1. Genetics and Genomics
Download icon

Cohesin residency determines chromatin loop patterns

  1. Lorenzo Costantino
  2. Tsung-Han S Hsieh
  3. Rebecca Lamothe
  4. Xavier Darzacq
  5. Douglas Koshland  Is a corresponding author
  1. Department of Molecular and Cell Biology, University of California, Berkeley, United States
Research Article
Cite this article as: eLife 2020;9:e59889 doi: 10.7554/eLife.59889
8 figures, 1 table and 3 additional files

Figures

Figure 1 with 1 supplement
Micro-C XL reveals positioned loops in mitotic chromosomes.

(A) Micro-C XL maps chromatin conformation by detecting nucleosomes that interact in vivo. Schematic illustrates how two interacting nucleosomes (e.g. 2 and 6) are visualized on the contact map. (B) Illustration of different chromosomal structures and their corresponding signal on the contact map. Unstructured chromosomes: only neighboring nucleosomes will be crosslinked, producing a linear signal along the diagonal. Interaction Domain: the nucleosomes within a domain will be also crosslinked, forming a square along the diagonal. Positioned loop: the nucleosomes at the base of the loop (5’ and 3’ anchors) will be crosslinked, forming a spot away from the diagonal. These different structures can form concomitantly on chromosomes producing contact maps with squares and spots along the diagonal. (C) Comparison of contact maps from mitotically arrested cells produced with Micro-C XL (top) and Hi-C (bottom) (Schalbetter et al., 2019). Contact maps show successive zoom-ins of chromosome VII (from the full 1090 kb to a 90 kb region), across multiple resolutions (from 1.6 kb to 100 bp). All contact maps throughout the manuscript have the standard colormap scheme that uses the shades of red from white (no detectable interactions) to black (maximum interactions detected) in log10 scale. (D) Micro-C XL uncovers chromatin interactions below the kilobase range. Interactions-versus-distance decaying curve shows the normalized contact density (y-axis) against the distance between the pair of crosslinked nucleosomes from 100 bp to 1 Mb (x-axis) for Micro-C XL and Hi-C (Schalbetter et al., 2019). (E) Micro-C XL detects abundant mitotic positioned loops and their corresponding loop anchors. Venn diagrams show how Micro-C identifies 1276 anchors for defined loops that include most of the loop anchors identified by Hi-C. (F) Genome-wide average heatmaps show a sharp enrichment of positioned loop signal in Micro-C XL data. Heatmaps were plotted using the 200 bp resolution data for Micro-C XL and Hi-C. The plot is a piled-up of the ±5 kb region surrounding the loop anchors (black arrows) of every loop identified in Micro-C XL. Numbers in the corners represent the fold-change of the signal enrichment of the center pixel over the indicated corner pixels (142 pixels). The contact map was normalized by matrix balancing and distance, using a diverging colormap with positive signal enrichment in red and negative signals in blue (gradient of normalized contact enrichment is in log2).

Figure 1—figure supplement 1
Comparison of Micro-C XL and Hi-C mapping in budding yeast.

(A) Micro-C XL in mitotically arrested cells detects more positioned loops than Hi-C. Comparison of contact maps from mitotically arrested cells produced with Micro-C XL (top), Hi-C (center) (Schalbetter et al., 2019), and Hi-C (bottom) (Paldi et al., 2020). Contact maps show contacts between chromosomes VII to X at 6.4 kb resolution (first column). Contact maps show successive zoom-ins of chromosome VII (from the full 1090 kb to a 90 kb region), across multiple resolutions (from 1.6 kb to 100 bp). Standard colormap scheme that uses the shades of red from white (no detectable interactions) to black (maximum interactions detected) in log10 scale. (B) Micro-C detects positioned loops in mitotically arrested, but not in asynchronous cells. Comparison of contact maps from mitotically arrested cells with Micro-C XL (top), and unsynchronized cell with Micro-C XL (bottom) (Hsieh et al., 2016), Contact maps show contacts between the whole genome (first column), followed by contacts from chromosomes XII to XV at 6.4 kb resolution (second column). The following contact maps show successive zoom-ins of chromosome XII (from the full 924 kb to a 60 kb region), across multiple resolutions (from 1.6 kb to 100 bp). Standard colormap scheme uses the shades of red from white (no detectable interactions) to black (maximum interactions detected) in log10 scale.

