Sister chromatid cohesion is mediated by individual cohesin complexes
Data files
Feb 22, 2024 version files 88.95 GB
Abstract
Eukaryotic genomes are organized by loop extrusion and sister chromatid cohesion, both mediated by the multimeric cohesin protein complex. Understanding how cohesin holds sister DNAs together, and how loss of cohesion causes age-related infertility in females, requires knowledge as to cohesin’s stoichiometry in vivo. Using quantitative super-resolution imaging, we identify two discrete populations of chromatin-bound cohesin in post-replicative human cells. While most complexes appear dimeric, cohesin localized to sites of sister chromatid cohesion and associated with Sororin is exclusively monomeric. The monomeric stoichiometry of Sororin:cohesin complexes demonstrates that sister chromatid cohesion is conferred by individual cohesin rings, a key prediction of the proposal that cohesion arises from their co-entrapment of sister DNAs.
README: Sister chromatid cohesion is mediated by individual cohesin complexes
[Access this dataset on Dyrad] (DOI:10.5061/dryad.3r2280gpg)
Eukaryotic genomes are organized by loop extrusion and sister chromatid cohesion, both mediated by the multimeric cohesin protein complex. Understanding how cohesin holds sister DNAs together, and how loss of cohesion causes age-related infertility in females, requires knowledge as to cohesin’s stoichiometry in vivo. Using quantitative super-resolution imaging, we identify two discrete populations of chromatin bound cohesin in post-replicative human cells. While most complexes appear dimeric, cohesin localized to sites of sister chromatid cohesion and associated with sororin is exclusively monomeric. The monomeric stoichiometry of sororin:cohesin complexes demonstrates that sister chromatid cohesion is conferred by individual cohesin rings, a key prediction of the proposal that cohesion arises from their co-entrapment of sister DNAs.
Description of the data and file structure
Imaging data is structured by figures and figure subpanels.
Fig.1 Chromatin bound cohesin exists in a monomeric and a dimeric state
(A) Schematic of 3D-SIM imaging approach of endogenously tagged RAD21 in G2 phase and on chromatin. (B) Representative widefield and super-resolution 3D-SIM images of chromatin bound RAD21-Halo. Single nucleus z-slice shown or copped region. False coloring by RAD21-Halo signal intensity shows different cohesin populations. (C) Frequency distribution of chromatin bound RAD21-Halo signal volumes with fitted non-linear regressions, n> 60.000, RAD21-Halo signals from 2 independent biological experiments. (D) Sub-labeling of chromatin bound RAD21-Halo with 50 pM JFX554 dye to achieve single molecule labeling, n> 900 RAD21-Halo signals. (E) Photobleaching of sub-labelled RAD21-Halo, example single-step bleach curve is shown corrected for background signal. Quantification in Fig. S1C. (F) Frequency distribution of single RAD21-Halo molecule volumes obtained from (D) and (E) was overlaid with frequency distribution of chromatin bound cohesin volumes as shown in (C). Quantification of absolute distributions of chromatin bound cohesin yielded the following frequencies: 22% monomer, 61% dimer, 17% trimer. Quantification is based on non-linear regression analysis. Experiments were carried out in two biological replicates (B-G).
File titles contain information on the experimental day, treatment, staining, replicate, image number, and processing steps as: DATE_TREATMENT_STAINING (488)_STAINING (568)_STAINING (647)_REPLICATE_IMAGE NUMBER_PROCESSING STEPS.tif. An exception is Fig.1E, where the file is labelled as: EXPOSURE TIME_LASER POWER_STAINING_IMAGE NUMBER.tif.
Fig.3 Cohesin complexes associated with sororin specifically mark sites of sister chromatid cohesion
(A) Schematic depiction of RASER-FISH approach for imaging of sites of sister chromatid cohesion. Two possible outcomes are shown; the cohesed sister chromatids are visible as one FISH signal or non-cohesed sister chromatids are visible as “split dots” due to their spatial separation. This figure was made with Biorender.com. (B) Frequencies of “split dots” (separated sister chromatids) measured by FISH and conventional microscopy in G2 cells. Probe locations are midpoints of cosmids in kb from chromosome 16 p telomere. TA; human lymphoblastoid cell line established from a normal individual, MEJY JY5.4; mouse erythroleukaemia hybrid cell line containing a normal human chromosome 16. N> 200 alleles analyzed per cosmid, averaged from 2 independent biological repeats. (C) RASER-FISH outcomes of two probes, marking sites of sister chromatid cohesion are shown, ether in wildtype conditions or conditions of cohesin loss induced by PROTAC3-mediated degradation of RAD21 (8h treatment). G2 cells were identified by sororin mean intensity (wild type) or Cyclin A mean intensity (no cohesin). Example images of G2 cells shown (left) and quantification shown (right). Experiments were carried out in biological duplicates. N=57 HS443D9 WT, n=43 HSD443D9 no cohesin, n=28 HS306A4 WT and n=31 HS306A4 no cohesin. Kruskal-Wallis test yielded p=0.0095, mean and standard deviation (SD) are plotted. (D) Co-localization of sororin to cohesed (top) and non-cohesed (bottom) sites and absolute quantification of sororin occupancy of the respective sites. Samples were prepared under pre-extraction conditions to visualize sororin, which maintained cohesion. Experiments were carried out in biological duplicates. N=48 HS443D9 (top), n=28 HSD443D9 (bottom), n=31 HS306A4 (top) and n=20 HS306A4 (bottom). Kruskal-Wallis test yielded p=0.0075 (top) and p=0.0086 (bottom), mean and SD are plotted. (E) Distance analysis between sister chromatids (distance between intensity maxima) at sites occupied by or void of sororin. Experiments were carried out in biological duplicates and n=49 for sororin positive sites and n=55 for sororin negative sites. Mann-Whitney U test yielded p<0.000001, mean and SD are plotted. Graphs are artificially jittered in x to show distributions.
