Nanoscale DNA tracing reveals the self-organization mechanism of mitotic chromosomes

Abstract

How genomic DNA is folded during cell division to form the characteristic rod-shaped mitotic chromosomes essential for faithful genome inheritance is a long-standing open question in biology. Here, we use nanoscale DNA tracing in single dividing cells to directly visualize how the 3D fold of genomic DNA changes during mitosis at scales from single loops to entire chromosomes. Our structural analysis reveals a characteristic genome scaling minimum of 6-8 megabases in mitosis. Combined with data-driven modeling and molecular perturbations, we can show that very large and strongly overlapping loops formed by condensins are the fundamental structuring principle of mitotic chromosomes. These loops compact chromosomes locally and globally to the limit set by chromatin self-repulsion. The characteristic length, density, and increasingly overlapping structure of mitotic loops we observe in 3D fully explain how the rod-shaped mitotic chromosome structure emerges by self-organization during cell division.

Keywords: cell division; chromatin tracing; chromosome compaction; condensins; genome organization; loop extrusion; mitosis.

Figure 1.. Multiscale DNA tracing from interphase to metaphase

(A) Schematic of non-denaturing FISH scheme and multiscale probe libraries targeting chromosome (chr) 2, which was decoded using a combination of 15-Mb segment barcode and 1-Mb resolution spot barcodes. Full chromosome libraries were supplemented with intermediate scale (30-kb spots every 200 kb for 10–12 Mb) and high-resolution (12-kb spots contiguously tiled for 1.2 Mb) libraries targeting the same and different (chr5, chr14, chr18) chromosomes. (B) Schematic of multiscale chromosome trace acquisition. Whole chromosomes were segmented based on the full library signal, while sequential images of 15-Mb segments and 100-kb spots were used to decode and uniquely identify and 3D localize all detected spots. Intermediate and high-resolution libraries were segmented by regional barcodes and traced by sequential spot fitting in 3D. (C) Exemplary maximum-intensity projected micrographs of cell nuclei (DAPI, gray) with full chromosome traces (color-coded sequential 15-Mb segments). Arrowheads indicate fiducial beads used for drift correction. (D) Reconstructed 3D traces (multi-colored) from libraries targeting whole chr2 and intermediate- and high-resolution regions. Data representative of in total 8,879 traces in 1,396 cells in 5 independent experiments. See also Figure S1 and Video S1.

Figure 2.. Mitosis-specific chromatin loop signatures

(A) Illustration of single-trace metrics. Euclidean distance-based filtering of close contacts (<100 nm) with a genomic distance over 30 kb defines “loops.” Based on their overlap with other loops or loop bases, contacts are classified into base loops, nested loops (when 2 or more base loops coincide to form a second, larger loop), or Z-loops (2 loops partially overlap). Radius of gyration is a measure of overall compaction of the trace. (B) Trace metrics calculated for the high-resolution (12 kb) 1.2-Mb region on chr5:149,500,723–150,699,962. Trace metrics were calculated from traces that were more than 80% complete. Data from n = 278 (510), n = 22 (44), n = 124 (297), and n = 152 (398) for interphase, prophase, prometaphase, and metaphase cells (traces) from 3 independent experiments. Median, quartiles, and whiskers are shown in the plots. See also Figure S2.

Figure 3.. Mitosis-specific genomic distance scaling signatures

(A–C) Traces of a high-resolution region (chr5, A), intermediate scale (chr2, B), and whole chr2 (C) and corresponding pairwise distance scaling plots. Exemplary pairwise distances (black) for a single spot (red) within the DNA trace are highlighted. Inset with arrow highlights the characteristic scaling dip at 6–8 Mb. Plots show median ± standard error of the mean. Data from (A) n = 400 (824), n = 39 (109), n = 241 (609), and n = 244 (644), from 3 experiments; (B) n = 165 (237), n = 27 (44), n = 175 (325), and n = 187 (346), from 2 experiments; and (C) n = 331 (417), n = 28 (46), n = 2 41 (414), and n = 212 (357), from 3 experiments; interphase, prophase, prometaphase, and metaphase cells (traces). As the scaling data become very sparse at maximal genomic trace distance, the scaling plots were cropped to 1, 10, and 100 Mb, respectively. See also Figure S3.

