to be a characteristic of other amyloid forming IDP's.19 We detected the influence of aggregates by checking the ratio of the value of the forward scattering intensity I0 against the protein concentration (c) determined from the absorbance at 280 nm acquired from the UV detector on the SEC using the absorption coefficient of 5120 M−1 cm−1 (Table 1). This was linear over a range of concentrations for all samples (Fig. S2, ESI†). We also estimated the linearity of the Guinier plot, which would be sensitive to the presence of oligomers and aggregates. However, this linearity only holds over a restricted region of the scattering profile where qRg < 1.0, or 1.3 for well-folded proteins. For IDPs with their multiple sizes, this region can be restricted to qRg < 0.8 or less,20 limiting the accuracy of the measurement because of the small number of useable points in the SAXS profile. We adopted qRg = 0.8 as a cut-off for deciding whether to include a given profile in the number to be averaged for each sample.
Sample | Low range peak (%) | High range peak (%) | ½ height peak width low range Rg (Å) | ½ height peak width high range Rg (Å) | Low range max cut-off Rg (Å) | High range min cut-off Rg (Å) | High range max cut-off Rg (Å) |
---|---|---|---|---|---|---|---|
WT | 60 | 40 | 8 | 8 | 42 | 49 | 69 |
A30P | 100 | 0 | 5 | — | 38 | — | — |
E46K | 55 | 45 | 7 | 6 | 36 | 44 | 60 |
A53T | 40 | 60 | 8 | 8 | 36 | 46 | 60 |
WT + Cu2+ | 90 | 10 | 8 | — | 36 | — | — |
A30P + Cu2+ | 100 | 0 | 9 | — | 37 | — | — |
A53T + Cu2+ | 80 | 20 | 10 | 20 | 40 | 25 | 60 |
4M4A | 70 | 30 | 10 | 8 | 44 | 42 | 61 |
4M4A + Cu2+ | 100 | 0 | 6 | — | 33 | — | — |
WT + VK7 | 70 | 30 | 9 | 13 | 37 | 35 | 62 |
All two-dimensional scattering patterns obtained were reduced to one-dimensional profiles of intensity I versus q using the ScatterBrain software package on line. The 280 nm absorbance was also displayed on a spread-sheet in real time so that the protein peak could be matched to the scattering profiles. Radius of gyration (Rg) calculation, Kratky and Guinier analyses were made using the ATSAS suite of programs,21–25 2.5.0–2 release available from http://www.embl-Hamburg.de/ExternalInfo/Research/Sax/index.html. The Kratky analysis was used to confirm that all of the scattering profiles used indicated essentially unstructured protein. The MM of each sample was determined from I0 calculated on the absolute scale by employing the method of Orthaber et al.,26 where the relative I0 of the protein is divided by the experimental scattering of water and then multiplying by the absolute scattering of water. The MM = [NAI0/c]/ΔρM2 where NA is Avogadro's number, I0/c is the absolute scale forward scattering normalised by concentration and ΔρM2 is the scattering contrast per mass.
EOM (Advanced EOM 2.0 (ref. 27 and 28) available in the ATSAS suite) was used for the analysis of the flexibility and size distribution of possible structures of the conformers of the protein contributing to the experimental scattering pattern. Both “Native” and “random” options were used for each run. The outputs were very similar and the “random” option data are presented here. This approach is based on the generation of a large pool (typically 10000) of theoretical structures derived with side-chain interaction constraints from the primary sequence of the protein. The theoretical X-ray scattering profiles calculated from these structures are then matched for fit using a genetic algorithm against the experimental scattering profile to create an ensemble of best fit structures. The structure-related outputs of the EOM analysis are the Rg and Dmax (the maximum distance between two Cα atoms within a conformer). A wide distribution of Rg and Dmax indicates a random distribution of structures that could fit the experimental SAXS profile, and a multi-modal distribution indicates the existence of a number of dominant conformers.29,30
Fig. 1 R g distributions from ensemble optimisation modelling of (A) WT, (B) A30P; (C) E46K; (D) A53T familial mutants of α-syn. Solid lines, ensemble; dotted lines, pool. |
The fitted SAXS profiles, Dmax distribution, SEC Abs280, Kratky and Guinier plots are available in the Fig. S4 (ESI†). The Dmax distributions closely followed those of the Rg's and SEC analysis showed that the samples were monodisperse. The values of average Rg, average ensemble Dmax, and χ for each profile fit are available in Table S1 (ESI†). All the profiles gave a reasonably good fit with an even distribution of errors. The average Rg value for the WT ensemble of 44.34 Å is close to those reported previously9,10 and gives some comfort as to the reproducibility of the scattering data.