Figure 2 with 2 supplements
Mitotic positioned loops are cohesin-dependent.

(A) Cohesin peaks colocalize with the loop anchors of positioned loops. Contact map showing the interactions in the 290–400 kb region of chromosome X overlays with Mcd1p ChIP-seq tracks on top and left. The dashed lines indicate the colocalization of anchors of positioned loops and Mcd1p peaks. (B) Cohesin binding is enriched at anchors for positioned loops genome-wide. Heatmap (bottom panel) shows the enrichment of the Mcd1p ChIP-seq reads at the loop anchors genome-wide (732 rows). Pairs of loop anchors were rescaled to the same length. Genome-wide average ChIP-seq signal over the rescaled loop anchors is plotted (top panel). (C) Mcd1p peaks center on the anchors for positioned loops. Heatmaps were plotted with 200 bp resolution of contact map signal in ±4 kb regions surrounding the paired Mcd1p peaks. The contact map is colored by the contact probability with >95% in black, 90–95% in red, 50–90% in yellow, 0–50% in shades of green. The average Mcd1p peak is overlaid with contact maps and centered at the loop bases. (D) The positioned loop signal on the contact map is lost upon cohesin depletion. Contact maps for WT and MCD1-AID Micro-C data show substantial loss of loop signal over chromosome X: 200–450 kb upon Mcd1p degradation. (E) Positioned loops are gone upon Mcd1p-depletion. The bar chart shows the loop number called by the HiCCUPS program in WT (733) and MCD1-AID (5). (F) Genome-wide analysis confirms that the positioned loop signal is gone in MCD1-AID cells. Heatmaps were plotted with 200 bp resolution data for WT and MCD1-AID in ±5 kb regions surrounding the wild-type loop anchors.

Figure 2—figure supplement 1
Mitotic loops are cohesin-dependent.

(A) Cohesin and condensin are efficiently depleted upon auxin addition. The depletion of Mcd1p and Brn1p tagged with AID upon auxin addition to the media is shown by western-blot analysis. Tubulin is used for loading control. (B) The majority of the positioned loops are cohesin-dependent. Contact maps from mitotically arrested cells were produced with Micro-C XL on wild-type (WT), cohesin Mcd1p-depleted (MCD1-AID), and condensin Brn1p-depleted (BRN1-AID) cells. Contact maps show contacts between the whole genome (first row), followed by contacts from chromosomes VIII to XI at 6.4 kb resolution (second row). The following contact maps show chromosome X and a zoomed-in region (from the full 745 kb to a 250 kb region), with 1.6 kb and 800 bp resolution, respectively. Standard colormap scheme that uses the shades of red from white (no detectable interactions) to black (maximum interactions detected) in log10 scale. (C) Almost no positioned loops are detected upon cohesin depletion, while condensin depletion presents a small reduction in the number of positioned loops. Bar chart shows the loop number called by the HiCCUPS program in wild-type (WT), cohesin Mcd1p-depleted (MCD1-AID), and condensin Brn1p-depleted (BRN1-AID) cells. (D) Genome-wide analysis confirms positioned loop signal is gone in cohesin-depleted cells and slightly reduced upon depletion of condensin. Heatmaps were plotted with 200 bp resolution data for wild-type (WT), cohesin Mcd1p-depleted (MCD1-AID), and condensin Brn1p-depleted (BRN1-AID), and cohesin- and condensin-depleted (MCD1-AID BRN1-AID) cells in ±5 kb regions surrounding the wild-type anchors for positioned loops. Numbers in the corners represent the fold-change of the signal enrichment of the center pixel over the indicated corner pixels (142 pixels). The contact map was normalized by matrix balancing and distance, using a diverging colormap with positive signal enrichment in red and negative signals in blue (gradient of normalized contact enrichment is in log2). (E) Cohesin depletion shows fewer contacts in the size-region corresponding to the size of positioned loops, while condensin depletion does not. Interactions-versus-distance decaying curve shows the normalized contact density (y-axis) against the distance between the pair of crosslinked nucleosomes from 100 bp to 1 Mb (x-axis) for wild-type (WT) in blue, condensin Brn1p-depleted (BRN1-AID) in yellow, and cohesin Mcd1p-depleted (MCD1-AID) in red. (F) Cohesin depletion results in more inter-chromosomes contacts. Bar chart shows the number of inter-chromosomes contacts in wild-type asynchronous (Async), wild-type (WT), cohesin Mcd1p-depleted (MCD1-AID), condensin Brn1p-depleted (BRN1-AID), and cohesin- and condensindepleted (MCD1-AID BRN1-AID) all arrested in mitosis.

