Cell culture
Cell lines RPE1-hTERT (Flp-In) (a gift from the laboratory of P. Jallepalli), Caco-2 (a gift from the laboratory of H. Clevers), HeLa (a gift from the laboratory of M. Vermeulen), HT-29 (a gift from the laboratory of H. Clevers), U2OS (a gift from the laboratory of S. Lens) and WiDr (a gift from the laboratory of H. Clevers) were cultured in DMEM/F12 and GlutaMAX supplement (Gibco), supplemented with 9% foetal bovine serum (FBS, Sigma-Aldrich) and 1% penicillin/streptomycin (Sigma-Aldrich). Cell lines U2OS DamID, U2OS CENPA and BJ-hTERT (a gift from the laboratory of R. Medema), HCT116 (a gift from the laboratory of H. Clevers) and HT1080 were cultured in DMEM, high-glucose GlutaMAX supplement and pyruvate (Gibco), supplemented with 10% FBS and 1% penicillin/streptomycin. Human small intestine duodenum and ileum organoids (a gift from the laboratory of H. Clevers) were cultured as described previously51. Rather than WNT conditioned medium, WNT surrogate was used (0.15 nM, U-Protein Express). DLD1 cells (a gift from D. Cimini) were cultured in RPMI and GlutaMAX supplement (Gibco), supplemented with 9% FBS and 50 μg ml–1 penicillin/streptomycin. To generate RPE1-hTERT Flp-in H2B-mNeon cells, cells were transduced with a lentivirus containing an H2B-mNeon-IRES-puromycin construct. Selection was performed with 10 μg ml–1 puromycin (Sigma-Aldrich) for 48 h. Organoids were transduced with the same construct without selection. Tetraploid RPE1-hTERT cells were generated by treatment of original RPE1-hTERT cells with 62.5 nM Cpd-5 for 48 h every 7 days for 4 weeks, after which tetraploid colonies grew out for an additional 22 weeks. Monoclonal RPE1-hTERT dCas9-3×GFP and DLD1 dCas9-GFP-3×FKBP FRB-mCherry-LaminB1 lines were generated by transduction with a dCas9-3×GFP or dCas9-GFP-3×FKBP lentivirus, followed by single-cell sorting. Next, FRB-mCherry-LaminB1 was lentivirally introduced in DLD1 cells. HT1080 cells containing a LacO-array in chromosome 11 (a gift from W. Bickmore) were transduced with LacI-GFP-FKBP and FRB-mCherry-LaminB1 and cloned by single-cell sorting. Cell lines were tested for mycoplasma contamination and not authenticated.
scKaryo-seq
RPE1-hTERT Flp-in cells were plated in a six-well plate (Corning) at 40% confluency and treated with palbociclib (250 nM; Selleck Chemicals). After 24 h, cells were washed three times with warm medium and treated with RO-3306 (5 μM; Tocris Bioscience). After 16 h, cells were washed three times for 5 min at 37 °C with warm medium containing DMSO, Cpd-5 (62.5 nM; a gift from R. Medema) or monastrol (200 μM; Sigma-Aldrich). Cpd-5-treated cells were cultured for a further 4 h before harvesting. Monastrol-treated cells were washed three times with warm medium containing 62.5 nM Cpd-5. Mitotic cells were collected by shake-off and plated in a new well of a six-well plate for 4 h. BJ-hTERT cells were plated in a six-well plate at 40% confluency and treated with 31.25 nM Cpd-5 for 16 h. All cells were trypsinized and stored at −20 °C for further processing. Single G1 nuclei of RPE1-hTERT Flp-in cells or single nuclei of BJ-hTERT cells were sorted as described previously8. Human intestinal organoids were plated 1 day before treatment for 16 h with 5 μM ZM447439 (Selleck Chemicals) or 10 μM EdU (Thermofisher) for 3 h, washed three times for 5 min with warm medium, incubated with 62.5 nM Cpd-5 for 16 h and fixed using 70% ice-cold ethanol. Ethanol was removed by one wash with PBS, and cells were incubated for 10 min with the Click-iT reaction cocktail (Click-iT EdU proliferation assay). The reaction cocktail was washed away and replaced with a PBS/DAPI mix. Single G1 nuclei in the case of ZM447439 or EdU-positive G1 cells were sorted in 384-well plates. Tetraploid RPE1-hTERT cells were plated at 40% confluency and treated with 62.5 nM Cpd-5 for 24 h. G1 nuclei were sorted. HCT116 cells were synchronized for 16 h using monastrol, released and treated with Cpd-5 as described for RPE1-hTERT cells. Plates were stored at −20 °C. NlaIII-based library preparation was performed as described previously, with several modifications8. Cell lysis was performed for 2 h at 55 °C with 8 mg ml–1 Proteinase K (Fisher Scientific) in 1× CutSmart (New England Biolabs) and heat inactivation at 80 °C for 10 min. Adaptors were ligated with 100 nl of 100 nM barcoded, double-stranded NLAIII adaptors and 400 nl of 10 U T4 DNA ligase (New England Biolabs) in 1× T4 DNA ligase buffer (New England Biolabs), supplemented with 3 mM ATP (Invitrogen) at 16 °C overnight. Samples were sequenced on an Illumina NextSeq500 or 2000 at 1× 75 or 1× 100 base pairs (bp), respectively. After sequencing, mapping (bwa aln 0.7.12 and python 2.7.5) and Aneufinder (v.1.2.0) plotting and copy number variations of whole and partial chromosomes were determined manually. Chromosome 8 of human intestinal organoids was not quantified because this chromosome was heterogeneously aneuploid under the control condition.