The Rg values calculated from a Guinier plot of the experimental scattering profiles, the absolute scale I0, the protein concentration determined from the 280 nm absorbance and the derived molecular masses (MM) are given in Table 2. The latter values are within the accepted 10% error.26 The I0 and corresponding concentrations are averages of 5 readings taken around the UV absorption maximum, although more values were taken from each side of it to confirm that I0/c remained reasonably constant. These are shown as scatter plots in Fig. S2 (ESI†).
Sample | R g (Å) | I 0 (corrected26) | c (mM) | I 0/c | MM (kD) |
---|---|---|---|---|---|
WT | 42.64 ± 0.97 | 137.84 | 0.2 | 685 | 15.8 |
A30P | 43.64 ± 0.91 | 125.46 | 0.18 | 689 | 16.0 |
E46K | 27.57 ± 0.40 | 130.57 | 0.19 | 687 | 15.9 |
A53T | 27.23 ± 0.86 | 148.24 | 0.22 | 674 | 15.8 |
WT + Cu2+ | 26.49 ± 0.61 | 132.48 | 0.23 | 669 | 15.6 |
A30P + Cu2+ | 31.37 ± 0.35 | 134.44 | 0.20 | 673 | 15.7 |
A53T + Cu2+ | 31.87 ± 0.22 | 172.64 | 0.26 | 664 | 15.4 |
4M4A | 30.99 ± 0.03 | 155.25 | 0.23 | 675 | 15.7 |
4M4A + Cu2+ | 21.11 ± 1.0 | 174.02 | 0.26 | 673 | 15.7 |
WT + VK7 | 29.16 ± 0.24 | 94.08 | 0.14 | 672 | 15.6 |
Fig. 2 R g distributions from ensemble optimisation modelling after the addition of 1 M/M CuGly to (A) WT and (B) A30P; (C) A53T familial mutants of α-syn. Solid lines, ensemble; dotted lines, pool. |
The fitted SAXS profiles, Dmax, Guinier plot and the Abs280 plots from the SEC are available in Fig. S6 (ESI†), and the values of average Rg, average ensemble Dmax, and χ for each profile fit are available in Table S1 (ESI†). The plot of I0/c, was linear for each set of profiles (Fig. S2, ESI†).