Figure 2—figure supplement 2
Comparison of chromosome contacts upon cohesin depletion of Micro-C XL versus Hi-C.

(A) Micro-C detects more cohesin-dependent chromosome contacts with a smaller size-range than Hi-C. Interactions-versus-distance decaying curve shows the normalized contact density (y-axis) against the distance between the pair of crosslinked chromosomal loci from 3 kb to 1 mb (x-axis) for wild-type arrested in mitosis (WT in blue), and cohesin-inactivated cells arrested in mitosis (cohesin-inactivated in red) in our Micro-C XL, Hi-C in Schalbetter et al., 2017, and Hi-C in Dauban et al., 2020 (top). Plot of the variation of the slope in the interactions-versus-distance decaying curve for the same datasets (bottom). (B) Comparison of the plots of the variation of the slope in the interactions-versus-distance decaying curve for our Micro-C XL in wild-type (WT in blue) and without cohesin (cohesin-inactivated in red), and Hi-C in Dauban et al., 2020 in wild-type (WT in grey) and without cohesin (cohesin-inactivated in green). (C) Differential slopes of contact decay curves between cohesin inactivation and wild-type. Differential slopes were obtained by the subtraction of the slopes of cohesin inactivation to wild-type (y-axis) across the range of 1 kb to 50 kb (x-axis), showing the data from Micro-C XL in purple, Schalbetter et al., 2017 in green, and Dauban et al., 2020 in yellow.

Ribosomal DNA (RDN) structure in mitosis is dependent on cohesin and condensin.