File titles contain information on the experimental day, treatment, staining, replicate, image number, and processing steps as: DATE_TREATMENT_STAINING (488)_STAINING (568)_STAINING (647)_REPLICATE_IMAGE NUMBER_PROCESSING STEPS.tif. in Fig. 3C & Fig. S5C, sub-folders indicate probes used, treatment conditions and replicates.
Note: this folder also contains the imaging files for Fig. S5C, as the same images underlie the quantifications in Fig.3C and Fig. S5C.
Fig.4 Sororin is a monomer and associated with <1/3 of chromatin bound cohesin
(A) Schematic depiction of 3D-SIM imaging approach of endogenously tagged RAD21 together with sororin in G2 phase and on chromatin. (B) Representative widefield and 3D-SIM images of chromatin bound sororin. Single nucleus z-slice shown or cropped region shown. False coloring by sororin signal intensity shows uniform population. (C) Frequency distribution of chromatin bound sororin signal intensities with fitted non-linear regression (Gaussian distribution); SumInt; Sum Intensity; A.U., arbitrary units, n> 140.000 sororin signals from two biologically independent experiments. (D) Frequency distribution of signal intensities of chromatin bound sororin in Fig. 4C is overlaid with non-specific antibody conjugate intensities observed on imaging slides outside nuclei as shown in (E). N>3700 antibody conjugates and data are from two biologically independent experiments. (E) Example image showing analysis of single sororin intensities. On imaging slides, primary and secondary antibody conjugates, which occur through non-specific binding, were segmented, and their frequency distribution analyzed. Arrows indicate non-specific antibody conjugates on slide (blue) and specific sororin signals (green). Antibody specificity in cells has been controlled for in Fig. S5A. Quantification of absolute distributions of chromatin bound sororin from Fig. 4C yielded the following frequencies: 74.4% (monomer), 15.75% (dimer) and 9.85% (multimer). Quantification is based on non-linear regression analysis. (F) Example image of RAD21-Halo and sororin colocalization in G2 phase cells. Maximum projections of 5 z-slices are shown, as entire nucleus and cropped region. Overlaps are indicated in white circles and numbered lines mark the normalized intensity profiles show measurements. (G) Colocalization analysis of RAD21 at sites of sister chromatid cohesion marked by sororin. Bar charts consist of dots, each representing an analyzed RAD21 signal. Colocalization was analyzed in two biologically independent experiments with >38.000 RAD21 molecules analyzed per repeat. Absolute numbers for RAD21 colocalization with sororin are 28.47% (repeat 1) and 28.56% (repeat 2). Two-tailed student t-test determined p<0.0001, means are shown.
File titles contain information on the experimental day, treatment, staining, replicate, image number, and processing steps as: DATE_TREATMENT_STAINING (488)_STAINING (568)_STAINING (647)_REPLICATE_IMAGE NUMBER_PROCESSING STEPS.tif. An exception is Fig. 4F, where the word MAX indicates that a z projection is is shown.
Fig.5 Sister chromatid cohesion is maintained by individual cohesin rings
(A) Schematic depiction of labeling approach. U2OS RAD21-Halo cells were incubated with two different HaloTag ligands, each labeling ~50% of RAD21-Halo (see Fig. S7A-C for controls) The table shows expected frequencies of observing one Halo dye at a sororin site, or the other, or both at the same time, for monomeric, dimeric and multimeric cohesin. (B) Example 3D-SIM image of G2 cell after incubation with JFX554 and JFX650 against RAD21-Halo (each labeling ~50%) and immunostaining with sororin antibody. A single z-slice of a nucleus is shown and quantification in (D) and (H). (C) Examples of classes I-III (sororin-containing signals) detected in G2 cells. Single z-slices of 3D-SIM example crops are shown. (D) Quantification of (C). Three-way colocalization was measured in two biologically independent experiments, with 1200 sororin sites analyzed per repeat. Means are 8.21% (sororin without RAD21-Halo), 45.00% (sororin with RAD21-Halo JFX554), 44.46% (sororin with RAD21-Halo JFX650) and 2.33% (sororin with RAD21-Halo JFX554 and RAD21-Halo JFX650). Dots represent individual cells. Statistical testing by two-tailed student t-test yielded p<0.000001. (E) Examples of classes IV-VI (cohesin at sororin absent sites) detected in G2 cells. Single z-slices of 3D-SIM example crops are shown. (F) Cohesin at sites without sororin was analyzed in two biologically independent experiments, with 1200 cohesin signals analyzed per repeat. Means are 26.67% (2 x RAD21-Halo JFX554), 24.79% (2 x RAD21-Halo JFX650) and 48.54% (RAD21-Halo JFX554 and RAD21-Halo JFX650). Dots represent individual cells. Statistical testing by two-tailed student t-test yielded p<0.000001. (G) RASER-FISH example of probe HS443D9 co-localizing with RAD21-Halo in a dye mixing experiment (left) and quantification (right). N=34 probe signals from two biologically independent experiments. Mann-Whitney U test yielded p=0.33.
File titles contain information on the experimental day, treatment, staining, replicate, image number, and processing steps as: DATE_TREATMENT_STAINING (488)_STAINING (568)_STAINING (647)_REPLICATE_IMAGE NUMBER_PROCESSING STEPS.tif.