Centromere FISH
Cells were plated on 12-mm round glass coverslips (Superior Marienfeld). To validate scKaryo-seq segregation error bias, cells were synchronized and treated with Cpd-5 as described above. Cells were fixed 45 min after release from RO-3306, at −20 °C with 75% methanol and 25% acetic acid. To determine the distance of chromosomes from the centre of the nucleus, cells were plated 1 day before fixation. To determine nuclear chromosome territories of monastrol-treated mitotic cells, cells were synchronized as described above and incubated for 4 h in monastrol, then subsequently fixed. After fixation, coverslips were air-dried and incubated for 2 min with 2× saline-sodium citrate (SSC) at room temperature. Coverslips were washed in series with 70%, 85% and 100% ethanol and air-dried. Next, 1.2 μl of a red and green satellite enumeration probe (Cytocell) and 1.6 μl of hybridization solution per coverslip were spotted on a glass slide. Coverslips were placed upside down on the probe solution and incubated at 75 °C for 2 min. Coverslips were incubated at room temperature for 4–16 h, followed by 2 min incubation at 72 °C with 0.25× SSC (pH 7.0). Coverslips were washed for 30 s with 2× SSC 0.5% Tween-20 at room temperature, incubated with DAPI and mounted using ProLong Gold antifade (Molecular Probes).
Image acquisition was done on a DeltaVision RT system (Applied Precision/GE Healthcare) with a ×1.40/100 numerical aperture (NA) UplanSApo objective (Olympus) as z-stacks at 0.5 μm intervals. For deconvolution, SoftWorx (Applied Precision/GE Healthcare, v.6.5.2) was used. Image analysis and quantification was done using Fiji ImageJ (v.2.0.0).
FISH segregation error frequencies were determined by counting the number of mis-segregating FISH-positive chromosomes and dividing that by the total number of mis-segregating chromosomes.
Chromosomes in low-dose nocodazole were considered misaligned when FISH-positive chromosomes were physically separated from the metaphase plate; this number was then divided by the total number of FISH-positive chromosomes.
To measure the distance of chromosomes from the centre of the nucleus, we determined the centroid X and Y coordinates of the three different thresholded channels (DAPI, red probe and green probe). The centre of monastrol-treated cells was determined using a custom ImageJ script, which measures the centre of mass of thresholded DAPI particles.
Live imaging
To time mitotic phases, RPE1-hTERT Flp-in H2B-mNeon cells were plated in a black, glass-bottom, 96-well plate (Corning) at 40% confluency and synchronized as described for scKaryo-seq. Cells were imaged on an Andor CSU-W1 spinning disk (50 µm disk) with a ×0.75/20 NA dry objective lens (Nikon). A 488 nm laser was used for sample excitation, with filters between 540 and 50 nm bandpass for emission. Images were acquired using an Andor iXon-888 EMCCD camera. Nine z-slices of 2 μm were imaged for 4 h every 1 min. NEBD was defined as one frame before extensive chromosome movement. Images were acquired using NIS-elements (Nikon, v.5.30.04).
To determine the time from condensation to anaphase onset and segregation errors, we used a Nikon Ti-E motorized microscope equipped with a Zyla 4.2Mpx sCMOS camera (Andor) and a ×1.3/40 NA oil objective lens (Nikon). Fluorescence excitation was done using a Spectra X LED illumination system (Lumencor) and Chroma-ET filter sets. Nine z-slices of 2 μm were imaged every 4 min for 4 h. The same videos were also used to determine cell survival.