(A) Model for RDN structure in interphase and mitosis. RDN (purple) is visualized as a dispersed puff separated from the rest of the genome (black). In mitosis, the RDN structures in a thin line. (B) Cohesin and Condensin accumulate in the 3’ region of each rDNA repeat. ChIP-seq signal of condensin (Brn1p) in yellow shows how condensin accumulates on the NTS1 region. ChIP-seq cohesin (Mcd1p) in red shows cohesin accumulation on the NTS1 and 2. Below a cartoon for the genomic organization of the rDNA repeat, where each repeat is made of two transcribed regions (RDN37 and RDN5) and two non-transcribed regions (NTS1 and NTS2). (C) Mitotic RDN structure resembles interphase upon cohesin or condensin depletion. Contact maps from one rDNA repeat from wild-type (WT) in interphase, followed by mitotically arrested cells in wild-type (WT), condensin Brn1p-depleted (BRN1-AID), cohesin Mcd1p-depleted (MCD1-AID), and cohesin- and condensin-depleted (MCD1-AID BRN1-AID) cells. Standard colormap scheme that uses the shades of red from white (no detectable interactions) to black (maximum interactions detected) in log10 scale. The contact fold-changes between mutants and wild-type were plotted using a diverging colormap with gaining contacts in mutants showing in red and losing contact in mutants showing in blue. Below the contact map, a cartoon for the genomic organization one RDN repeat and the ChIP-seq signal of condensin (Brn1p) in yellow and cohesin (Mcd1p) in red. (D) Cohesin and condensin depletion makes the rDNA less isolated from the rest of the genome. Normalized contacts between the rDNA and the rest of the genome in wild-type (WT), condensin Brn1p-depleted (BRN1-AID), cohesin Mcd1p-depleted (MCD1-AID). (E) Cohesin and condensin depletion increases the number of contacts between the RDN. Reads per million mapped reads (RPM) contacts between rDNA in wild-type (WT), condensin Brn1p-depleted (BRN1-AID), cohesin Mcd1p-depleted (MCD1-AID). (F) Cohesin and condensin depletion shows more long-range contacts on RDN. Interactions-versus-distance decaying curve shows the normalized contact density (y-axis) against the distance between the pair of crosslinked nucleosomes from 100 bp to 10000 bp (x-axis) for wild-type (WT), condensin Brn1p-depleted (BRN1-AID), cohesin Mcd1p-depleted (MCD1-AID), and condensin- and cohesin-depleted cells (BRN1-AID MCD1-AID).

Figure 4 with 1 supplement
The organization of cohesin-dependent loops.

(A) Features of chromatin loops organization from the contact map. Contact map showing the interaction in the 40–160 kb region of chromosome I overlay with the tracks for Mcd1p ChIP-seq signal and peaks. Squares indicate loops from adjacent CARs. Circles indicate loops from distal CARs. Dashed lines indicate the position of barriers that insulate from loop expansion. (B) Model for loops organization. Each CAR (Mcd1p peak) forms a loop with the CAR that follows and a loop with the CAR that precedes (Bidirectional loops from adjacent CARs). Some CAR can form loops with distal peaks (CAR0 with CAR+2, and CAR0 with CAR-2) (loop expansion). The high-residency CAR (peaks with the highest Mcd1p signal) are barriers to loop expansion. (C) The distribution of the size of positioned loops is correlated to the cohesin peak interval. The histogram shows the probability of the length distribution for positioned loop size and Mcd1p peak interval in the range from 0 to 40 kb. The inset table highlights the median length. (D) The positioned loop signal is detected till the +5 CARs interval genome-wide. Heatmaps were plotted with 200 bp resolution data for WT. A ±5 kb region surrounding each loop anchor and the corresponding CARs interval from +1 to +10. (E) The levels of cohesin binding in a loop anchor influence the strength of the corresponding loop and the ability to function as a barrier. Heatmaps were plotted with 200 bp resolution data for WT. We plotted a ±5 kb region either surrounding the high-residency CARs and the corresponding two-interval CARs (first panel); or surrounding the low-intensity CARs and the corresponding two-interval CARs (second panel); or surrounding a CAR and the corresponding two-interval CAR with a high-residency CAR in the middle (third panel).

Figure 4—figure supplement 1
Cohesin-dependent organization of loops.