Figure 4.. Mitotic chromosome structural features depend on condensins

(A) Wild-type (WT) HeLa Kyoto cells and HeLa Kyoto cells with SMC4-mAID-Halo acutely depleted for 3–4 h with 5-Phenyl-indole-3-acetic acid (5-Ph-IAA) before mitotic entry (ΔSMC4). Exemplary maximum-intensity projected micrograph of metaphase chromatin (DAPI, gray) with full chr2 traces (DNA-FISH, multi-colored). Data representative of 212 WT metaphase traces from 3 independent experiments and 84 metaphase ΔSMC4 traces from 2 independent experiments. (B) Trace metrics from chr5:149,500,723–150,699,962 (1.2 Mb, 12-kb resolution) for WT and ΔSMC4 cells. Data from 152 (398) WT cells (traces), 3 independent experiments, and 153 (265) ΔSMC4 cells (traces), 2 independent experiments. Median, quartiles, and whiskers are shown in the plots. (C) 3D DNA traces for chr5, 1-Mb scale (12-kb resolution); chr2, 10-Mb scale (200-kb resolution); and whole chr2 (1-Mb resolution) from WT and ΔSMC4 cells at metaphase. Data representative of chr5, 1 Mb: 152 (398) WT cells (traces), 3 independent experiments, and 153 (265) ΔSMC4 cells, 2 independent experiments; chr2, 10 Mb: 159 (286) WT cells (traces), 2 independent experiments, and 106 (153) ΔSMC4 cells (traces), 2 independent experiments; chr2 whole: 130 (212) WT cells (traces), 3 independent experiments, and 51 (62) ΔSMC4 cells (traces), one experiment. (D) Distance scaling plots for chr5, 1-Mb scale; chr2, 10-Mb scale; and whole chr2 at metaphase. Data from chr5, 1 Mb: n = 400 (824) WT interphase cells (traces) and 244 (644) WT metaphase cells (traces), 3 independent experiments, and 212 (608) ΔSMC4 metaphase cells (traces), 2 independent experiments; chr2, 10 Mb: n = 165 (237) WT interphase cells and 187 (346) WT metaphase cells (traces), 3 independent experiments, and 124 (193) ΔSMC4 metaphase cells (traces), 2 independent experiments; chr2, whole: n = 331 (417) WT interphase cells (traces) and 212 (357) WT metaphase cells (traces), 3 independent experiments, and 206 (355) ΔSMC4 metaphase cells (traces), 2 independent experiments. As the scaling data become very sparse at maximal genomic trace distance, the scaling plots were cropped to 1, 10, and 100 Mb, respectively. See also Figure S4.

Figure 5.. Loop extrusion and self-repulsion correctly predict mitotic chromosome structure

(A) Polymer simulation of mitotic progression for a 100-Mb chromosome sampled at the indicated time points, displayed at 2-kb resolution (multi-colored, thin line) and as a 1-Mb rolling average (gray, thick line). Data representative of 20 dynamically simulated chromosomes. (B) Condensin I (magenta) and condensin II (green) loop lengths shown for a 10-Mb stretch of the simulation example shown in (A). Loop height scales with length for clarity. (C) Reconstructed traces from simulated regions from 100-Mb chromosomes at metaphase (40 min) and corresponding experimental data (chr5 1-Mb scale, chr2 10-Mb scale, chr2 100 Mb from q-arm). Simulated traces were sampled as experimental data (12-kb tiled probes, 30-kb probes with 200-kb resolution and 100-kb probes with 1-Mb resolution). Simulated data representative of 20 simulated 100-Mb chromosomes and experimental data representative of chr5, 1 Mb: 152 (398) cells (traces); chr2, 10 Mb: 159 (286) cells (traces); chr2, 100 Mb: 130 (212), 3 independent experiments. (D) Distance scaling plots for simulated metaphase chromosomes and corresponding experimental data. Simulated chromosomes were sampled as in (C), either 10 times (10- and 100-Mb scales) or 60 times (1-Mb scale) with different starting positions in each of 20 100-Mb simulated chromosomes for representative sampling. Experimental data from chr5, 1 Mb: n = 244 (644) cell (traces), 3 independent experiments; chr2, 10 Mb: n = 187 (346) cell (traces), 2 independent experiments; and chr2q, 100 Mb: n = 212 (686) cell (traces), 3 independent experiments. See also Figure S5 and Video S2.

Figure 6.. Global compaction is influenced by self-repulsion

(A) WT HeLa Kyoto cells and HeLa Kyoto cells treated with TSA before mitotic entry. Exemplary maximum-intensity projected micrograph of metaphase chromatin (DAPI, gray) with full chr2 traces (DNA-FISH, multi-colored). Data representative of 212 WT metaphase cells from 3 independent experiments and 205 prometaphase/metaphase TSA-treated cells from 2 independent experiments. (B) Trace metrics from chr5:149,500,723–150,699,962 (1.2 Mb, 12-kb resolution) for WT and TSA-treated cells. Data from 152 (398) WT cells (traces), 3 independent experiments, and 67 (168) TSA-treated cells (traces), 2 independent experiments. Median, quartiles, and whiskers are shown in the plots. (C) Reconstructed traces for chr5, 1-Mb scale (12-kb resolution); chr2, 10-Mb scale (200-kb resolution); and whole chr2 (1-Mb resolution) from WT and TSA-treated cells at metaphase. Data representative of chr5, 1 Mb: 152 (398) WT cells (traces), 3 independent experiments, and 67 (168) TSA-treated cells, 2 independent experiment; chr 2, 10 Mb: 159 (286) WT cells (traces), 2 independent experiments, and 69 (126) TSA-treated cells (traces), 2 independent experiments; chr2 whole: 130 (212) WT cells (traces), 3 independent experiments, and 19 (31) TSA-treated cells (traces), two independent experiments. (D) Distance scaling plots for chr5, 1-Mb scale; chr2, 10-Mb scale; and whole chr2 at metaphase of TSA-treated cells together with corresponding WT and ΔSMC4 data. Data from chr5, 1 Mb: 71 (198) TSA-treated cells (traces), 2 independent experiments; chr2, 10 Mb: 73 (135) TSA-treated cells (traces), 2 independent experiments; and chr2, whole: 24 (69) TSA-treated cells (traces), 2 independent experiments. See Figures 3 and 4 for details on WT and ΔSMC4 data. As the scaling data become very sparse at maximal genomic trace distance, the scaling plots were cropped to 1, 10, and 100 Mb, respectively. (E) Distance scaling plots from simulated 100-Mb WT and TSA-treated chromosomes, sampled as the corresponding experimental data (see D for details). TSA treatment was simulated by increasing the repulsive potential between monomers. Simulated data from 20 dynamically simulated chromosomes per condition, sampled at metaphase (40 min). See also Figure S6.