Fig. S1 Molecular counting of chromatin bound RAD21-Halo
(A) Representative widefield and super-resolution 3D-SIM images of chromatin bound RAD21-Halo as shown in Fig. 1B with added orthogonal views for illustration of 3D resolution increase by SIM. Single nucleus z-slice shown or copped region. (B) HILO-photobleaching of sub-labeled chromatin bound RAD21-Halo. Example cell nucleus with sparse RAD21-Halo signals is shown as well as crops at consecutive time points to visualize single step bleach reactions of events (highlighted with magenta arrows). (C) Quantification of single step-wise photobleaching of 10 example signals of chromatin bound RAD21-Halo after sub-labeling. Data is background corrected using regions of identical sizes. Bleaching occurs at different timepoints due to Gaussian nature of HILO laser beam. (D) Quantification of steps in bleach events observed in RAD21-Halo after sub-labeling and on chromatin. The experiment was carried out in two independent biological repeats with >90 bleach reactions analyzed per repeat. Dots indicate independent technical repeats per biological repeat. Mann-Whitney U test yielded p=0.0022.
Files are labelled as: EXPOSURE TIME_LASER POWER_STAINING_IMAGE NUMBER.tif.
Fig. S2 Generation of cell line for endogenously tagged cohesin complex
(A) Sequence alignment for SMC3 hinge region from H. sapiens to S. cerevisiae colored by amino acid conservation (60% threshold). The loop targeted for endogenous tagging surrounding amino acid residue E602 is highlighted in magenta. (B) M. musculus cohesin hinge with loop targeted for endogenous tagging of SMC3 highlighted in magenta (PDB: 2WD5) (55)(C) PCR showing homozygous tagging of SMC3-E602 with Spot-tag in two independent clones. (D) Western blot of SMC3-E602-Spot clones 1 and 2 compared to parental cell lines to show homozygous tagging. Membrane was immunostained for Spot, RAD21, SMC1 and Lamin-B1. Halo-tag was detected in-gel with JF549 and SMC3 was detected on overnight midi-gel to visualize 1.4 kDa size difference of Spot-tag insertion. (E) IP of SMC3-E602-Spot clones 1 and 2 by a-SMC1 antibody, immunostained for SMC1, SMC3 and RAD21. Input is 1% total sample. (F) Scatter plots showing the cell cycle distribution pattern of chromatin bound SMC3 detected in the endogenously tagged SMC3-E602-Spot cell line with Spot antibody (left) versus the cell cycle distribution pattern of chromatin bound SMC3 detected with SMC3 antibody in RAD21-Halo cells. m; mouse, rb; rabbit, ab; antibody. N=9700, the experiment was repeated twice biologically independently. (G) Quantification of metaphase spreads in RAD21-Halo parental cells and SMC3-E602-Spot clones 1 and 2. Spreads were counted as fully cohered, and railroad as well as prematurely separated chromosomes were counted as loss of cohesin phenotype. Two biologically independent experiments were carried out, with >20 metaphase spreads analyzed per cell line and repeat. Mean and SD are shown. (H) Western blot showing siRNA depletion efficiency of PDS5A and B in RAD21-Halo U2OS cells. Vermicelli were induced by combined siRNA depletion of PDS5A and B. (I) Example images of RAD21-Halo and SMC3-E602-Spot clones 1 and 2 after control or PDS5A/B siRNA treatment. RAD21-Halo was visualized in all cell lines by JFX554 dye and DNA by DAPI staining. Images were acquired with a spinning disk microscope and false-colored. (J) Vermicelli were categorized as wild type, mild, moderate and severe chromatin compaction phenotypes by automated ScanR analysis. Mean and SD are shown. Two biologically independent experiments were carried out, with >1000 cells analyzed per cell line, condition and repeat.
File titles contain information on the experimental day, treatment, staining, replicate, image number, and processing steps as: DATE_TREATMENT_STAINING (488)_STAINING (568)_STAINING (647)_REPLICATE_IMAGE NUMBER_PROCESSING STEPS.tif. An exception is Fig. S2F where imaging files from QIBC are labelled as WELL NUMBER_POSITION NUMBER_TIMEPOINT_STAINING.tif. In Fig. S2G, sub-folders indicate the different cell lines used.
Fig S3 SIMinspector, an image analysis tool for 3D super-resolution microscopy
(A) Schematic depiction of the SIMinspector workflow. 3D-SIM images are acquired and processed. Calibration samples are analyzed to determine microscope- and sample-specific parameters. Script 1 (SIMinspector_mainobject_Fiji) segments the main object based on intensity thresholding. Script 2 (SIMinspector_subobject_Fiji) segments subobjects by intensity thresholding and watershedding and exports intensity, volume, pairwise colocalization measurements and center positions. 3-way colocalization is determined by 2x two-way colocalization. (B) For calibration, U2OS cells were incubated with 100 nm 3-colour beads for 48 hours, which led to nuclear incorporation. This allowed calibration measurements in same cellular environment as actual samples. (C) Example 3D-SIM images of DAPI stained nucleus with incorporated 3-colour beads (5 slice maximum projection). (D) Full width at half maximum (FWHM) measurements in lateral and axial dimensions for monomeric beads within U2OS cells for all 3 channels, n=10 single bead signals, shown are median (middle line) and interquartile range (IGQ; box), whiskers are 1.5* IQR. (E) Measurement of two-way colocalization between green and red emission signals (left) and between green and far-red emission signals (right) detected from three-color 100 nm beads at a varying distance threshold, solid line represents sigmoid fit. N=80 beads. (F) Three-way colocalization analysis of beads within U2OS cells. At 99 nm distance threshold, >95% of bead signals from 3 channels colocalize. N=80 beads, experiment was carried out twice biologically independently. Kruskal-Wallis test determined p<0.0001, mean and SD are shown.