(A) Genome-wide analysis confirms loop expansion in wild-type. Bar chart shows the percentage of the number of positioned loops per anchor in wild-type (WT) arrested in mitosis. (B) Around half of the visibly positioned loops are missed by HiCCUPS loop calling program. Contact map showing the interaction in the 40–160 kb region of chromosome I overlay with the tracks for Mcd1p ChIP-seq signal and peaks. Squares indicate loops called by HiCCUPS. Circles indicate visibly distinct positioned loops missed by HiCCUPS. Dashed lines indicate the position of barriers that insulate from loop expansion. (C) The positioned loop signal is detected till the +5 loop anchors interval genome-wide. Quantification of the heatmap analysis in Figure 3D. The chart shows the loop enrichment (Log2 of the ration of the loop signal in the center of the heatmap divided for the background signal in the corner) on the y-axis present at each CAR interval (x-axis) in wild-type (WT) cells in mitosis. (D) The positioned loop intensity correlates with the intensity of the Mcd1p ChIP-seq signal at the corresponding loop anchors. Loop intensity was sorted into quartiles from high to low. Mcd1p ChIP-seq signal was plotted against the loop anchors at the center. (E) Scatter plot showing a positive correlation of positioned loop intensity and Mcd1p ChIP-seq signal. Averaged ChIP-seq peak signals at the positioned loop anchors and the contact enrichment of loops (quantified by HiCCUPS or Chromosight) were normalized with z-score, and then plotted with Mcd1p signal at y-axis and loop intensity at x-axis. The scatter plot shows a positive correlation between two measurements with a Spearman coefficient at 0.63.

Figure 5 with 1 supplement
Chromatin loops are cell cycle dependent.

(A) Chromatin positioned loops emerge upon cohesin deposition during DNA replication. The figure shows Micro-C (top), Mcd1p ChIP-seq (middle), and Flow cytometry profile (bottom) data across a time course after release from G1 arrest (15 to 105 min). Heatmaps were plotted with 200 bp data resolution across a time course in ±5 kb regions surrounding the loop anchors identified in the mitotically arrested wild-type data. Average Mcd1p peak intensity was plotted across ±5 kb regions around the center of loop anchors at each time point. Flow cytometry of 20,000 cells per time point. (B) Loop formation peaks in G2 (75 min time point after G1 release). Bar chart represents the percentage of positioned loop detected at each time point compared to the number of positioned loops found upon mitotic arrest. (C) Positioned loops detected during the time course are less defined on the contact map. Snapshots for cells arrested in mitosis (left) and after 75 min release from G1 (right) represent the chromatin interaction in the 20–150 kb region of chromosome XII. Note that snapshot for the 75 min time point is in a higher color contrast for a better loop visualization. (D) Only a small subset of loop anchors form positioned loops with distal anchors at the 75 min time point. The pie chart shows the percentage of loop anchors that form positioned loops either with the corresponding neighboring anchors (+1) or with distal anchors (+2) in the 75 min time point. (E) Cohesin-dependent chromosome contacts peak at the same time points where positioned loops were detected. Interactions-versus-distance decaying curve shows the normalized contact density (y-axis) against the distance between the pair of crosslinked nucleosomes from 100 bp to 500 kb (x-axis) for wild-type (WT arrested in mitosis), Mcd1p-depleted (MCD1-AID arrested in mitosis), and cells from each time point of the time course. On the right, the derivative slopes for each data were plotted against the distance from 3 to 500 kb.

Figure 5—figure supplement 1
Loop formation during the cell cycle.

(A) Positioned loops detected during the time course are less defined on the contact map. Snapshots for cells for each time point after G1 release at 30°C and then cells arrested in mitosis (right). It represents the chromatin interaction in the 20–150 kb region of chromosome XII. (B) Chromatin loops emerge upon cohesin deposition during DNA replication. The figure shows Micro-C (top) Mcd1p ChIP-seq (middle), Flow cytometry profile data across a time course after release from G1 arrest (15 to 90 min) at 30°C. Heatmaps were plotted with 200 bp data resolution across a time course in ±5 kb regions surrounding the loop anchors identified in the mitotically arrested wild-type data. Average Mcd1p peak intensity was plotted across ±5 kb regions around the center of loop anchors at each time point. Flow cytometry profile of 20,000 cells per time point. (C) Loop formation peaks in G2 (60 min time point after G1 release at 30°C). Bar chart represents the percentage of the detected positioned loop at each time point compared to the number of positioned loops found upon mitotic arrest. (D) G2 cells present more contacts in the loop size-range. Interactions-versus-distance decaying curve shows the normalized contact density (y-axis) against the distance between the pair of crosslinked nucleosomes from 100 bp to 1 Mb (x-axis) for 15 min (G1) versus 60 min (G2) at 30°C (left panel). Decaying curve comparing 15 min (G1) versus 75 min (G2) at 23°C (right panel). (E) The positioned loops are formed mainly between adjacent CARs genome-wide during the time course. Heatmaps were plotted with 200 bp resolution data for each time point at 30°C (left) and 23°C (right). A ±5 kb region surrounding each loop anchor and the corresponding CARs interval from +1 to +5.