To examine cell survival for MN-seq, RPE1-hTERT Flp-in and BJ-hTERT cells were plated at 40% confluency. Cells were imaged on the same microscope used for determination of segregation errors. DIC and a ×0.45/10 NA objective lens (Nikon) were used to visualize cells every 3–5 min for 16 h.
Human intestinal organoids were imaged as described previously8.
To determine mis-segregations in cells treated with low-dose nocodazole, RPE1-hTERT H2B-eYFP cells were plated at 40% confluency 1 day before imaging. Next, cells were treated with nocodazole (48 nM; Sigma-Aldrich) and imaged on an Andor CSU-W1 spinning disk (50 µm disk) with a ×1.45/100 NA oil objective lens (Nikon). A 488 nm laser was used for sample excitation and filters between 540 and 50 nm bandpass for emission. Images were acquired using an Andor iXon-888 EMCCD camera. Nine z-slices of 2 μm were imaged for 16 h every 3 min.
To compare the behaviour of polar and non-polar chromosomes, RPE1-hTERT cells stably expressing both CENPA-GFP and Centrin1-GFP (a gift from A. Khodjakov) were imaged on the Expert Line easy3D STED microscope system (Abberior Instruments) using Prairie View (5.4.64.500) and Imspector (Abberior Instruments, v.16.3) with 485 and 640 nm lasers using a ×60/1.2 UPLSAPO 60×W water objective (Olympus) and an avalanche photodiode (APD) detector. Low-dose (1:100,000) SPY-595-DNA was added to detect the moment of nuclear envelope breakdown, and low-dose (1:50,000) SPY-640-tubulin (Spirochrome, AG) was added to distinguish between poles and kinetochores, as well as to enable pole tracking when the Centrin1 signal was not easily detectable in a specific frame. Six z-slices of 1 μm were taken every 20 s. Immediately after nuclear envelope breakdown, the edges of the nucleus were manually drawn to determine the relative nuclear position of tracked chromosomes by dividing the nucleus into three equally spaced concentric areas. Chromosomes were considered central if they resided in the two innermost shells or were touching the second-most outer ring. Positions of both centrosomes were also determined at that point. Each kinetochore pair was followed manually in a maximum-intensity projection. The positions and trajectories of the kinetochore pairs were additionally verified in single z-planes of a z-stack in Fiji (v.1.53f51/1.53s30/1.53r), as well as in Imaris 3D Viewer (v.9.8.0). One pair each of polar and non-polar peripheral chromosomes with the same distance to the metaphase plate were selected from the same cell.
U2OS kinetochore tracking experiments were performed with a U2OS cell line stably expressing CENPA-GFP, mCherry-α-tubulin and photoactivatable-GFP-α-tubulin (a gift from M. Barisic and H. Maiato). Cells were imaged using a Bruker Opterra I multipoint scanning confocal microscope system, as previously described52. Image acquisition was performed at 1 min intervals with z-stacks of 15 slices at 1 μm spacing. Misaligned kinetochores included all pairs of kinetochores displaced from the metaphase plate in the frame when elongation of the prometaphase spindle reached its peak, which was defined as the final point at which the separation of two centrosomes showed a continuous increase in spindle length for two consecutive frames >1 μm. Spatial x and y coordinates of unaligned kinetochores were extracted in every time frame using the Low Light Tracking Tool (v.0.10), an ImageJ plugin, as previously described53. The tracking of kinetochores in x and y planes was performed on individual imaging z-planes. Around 10–15% of unaligned kinetochore pairs could not be successfully tracked in all frames, mainly owing to cell and spindle movements in the z-direction over time. Spindle poles were manually tracked with points placed in the centre of the pole structure, in the z-plane in which the tubulin signal was highest. Aligned kinetochore pairs were manually tracked in two dimensions. All unaligned pairs in the NEBD frame were double-checked as being ‘behind spindle poles’ using a 3D Imaris Viewer. Lagging chromosomes were defined as a single kinetochore that was stuck and stretched between the separating mass of kinetochores during early anaphase. Chromosome bridges included cells with a kinetochore pair that was well separated but remained between the separating mass of kinetochores during early anaphase. Misalignments included cells that had at least one pair of kinetochores at the pole during anaphase, and the ‘no error’ phenotype was defined as a cell with absence of the aforementioned phenotypes. Multipolar cells (one out of 190) were not included in the analysis. Quantitative analysis of all parameters was performed using custom-made MATLAB (MatlabR2021a 9.10.0) scripts.