File titles contain information on the experimental day, treatment, staining, replicate, image number, and processing steps as: DATE_TREATMENT_STAINING (488)_STAINING (568)_STAINING (647)_REPLICATE_IMAGE NUMBER_PROCESSING STEPS.tif. An exception is Fig. S3C, where the word MAX indicates that a z projection is is shown.
Fig. S4 Chromatin bound RAD21 marks cohesin complexes
(A) Schematic depiction of 3D-SIM imaging approach of endogenously tagged cohesin complexes in G2 phase and on chromatin. RAD21 is Halo-tagged and SMC3 Spot-tagged at the cohesin hinge. (B) Chromatin-bound cohesin visualized by 3D-SIM microscopy, nucleus shown as single z-slice, colocalization shown in cropped regions of single z-slices and intensity line profiles. RAD21-Halo was visualized by incubation with JFX554 dye and SMC3-E602-Spot with Spot antibody. Images were false colored. (C) Quantification of colocalization between RAD21-Halo and SMC3-E602-Spot as detected in (B). 2 biologically independent repeats are shown in overlay, single spots represent single RAD21-Halo signals analyzed. N>90.000 analyzed RAD21 signals per repeat, the experiment was repeated twice in biologically independent manner. Statistical testing by two-tailed student t-test yielded p=0.0004.
File titles contain information on the experimental day, treatment, staining, replicate, image number, and processing steps as: DATE_TREATMENT_STAINING (488)_STAINING (568)_STAINING (647)_REPLICATE_IMAGE NUMBER_PROCESSING STEPS.tif.
Fig S6 Sororin is a monomer and its recruitment to chromatin coupled to DNA replication
(A) Cell cycle analysis of sororin binding to chromatin by immunofluorescence. Cell cycle stages were identified based on EdU (5-ethynyl-2’-deoxyurdine, thymidine analogue, S phase marker), MCM2 antibody (marker for pre-replicative chromatin) and DAPI (DNA marker). Example images for each cell cycle stage were acquired on a Spinning Disk microscope and quantification shown in (B). Cells were false-colored. (B) QIBC of chromatin bound sororin. Scatter plots shows single cells as dots over cell cycle by plotting of EdU over DAPI nuclear intensities. This delineates G1 phase (bottom left cloud with 2N DNA), S phase cells (arch) and G2 phase cells (bottom right cloud with 4N DNA). The color code shows the mean intensity of chromatin bound sororin signal per cell. 3400 cells are shown, the experiment was repeated twice in a biologically independent manner. (C) Western blot of sororin depletion from U2OS wild type cells with two independent siRNAs (Sor-1 and Sor-2) over 72 or 96 h, with immunostaining for sororin and total protein extract. ST, single transfection; DT; double transfection; Con, control siRNA; Sor-1, siRNA1 against sororin; Sor-2; siRNA2 against sororin. (D) PCR showing homozygous tagging of sororin with SNAP-tag. (E) Western blot showing endogenous tagging of sororin with SNAP-tag. Western blot was immunostained for Lamin-B1, RAD21, SMC1, SMC3, sororin and sororin-SNAP detected in gel with JFX650. Cell lines used were U2OS RAD21-Halo (RAD21-Halo), U2OS RAD21-Halo SMC3 E602-Spot (SMC3-E602-Spot), U2OS RAD21-Halo, SMC3-E602-Spot, sororin-SNAP (sororin-SNAP). (F) QIBC analysis of sororin antibody (left) and endogenously tagged sororin-SNAP (right) serves as antibody specificity control. 3500 cells are shown per condition. The experiment was repeated twice independently. (G) Example of 1-step photobleaching of chromatin bound sororin-SNAP incubated with JF549 by HILO. (H) Quantification of photobleaching of chromatin bound sororin-SNAP incubated with JF549 dye. The experiment was carried out in two independent biological repeats with >80 bleach reactions analyzed per repeat. Dots indicate independent technical repeats per biological repeat. Mann-Whitney U test determined p=0.002165, mean and SD are shown.
File titles contain information on the experimental day, treatment, staining, replicate, image number, and processing steps as: DATE_TREATMENT_STAINING (488)_STAINING (568)_STAINING (647)_REPLICATE_IMAGE NUMBER_PROCESSING STEPS.tif. Files is Fig. S6B and Fig. S6F are labelled as WELL NUMBER_POSITION NUMBER_TIMEPOINT_STAINING.tif. Files in Fig. S6G &H are labelled as: EXPOSURE TIME_LASER POWER_STAINING_IMAGE NUMBER.tif.
Fig S7 HaloTag dyemixing allows molecular counting
(A) Titration of RAD21-Halo dyes JFX554 and JFX650 in U2OS RAD21-Halo cells. EC50 was determined to 3.29 nM (JFX554) and 1.05 nM (JFX650). (B) Labeling approach (top) and quantification of signal frequencies (bottom). Samples were incubated with varying concentrations of mixed dyes JFX554 and JFX650 for 1.5 h, washed and chased with a third dye JF479. Samples were imaged with 3D-SIM and single RAD21-Halo signals analyzed. N>28.000 RAD21-Halo analyzed per condition. W; wash. Mean and SD are shown. (C) To validate labeling efficiency of JF479 dye used for chasing in Fig. S7B, cells were incubated with dye JFX554 for 1 h, chased by JF479 and vice versa. RAD21-Halo labelling frequency was obtained from normalized intensity distributions for n>4000 cells imaged by QIBC. The experiment was repeated twice in a biologically independent manner. W; wash. Mean and SD are shown. (D) 3D-SIM image crop of 2 sororin signals overlapping with two RAD21-Halo molecules, one labelled with JFX554 (magenta) and one labelled with JFX 650 (cyan). A single z-slice is shown and intensity quantification below. (E) Overlay of frequency distribution of chromatin bound RAD21 monomer volumes shown in Fig. 1F with volumes of sororin-associated RAD21. N>33.000 sororin-cohesin complexes and the experiment was carried out twice in biologically independent manner.
