SMLM Imaging

Single molecule measurements were performed on a custom built wide-field setup. Continuous-wave lasers of 405, 647 (Toptica Photonics, Germany), and 532 nm (Cobolt, Sweden) were used for activation and excitation. The laser lines were combined in an Acousto-optical tunable filter (AOTF; Gooch & Housego, United Kingdom), passed through a clean-up filter (AHF, Germany), and focused on the back focal plane of an oil immersion objective (60× APO TIRF, NA 1.49 Oil, Nikon) to create an oblique illumination. The fluorescence emission was separated first from the excitation light with dichroic mirrors and later into two different detection channels by combination of dichroic mirrors and emission filters (all from AHF Analysentechnik, Germany), which were then measured on separate EMCCD cameras (iXon 897 and iXon Ultra 897 Andor Technology, United Kingdom). The z-drift of the samples was corrected with a home built auto-focus system (Hellen and Axelrod, 1990). The samples used for SMLM were always cultured on a high precision glass coverslip of 170 μm thickness (Carl Roth).

The method of dSTORM was used for obtaining two-color sub-diffraction resolution images. For imaging, the samples were placed in a special oxygen scavenger buffer with a pH of 7.4 containing 100 U/ml glucose oxidase, freshly prepared 100 mM cysteamine, 400 U/ml of catalase, and 40 mg/ml of glucose (all Sigma–Aldrich, Germany) mixed in degassed PBS. For both channels, a series of 30,000 images were recorded at 20 ms exposure time, with the red channel (Alexa Fluor^®647) measured first followed by the green (Alexa Fluor^®532).

Material preparation and DNA-points accumulation for imaging in nanoscale topography (DNA-PAINT) techniques described in Schnitzbauer et al. (2017) were adapted for this study. For SR color multiplexing, samples were immunolabeled with primary antibody as described in the section “Immunocytochemistry.” Secondary antibodies against guinea-pig and mouse were tagged with different DNA docking strands and their respective complimentary sequences, labeled with Cy3B dye, were used as imager strands. The DNA sequence P6 (5′-TTTTAGGTAAA-3′) and P9 (5′-TTAATTAGGAT-3′) were used, respectively, for mouse and guinea-pig antibodies (Schnitzbauer et al., 2017). In order to choose FUS-positive synapses, FUS was labeled with a normal anti-rabbit secondary antibody with Alexa Fluor^®647 and imaged with the above mentioned dSTORM technique (the section “dSTORM Technique”). Before measurement samples were incubated with gold beads (80 nm gold beads, BBI solutions, United Kingdom) for tracking the drift during imaging. First the SR images of the FUS protein in the red channel were obtained. The oxygen scavenger buffer was exchanged with an imager buffer (PBS with 500 mM NaCl, pH 7.2) containing the P6 imager strand and then 40,000 images were measured at 200 ms exposure time. The P6 imager strand was washed out with washing buffer (PBS, pH 7.2) till no trace of fluorescence was visible. The third color was measured in a similar manner with P9 imager strand.

All the single molecule data reconstruction was done with the MATLAB-based FIRESTORM software as previously described (Schoen et al., 2016). Briefly, the center of every single molecule localization was determined by their COM and several such signals were filtered based on intensity, asymmetry, and width thresholds. The selected localizations were reconstructed in a 10 nm pixel raster weighted by their intensity. Green channel images were then adjusted to the red channel based on a transformation function obtained from fluorescent beads imaged in both channels (100 nm four-color labeled beads, TetraSpek Microspheres, Life Technologies). For the dSTORM images, the xy-drift was corrected in the post-processing with a redundant cross correlation algorithm implemented in the FIRESTORM software (Wang et al., 2014). A drift correction function based on bead-tracking was used for the DNA-PAINT measurements.