Self-Blinking Thioflavin T for Super-resolution Imaging

Thioflavin T (ThT) is a typical dye used to visualize the aggregation and formation of fibrillar structures, e.g., amyloid fibrils and peptide nanofibrils. ThT has been considered to produce stable fluorescence when interacting with aggregated proteins. For single-molecule localization microscopy (SMLM)-based optical super-resolution imaging, a photoswitching/blinking fluorescence property is required. Here we demonstrate that, in contrast to previous reports, ThT exhibits intrinsic stochastic blinking properties, without the need for blinking imaging buffer, in stable binding conditions. The blinking properties (photon number, blinking time, and on–off duty cycle) of ThT at the single-molecule level (for ultralow concentrations) were investigated under different conditions. As a proof of concept, we performed SMLM imaging of ThT-labeled α-synuclein fibrils measured in air and PBS buffer.


Preparation of ThT labelled α-Syn fibrils.
α-Syn was expressed in Escherichia coli BL21-D3 up to OD600 of 0.5 and then induced with 0.5 mM IPTG, post induction grows the cells o/n for 15-18 hours at 20 °C.Harvest the cells at 8000g.Briefly, α-Syn WT plasmid containing E. coli cells were grown in Luria broth (LB) overnight in the presence of 100 µg/mL ampicillin.Upon reaching the OD600 of 0.5, the cells were induced to express α-Syn using 1 mM IPTG for 4 h, while the culture flasks were incubated in 37 °C and shaken at 180 rpm.Then, the cells were harvested by centrifugation at 6,000 rpm for 15 min at 4 °C to obtain the cell pellet.Following re-suspension of the cell pellet in lysis buffer (20 mM Tris base pH 8.0), sonication at 50 W (12 minutes, 6sec on 9 sec off,) was performed.Then, the cells suspended in the lysis buffer were placed in boiling water for 10 min followed by centrifugation at 18,000 ×g for 30 min.The supernatant was collected and ammonium sulfate was slowly added until reaching 0.36 g/ml.Following stirring for 30 min at 4 °C, the suspension was centrifuged at 18,000 ×g for 20 min at 4 °C.The pellet was re-suspended in 20 mM Tris buffer (pH 8.0) and loaded onto an anion exchange using Q Sepharose.α-Syn was eluted at 300 mM NaCl and its purity was confirmed by SDS-PAGE.Purified α-Syn was dialyzed against PBS 1X pH7.4 (overnight) @ 4 °C using 3K MWCO membrane and its concentration was determined by measuring its absorbance at 275 nm (ε275 = 5600 M -1 cm -1 ).
Purified protein at 467 µM was aliquoted and stored at -80°C.
For fibrillation in Eppendorf tubes: 70µM of α-Syn in PBS pH 7.4 was incubated at 37 °C for 48 hour under continuous shaking at 1000 rpm using a Thermomix.
For SMLM fibril imaging in air: ThT was added to pre-formed fibrils (final concentration of 20 μM for fibrils and 2 nM for ThT) and allowed to incubate for 15 minutes in dark place at room temperature.10 μL of this solution was added to cleaned coverslips and allowed to dry in ambient, dark conditions.
For SMLM fibril imaging in PBS: Cleaned coverslips were treated with 50 µL polylysine solution (0.1% in H2O w/v) to render them adhesive.After 30 min, the polylysine was washed off with Milli-Q water, and the cover slip was dried with N2 flow.20 μL of the fibrils were dropped onto the treated surface.After air-drying in the dark, the coverslip was washed with Milli-Q water for 3 times and dry in the air room temperature overnight.The sample was covered in PBS, sealed, and stored for imaging in the dark at room temperature.