File titles contain information on the experimental day, treatment, staining, replicate, image number, and processing steps as: DATE_TREATMENT_STAINING (488)_STAINING (568)_STAINING (647)_REPLICATE_IMAGE NUMBER_PROCESSING STEPS.tif. Files is Fig. S7C are labelled as WELL NUMBER_POSITION NUMBER_TIMEPOINT_STAINING.tif, and sub-folders indicate dye chase conditions and replicates.
Please see the below overview of abbreviations for processing steps:
ALN: channel aligned file
LUT: false-coded by look up table
MCF_THR: MCNR corrected file
PWF: pseudo-widefield image
SIR: reconstructed SIM image
THR: 16-bit conversed and thresholded file
For information on how these processing steps were executed, please see the SIM section in methods here on Dyrad.
All images can be opened with Fiji/ImageJ.
Code/Software
We have deposited the code for SIMinspector, our SIM image processing pipeline, as well as Jupyter notebooks for the analysis of ChIP and sister chromatid sensitive-HiC (scs-HiC) data.
SIMinspector
For SIMinspector two files have been deposited, called SIMinspector_Mainobject_Fiji and SIMinspector_Subobject_Fiji, which are Fiji plug-ins.
SIMinspector is a custom image analysis tool for the segmentation of sub-cellular features and analysis of colocalization in multicolor 3D-super-resolution microscopy images.
SIMinspector consists of 2 Fiji scripts that are explained below and can be executed in the following order:
1. SIMinspector_Mainobject_Fiji: This Fiji script segments the main object, e.g. cell nucleus, based on intensity thresholding. For this, a selected channel (“Channel for mask creation”) is used. Objects are detected based on intensity thresholding (“Threshold method”) and all objects but the largest are excluded. It produces segmented and cropped 3D images in the specified output directory.
2. SIMinspector_Subobject_Fiji: This Fiji script segments subobjects from 2 channels, measures volumes, intensities and centers as well as colocalization based on voxel overlap, center-to-center distances and center-to-edge distances. To this end, images are made isotropic and channels are thresholded (“Segmentation algorithm”). Maxima are detected (“Maxima radius”) and made into labels. Maxima-derived labels were used for marker-controlled intensity-based watershed filtering. Optional, resulting labels can be eroded to optimize segmentation (“Label erosion channel”). Afterwards images were resliced and objects below minimum and above maximum volumes excluded (“Minimum spot volume in voxels”, “Maximum spot volume in voxels”). Resulting stacks are converted to 16-bit, original pixel sizes restored, and statistics of labels computed. Computed centers of mass are mapped. Fiji’s plugin DiAna (Gilles et al., 2017) is used to calculate voxel overlaps, center-to-center and center-to-edge distances for the N nearest neighbors. The script produces 4 .csv files. The file ending on “StatisticsOfLabelmap_C1” contains all measurements for channel 1, the file ending on “StatisticsOfLabelmap_C2” contains all measurements for channel 2 and the files “AdjacencyResults_C1vsC2” contain the colocalization results originating from either center of mass or geometric center values.
The dependencies for these scripts are 3D ImageJ Suite(Ollion et al., 2013), Java 8, CLIJ (Haase et al., 2020), CLIJ2, clijx-assistant, clijx-assistant extension, IJPB (Legland et al., 2016) and DiAna (Gilles et al., 2017).
ChIP/scs-HiC
We have also deposited all code for the alignment of ChIP and sister chromatid sensitive Hi-C, which can be found as 6 files:
001_Generate_intervals.ipynb
002_Generate_stacks_SMC3_w_Sororin.ipynb
002_Generate_stacks_SMC3_wo_Sororin.ipynb
003_HiC_aggregate_maps_SMC3_w_Sororin.ipynb
003_HiC_aggregate_maps_SMC3_wo_Sororin.ipynb
004_Generate_HiC_line_profiles_SMC3_w_Sororin.ipynb
004_Generate_HiC_line_profiles_SMC3_wo_Sororin.ipynb
All files are Jupyter notebooks and are differentiated based on cohesin (SMC3) with or without co-localising Sororin, as described in the manuscript Fig.2.
Details on the processing of ChIP and scs-HiC datasets can be found below.
SMC3 and sororin ChIP-Seq heatmaps
ChIP-Seq data for SMC3 and sororin from (Ladurner et al., 2016) (https://www.ebi.ac.uk/ena/browser/view/PRJEB12214) were mapped using bwa (v0.7.17) against human genome assembly GRCh37/hg19. Raw read counts were normalized to counts per million (CPM) values and log-transformed using a base two logarithm. The peaks were called using macs2 (54) using a p-value threshold of 1e‑10. For each detected SMC3 peak the average sororin signal was measured using pybbi package version 0.3.2 (https://github.com/nvictus/pybbi). SMC3 peaks with a mean sororin signal higher than 0.1 CPM were marked as “SMC3 with sororin” (20943 peaks). The remaining SMC3 peaks were marked as “SMC3 without sororin” (13848 peaks). Heatmap plots centered at SMC3 sites with or without sororin were calculated using a custom ipython notebook. Samples come from wild type HeLa Kyoto cells synchronized to G2, and from two biological replicates. SMC3 and sororin ChIP-Seq data from (Ladurner et al., 2016) can be accessed in the European Nucleotide Archive (ENA) with the accession number PRJEB12214.
