Optical Structural Analysis of Individual α‐Synuclein Oligomers

Abstract Small aggregates of misfolded proteins play a key role in neurodegenerative disorders. Such species have proved difficult to study due to the lack of suitable methods capable of resolving these heterogeneous aggregates, which are smaller than the optical diffraction limit. We demonstrate here an all‐optical fluorescence microscopy method to characterise the structure of individual protein aggregates based on the fluorescence anisotropy of dyes such as thioflavin‐T, and show that this technology is capable of studying oligomers in human biofluids such as cerebrospinal fluid. We first investigated in vitro the structural changes in individual oligomers formed during the aggregation of recombinant α‐synuclein. By studying the diffraction‐limited aggregates we directly evaluated their structural conversion and correlated this with the potential of aggregates to disrupt lipid bilayers. We finally characterised the structural features of aggregates present in cerebrospinal fluid of Parkinson's disease patients and age‐matched healthy controls.

Abstract: Small aggregates of misfolded proteins play ak ey role in neurodegenerative disorders.Suchspecies have proved difficult to study due to the lackofsuitable methods capable of resolving these heterogeneous aggregates,w hich are smaller than the optical diffraction limit. We demonstrate here an alloptical fluorescence microscopym ethod to characterise the structure of individual protein aggregates based on the fluorescence anisotropyo fd yes such as thioflavin-T,a nd show that this technology is capable of studying oligomers in human biofluids such as cerebrospinal fluid. We first investigated in vitro the structural changes in individual oligomers formed during the aggregation of recombinant a-synuclein. By studying the diffraction-limited aggregates we directly evaluated their structural conversion and correlated this with the potential of aggregates to disrupt lipid bilayers.W ef inally characterised the structural features of aggregates present in cerebrospinal fluid of Parkinsonsd isease patients and agematched healthy controls.
The formation and spread of small aggregates of proteins such as a-synuclein, b-amyloid or tau is reported in aw ide range of neurodegenerative diseases. [1,2] Of these,Parkinsons disease (PD) is characterised by the accumulation of am isfolded and aggregated protein called a-synuclein within neurons to form Lewy neurites and Lewy bodies. [3] Genetic and pathological evidence suggests that the protein asynuclein is central to neurodegeneration in PD. [4] Specifically,t he transition from an intrinsically disordered asynuclein monomer through aseries of oligomeric intermediates (with varying structures and size) to ah ighly structured filament [5,6] is recognised to drive pathogenesis in a-synucleinopathies.F urthermore,a ggregates of a-synuclein exhibit cell-cell transfer, leading to seeding and recruitment of more protein molecules to form additional aggregates that can generate new seeds in an exponential way, [7] leading to the region-region spread of disease.T he distinct structure of asynuclein aggregates has arole in its pathogenic properties,in particular,the toxicity of the aggregate,the cell type affected, seed competency, and the regional transfer of pathology. [8,9] This dramatic effect of the structure of the ordered assembly on the pathogenic pathway in the brain underpins the importance of understanding the order/structure of a-synuclein aggregates.
Previous studies have shown aggregates to be very diverse in terms of their mechanisms of formation, size and structure. [10][11][12][13] Bulk measurements obtained with conventional, ensemble-based, biophysical techniques are able to characterise many features of these heterogeneous aggregates, [14,15] but new quantitative tools are needed to specifically characterise in greater detail the structural features of individual aggregates,particularly in human tissue and biological fluids.
