DNA-PAINT Imaging.

DNA-PAINT imaging was performed on a commercial Nikon N-STORM 3.0 microscope featuring a Perfect Focus System and a motorized TIRF illuminator. DNA-PAINT was performed by using HILO with 15–30% of a 200-mW, 561-nm laser (Coherent Sapphire) using a CFI Apo TIRF 100× oil (N.A. 1.49) objective at an effective power density of ∼0.5–1 kW/cm². The 561-nm laser excitation light was passed through a clean-up filter (Chroma ZET561/10) and directed to the objective by using a multiband beam splitter (Chroma ZT405/488/561/647rpc). Emission light was spectrally filtered (Chroma ET600/50m) and imaged on an EMCCD camera (Andor iXon ×3 DU-897) with 16-µm pixels by using a CCD readout bandwidth of 3 MHz at 14 bit, 5.1 preamp gain, and no electron-multiplying gain on the center 256 × 256 pixels, resulting in an effective pixel size of 160 nm. A total of 15,000 100-ms frames were acquired for each image by using 1–3 nM of Cy3B-labeled 10mer oligo in 1× PBS solution + 125–500 nM NaCl. Gold nanoparticles (40 nm; no. 753637; Sigma-Aldrich) were used as fiducial markers to facilitate drift correction. Single-molecule localization events were identified by using in-house MATLAB software (71) that calls a 2D fitting algorithm (72). Individual localization events were blurred with 2D Gaussian functions whose “sigma” parameter was set according to the global drift-independent localization precision as determined by NeNA (73). NeNA values were as follows: Xq28– 5.6 nm sigma, 13.2 nm supported resolution; Xist RNA– 5.1 nm sigma, 12.0 nm supported resolution. One- and two-component Gaussian fits of the line traces presented in Fig. 4 E–H were calculated by using the “Gaussian Mixture Model” module in scikit-learn (55).