Nanobodies raised against monomeric ɑ-synuclein inhibit fibril formation and destabilize toxic oligomeric species

Background The aggregation of the protein ɑ-synuclein (ɑS) underlies a range of increasingly common neurodegenerative disorders including Parkinson’s disease. One widely explored therapeutic strategy for these conditions is the use of antibodies to target aggregated ɑS, although a detailed molecular-level mechanism of the action of such species remains elusive. Here, we characterize ɑS aggregation in vitro in the presence of two ɑS-specific single-domain antibodies (nanobodies), NbSyn2 and NbSyn87, which bind to the highly accessible C-terminal region of ɑS. Results We show that both nanobodies inhibit the formation of ɑS fibrils. Furthermore, using single-molecule fluorescence techniques, we demonstrate that nanobody binding promotes a rapid conformational conversion from more stable oligomers to less stable oligomers of ɑS, leading to a dramatic reduction in oligomer-induced cellular toxicity. Conclusions The results indicate a novel mechanism by which diseases associated with protein aggregation can be inhibited, and suggest that NbSyn2 and NbSyn87 could have significant therapeutic potential. Electronic supplementary material The online version of this article (doi:10.1186/s12915-017-0390-6) contains supplementary material, which is available to authorized users.

solutions were prepared and incubated under the same conditions as for the sm-FRET and bulk ThT experiments, either alone or with the addition of 2 molar equivalents of nanobodies (35 µM ɑS, 70 µM nanobodies, 20-h incubation). For imaging, the solutions were diluted to 200 nM (ɑS and ɑS with NbHul5g) or 500 nM (ɑS with NbSyn2 and ɑS with NbSyn87) concentration into PBS buffer (PBS, 00-3002, ThermoFisher scientific), containing 5 µM ThT and 2 nM Nile red (NR, N1142, ThermoFisher scientific), to enable the simultaneous staining of ɑS aggregates with ThT and NR dyes. Fluorescence imaging was carried out using custom-built TIRFM setup described previously [6]. Images were acquired at a frame rate of QCM experiments to analyze the elongation rates of pre-formed ɑS fibrils in the presence of NbSyn87 and NbSyn2. Quartz crystal microbalance (QCM) measurements were carried out with E4 QCM D instrument (Q-Sense, Biolin Scientific, Stockholm, Sweden), using gold-coated QSX 301 sensor. ɑS fibril preparation and all subsequent measurements were performed at 37 °C in PBS buffer, closely following previously reported protocols [7][8][9].
In order to outcompete the C-terminal region of ɑS with respect to binding to the nanobodies and regenerate the surface-bound ɑS fibrils, a synthetic peptide (Genemed Synthesis Inc., San Antonio, TX, USA) was used, based on the sequence of ɑS and comprising its residues 118-140 (118-VDPDNEAYEMPSEEGYQDYEPEA-140). QCM data are shown in Supplementary Fig. 1d. The rates of frequency decrease during the periods of incubation with ɑS monomer, or with an equimolar ratio of ɑS and NbSyn87 or NbSyn2 are included in Supplementary Tables 3 and 4, and presented as averages between the overtones N = 3,5,7.
The detailed interpretation of the measured values is in Supplementary Results. Nanobody labeling with AF dyes. Random lysine labeling was performed using N-Hydroxysuccinimde (NHS) linked AF647 or 488. 1.5 molar equivalents of AF647 (or 488) NHS ester (Life Technologies) were added to unlabeled nanobody solution in 100 mM sodium bicarbonate buffer (pH 8.3), and incubated with agitation for 3 h. The labeled protein was then separated from free dye using Sephadex G-25 loaded PD-10 desalting columns (GE Healthcare). The labeling efficiency was determined using UV-Vis absorbance measurements, and was above 90% (NanoDrop 2000c UV-Vis Spectrophotometer, Thermo Scientific). The stoichiometry of AF647 labeling of the nanobodies was determined according to the manufacturer"s specifications and confirmed by mass-spectrometry and was found to be close to a 1:1 stoichiometry. The presence of the dye did not interfere with ɑS binding as observed using total internal reflection fluorescence spectrometry and SPR (data not shown).

