STED imaging

Data from non-expanded samples were acquired using a Leica TCS SP8 STED 3X microscope with a pulsed (80MHz) white-light laser, HyD detectors and spectroscopic detection using HC PL APO 100×/1.40 Oil STED WHITE (Leica 15506378) oil-immersion objective. For Abberior STAR 635P and Alexa 594 we used 633 nm and 594 nm laser lines for excitation and a 775 nm synchronized pulsed laser for depletion, with a time gating range of 0.3–7 ns. For Alexa 488 we used 488 nm excitation, 592 nm continuous depletion laser line and time gate of 1.1–7 ns. Emission detection windows were 500–560 nm, 605–630 nm and 640–750 nm for Alexa 488, Alexa 594 and Abberior STAR 635P, respectively. No bleed-through was observed between the channels. For three-color cell body imaging (Figure 2, Figure 3), each fluorescent channel was imaged using the 2D STED configuration (vortex phase mask) in sequential z-stack mode from highest to lower wavelength, to prevent photobleaching by the 592 nm depletion laser line. For two-color imaging of dendrites (Figure 1), we used the Abberior STAR 635P/Alexa 594 combination and a single 775 nm depletion line and therefore acquired images in line-sequential mode. For the 3D STED imaging, we used a combined depletion PSF light path consisting of a mixture 60% Z-donut and 40% vortex phase mask, providing approximately isotropic resolution.

For the data shown in Figure 2, Figure 3, the size of the field-of-view was in the range of 30–50 µm and it was positioned to include the whole cell body of a neuron (soma) and the first 5–10 µm of dendrites emanating from it (Figure 2A, Figure 3A). For the data shown in Figure 1, the size of the field-of-view was in the range of 50–100 µm and it covered 30–50 µm of the proximal dendrites. The depth of z-stacks varied in the range from 3 to 6 µm for each acquisition and for all cases it was chosen to fully cover the dendrite’s thickness. The lateral pixel size was in the range of 27–30 nm with a distance between z-planes in the range of 150–160 nm. The z-stacks were subjected to a mild deconvolution using Huygens Professional software version 17.04 (Scientific Volume Imaging, The Netherlands) with CMLE (classic maximum likelihood estimation) algorithm with parameters of SNR (Signal-to-Noise Ratio) equal to 7 over 10 iterations. After the deconvolution, z-stacks of tyrosinated and acetylated channels were registered in 3D to total tubulin channel using maximum intensity projections in XY and XZ planes using Correlescence plugin v.0.0.4 (https://github.com/ekatrukha/Correlescence archived on Zenodo repository https://doi.org/10.5281/zenodo.4534715) for ImageJ.