Representative maximum‐intensity projection of a nucleus in a live zebrafish embryo (sphere stage), where Pol II was detected via fluorescently labeled antigen‐binding fragments (Fab) specific against Ser5 and Ser2 phosphorylation of the C‐terminal domain heptad repeat (Pol II Ser5P, Pol II Ser2P). Pol II Ser5P clusters representing the different apparent types of morphologies are marked. Single time point z‐stacks were recorded using an instantaneous Structured Illumination Microscope (instant‐SIM), raw data were processed by local background subtraction (both channels) and smoothed (Pol II Ser2P only).

Detail views of the clusters of the apparent morphology types i–iii, as marked in panel (B). The detail views are 2.7 μm across.

Examples of the varied morphologies of type iii clusters, shown as maximum‐intensity projections and corresponding volume renderings of the processed Pol II Ser5P signal. Morphologies are named by similarity to patisserie and candy items. Scale bar: 500 nm. 3D renderings: ImageJ Volume Viewer plugin.

Detail views of the marked clusters, representing the typical morphologies i–iii.

Area and solidity of individual clusters, with gate regions for the typical morphologies i–iii. Clusters were segmented based on Pol II Ser5P intensity, data obtained from a total of 52 mid‐nuclear sections from two different samples.

The Pol II Ser5P and Pol II Ser2P intensities (mean intensity across all pixels inside a given cluster's segmentation mask) of the clusters in the gates i–iii are plotted in color over the entire ungated cluster population (light gray). Mean Pol II Ser5P and Pol II Ser2P intensities were scaled by the median value for each nucleus, then pooled.

The median of the Pol II Ser5P and Pol II Ser2P levels of the gated clusters in panel D is plotted over the ungated population of clusters. Each cluster type is plotted with 95% bootstrapped confidence intervals in Pol II Ser5P and Pol II Ser2P direction (10,000 resamples).

Analysis of the placement of Pol II Ser2P spots relative to Pol II Ser5P clusters of type ii or type iii (spots segmented based on Pol II Ser2P channel). Lines represent histograms of the distances to the nearest surface of a Pol II Ser5P cluster. The Euclidean metric was used to calculate distance of Ser2P foci gated pixels from nearest cluster. To display relative enrichment against random placement, histogram bin counts are normalized by counts for ungated pixel distances, then scaled by the mean across bins.

Sketch of apparent morphology types of Pol II Ser5P clusters, placed by their levels of elongating and recruited Pol II.

Quantification of changes in Pol II Ser5P cluster morphology upon hexanediol treatment. Area displayed as standard boxplots, solidity and the number of clusters per nucleus are mean±SEM. *** indicates P < 0.001 (two‐tailed permutation test for differences upon hexanediol treatment from control, N = 5, 3 independent samples; area: P < 0.0001 with n_Cluster = 1,435, 841 clusters for area; number of large clusters (area > 0.2 μm²) per nucleus: P < 0.0001 with n_Nuc = 212, 124 nuclei; solidity: P < 0.0001 with n_Large = 401, 135 large clusters).

Pol II Ser2P channel micrographs of the same nuclei shown in panel A.

Quantification of changes in Pol II Ser2P spots. n.s. indicates no statistically significant changes (area: P = 0.76 with n_Spots = 3,394, 1,937 spots; number of clusters per nucleus (no size cut‐off): P = 0.80 with n_Nuc = 212, 124 nuclei; solidity: P = 0.21 with n_Spots = 3,394, 1,937 spots).

Representative time‐lapse recording of Pol II Ser5P Fab in a live embryo (no hexanediol treatment). Similar results were observed in two independent experiments, each performed on three different embryos. Images are maximum‐intensity projections, and images recorded by instant‐SIM.

Representative close‐up time‐lapse showing transient merging and separation events within a Pol II Ser5P cluster. Single z‐sections, images were bleaching‐corrected and local background was subtracted (radius 3.3 μm). Images from same data set as panel (E). Contour plots are obtained by application of a manually adjusted threshold to assist interpretation, same threshold for all time points.

Example view of a genomic region containing a gene with high Pol II Ser5P levels at the promoter (rnf19a).

Overview of Pol II Ser5P and H3K27ac ChIP‐seq levels for 50‐kb windows covering the entire genome, windows covering regions containing super‐enhancers, and symbols indicating regions covered by specific oligopaint probe sets.

Top row—representative optical section obtained by three‐color STED microscopy from a fixed zebrafish embryo, showing Pol II Ser5P and H3K27ac (immunofluorescence) and DNA (JF646‐Hoechst). Bottom row—Detail views as marked in the top row. The details views are 2.4 μm across.

Mean fluorescence intensities (normalized against median intensity of a given nucleus) within Pol II Ser5P clusters. Overall levels of H3K27ac and DNA inside clusters were determined (values for individual clusters in gray, median with 95% bootstrap confidence interval in black) as well as correlations between H3K27ac intensity and Pol II Ser5P intensity, cluster area, and cluster solidity (Pearson correlation coefficient, red line is a linear fit to guide the eye). n = 470 clusters and n = 130 large clusters (area > 0.08 μm², used for solidity analysis) were extracted from images of N = 29 nuclei obtained from five embryos.