For live tracking of individual chromosomes, RPE1-hTERT dCas9-3×GFP were transduced with lentiviruses containing single-guide RNAs targeting chromosome 1 (ATGCTCACCT) and chromosome 9 (TGGAATGGAATGGAATGGAA). 24 h post transduction, cells were plated in an optical-quality, plastic, eight‐well slide (IBIDI) at 50% confluency. After 16 h, asynchronous mitotic cells were treated with 62.5 nM Cpd-5 and immediately imaged using a ×1.4/40 NA oil PLAN Apochromat lens on a Zeiss Cell Observer microscope equipped with a AxioImager Z1 stand, a Hamamatsu ORCA‐flash 4.0 camera and a Colibri 7 LED. Images were acquired every 2.5 min for 4.2 h. Videos were subsequently processed and analysed using ZEN software (Zeiss, v.3.3).
Chromosome 9 tracking and tethering experiments were performed on the spinning-disk system as previously described, with several adaptations; 500 nM rapalog (Takara) was added 24 h before imaging of DLD1 cells and 62.5 nM Cpd-5 was added immediately before imaging. We used a ×1.20/60 NA water phase immersion oil lens, and 16 z-slices of 1 μm were imaged every 3 min overnight.
Cells were imaged at 37 °C in 5% CO2 for all imaging experiments.
MN-seq
RPE1-hTERT Flp-in cells were plated in a six-well plate at 40% confluency and treated with Cpd-5 or nocodazole for 16 h. Cancer cell lines were plated in a similar fashion, but were not treated with any drugs. Preparations for FACS were performed similarly to the method described previously33. In short, cells were incubated on ice for 30 min under light with PBS/2% FBS and 12.5 μg ml–1 EMA (ThermoFisher). EMA was washed four times using PBS, and (micro)nuclei were harvested from cells with the same nuclear staining buffer used for scKaryo-seq. EMA-negative and Hoechst-positive (micro)nuclei were sorted in bulk in a PCR strip containing mineral oil and stored at −20 °C for further processing. Library preparation was performed similarly to scKaryo-seq, but with several modifications. Every 5 μl of sorted (micro)nuclei was incubated with 5 μl of lysis buffer (final concentration, 0.02 U Proteinase K μl–1 (NEB) in 1× CutSmart Buffer (NEB)) for 2 h at 55 °C and 10 min at 80 °C. Genomic DNA was digested by incubation of (micro)nuclei with 10 μl of digestion mix (final concentration, 0.5 U NLAIII μl–1 (New England Biolabs) in 1× CutSmart Buffer) for 2 h at 37 °C, followed by 20 min at 65 °C. Genomic DNA fragments were subsequently ligated to adaptors by the addition of 20 μl of ligation mix (final concentration, 20 U μl–1 T4 DNA ligase (New England Biolabs), 0.5 mM ATP (ThermoFisher) and 25 nM adaptor in 0.5× T4 DNA ligase buffer (New England Biolabs), with incubation at 16 °C overnight. After ligation, the remainder of library preparation, sequencing and analysis was performed as described for scKaryo-seq. To determine the percentage of reads per chromosome, all reads mapped to a specific chromosome were summed and normalized by dividing this by the number of bins for that specific chromosome. The percentage of reads for chromosome 10 in RPE1-hTERT cells was normalized using bulk-sequenced nuclei, because the q-arm of this chromosome is present in three copies.
DamID
U2OS DamID sequencing data were generated in bulk from clonal cell lines stably expressing Dam-LaminB1 or untethered Dam protein. DamID data from Shield1-inducible DamID U2OS cells were derived by transfection of Dam-LaminB1 or Dam constructs (cloned into the pPTuner IRES2 vector (Clontech, Takara)), antibiotic resistance selection with 500 µg ml–1 G418 (Gibco) and subsequent characterization of monoclonal cell populations. Selection of suitable clones was based on methylation concentrations at known LAD or iLAD genomic regions, measured by quantitative MboI-based PCR and DamID as previously described54. Stabilization of Dam proteins was achieved by the addition of Shield1 ligand (AOBIOUS) to the cell culture medium at 500 nM final concentration for 18–24 h before cell collection. Multiplexed DamID was performed as previously described54 and sequenced on an Illumina NextSeq 500 platform (1× 50 bp). Raw reads were demultiplexed by their library-specific index and sample-specific DamID barcode, universal DamID adaptor sequence was trimmed with cutadapt (v.1.16) and reads were aligned to reference genome hg19 using bowtie2 (v.2.3.4). Reads mapping to annotated GATC sites were counted and aggregated in genomic bins of 100 kb. Computation of observed over expected values per bin was performed as previously described55.
Statistics
Statistical analyses were performed using GraphPad Prism software (v.8.4.3). Superplots were used in many of the graphs in which each colour represents a replicate, the small dots individual measurements and large dots the mean of each replicate.
Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this paper.