Fig. 3: TAD anchors are brought together at rates of ~0.1 kb s⁻¹ in living cells.

a, Cohesin motor speed and CR are two different measures of extrusion speed. The cohesin motor speed is the number of DNA base pairs extruded by a single cohesin complex per unit time on DNA devoid of obstacles such as CTCF sites (left). The CR measures the rate at which the effective loop-free DNA between TAD anchors—the shortest 1D path between them (pink)—decreases due to extrusion (right). The CR reflects the action of single or multiple cohesin complexes extruding simultaneously, as well as cohesin stalling at internal CTCF sites. Cohesin complexes not involved in the shortest 1D path do not contribute to the CR. b, Distance time series of the T1 TAD without auxin treatment before (left) and after (right) random shuffling of time points. Indigo bars indicate segmented proximal state intervals. Dotted lines indicate alignment of the starting times of proximal states (\({t}_{{\rm{start}}}\)) across single time series. c, Fitting strategy to determine CRs. A piecewise linear model with three parameters (\({R}_{{\rm{plateau}}}^{2},\,{R}_{\mathrm{int}}^{2},\,{t}_{{\rm{extr}}}\)) is fitted to the mean squared 3D anchor–anchor distances as function of time after alignment to \({t}_{{\rm{start}}}\). The slope of the linear decrease defines the CR. d, Aligned time series of mean-squared 3D distances measured on the T1 TAD, weighted by localization precisions, before (red, n = 46 time series) or after (olive, n = 19 time series) random shuffling. The piecewise linear fit is shown as a dashed line. The shaded area represents the weighted s.e.m. e, CRs estimated from time series of untreated cells. Box plot lines show the median, lower and upper quartiles, and whiskers extend from 10th to 90th percentiles; n indicates the number of bootstrap samples including at least 15 time series and associated with nonconstant distances out of a total of 5,000 bootstrap samples.