Figure 6 with 1 supplement
Wpl1p- and Pds5p-depletion perturb the looping pattern.

(A) Wpl1p-depletion causes excessive loop expansion on the contact map. Contact maps for WT (top) and WPL1-AID (bottom) show the chromatin interactions over the 200–450 kb region of chromosome X. Mcd1p ChIP-seq data for WT and WPL1-AID are overlaid across the same region. The scale for ChIP-seq is 0–200. (B) Wpl1p-depletion doubles the number of positioned loops detected. Bar chart shows the total number of positioned loops called by HiCCUPS in WT and WPL1-AID. (C) Genome-wide analysis confirms loop expansion upon Wpl1p-depletion. Bar chart shows the percentage of the number of positioned loops per anchor for WT (black) and WPL1-AID (green). (D) Wpl1p-depletion results in the expansion of the loop signal to further distal loop anchors. Heatmaps were plotted with 200 bp data for WT and WPL1-AID in ±5 kb regions surrounding the wild-type loop anchors and the corresponding CARs interval from a +1 to a +10. (E) Pds5p-depletion results in positioned loops between each centromere and surrounding CARs. Contact maps for WT (top) and PDS5-AID (bottom) shows the chromatin interactions over a chromosome arm on the 50–175 kb region of chromosome VII (left), and the centromeric region at 430–560 kb on chromosome VII (middle), or another centromeric region at 10–240 kb of chromosome XII (right). Mcd1p ChIP-seq data for WT and PDS5-AID are overlaid across the corresponding region in each panel. The scale for ChIP-seq is 0–400. (F) Pds5p-depletion results in loss of signal at wild-type anchors for positioned loops. Heatmaps were plotted with 200 bp data resolution for WT and PDS5-AID in ±5 kb regions surrounding the wild-type positioned loop anchors. (G) Pds5p-depletion causes cohesin depletion from chromosome arms, but not from pericentric regions. Average Mcd1p peak intensity for WT and PDS5-AID were plotted over ±5 kb regions around the peak center. Peaks were sorted into two groups: (1) peaks on the chromosome arms (>30 kb away from the centromeres) (left); (2) pericentric peaks (<30 kb away from the centromeres) (right).

Figure 6—figure supplement 1
Wpl1p- and Pds5p-depletion alters the pattern of loops genome-wide.