By means of single-molecule fluorescence resonance energy transfer (smFRET) experiments,asubpopulation of aggregates formed with fluorescently labelled a-synuclein has previously been shown to undergo as low structural rearrangement before growing into fibrils. [12] This conversion can generate more cross-b structure and the resulting aggregates were reported to be both, more resistant to proteinase-K and more toxic to cells.U nlabelled fibrils of amyloid-b or asynuclein can be imaged with total internal fluorescence (TIRF) microscopy and structurally specific dyes such as thioflavin T(ThT), [16,17] opening up the possibilities of studying aggregates in human biofluids. [18] At as ingle fibril level, dyes such as ThTo rC ongo Red have been shown to bind insulin fibrils in an ordered way,a nd by monitoring the intensity as af unction of the polarisation angle,t hese dye classes can be provide information on the structure of fibrils. [19] In this work, we have characterised structural features of aggregates formed by an amyloidogenic protein, by implementing ah ighly sensitive fluorescence anisotropy setup. Fluorophores absorb light with ap robability proportional to the square of the dot product of the local optical electric field and the molecular transition dipole moment. Thus when adye binds in ad efined orientation it emits highly polarised anisotropic fluorescence.W etherefore built abespoke anisotropy instrument to study the structure of spatially isolated amyloid aggregates by placing ap olariser in the detection path which rotates continuously during image acquisition ( Figure 1a). ThTi sawidely-used benzothiazole dye that increases its fluorescence quantum yield by several orders of magnitude upon binding extended cross-b structures.Experiments suggest that the dye preferentially binds with its long axis parallel to the long axis of fibrils, [20][21][22] but depending on the protein under study and the structure of the fibril other binding sites may exist. When single a-synuclein fibrils are imaged as af unction of the angle of the axis of polarisation, their fluorescence modulates sinusoidally between am aximum when the axis is aligned with the fibril and aminimum if the axis is perpendicular to the fibril (Figure 1b-d). This further confirms that the dominant binding site (or possibly sites) of ThTisaligned with the axis of the a-synuclein fibrils.
Am easure of the degree of extended ordered cross-b structure in an aggregate is the amplitude of the fluorescence signal of ThTa st he polariser rotates.T he normalised fluorescence intensity as af unction of polariser angular displacement of each individual protein aggregate was fitted to as inusoid (see Methods section and Figure S1 in the Supporting Information (SI)). Thef luorescence was fitted to y ¼ acos bx þ c ðÞ þ d,where a represents the amplitude of the signal, b the constant angular frequency( user-defined rotation velocity of the polariser), c the phase and d an offset (Figure 1e). Ther esponse to the anisotropy measurement is defined by: modulation ¼ 2a=ða þ dÞ.Asmall modulation value implies disordered b-sheet content in the aggregate, whereas alarger modulation value implies spatially aligned bsheet content. We used as ingle-molecule sensitive TIRF imaging mode to measure the structural arrangement of individual spatially isolated diffraction-limited aggregates which we have previously characterised using super-resolution techniques. [23] We performed an aggregation reaction for recombinant a-synuclein and focused on the kinetics of the lag phase of the aggregation. [24] Ther eaction was done at ac oncentration of 70 mm under agitation at 200 rpm in 25 mm Tris buffer (pH 7.4) supplemented with 0.1m NaCl and 0.01 %N aN 3 at 37 8 8C. Theaggregation reaction was performed in low binding polypropylene tubes to minimise protein adhering to the tubes.M ore specifically we analysed samples taken from the aggregation reaction at times between 1hand 4h.Atlonger times aggregates larger than the diffraction limit of optical light (i.e. % 170 nm) start to form. Ahistogram of modulation  values for each time point revealed that oligomers present at 1hand 2hhave low modulation values (typically lower than 0.5), while at 3hwe found adistinct population of oligomers that respond with high modulation values ( Figure 2). These data are indicative of the structural rearrangement from ar elatively amorphous oligomer into a" fibril-like" periodic structure.T he modulation measurement does not correlate with the fluorescence intensity of the aggregate (SI, Figure S2 and S3), suggesting that the number of ThTbinding sites can remain constant during as tructural re-arrangement. Although there is av ariability associated with the stochasticity of the nucleation process during the lag phase of the aggregation, independent experiments show the presence of the two populations of aggregates and the same trend for the evolution of the species (SI, Figure S4).
Combining all the detected aggregates,t he overall distribution of the degree of modulation can be fitted to two Gaussian distributions (Figure 3a). When compared to the modulation of long fibrils (for example formed after 24 ho f aggregation, yielding fibrils that are several mml ong), we observe that the fibrils typically have higher modulation values (Figure 3b). Thehistogram of oligomers and fibrils can be therefore fitted to three Gaussian distributions (Figure 3c). Thep opulation corresponding to low modulation values have some cross-b content (as ThTbinds to them), and behaves in asimilar way to fluorescent beads (SI, Figure S5), meaning that these aggregates are not structurally aligned. Thes catter plot of modulation vs.t he mean intensity of all measured aggregates suggested once again that disordered aggregates convert to fibril-like aggregates without an increase in integrated fluorescence intensity (Figure 3d). We observed that fibrils are characterised by high modulation and intensity values (Figure 3d). This strong fluorescence anisotropy response of fibrils can be achieved by labelling with other fluorescent dyes such as pentameric formyl thiophene acetic acid (pFTAA) which has also been shown to bind to cross-b structures [25] (SI, Figure S6).