Single-molecule localization microscopy.
The single-molecule localization measurements were performed using a super resolution ground state depletion (SR GSD) microscope (Leica).488 nm (300 mW) laser was selected for fluorescence reactivation.For the 488 nm laser, the excitation filter (483 nm -493 nm/400 nm -410 nm), the dichroic beam splitter (496 nm) and the emission filter (505 nm -605 nm/449 nm -451 nm) were used.Laser intensity (5.1 kW/cm 2 ) was obtained by dividing the laser power (82.2 mW was used, tested under the objective) by the area.The objective lens HCX PL APO 160x 1.43 NA Oil CORR-TIRF was selected for single-molecule measurements and super-resolution imaging.The microscope was equipped with an EMCCD camera (iXonDU-897, Andor).The camera settings were 10 MHz at 14 bit and a preamplification of 5.1.For super-resolution imaging, the camera exposure time was set to 30 ms and an EM gain of 100 was used.Please note here that the double bandwidth of the filters/beam splitter were chosen for 405 nm back pumping and in our experiments mentioned in this work, such back pumping was not used.

Data analysis
To obtain the number of points from the series frames images, first of all, time projection (maximum intensity projection) was done with all images, and then each point was localized through the ThunderSTORM-plugin in ImageJ 1 .Fluorescence intensity traces were extracted by first generating a maximum intensity projection of the recorded frames.Fluorescence signals in this projection were localized using the ThunderSTORM.We then calculated the intensity trace for each localization throughout all raw data frames as the total background corrected intensity in a 7 x 7 ROI around the localized coordinates.The local background for every localization in every frame was calculated within a 17 x 17 ROI.Pixel values exceeding 5 times the standard deviation within this ROI were excluded from background calculation as they were considered as fluorescence signal.Calculated total intensities within the ROIs were then plotted for every frame.
To calculate the photoelectrons on a camera pixel for an EMCCD, the equation is: Where N is the number of photoelectrons, p is the EMCCD sensitivity (in photoelectrons per A/D count), g is the EM gain for EMCCD camera, I is the image intensity (in A/D counts) and b is the baseline level (in A/D counts) from a dark image at the identical camera settings.For camera (iXonDU-897, Andor) used in this work, p is 12.09 while g was set to 100, b is 100.The photon numbers, blinking time, duty cycle and localization precision were analyzed following the reported method 2 .The analysis needs a two-step process.First, ThunderSTORM plugin 1 in ImageJ was used to localize molecules in every frame of the recorded imaging data.A result table including coordinates of localizations, sigma (standard deviation of the Gaussian fitted on the peak), intensity/photons and uncertainty was given.Localizations were then filtered according to the expected width of the sigma (calculated from diffraction limit, we used range of 75-125 for ThT).Localizations appearing in consecutive frames were then merged.As spatial constraint, we used a maximum distance of 80 nm, a rather large radius was chosen to allow localizations with low photon counts to be still properly merged.After merging, new column called detections can appear, which means number of frames where the peak has been detected and fitted.Second, the exported result was processed with matlab.Photon number and blinking                It is important to note that Dempsey et al. 2 reported the survival fraction over the first 700 seconds out of a total measurement of 2,000 seconds.In our calculations, we calculated the S15 survival fraction over a total testing duration of only 600 seconds.Many ThT molecules entered deactivation state and were not re-activated in the first 600 seconds, which in our analysis is defined as being photobleached.Therefore, the analytical results as shown in Figure S15 reported a relatively much lower survival fraction as compared with the ones reported in in Dempsey et al. 2 .Consistent with the time traces measured under different buffer conditions shown in Figure S14 of the previous revision, the highest survival fraction was calculated in O2 Free: glucose (1% vol) and glucose oxidase (GLOX, 1% vol) in DPBS. a ThT labelled on α-Synuclein; b ThT labelled on Aβ42.