Hi-C aggregate maps
Scs-Hi-C data for wild type HeLa Kyoto cells synchronized in G2 (GSE152373) were processed as described in (Mitter et al., 2020). Observed-over-expected scsHi-C aggregate maps were calculated using the cooltools package version 0.5.4 (https://github.com/open2c/cooltools). 1 Mb ICE-corrected snippets around SMC3 sites with or without sororin (“observed”) were extracted and each value in the snippets was normalized by the value of the scaling at that diagonal (“expected”). The pixel-wise average across the snippet was then calculated. The main diagonal and one neighboring diagonal were blanked out to avoid Hi-C artefacts. G2 maps were constructed from 1.7 billion Hi-C reads with 195 million unique reads from 11 biological repeats. HeLa scsHi-C data from (Mitter et al., 2020) can be access at Gene Expression Omnibus (GEO) with accession number GSE152373.
Stack-up analysis of line profiles for ChIP-Seq and scsHi-C data
Stack-ups of line profiles within 1 Mb windows centered at SMC3 sites with or without sororin were calculated using a custom iPython notebook. The contact density within a sliding diamond of 100 kbs was calculated along each region within the set of regions of length m for observed-over-expected matrices, resulting in a vector of size n for each region. Then, these vectors were stacked into an m × n matrix with m denoting the number of regions and n denoting the length of the line profile along each region. For display of observed-over-expected values, a pseudocount of 0.01 was added before log-transformation. Line profiles containing only invalid Hi-C bins were removed from the stack-up. Code underlying ChIP and scs-HiC analysis, can be downloaded from Dyrad with the identifier doi:10.5061/dryad.3r2280gpg.
References
- Ball G, Demmerle J, Kaufmann R, Davis I, Dobbie IM, Schermelleh L. 2015. SIMcheck: a Toolbox for Successful Super-resolution Structured Illumination Microscopy. Sci Rep 5:15915. doi:10.1038/srep15915
- Brown JM, Leach J, Reittie JE, Atzberger A, Lee-Prudhoe J, Wood WG, Higgs DR, Iborra FJ, Buckle VJ. 2006. Coregulated human globin genes are frequently in spatial proximity when active. J Cell Biol 172:177–187. doi:10.1083/jcb.200507073
- Brown JM, Ornellas SD, Parisi E, Schermelleh L, Buckle VJ. 2022. RASER-FISH: non-denaturing fluorescence in situ hybridization for preservation of three-dimensional interphase chromatin structure. Nat Protoc 17:1306–1331. doi:10.1038/s41596-022-00685-8
- Daniels RJ, Peden JF, Lloyd C, Horsley SW, Clark K, Tufarelli C, Kearney L, Buckle VJ, Doggett NA, Flint J, Higgs DR. 2001. Sequence, structure and pathology of the fully annotated terminal 2 Mb of the short arm of human chromosome 16. Hum Mol Genet 10:339–352. doi:10.1093/hmg/10.4.339
- Gilles J-F, Santos MD, Boudier T, Bolte S, Heck N. 2017. DiAna, an ImageJ tool for object-based 3D co-localization and distance analysis. Methods 115:55–64. doi:10.1016/j.ymeth.2016.11.016
- Haase R, Royer LA, Steinbach P, Schmidt D, Dibrov A, Schmidt U, Weigert M, Maghelli N, Tomancak P, Jug F, Myers EW. 2020. CLIJ: GPU-accelerated image processing for everyone. Nat Methods 17:5–6. doi:10.1038/s41592-019-0650-1
- Ladurner R, Kreidl E, Ivanov MP, Ekker H, Idarraga‐Amado MH, Busslinger GA, Wutz G, Cisneros DA, Peters J. 2016. Sororin actively maintains sister chromatid cohesion. EMBO J 35:635–653. doi:10.15252/embj.201592532
- Legland D, Arganda-Carreras I, Andrey P. 2016. MorphoLibJ: integrated library and plugins for mathematical morphology with ImageJ. Bioinformatics 32:3532–3534. doi:10.1093/bioinformatics/btw413
- Matsuda A, Koujin T, Schermelleh L, Haraguchi T, Hiraoka Y. 2020. High-Accuracy Correction of 3D Chromatic Shifts in the Age of Super-Resolution Biological Imaging Using <em>Chromagnon</em>. J Vis Exp. doi:10.3791/60800
- Mitter M, Gasser C, Takacs Z, Langer CCH, Tang W, Jessberger G, Beales CT, Neuner E, Ameres SL, Peters J-M, Goloborodko A, Micura R, Gerlich DW. 2020. Conformation of sister chromatids in the replicated human genome. Nature 586:139–144. doi:10.1038/s41586-020-2744-4
- Ochs F, Karemore G, Miron E, Brown J, Sedlackova H, Rask M-B, Lampe M, Buckle V, Schermelleh L, Lukas J, Lukas C. 2019. Stabilization of chromatin topology safeguards genome integrity. Nature 574:571–574. doi:10.1038/s41586-019-1659-4
- Ochs F, Somyajit K, Altmeyer M, Rask M-B, Lukas J, Lukas C. 2016. 53BP1 fosters fidelity of homology-directed DNA repair. Nat Struct Mol Biol 23:714–721. doi:10.1038/nsmb.3251
- Ollion J, Cochennec J, Loll F, Escudé C, Boudier T. 2013. TANGO: a generic tool for high-throughput 3D image analysis for studying nuclear organization. Bioinformatics 29:1840–1841. doi:10.1093/bioinformatics/btt276
- Stallings RL, Torney DC, Hildebrand CE, Longmire JL, Deaven LL, Jett JH, Doggett NA, Moyzis RK. 1990. Physical mapping of human chromosomes by repetitive sequence fingerprinting. Proc Natl Acad Sci 87:6218–6222. doi:10.1073/pnas.87.16.6218
- Szczurek AT, Dimitrova E, Kelley JR, Klose RJ. 2023. Polycomb sustains promoters in a deep OFF-state by limiting PIC formation to counteract transcription. bioRxiv 2023.06.13.544762. doi:10.1101/2023.06.13.544762
- Toledo LI, Altmeyer M, Rask M-B, Lukas C, Larsen DH, Povlsen LK, Bekker-Jensen S, Mailand N, Bartek J, Lukas J. 2013. ATR Prohibits Replication Catastrophe by Preventing Global Exhaustion of RPA. Cell 155:1088–1103. doi:10.1016/j.cell.2013.10.043
Methods
Below is the information on how samples were stained and images acquired, sorted by microscopy technique.
