Spatial organization of centromeres is cell-type specific in human cell lines.

a, Representative images of CENP-C (green) and DAPI (gray) stained nuclei in indicated human cell lines. Scale bar: 10 µm. b, c, Spatial organization of centromere quantified using Ripley K’s clustering score (b), CENP-C spot count (c). d, Nuclear area and e, mean radial distance in human cell lines. Statistical significance of difference between cell lines for clustering score, spot count, mean radial distance and nuclear area was tested using ANOVA (p-value or ‘Pr(>F)’ < 2e-16) following Tukey’s HSD test to compare means of all pairs of cell lines. Box plots represent the inter-quantile range (IQR) between first and third quantile (box), the median (horizontal bar), and the whiskers extend to the highest or lowest data point up to 1.5 times of IQR. Values are from one representative experiment with at least 7 technical replicates. At least 1000 cells were analyzed in each category per experiment.

Identification of the molecular determinants of spatial centromere distribution across cell types.

a, Schematics showing three stages of high-throughput imaging based arrayed CRISPR knockout screen employed to identify molecular determinants of spatial centromere distribution. Created with BioRender.com/6w483wa. b, Changes in spot count (mean Z-score of two replicates, y-axis) and clustering score (mean Z-score of two replicates, x-axis) for each of the 1064 sgRNAs. The most prominent hits were labelled and color coded as in Fig. 2d. Non-hits are colored in gray. c, Changes in clustering score in HCT116 (mean Z-score of two replicates, x-axis) and in RPE1 (mean Z-score of two replicates, y-axis) cells for each of the 1064 sgRNAs. Hits and non-hits are color coded and labelled as in b. A linear trend line (gray) was fitted to the data and Pearson’s correlation coefficient calculated is indicated at the top left corner of the plot. d, Network diagram with lines between 52 common hits drawn based on known physical and/or genetic interactions generated by the STRING database. Thickness of the lines indicates higher strength of data supporting the interaction. Broad categories are color coded as indicated. e, Plot shows changes in clustering (clustered or unclustered), count (higher or lower) and direction between two cell lines (same or opposite) for each of the common genes that are color coded based on their category as in Fig. 2d. Counts of genes in each subcategory are indicated. Values represent two biological replicates. Typically, 200-500 cells were analyzed for each target gene per experiment.

Changes in spatial organization of centromeres require progression through the cell cycle.

a, HCT116 and RPE1 cells at G1, S and G2/M phases stained with CENP-C (green), DAPI (gray) and EdU (red). Scale bar: 10 µm. b, EdU intensity (y-axis) and DAPI intensity (x-axis) showing separation between G1 (brown), S (gray), and G2/M (green) sub populations in cycling HCT116 cells. Comparison of cells at G1, S or G2/M stages (x-axis) for their clustering score (c), spot count (d), mean radial distance (e), or nuclear area (f). Statistical significance of differences was tested by pairwise t-test with Bonferroni correction. Asterisks indicate level of significance between a given pair reflecting the corresponding p-value of that comparison. g, A linear regression line (red) fitted through the single cell data for nuclear area (y-axis) and clustering score (x-axis) in cells at different cell cycle phases in HCT116 and RPE1 cells. Pearson correlation coefficient and respective adjusted p-values are indicated at the top of each panel. h, Experimental outline to test cell cycle stage-specific effect of knocking down select hits. Created with BioRender.com/g7atcbi. i, Effect of siRNA knockdown for a panel of genes (x-axis) using three individual siRNAs per gene in HCT116 cells that are either arrested at G/S and G2 or cycling. Two control siRNAs for siNCAPH2 are in blue and siScrambled in yellow. Mean value for siScrambled is depicted by a horizontal yellow dotted line. Statistical significance of differences was tested by performing pairwise t-tests with Bonferroni correction using siScrambled as control group. Box plots represent the inter-quantile range (IQR) between first and third quantile (box), the median (horizontal bar), and the whiskers that extend till the highest or lowest value up to 1.5 times of IQR. Values are from one representative experiment. Typically, 200 to 500 cells were analyzed in each category. Statistical significance of difference was denoted by stars where * indicates p ≤ 0.05, ** indicates p ≤ 0.01, *** indicates p ≤ 0.001 and **** indicates p < 0.0001.

Progression through S-phase in the absence of select clustering factors does not alter interphase genome organization

a, Experimental outline to compare centromere distribution during progression through S-phase in the presence or absence of clustering factors. b, c, Clustering score in all cells (b) or G2/M cells (c) in the presence (blue) or absence (red) of indicated clustering factors before and after S-phase release from G1/S arrest. Pair-wise comparisons were tested with t-test with Bonferroni correction, and the level of significance is indicated by asterisks if any. Pairs without significant difference are not labelled. Box plots represent the inter-quantile range (IQR) between first and third quantile (box), the median (horizontal bar), and the whiskers that extend till the highest or lowest value up to 1.5 times of IQR. Values are from one representative experiment with three technical replicates. Typically, 200 to 500 cells were analyzed in each category.