TCCD data acquisition and analysis.
To verify the absence of binding between fluorescently-labeled nanobodies and fluorescently-labeled ɑS at single-molecule concentrations in our experiments, control two-color coincidence detection (TCCD) was performed. Dual excitation in either 488/594 nm or 488/633 nm mode (depending on the AF label pairs of the analyzed samples) was used, and single-molecule confocal instrumentation and methodology as previously described in detail [10,11], utilizing a detection under fastflow as for the sm-FRET measurements. In TCCD experiments, in contrast to monitoring FRET signal between the dye pairs, direct fluorescence signal from both AF488 and AF594 (or AF647) was detected as a consequence of direct dual-color laser excitation. Bound species bearing two different fluorophores are expected to produce two fluorescence bursts that are coincident in time, while singly labeled molecules will produce non-coincident fluorescence bursts. The coincidence can be quantified using the association quotient Q, defined as: where A is the number of fluorescent bursts in the blue channel above the 15 photons bin -1 threshold, B is the number of fluorescent bursts in the red channel, C is the number of coincident events, and E is the number of chance-coincident events. Q value arising purely due to chance coincidence events was determined by analyzing free AF dyes in solution, as described below.
For the measurements, 1:1 stoichiometric ratio of AF-labeled ɑS and NbSyn87 was used.
This nanobody was chosen owing to its highest affinity for ɑS and therefore the highest potential chance for it to remain bound in our single-molecule experiments. The following equimolar combinations were tested: AF647-NbSyn87 + AF488-ɑS, AF488-NbSyn87 + AF647-ɑS, AF488-NbSyn87 + AF594-ɑS. In addition, samples of free AF647 + AF488 and aggregated AF647-ɑS + AF488-ɑS were analyzed as controls. For the measurements, the samples were diluted either in PBS buffer or deionized water up to single-molecule concentration suitable for the TCCD analysis. The dilution into different buffers served as a test for potential changes in the co-interaction of the nanobody and ɑS due to the changes in pH and ionic strength. Following the dilution, solutions were immediately introduced into a straight channel of a microfluidic device via a gel-loading tip (200 µL, Life technologies) and withdrawn at a constant rate of 1 cm s -1 via a syringe pump (PHD2000, Harvard Apparatus).
Overlapped laser beams were focused into the middle of the channel. For each sample, data were acquired for 600 s, with 100 µs bin-width, chosen according to the expected residence time in the excitation region at the chosen flow speed [12], 100,000 bins per frame and a total of 60 frames. The fluorescence photon traces in two separate channels, the emission from AF488 and AF594 dyes (or AF647, when the samples and the setup were in 488/647 mode), were collected simultaneously and outputted using custom-programmed field-programmable gate array (FPGA, Colexica). All measurements were made at ambient temperature around 20 °C, similarly to sm-FRET measurements.
The collected photon traces were analyzed using custom-written Igor Pro 6.22 (Wavemetrics) software analogous to previously described [2]. The data were corrected for autofluorescence and the crosstalk. Photon bursts with intensities greater than the threshold of 15 photons.bin -1 in the blue and in the red channels were selected according to previously established threshold selection approach that allows maximizing the detection of coincident events.
Simultaneous events in both channels above the threshold (the AND criterion [13]) were selected. To account for any possible coincident events due to chance, the desynchronization approach was used [14]. Time-bins in the blue channel were randomly re-numbered before the selection of simultaneous events in the two channels above the threshold. Using these outputs, the association quotient Q was estimated according to equation S1. The resulting values are summarized in Supplementary Table 1.

TCCD chance coincidence controls using free dyes in solution or dual-labeled ɑS.
To determine Q corresponding to a non-interacting system and resulting purely from the chancecoincidence, 1:1 stoichiometric mixture of free AF dyes were measured by TCCD, keeping the same detection conditions as for the protein measurements. The free unbound dyes were prepared according to a previously published protocol [15]. For the measurement, 1:1 molar ratio of AF488 and AF647 were used. The samples were recorded straight after preparation.
As positive controls, the samples of dual-labeled ɑS incubated under agitation at 2 µM for 72 h were also recorded. The data were analyzed in the same manner as the data recorded using the fluorescent proteins and the resulting Q values are presented in Supplementary Kinetic analysis. The kinetic analysis was as recently reported [16], with the additional inclusion of two explicit reverse reactions from high-FRET oligomers to low-FRET oligomers, and from low-FRET oligomers to monomers, as shown in Supplementary Fig. 3a.
The following kinetic moment equations were used to model the aggregation of ɑS in the absence and in the presence of nanobodies: Proteasome degradation assays. These assays, presented in Fig. 4b (main text) were carried out according to previously described protocol [18]. determined as the numbers of oligomers after the incubation divided by the numbers of oligomers immediately after mixing (Fig. 4b, main text).

ROS measurements. Mixed cultures of hippocampal or cortical rat neurons and glial cells
were prepared and cultured as described previously [19]. The experiments were carried out according to previously detailed protocols [1,20]. For the assays presented in Fig. 4c  NbSyn2 lead to a significant reduction in cell death (Fig. 4d). Incubation of cells with either nanobodies in the absence of ɑS was confirmed not to alter the basal rate of cell death. Fig. 4e (main text) were carried out using BV2 microglia, according to the previously reported protocol [18].