(A) Wpl1p and Pds5p are efficiently depleted upon auxin addition. The depletion of Wpl1p and Pds5p tagged with AID upon auxin addition to the media is shown by western-blot analysis. Tubulin is used for loading control. (B) Wpl1p-depleted cells present more loops and more loop expansion. Contact maps from mitotically arrested cells produced with Micro-C on wild-type (WT), cohesin Wpl1p-depleted (WPL1-AID). Contact maps show contacts between the whole genome (first column), followed by contacts from chromosomes VIII to XI at 6.4 kb resolution (second column). The following contact maps show chromosome X and a zoomed-in region (from the full 745 kb to a 250 kb region), with 1.6 kb and 800 bp resolution respectively. Standard colormap scheme that uses the shades of red from white (no detectable interactions) to black (maximum interactions detected) in log10 scale. (C) The loop signal at wild-type loops is still present upon Wpl1p-depletion. Heatmaps were plotted with 200 bp data resolution for wild-type (WT) and Wpl1p-depleted (Wpl1-AID) cells in ±5 kb regions surrounding the wild-type loop anchors. (D) Distribution of contacts upon Wpl1p-depletion and Pds5p-depletion. Interactions-versus-distance decaying curve shows the normalized contact density (y-axis) against the distance between the pair of crosslinked nucleosomes from 100 bp to 1 Mb (x-axis) for wild-type (WT in blue), Wpl1p-depleted (WPL1-AID in green), and Pds5p-depleted cells (PDS5-AID in purple) (left). Plot of the slopes derived from the interactions-versus-distance decaying curves for the same datasets along with Mcd1p-depleted cells (MCD1-AID in red) (right). The dotted line indicates the size of high contact enrichment (or the slowest contact decay with distance) for each strain. (E) Loop signal is detected till the +10 CARs interval upon Wpl1p-depletion. Quantification of the heatmap analysis in Figure 5D for Wpl1p-depleted (WPL1-AID in green) cells compared to wild-type (WT in blue) and cohesin-depleted (MCD1-AID in red). The chart shows the loop enrichment (Log2 of the ration of the loop signal in the center of the heatmap divided for the background signal in the corner) on the y-axis present at each CAR interval (x-axis). (F) Pds5p-depleted cells present no loops on chromosome arms. Contact maps from mitotically arrested cells produced with Micro-C on wild-type (WT), Pds5p-depleted (PDS5-AID). Contact maps show contacts between chromosomes VII to X at 6.4 kb resolution (first column). The following contact maps show chromosome VII and zoomed-in regions (from the full 1090 kb to a 125 kb region), from 1.6 kb to 400 bp resolution. Standard colormap scheme that uses the shades of red from white (no detectable interactions) to black (maximum interactions detected) in log10 scale.

Figure 7 with 1 supplement
Cohesin depletion alters the domain landscape.

(A) Contact map of mitotic chromatin in wild-type cells reveals the presence of CAR domains and their boundaries. Contact map showing the interaction in the 100–550 kb region of chromosome V overlays with the tracks for Mcd1p ChIP-seq signal. Triangles indicate the CAR domains’ position. Dashed lines indicate the position of the CAR domain boundaries. (B) The signal composing CAR domains in wild-type is lost upon cohesin depletion. The wild-type (WT) domains are rescaled to the same length with left and right boundaries aligned on the plot (black bar). Genome-wide average domain/boundary strength for WT, BRN1-AID, MCD1-AID, and asynchronous cells were plotted as distance-normalized matrices over the CAR domains called in the wild-type (n = 306). (C) Cohesin is enriched at the boundaries of CAR domains in wild-type, but not at domains formed upon cohesin depletion. Mcd1p ChIP-seq data from wild-type cells were plotted separately over ±5 kb region (top) and cumulative curves were plotted separately as the function of log2 ratio of Mcd1p ChIP-seq signal (bottom) around the boundaries of CAR domains in wild-type (WT) (blue), the boundaries present upon Mcd1p-depletion (pink), or the boundaries present in asynchronous cells (grey). (D) CARs at wild-type domain boundaries have higher levels of cohesin binding than other CARs genome-wide. Cumulative curves show the probability distribution of Mcd1p ChIP-seq signal (log2) from CARs present at the wild-type CAR boundaries (WT) (blue), and from the other CARs (purple). Below the curves are plotted the corresponding Mcd1p ChIP-seq values of each CAR (blue dots for CARs at the CAR boundaries and purple dots for the other CARs). Box and whiskers plot indicates the median values and the quartiles distribution.

Figure 7—figure supplement 1
Cohesin depletion alters the chromatin domain landscape.