S16
Figure S1.Wide-field images of ThT embedded in PS film prepared in a dark environment.(a) The first frame and (b) reconstruction (time projection) image of 20,000 frames.Scale bar: 5 μm.Color bar: photons per pixel (100 nm).Figures S1a and 1b show the first frame of the wide-field image and the reconstruction (time projection) image of 20,000 frames of the same imaging area of ThT embedded in PS film prepared in dark environment.The sample was prepared on an Ibidi gridded coverslip, with which the focal plane can be easily find under bright field with weak lamp.Then switch to fluorescence mode to collect data.The reconstruction (time projection) image shows more count (number of fluorescent spots) and higher fluorescent intensity.Comparing Figures S1a and 1b, we note that 56% of the spots are detected in the first frame -910 (0.57 molecules/µm 2 ) vs. 1635 (1.02 molecules/µm 2 ), indicating that 56% of ThT molecules are in the fluorescent ON state once excited.Compared with the samples in the Figure 1 (prepared under room light), the samples prepared in the dark show more points are in the fluorescent ON state once excited, indicating that some ThT molecules can be illuminated and pushed into fluorescent OFF state by room light during the sample preparation or the weak laser intensity used to find the imaging focal plane.

Figure
Figure S2.Wide-field images of pure PS film.(a) The first frame and (b) time projection image of 20,000 frames.Scale bar: 5 μm.Color bar: photons per pixel (pixel size 100 nm).

Figure
Figure S3.Wide-field images of ThT on coverslip (sealed in N2) with samples exposed to light.(a) The first frame and (b) time projection image of 20,000 frames.Scale bar: 5 μm.Color bar: photons per pixel (pixel size 100 nm).Figures S3a and 3b show the first frame of the wide-field image and the reconstruction (time projection) image of 20,000 frames of the same imaging area of ThT prepared on coverslip in nitrogen.The reconstruction (time projection) image shows much more count (number of fluorescent spots) and higher fluorescent intensity.Comparing Figures S3a and 3b, we note that only 22% of the spots are detected in the first frame -271 (0.17 molecules/µm 2 ) vs. 1255 (0.78 molecules/µm 2 ), indicating that only 22% of ThT molecules are in the fluorescent ON state once excited.

Figure S4 .
Figure S4.Photophysical properties of ThT on coverslip (sealed in N2).(a) Histogram of detected photons per switching event and single-exponential fit, mean photon numbers were determined by the exponential fit; (b) on-off duty cycle (fraction of time a molecule resides in its fluorescent state) of ThT calculated from single-molecule fluorescence time trace.

Figure S6 .
Figure S6.Photophysical properties of ThT embedded in PS film (sealed in N2).Histogram of blinking on time per switching event and single-exponential fit, mean blinking time were determined by the exponential fit.

Figure S8 .
Figure S8.Representative wide-field images of ThT measured in different environments.(a, b) on coverslip measured in air, (c, d) embedded in PS film, (e, f) on coverslip sealed in N2, (g, h) on coverslip measured in PBS.(a, c, e, g) The first frame and (b, d, f, h) reconstruction (time projection) image of 20,000 frames.Scale bar: 5 μm.Color bar: photons per pixel (pixel size 100 nm).It should be noted here, the samples used in this figure were prepared using the same

Figure S9 .
Figure S9.Fourier ring correlation (FRC) analysis for α-Syn fibrils labelled with ThT measured in air (a-d) and PBS (e-h).Odd (a,e) and even (b,f) frames produced independent superresolution reconstruction images, each from 10000 frames.FRC analysis (c,g) and local mapping of FRC values (d,h) for super-resolution images in a,b and e,f.For the FRC analysis, imageswere splitted into odd and even frames, which can produce two independent superresolution reconstructions.Within the FRC plugin in ImageJ, the two data reconstructions are divided into blocks and for each block the FRC value is calculated with algorithm described in4 .NanoJ-SQUIRREL plugin in ImageJ was used to generate FRC mapping5 and for each block the FRC value is calculated as c and g.The mean FRC values from FRC mapping for ThT labelled on α-Syn fibrils measured in air and PBS are 73 nm (std 16 nm) and 89 nm (std 13 nm), respectively.