In order to understand further the evolution of aggregates we globally fitted two Gaussian functions to each time point, pooling data obtained in three independent aggregation reactions to describe better the landscape of aggregates (Figure 3a). Thei ntegrated areas of the Gaussians give an estimate of the number of aggregates in each population (Figure 4a). Theresults show that non-modulating aggregates appear before modulating ones,a nd at as lower rate,  suggesting that there is ac onversion of non-modulating to modulating aggregates.I nc ontrast to these well-defined populations,a ggregates formed during an aggregation reaction of a-synuclein at low monomer concentration (1 mm for 1month) display ab roader distribution of modulation responses,s uggesting that aw ider variety of species are formed over long periods of time (SI, Figure S7).
To correlate the structural information of the aggregated species with the ability of these aggregates to generate toxic effects in cells we evaluated their capability to disrupt membranes with at echnique that uses vesicles filled with aC a 2+ sensitive dye. [26] Upon the interaction of ap rotein aggregate with the vesiclesm embrane,C a 2+ ions enter the vesicle from the surrounding solution and hence becomes fluorescent. This change in fluorescence can be detected using TIRF microscopy.W ei maged individual liposomes in the presence of Ca 2+ buffer (blank), followed by the addition of an aliquot of a-synuclein aggregates at ac oncentration of 50 nm and subsequent addition of ionomycin. In the presence of only Ca 2+ buffer, the fluorescence intensity of the vesicles was low and comparable to that of background noise due to minimal Ca 2+ presence within the vesicle. [26] After incubation (for 10 minutes) with a-synuclein samples,w ed etected an increase in the localised fluorescence intensity of the vesicles showing that Ca 2+ ions could enter the vesicles as ac onsequence of the aggregates induced membrane permeability. Subsequent addition of ionomycin, an ionophore enabling Ca 2+ to enter the vesicles,r esults in the saturation of all vesicles with Ca 2+ ions,a llowing us to quantify the extent of membrane disruption (Figure 4c).
Thep ermeabilisation assay showed al inear increase of the Ca 2+ influx after incubation with aggregates (previously aggregated for 1hto 4h), suggesting that both disordered and fibril-like aggregates induce calcium influx in the liposomes (Figure 4b). Given that the trend of modulating species dominates the later time points while the abundance of nonmodulating aggregates increases at aslower rate (Figure 4a), our results suggest that modulating species have ah igher ability to disrupt membranes.
To demonstrate the broad applicability of our technique, we also applied the method to samples of human cerebrospinal fluid (CSF), comparing aged-matched healthy controls (HC) to Parkinsonsd isease (PD) patients.B ya nalysing the modulation of individual species in each group (obtained from 4H Ca nd 4P Ds amples), we found that the large majority of species show non-modulating behaviour, ( Figure 5a,b) meaning that they are disordered. Only as mall fraction ( % 1%)o fs pecies showed modulating behaviour in both HC and PD groups (Figure 5c). Species with higher modulation values (modulation over 0.45) in PD patients have am ean of 0.57 compared to HC with am ean of 0.48 (inset Figure 5c). Theabundance of these modulating species in CSF is very low and therefore prevents us from ar obust statistical analysis.F urther studies need to be done to characterise these ordered species,a st hey are candidates to be involved in toxicity and spreading mechanisms.H owever, this does demonstrate the technique is capable of making structural measurements in human CSF.
In summary,w eh ave demonstrated that by combining sensitive TIRF microscopy with anisotropy measurements, one can directly characterise the structural features of individual oligomers.T his method is highly flexible as it does not require protein labelling,b ut rather ad ye that recognises cross-b motifs.W eh ave shown the conversion from disordered aggregates of a-synuclein to fibrillar aggregates,i na greement with previously reported smFRET measurements.F urthermore,o ur experiments suggest that modulating aggregates have ahigher capacity to disrupt lipid membranes.Our results provide clear evidence that most ThT active species in CSF are disordered, but do,however,contain cross-b sheet structure.O ur ability to analyse single aggregates individually allowed us to detect an ultra-low abundance of fibril-like species in human CSF.This methodology is compatible with other proteins whose aggregation has been associated with human disorders such as amyloid-b,t au, lysozyme or insulin. Overall this approach provides an ew method to characterise the degree of fibrillation in individual protein aggregates,c ontributing to the set of biophysical methods needed to understand some of the most fundamental mechanisms in neurodegeneration.