Immunofluorescence staining
Cells were grown to 80% confluency on 22x22 mm #1.5H high-precision coverslips (thickness 0.170 ± 0.005 mm, Marienfeld Superior). Cells were washed in PBS, pre-extracted in ice-cold PBS 0.2% Triton X-100 (Sigma-Aldrich) for 1.5 min on ice and fixed in 4% formaldehyde for 15 min. For non-pre-extracted samples, cells were fixed and subsequently permeabilized in PBS 0.2% Triton X-100 for 5 min. Primary and secondary antibodies were diluted in antibody diluent (DMEM medium containing 10% FBS, 0.05% sodium azide, sterile filtered 0.2 mm). Samples were incubated in primary antibody for 1.5 h and for 1 h in secondary antibody solution supplemented with 4’,6’-diamidino-2-phenylindole-dyhydrochloride (DAPI, 0.5 mg/ml). In between primary and secondary antibody staining, samples were washed 3x in PBS 0.2% Tween. After secondary antibody incubation, samples were washed 3x again, samples were post-fixed in 4% formaldehyde for 15 min and samples were washed in PBS and distilled water. For QIBC, samples were mounted in Mowiol-based mounting medium (Mowiol 488 (Calbiochem)/glycerol/Tris-HCl, pH 8.5) and in non-hardening Slowfade Diamond (Thermo Fisher Scientific, S36963) for Spinning Disk and 3D-SIM imaging. For EdU staining, cells were incubated with 10 mM EdU 20 min before pre-extraction and EdU detection was performed before primary antibody incubation according to the manufacturer’s instructions (Thermo Fisher Scientific). For detection of endogenously tagged proteins with chemical dyes, samples were incubated for 1.5 h with Halo- or 3 h with SNAP-ligands, washed three times in fresh media, and incubated for 30 min in the incubator in fresh media for exit wash of unbound dye. For experiments with two Halo dyes, cells were incubated with both dyes for 1.5 h, then washed and chased with a third dye as indicated. For calibration purposes, U2OS cells were incubated with 100 nm 4-colour Microspheres (fluorescent blue, green, orange, dark red; Thermo Fisher Scientific) for 3 days at a density of 30 ml bead slurry / 2x105 cells. For G2 analysis, cells with highest mean intensities for Cyclin A (fixed cells) or sororin (pre-extracted cells) were chosen.
DNA FISH
Methanol-acetic acid–fixed cells were prepared and hybridized as described in (Brown et al., 2006). Hybridized slides were examined using a fluorescence microscope (Olympus BX60). Hybridization efficiency was assessed by scoring FITC signals in ten metaphase spreads and only those hybridizations with greater than 80% efficiency were scored. 200 FITC signals were scored as either single or split dots for each hybridization. The percentage of split dot values for each cosmid plotted represent the average of two independent biological experiments.
RASER-FISH
RASER-FISH allows DNA sequence detection in intact 3D cell nuclei as it maintains fine-scale chromatin structure by replacement of heat denaturation with exonuclease III digestion after UV-generation of DNA nicks. RASER-FISH was carried out as described in (Brown et al., 2022; Ochs et al., 2019). In brief, U2OS cells were seeded on 22x22 mm #1.5H high-precision coverslips (thickness 0.170 ± 0.005 mm) and labelled for 18 h with 10 mM BrdU/BrdC mix (3:1). Cells were either pre-extracted and fixed as described above or fixed in 4% formaldehyde and permeabilized in PBS 0.2% Triton-X-100 for 5 min. Immunofluorescence staining for sororin or Cyclin A was carried out as described above. After incubation with 0.5 mg/ml DAPI for 15 min for UV sensitization, cells were treated with UV light (254 nm) for 15 min followed by incubation with 5 U/ml exonuclease III (NEB) at 37°C for 15 min. Biotin labelled probes were denatured in hybridization mix at 90°C for 10 min, pre-annealed with human Cot-1 DNA (Invitrogen) at 37°C for 15 min and hybridized to samples overnight at 39°C. Coverslips were washed twice in 1x SSC buffer at 37°C for 30 min, once in 1x SSC at room temperature, once in PBS, and biotin was detected by incubation with streptavidin-Alexa488 (1:500, Thermo Fisher). Samples were post-fixed, rinsed in PBS and MilliQ and mounted in Slowfade Diamond. For RAD21 degradation, cells harboring RAD21-Halo were incubated with 2.5 mM HaloPROTAC3 (Promega) for the last 8 h of BrdU/BrdC incubation. For dye mixing experiments, dye mixing was carried out as described above for the last 2 h of BrdU/BrdC incubation. The probes for all FISH assays in this study were obtained from a chromosome 16p specific cosmid library based on genome assembly NCBI36 (Stallings et al., 1990) and have been characterized in (Daniels et al., 2001) for mapping of the terminal 2 Mb of chromosome 16 p-arm. The GenBank accession number for the entire map is AE005175 and the specific probes used in this study have EMBL IDs HS306A4 (AL008727) and HS443D9 (Z92845). Probes were biotin-labelled.