Figure S10 .
Figure S10.Histogram of the number of localizations per bin (20 × 20 nm 2 ) corresponding to Figure 2a and f, α-Syn fibrils labelled with ThT measured in air (a) and PBS (b).The number of localizations was given by ThunderSTORM as intensity of image by the visualization mode 'Histogram'.Number of localizations/length of fibril was also estimated.First, 'Apply_DOG_Filtering.py script' from GitHub was used to enhance the image and highlight the strands.Then 'ridge detection' plugin was used to estimate the length of fibrils.Numbers of

Figure S11 .
Figure S11.Wide-field images of fibrils without dye labelling measured in air (a, b) and in PBS buffer (c, d).(a, c) the first frame and (b, d) reconstruction (time projection) image of 20,000 frames.Scale bar: 5 μm.Color bar: photons per pixel (pixel size 100 nm).

Figure S12 .
Figure S12.Histograms of blinking on time per switching event of ThT that has labelled α-Syn fibrils measured in air (a) and in PBS buffer (b) and single-exponential fit, mean blinking time were determined by the exponential fit.

Figure S13 .
Figure S13.Blinking properties of ThT that has labelled α-Syn fibrils measured in PBS buffer where fibrils were not washed properly.(a) Histogram of detected photons per switching event and single-exponential fit.(b) On-off duty cycle of ThT calculated from single-molecule fluorescence time traces; (c) histogram of blinking on time per switching event and singleexponential fit, mean blinking time were determined by the exponential fit.(Compare to Fig. 3 c,d in main text).

Figure S15 .
Figure S15.Survival fraction of ThT molecules measured in different environment.The first two rows are from individual ThT molecules (10 -12 M), while the latter two rows are from ThT molecules on fibrils.A single molecule was defined to be bleached/entering long term fluorescence-off state after its final on-switch.The time-dependent survival fraction was calculated as the number of molecules that are not yet bleached/ entering long term fluorescenceoff state divided by the total number of detected molecules.Note that we report the survival fraction for the total imaging time of 600 seconds long.

Figure S15 .
Figure S15.Upper: Representative wide field images at three selected time point as pointed by the red lines; lower: number of detected points per image frame versus the time.The sample (ThT embedded in PS film in N2) was illuminated continuous with 488 nm laser (5.1 kW cm −2 , green line), camera exposure time 30 ms.After about 600 s imaging, an additional 405 nm excitation (0.30 kW cm −2 , blue line) was administered and repeated stochastically.Scale bar: 5 μm.Color bar: photons per pixel (100 nm).

Figure S18 .
Figure S18.Number of detected points per image frame versus the time.The sample (ThT embedded in PS film in N2) was illuminated with 488 nm laser (2.23 kW cm −2 , green line), camera exposure time 30 ms.After about 900 s continuous imaging, 488 nm laser was turned off for a period of time (duration of 14 s, 23 s, 99 s and 197 s) and then was turned on for imaging again (imaging time of 5 -8 s), and about 7%, 9%, 15% and 17% of molecules were detected compared with the first frame, respectively.

Table S1 .
molecules.Comparing the first frame and reconstruction image, the ratio of points detected in the first frame was 22% It should be noted here, the samples used in this figure were prepared using the same protocol as described in SI ''Preparation of ThT single molecule samples'', but with ~1/5 concentration of ThT to avoid possibility of aggregation/overlap of molecules.From the summarized results in TableS1below, the ratio of points detected in the first frame are not affected by the oxygen too much.Meanwhile, the blinking properties (photons and duty cycle) of ThT vary a little bit in different environments, but the impact on SMLM imaging quality is negligible.Summary of blinking properties of ThT measured in different environments.
a ThT embedded in PS film prepared in N2 glove box, average of 3 measurements.b Averaged of 4 measurements.c From 1 measurement.

Table S2 .
Summary of blinking properties of ThT measured in different environments.Air a Air b O2 free b Thiol b Radical scavenger b