(A) Chromosome domains are reorganized after cohesin removal. Wild-type (WT) (top), cohesin-depleted cells (MCD1-AID) (middle), and asynchronous cells (bottom) Micro-C XL contact maps are plotted at a 2 kb resolution across chromosome V: 100–450 kb region. Above each contact map, the corresponding domains called by hicexplorer are indicated with blue/yellow bars, and their respective boundaries are indicated with a red line above. Finally, Mcd1p ChIP-seq signal is overlaid with the contact maps in the same region at the very top of the figure. (B) Comparison of domain strength between wild-type, MCD1-AID, BRN1-AID, and asynchronous cells. The score of domain strength in wild-type was plotted at x-axis against the score in each condition at y-axis, respectively. Colormap codes the density of scatter dots. (C) Box plot showing the distribution of domain strength in each condition. On each box, the central mark indicates the median, and the bottom and top edges of the box indicate the 25th and 75th percentiles, respectively. The whiskers extend to the most extreme data points not considered outliers, and the outliers are plotted individually using the 'o' symbol. (D) CAR domain boundaries are enriched at terminators, while domains in cohesin-depleted and asynchronous cells are enriched at promoters. Bar chart shows the log2 enrichment of domain boundaries from wild-type (WT) in blue, cohesin-depleted cells (MCD1-AID) in red, and asynchronous cells (Async) in grey, at promoter or terminator.

Cohesin residency model for loop formation.

Loop-extrusion with a heterogenous cohesin residency gives rise to a heterogeneous pattern of chromatin looping and domains. The intensity of a cohesin ChIP-seq peak reflects the probability of the CAR being occupied by cohesin in a cell population. ChIP-seq peaks with high signals reflect CARs that are occupied with high probability by cohesin in a cell population. Low-signal peaks are CARs that are occupied by cohesin only in a fraction of cells. Cohesin with loop extrusion activity will start looping and stop when encountering two cohesin-bound CARs, creating a positioned loop. Every loop will result in a spot on the contact map from the stable interaction of the two CARs that function as loop anchors. Furthermore, the sequences inside the loop will have more probability of interacting with each other creating a loop domain visualized as a square in the contact map with spots at its vertices. Each spot and square signal from different positioned loops coming from individual cells will pile-up producing the final contact map signal detected by Micro-C XL with CAR domains with two high-intensity CAR as domain barriers and spots inside.

Tables

Table 1
Strains.
Strain name in the paperGenotypeRequest codeReference
WTMATa trp1∆::pGPD1-TIR1-CaTRP1 lys4::LacO(DK)-NAT leu2-3,112 pHIS3-GFPLacI-HIS3:his3-11,15 ura3-52 bar1VG3620Çamdere et al., 2015
MCD1-AIDMATa MCD1-AID-KANMX6 ADH1-OsTIR1- URA3::ura3-52 lys4::LacO(DK)-NAT trp1-1 GFPLacI-HIS3:his3-11,15 bar1 leu2-3,112DK5542Eng et al., 2014
BRN1-AIDMATa BRN1-D375-3V5-AID2-HPHMX trp1∆::pGPD1-TIR1-CaTRP1 lys4::LacO(DK)-NAT leu2-3,112 pHIS3-GFPLacI-HIS3:his3-11,15 ura3-52 bar1RL401Lamothe et al., 2020
BRN1-AID MCD1-AIDMATa MCD1-AID-KANMX6 BRN1-D375-3V5-AID2-HPHMX ADH1-OsTIR1- URA3::ura3-52 lys4::LacO(DK)-NAT trp1-1 GFPLacI-HIS3:his3-11,15 bar1 leu2-3,112RL406Lamothe et al., 2020
PDS5-AIDMATa PDS5-3V5-AID2-KANMX6 lys4::LacO(DK)-NAT pHIS3-GFP-LacIHIS3::his3-11,15 trp1-1 ura3-52 bar1TE228Eng et al., 2014
WPL1-AIDWPL1-3V5-AID-G418 TIR1-CaTRP1 bar1
LacO-NAT::lys4 leu2-3,112 GAL+ pHIS3-GFPLacI-HIS3:his3-11,15 ura3-52
VG3629-3Bthis work

Additional files

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)