Quantitative image-based cytometry (QIBC)
QIBC (Ochs et al., 2016; Toledo et al., 2013) was performed on a motorized Olympus IX83 inverted wide-field microscope equipped with Semrock DAPI/FITC/Cy3/Cy5 Quad LED filter set, Hamamatsu Orca Fusion B CMOS camera and Lumencor SPECTRA X light source. Automated unbiased image acquisition was carried out with the proprietary ScanR acquisition software. Identical exposure times were used for all samples within one experiment and settings were chosen for maximum dynamic range under non-saturating conditions. Most data was obtained with the Olympus UPLXAO 20x, NA 0.8 air objective. Depending on cell confluency and type of image analysis, 49-100 images were acquired, aiming for at least 2000 cells per analysis after gating. After acquisition, images were analyzed using the ScanR analysis software. DAPI signal was used for segmentation of nuclei based on intensity threshold. This nucleus mask was then used to quantify pixel intensities in the different channel for each individual nucleus. After segmentation and pixel quantification, measured parameters were extracted (mean and total intensities, area, circularity, well). For automated vermicelli quantification, edge detection was used, and 4 arbitrary classes were defined based on increasing edges (none, mild, moderate, and severe chromatin compaction). All data were exported to Tibco Spotfire, which was used to quantify average/median values in cell populations and to generate color-coded scatter plots in a flow-cytometry fashion, or to GraphPad Prism to generate stacked bar charts.
Spinning disk microscopy
Representative pseudo-confocal images were acquired with an UltraView Vox spinning-disk microscope (Perkin Elmer) and Volocity software (version 6.3.1) with a 60x, 1.42 NA Plan-Apochromat oil-immersion objective. Images were captured with a Hamamatsu EMCCD 16-bit camera at a spatial resolution of 121x250 nm. Single z-slices of whole nuclei are shown, brightness and contrast were linearly adjusted for optimal display. Color-channels were false-colored.
Structured illumination microscopy (SIM)
3D-SIM images were acquired with a DeltaVision OMX SR system (GE Healthcare) equipped with a 60x, 1.5 NA UPLAPO60XOHR oil immersion objective (Olympus), pco.edge 4.2 sCMOs cameras (PCO), and 405 nm, 488 nm, 568 nm and 640 nm lasers. 3D image stacks were acquired over the whole nuclear volume in z with 15 raw images per plane (3 angles, 5 phases). Spherical aberration was minimized using immersion oil with refractive index 1.514 and an objective collar setting of 0.140 for image acquisition. Raw data was computationally reconstructed with SoftWoRx 7.2.0 (GE Healthcare) using channel-specific optical transfer functions (OTFs) recorded using immersion oil with RI 1.516 and Wiener filter setting 0.0030. All SIM data were routinely and meticulously quality-controlled for effective resolution and absence of artifacts using SIMcheck (Ball et al., 2015). Multichannel acquisitions were aligned in 3D with Chromagnon software (Matsuda et al., 2020) using 3D-SIM acquisitions of multicolor EdU-labelled C127 cells as colocalization reference. Images were thresholded based on the MCNR function of SIMcheck, which generates a metric of local stripe modulation contrast in different regions of the raw data and directly correlates this with the level of high-frequency information content in the reconstructed data. Only immunofluorescent signals with underlying MCNR values that exceeded a stringent quality threshold were considered for further analysis, while localizations with low underlying MCNR values were discarded. Thereby, any SIM signal, which falls below reconstruction confidence and is considered to be a labelling/imaging artifact, is excluded from further data interpretation (Ball et al., 2015). For representative images, single z-slices, cropped regions of whole nuclei or partial z-stacks are shown as indicated in figure legends. Brightness was linearly adjusted for optimal presentation. Channels were false-colored.
HILO photobleaching
Cells were seeded and labelled on 35 mm diameter, No. 1.5 MatTek dishes, pre-extracted and fixed as described above, and imaged in PBS. For photobleaching, a custom-built total internal reflection fluorescence (TIRF) microscope (described in detail in (Szczurek et al., 2023)) was used with a 100x 1.4 NA oil objective (Olympus) and with an iChrome MLE MultiLaser engine (Topica Photonics). Photobleaching was performed using highly inclined laminar optical sheet illumination (HILO). Emission was projected onto the central region of an iXon 897 EMCCD camera (Andor, 512x512 pixels). Pixel size in acquired images was 96 nm. Sample position and focus were controlled with a motorized stage and z-motor (ASI). For photobleaching samples were exposed to 30 % (RAD21-Halo) and 10-20% (sororin-SNAP) 561 nm laser and imaged using 100 ms camera interval for up to 1.5 min. Obtained images were analyzed with the Fiji plugin “Time Series Analyser” (https://imagej.nih.gov/ij/plugins/time-series.html) and background subtraction over time was performed using parallel background measurements.