Assessment of Pol II Ser5P and Pol II Ser2P immunofluorescence signal intensities at the location of super‐enhancer probe sets. Genomic target regions are sorted left‐to‐right by descending Pol II Ser5P signal. Intensities are normalized by the whole nucleus median level, shown is the median with 95% bootstrap confidence intervals. N = 2 embryos imaged per probe set, yielding n = 67, 66, 91, 63, 53, 70, 93, 63, 71, 78, 59, 58, 103, 55, 65, 127 oligopaint foci for the probe sets SE1, SE2, SE3, SE4, SE5, SE6, SE7, SE8, SE9, SE10, SE11, SE12, cdc25b, celf1, crsp7, rnf19a, respectively. Each probe set contained approximately 500 probe sequences, covering 50 kb sequence length, images acquired by instant‐SIM (genomic regions see Appendix Fig S8, probe sets see Materials and Methods).

Representative images of indicated oligopaint probe sets and immunofluorescence signal, two examples are shown for SE5 and crsp7 to illustrate positioning outside a Pol II Ser5P cluster and inclusion into a cluster. Shown are single z‐sections with segmentation outlines of the oligopaint signal.

Examples of lattice configurations obtained from simulations containing only red particles with increasing self‐affinity (w_S5P‐S5P, as indicated). Lattice simulations containing a polymer chain of length L_polymer = 20 with N_IC = 12 black monomers (black‐black affinity w_IC‐IC = −0.5) and N_RC = 8 blue monomers (blue‐red affinity w_RC‐S5P = −0.5, adjustment see Appendix Fig S10), same w_S5P‐S5P values as for simulations without polymer chain. All simulations on 25‐by‐25 lattices, N_S5P = 100 red particles, for adjustment of N_S5P, see Appendix Fig S10A.

Interaction matrix for different species in the lattice model. Affinity is represented by negative and repulsion by positive values.

Long time behavior (total of 1 × 10⁷ iteration steps) of the model shown as lattice output and synthetic microscopy images, 30‐by‐30 lattice, N_S5P = 140, nine chains, extent of blue and gray regions chosen randomly for each chain at initialization.

Time‐lapse showing transient separation and merging events of a S5P cluster. 25‐by‐25 lattice, N_S5P = 100, four chains.

Simulated hexanediol treatment. 25‐by‐25 lattices, N_S5P = 100, three chains, N_RC = 8, N_AC = 6, modified parameter values in hexanediol simulations: w_S5P‐S5P = −0.15, w_S5P‐RC = −0.25.

Correlations between H3K27ac intensity (RC) and cluster area (in units of lattice cells, n = 1,132) and cluster solidity (only including clusters with area greater 50, n = 551, ρ is the Pearson correlation coefficient, red line—linear fit to guide the eye). Analysis based on 30 simulations of 25‐by‐25 lattices, N_S5P = 100, four chains per lattice, extent of blue and gray regions randomly assigned per chain at initialization of each simulation.

Area and solidity of individual clusters, with gate regions for the typical morphologies i–iii. Clusters were segmented based on Pol II Ser5P intensity (total 11,248 clusters). For each simulation, four chains with regions of randomly assigned length of N_RC∈{0,2,4,6,8} blue monomers and N_AC∈{0,3,6} gray monomers were placed, total chain length L_polymer = 20. 2,808 clusters in gate i, 1,021 clusters in gate ii, and 2,469 clusters in gate iii.

The Pol II Ser5P and Pol II Ser2P intensities (mean intensity across all pixels inside a given cluster's segmentation mask) of the clusters in gates i–iii (color), ungated cluster population in gray. Intensities were scaled by the population median.

Pol II Ser5P and Pol II Ser2P levels of the gated clusters in panel (C) plotted over the ungated population of clusters, median with 95% bootstrap confidence interval. For all types (i–iii), the confidence interval is hidden by the median data point.

Pol II Ser5P and Pol II Ser2P intensities at Pol II Ser5P clusters and throughout entire nuclei, standard boxplots. *** indicates P < 0.0003, * indicates P < 0.017, n.s. indicates P ≥ 0.017, significance levels Bonferroni‐corrected for multiple testing (two‐tailed permutation test for differences from control, data obtained from three independent sets of experiments; Pol II Ser5P at clusters: P = 0.08, P < 0.0001, n = 1,534, 716, 1,682 clusters; Pol II Ser2P at clusters: P < 0.0001, P < 0.0001, n = 1,534, 716, 1,682 clusters; Pol II Ser5P throughout nuclei: P = 0.17, P = 0.014 with n = 165, 148, 118 nuclei; Pol II Ser2P throughout nuclei: P < 0.0001, P = 0.0001 with n = 165, 148, 118 nuclei; data obtained from three independent experiments). For additional properties and actinomycin D treatment, see Appendix Fig S14A.

Examples of lattice configurations obtained from simulations with modifications that mimic inhibitor treatments. For flavopiridol treatment, no gray regions were assigned. For triptolide, no gray regions were assigned, the Pol Ser5P self‐affinity (w_S5P‐S5P = −0.25) and affinity to regulatory regions (w_RC‐RC = −0.25) were reduced, and the number of red particles was increased from 100 to 120.

Cluster solidity and the number of clusters obtained from lattice simulations and microscopy images of cell cultures, mean±SEM. Quantification was based on synthetic microscopy images, n = 1,000 images analyzed per condition. For lattice simulations, *** indicates P < 0.0005; for cell cultures, *** indicates P < 0.0003, ** indicates P < 0.003 (simulations, solidity: P < 0.0001, P < 0.0001 with n = 1,000, 1,000, 1,000; simulations, number of clusters: P < 0.0001, P < 0.0001 n = 1,000, 1,000, 1,000; cell culture, solidity: P < 0.0001, P = 0.0007 with n = 1,514, 703, 1,631 clusters; cell culture, number of clusters: P < 0.0001, P = 0.001 with n = 165, 148, 118 nuclei).