Analysis of QCM experiments: elongation of ɑS fibrils in the presence of NbSyn87 and
NbSyn2.
The data for both nanobodies show that the elongation rate is reduced directly after incubation with nanobodies, or, even more clearly, in the presence of nanobodies. In both cases the reduction in measured rate of frequency change is ca. 50%. Given that stoichiometric amounts of ɑS monomer and nanobody are added, the elongation likely proceeds through the addition of ɑS-nanobody complex. This complex has a mass of the order of 24 kDa, compared to the molecular mass of ɑS alone of ca. 14 kDa. This difference is relevant, as QCM senses the change in mass of the surface-bound material. If the signal in the QCM experiment would only stem from the dry mass of the added protein, a decrease of the rate of frequency change by 50% would translate therefore into an inhibition of 70%, representing an upper bound of the observed inhibition.
We have shown in the past that the mass sensitivity of QCM for fibril growth is ca. 4 times as high as predicted by the Sauerbrey equation [8], using the dry mass. Therefore, only ca. 25% of the signal stem from the dry mass, and 75% stem from water that is dragged along with the fibrils. If the frequency shift were proportional to the dry mass, i.e. that the dry mass always contributes the same fraction to the frequency shift, then this would yield the same degree of inhibition as calculated above (70%). If, however, we assume that the water contribution is independent of the mass of the elongating unit (and rather defined by the fibril geometry, which is probably not strongly dependent on the binding of the nanobody), we obtain a lower bound for inhibition, corresponding to 58%.
Therefore, based on the QCM experiments, the elongation rate of pre-formed ɑS fibrils in the presence of nanobody corresponds to 30-40% of the rate in the absence of nanobodies.

Kinetic analysis of ɑS aggregation in the presence of nanobodies.
Our experimental results showed that the presence of both ɑS-specific nanobodies slowed down the aggregation of ɑS. Using sm-FRET measurements, we characterized the impact of nanobodies on the earliest steps of the aggregation reaction and observed a clear inhibition of high-FRET oligomer formation. To quantitatively characterize this inhibition in the presence of the nanobodies, we applied kinetic analysis, using the nucleation-conversion-polymerization model that we have recently reported for the aggregation of ɑS alone [16], but additionally including two explicit reverse reactions, as is schematically outlined in Supplementary Fig. 3a. According to this model, monomeric ɑS molecules assemble into low-FRET oligomers, which subsequently convert into high-FRET oligomers. These in turn convert into fibrils, which then grow via a succession of monomer addition steps. As previously [16], within this description, oligomer conversion and fibril elongation steps are treated as size-independent, and any late-stage processes such as fibril fragmentation do not enter the analysis. To account for the observed inhibitory effects of the nanobodies on oligomer conversion, we introduced two explicit reverse reaction steps, from high-FRET oligomers to low-FRET oligomers and from low-FRET oligomers to monomers. The inclusion of these steps enabled the overall rate of conversion to high-FRET oligomers to be reduced in the presence of the ɑS-specific nanobodies, assuming the forward reactions to be unaltered. It was confirmed that the addition of these two steps was necessary for the analysis, because either decreasing the rates of forward reactions or solely decreasing one of the two reverse reaction rates resulted in large deviations between the predicted and observed ranges of aggregate concentrations.
We used the modified model to globally fit the kinetic profiles of monomer depletion, oligomer formation and fibril formation. The rate constants reported in Supplementary Fig. 3 result from the best fit to monomer, oligomer and fibril kinetics simultaneously, and the fits are shown in Supplementary Fig. 3b-e. All resulting parameters of the forward reactions are consistent with our previous results 23 . The estimated rate constants for the conversion from high-FRET to low-FRET oligomers, ̃ , were 1± 0.5 h -1 for NbSyn2 and 10±5 h -1 for NbSyn87, and both values are consistent with our independent experimental observation of the rapid conversion within minutes of pre-formed high-FRET oligomers to low-FRET oligomers upon the addition of nanobodies (shown in main text, Fig. 3). Using the derived parameters, we predicted the kinetics of low-FRET and high-FRET oligomer production during the aggregation reaction over longer time-period up to 100 h. According to these predictions shown in Supplementary Fig. 3f, the concentration of low-FRET oligomers would decrease slower in the presence of the ɑS-specific nanobodies than in their absence.
Furthermore, the formation of high-FRET oligomers would be inhibited, in good agreement with the sm-FRET findings. These effects are predicted to be stronger in the presence of NbSyn87, consistent with its higher affinity for ɑS in comparison to NbSyn2.