The reported variations in conformer distribution make it clear that sample preparation conditions and measurement techniques influence the result. A recent advance on SAXS/WAXS beamlines has been to install SEC in line with the capillary in the X-ray beam,32,33 permitting precise control of conditions including matching concentration of the subtraction buffer, an important factor in the study of IDPs whose long range interactions may be disturbed or masked by slight changes in buffer conditions. In the present study SAXS profiles were measured 900 s after the protein left the GuHCl chaotrope behind in the SEC. This approach enabled the early establishment of intra-molecular contacts creating the conformer distribution. Deconvolution showed that with the WT there was a sufficient distinction between the maximum cut-off Rg value of the low range peaks and the minimum value of the high range to justify the conclusion that this early distribution was bi-modal. It could be argued that the broader distributions seen earlier have undergone further conformer development occurring during sample preparation. The increase in the proportion of extended conformers in the ensembles of the rapid fibrillisation-prone E46K and A53T familial PD mutants should be viewed in the light of earlier data suggesting that these conformers represent the pool from which fibrils are derived.3 Although the mutation at residue 46 would alter the balance of charges and could favour the extended conformer13 it is less clear that a decrease in hydrophobicity could lead to a similar effect. However, the A30P mutant, which is also a pathogenic mutation in α-syn, displays no extended conformers and a lower Rg distribution cut-off. A comparison of the oligomerisation and fibrillisation propensities of the A30P and A53T showed that the former tended to form oligomers, particularly when mixed with the WT.34,35 Bertoncini et al.36 showed that residual dipolar couplings were both considerably reduced at both the A30P and A53T mutation sites and, taking into account site-directed spin labelling and paramagnetic broadening data, concluded that both could more easily overcome the barrier to oligomerisation. However, they suggested that the presence of Pro at position 30 could hinder the incorporation of the N-terminus into the β-sheet structure needed to form fibrils. The A30P mutant has been found to bind less thioflavin T during fibrillisation experiments than the WT.37,38 These results suggest that the extended conformer is the basis for fibril formation, as those mutations that have less of it have significantly reduced propensity to aggregate into amyloid.
While pathogenic mutations do cause a significant proportion of clinical cases of PD, the disease is more often associated with a sporadic presentation. This may be due to external factors, which can also affect α-syn folding, such as metal ions, small compounds and other biological macromolecules. One such factor is copper (Cu2+). Cu2+ dysregulation has been proposed as one of the factors in the development of PD (reviewed ref. 39), and Brown et al.40 showed that the addition of Cu2+ favoured the formation of stellate-shaped toxic oligomers, adding to other findings that the presence of Cu2+ would favour oligomer formation under various conditions.41,42 These findings suggested that Cu2+ may have a significant effect on the initial distribution of conformers, possibly by coordinating with His 50 and other residues, altering the balance of charges.13 Indeed, the addition of Cu2+ to the WT, A30P and A53T mutants and the Met to Ala-substituted protein produced a narrow distribution, suggesting that these very compact conformers constitute the ensemble from which the Cu2+-induced toxic oligomers40 might be derived. The narrower Dmax and Rg distributions of the A30P mutant, even in the absence of Cu2+, support the suggestion that the compact conformers are at the commencement of the toxic oligomer pathway. Whether this compact conformer is initiated by a mutation favouring a compact distribution, by the presence of Cu2+ or by some means yet to be described is unfortunately unclear at this time.
Since Cu2+ has such a significant effect on the conformer distribution, which may lead to deleterious effects, there is hope that other factors, such as small molecules, may possibly influence the starting conformation in a beneficial fashion. To test this hypothesis, data was acquired for α-syn in the presence of a platinum based agent, VK7. This compound has previously been observed to be non-toxic and to reduce the amyloid plaque burden in human Aβ transgenic mice17 and has recently been observed to inhibit α-syn fibrillisation (Kench, unpublished). The Rg and Dmax distribution induced by the Pt compound shows significant reduction in the proportion of elongated structures, indicating induction of a compact form. However, this distribution is significantly broader then the narrow range of ensemble parameters induced by Cu2+, suggesting that a wider range of conformations are present in the sample. These results suggest that the compact region of the distribution may represent the starting pool for both toxic and non-toxic oligomers.
Given the observations described here, altering the balance of long-range interactions that promote the creation of the compact and extended formations in favour of those promoting the conformations with Rg values in the range of 35–45 Å appears to represent a potential therapeutic approach to the misfolding of α-syn, and may form the basis of a targeted compound screen. In addition, the study of other IDPs, such as β2-microglobulin,43,44 amylin44,45 and transthyretin mutants involved in familial amyloidotic neuropathy46 where alteration of long range interactions could have pathological consequences may also benefit from the SEC-EOM-SAXS approach.
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4mb00356j |
‡ Present address: Biological Small Angle Scattering Group European Molecular Biology Laboratory c/o DESY Notkestrasse 85, Geb. 25a22603 Hamburg, Germany. |
This journal is © The Royal Society of Chemistry 2015 |