(A) Sketch illustrating stroboscopic paSMT. Sketch illustrating labeling Halo-tagged proteins, e.g. CTCF or Rad21, with PA-JF646. This dye remains dark until 405 nm activation, which converts it to regular fluorescent JF646. The advantage is that thousands of single-molecule trajectories can be recorded from a single cell at a density of ~0.5 fluorescent molecules per nucleus per frame, which makes tracking unambiguous, by using very low intensity 405 nm excitation. Since high 633 nm laser powers are used, most molecules bleach within 3–8 frames. We use PA-JF646 instead of PA-JF549 since the red-shifted 633 nm laser induces less photo-toxicity, although the displacement histograms were identical between PA-JF549 and PA-JF646. Moreover, we never record for more than 2 min per cell. We observed no visible signs of photo-toxicity after 2 min of paSMT. Below, sketch illustrating stroboscopic illumination. To minimize ‘motion-blurring’ of fast-diffusing molecules, we used pulsed 633 nm excitation with 1 ms pulses. The camera integration time was 4 ms + ~0.447 ms (frame-transfer mode) resulting in a frame rate of roughly 225 Hz. Below, raw microscopy images demonstrating that even fast-diffusing molecules can be imaged and tracked (red lines) at high-signal-to-background. (B) Overview of two-state dynamic displacement model. Full details are provided in the Materials and methods. Briefly, the model assumes molecules can exist in either a chromatin associating (specific and non-specific) state called ‘bound’ or in a free 3D diffusion state called ‘free’. A mathematical model describing how the distribution of displacements, r, depends on the time delay, fraction bound, diffusion constants, localization error and axial detection slice is shown below. Overall, the model contains three fitted parameters, which were estimated using least squares fitting to the raw displacements considering the first seven displacements (Δτ ~4.5 ms to 31.5 ms). For ease of visualization, we show displacement histograms in (C–F), but the fitting was performed on cumulative distribution functions (CDFs) to minimize binning artifacts. (C–F) displacement histograms for various cell lines all measured using the approach in (A–B). For ease of visualization, the displacement histograms are cut off at 1050 nm, but longer trajectories were included in the model fitting. (C) shows Halo-only and Halo-3xNLS in mESCs, which show negligible binding. Note, that most fast-diffusing molecules eventually move out of the focal plane. (D) shows various Halo-mCTCF constructs. C59 and C87 are endogenous Halo-mCTCF knock-ins. pL30-wt-Halo-mCTCF was transiently expressed using a weak promoter (L30). Compared to Halo-mCTCF overexpressed by the strong CMV promoter, overexpressing Halo-mCTCF using a weak promoter (L30) causes only a minor (10 percentage points; likely due to saturation of binding sites) underestimation of the fraction bound. Right, two transiently transfected Halo-mCTCF mutants: 11ZF-mut-Halo-mCTCF is a CTCF mutant with an essential His amino acid in all 11 zinc-fingers mutated to Arg, which should abolish specific DNA-binding. We used this mutant to estimate the non-specifically bound fraction. ΔZF-Halo-mCTCF has the entire 11-zinc-finger domain deleted. We used this mutant to verify that the zinc-finger domain solely is responsible for chromatin association. (E) H2B-Halo and Rad21 experiments in mESCs. We used H2B-halo as a control for a protein that is almost exclusively bound. Note that since we use an EF1a promoter to express H2B, which is not cell-cycle regulated, some H2B-molecules do show free diffusion. mRad21-Halo in S/G2 and G1 are also shown, as is transiently transfected wt-mRad21-Halo expressed using the low-expression promoter, L30. Even though mRad21-Halo is only weakly overexpressed, most molecules show free 3D diffusion. The overexpression artifact may be caused by the fact that without similar overexpression of Smc1, Smc3 and SA1/2, most Rad21 cannot form cohesin complexes. Finally, a Rad21-mutant (F601R, L605R, Q617K), which cannot form cohesin complexes was used to estimate the non-specifically bound Rad21 fractions. (F) Stroboscopic paSMT experiments in human U2OS cells. H2B-Halo and Halo-3xNLS were used as controls for mostly bound and free molecules and the same zinc-finger mutants as in mESCs were transiently transfected as control for non-specific chromatin association. Note that C32 Halo-hCTCF shows slightly more free diffusion than C59 and C87 in mESCs.