Orderly progression through mitosis is required to re-establish centromere distribution.

a, Experimental outline to compare centromere distribution during mitotic progression in the presence or absence of clustering factors. b, c, Clustering score in all cells (b) or G1 cells (c) in the presence (blue) and absence (red) of indicated clustering factors before (0 h) and after (6 hrs) mitotic release from G2 arrest. Pair-wise comparisons were tested with t-test with Bonferroni correction, and the level of significance is indicated by asterisks. Pairs without significant differences are not labelled. d, Representative images showing G1 nuclei stained with DAPI (gray) and CENP-C (green) in the presence or absence of indicated factors. Scale bar: 10 µm. e, Schematics for co-depletion of indicated factors. f, Clustering score (y-axis) in G1 cells after siRNA knockdown of indicated factors (x-axis) in presence (blue) or absence (red) of SPC24, KI67 or NCAPH2 as indicated. Statistical significance of difference between indicated pairs were tested by performing t-test with Bonferroni corrections for multiple comparisons and denoted by stars, where * indicates p ≤ 0.05, ** indicates p ≤ 0.01, *** indicates p ≤ 0.001 and **** indicates p < 0.0001. Comparisons between individual siRNA groups (x-axis) and inter-group comparisons in the presence and absence of the given degron tagged protein are shown in black, blue (presence) and red (absence), respectively. g, Clustering score (y-axis) in cells that were depleted (red) or depleted of indicated factors and then re-expressed (green) or remained unperturbed (blue). Statistical significance of difference between indicated pairs were tested by performing t-test with Bonferroni corrections for multiple comparisons and denoted by stars, where higher number of stars indicate higher confidence levels. Box plots represent the inter-quantile range (IQR) between first and third quantile (box), the median (horizontal bar), and the whiskers that extend till the highest or lowest value up to 1.5 times of IQR. Values are from one representative experiment containing three technical replicates. Typically, 200 to 500 cells were analyzed for each category. Statistical significance of difference was denoted by stars where * indicates p ≤ 0.05, ** indicates p ≤ 0.01, *** indicates p ≤ 0.001 and **** indicates p < 0.0001.

Mitotic events shape interphase genome organization

A model showing defective loading of outer kinetochore (inset; green and red box) in late G2 leads to uncoordinated metaphase alignment and aberrant migration towards spindle pole during anaphase that lowers chances of interactions between centromeres during telophase and the lack of homotypic adhesion results in dispersion of centromeres in the daughter nuclei. Created with BioRender.com/j8y5xk9.

Quantification of centromere clustering using CENP-A and CENP-C as centromere markers.

a, Co-staining of HCT116 cells with CENP-A (red), CENP-C (green) and DAPI (gray). Scale bar: 10 µm b, Representative image showing segmentation of DAPI stained nuclei (gray) and CENP-C-stained centromere spots (green) in high-throughput imaging data using HiTIPS. Red lines around the DAPI stained nuclei indicate nuclear segmentation, the green circles around CENP-C spots indicate segmentation of centromeres. Zoomed images of the same nucleus are shown with yellow and pink borders indicate before and after spot segmentation was applied. Scale bar: 10 µm. c, Quantification of spot count and d, Clustering Score using CENP-A (red) and CENP-C (blue) as centromere markers in HCT116 cells. Values are from two replicates, with at least 2000 cells analyzed for each experimental condition.

CRISPR knockout screens for centromere distribution phenotypes in HCT116 cells are reproducible.

a, Mean and standard deviation for phenotypic separation between control sgRNAs with individual data points representing mean value per well for number of spots per nucleus and b, Clustering Score in two biological replicates. c, scatter plot showing changes in Clustering Score or d, spot count for replicate 1 (x-axis) and replicate 2 (y-axis) in HCT116 cells for each of the 1068 sgRNAs. A linear regression line (gray) was fitted to the data and Pearson’s correlation coefficient calculated is indicated at the top left corner of the plot. e, scatter plot and linear regression line correlating changes in Clustering Score (y-axis) and f, spot count (y-axis) with nuclear area (x-axis). Pearson’s correlation coefficients and corresponding p-values are indicated at the top left corner. Values are from 2 biological replicates. Typically, 200 to 500 cells were imaged for each target gene per experiment.

Identification of the molecular determinants of spatial centromere distribution in RPE1 cells.

Changes in spot count (mean Z-score of two replicates, y-axis) and Clustering Score (mean Z-score of two replicates, x-axis) for each of the 1068 sgRNAs. The most prominent hits were labelled and color coded as in Fig. 2d. Non-hits are colored in gray. Values are from two biological replicates. Typically, 200 to 500 cells were analyzed for each target gene per experiment.

Validation of screen hits, cell cycle analysis of clustering factor knockdown and their effect on Clustering Score.

a, Clustering scores of targets (x-axis) after siRNA knockdown in HCT116 cells. Two control siRNAs for siNCAPH2 are in blue and siScrambled is shown in yellow. Upon siRNA knockdown, the Clustering Score or spot count for the targets (x-axis) labelled in green change in the same direction as in the CRISPR-KO screens and are compared to the mean value for siScrambled as depicted by a horizontal yellow dotted line by performing pairwise t-tests with Bonferroni correction. Significantly different pairs are labelled with stars, where * indicates p ≤ 0.05, ** indicates p ≤ 0.01, *** indicates p ≤ 0.001 and **** indicates p < 0.0001. Targets in gray disagree with either Clustering Score or spot count or both parameters compared to data in CRISPR-KO screens. Three separate siRNA were used per target. b. Fraction of cells in each cell cycle stage (x-axis) after knockdown of select targets as indicated (y-axis). Individual sub-populations of the cell cycle are color-coded as identified using DAPI and EdU fluorescence intensity measurement, and their percentages are indicated. c, Bar plots showing the fraction of G1, S, and G2/M cells (y-axis) after siRNA knockdown of select targets (red) and scrambled siRNA control (blue). Statistical significance of difference (p<0.05) was tested using t-test with FDR correction as compared to the scrambled control and significantly different targets are labelled as stars, where * indicates p ≤ 0.05, ** indicates p ≤ 0.01, *** indicates p ≤ 0.001. Higher number of stars indicate lower p-value. d, Clustering score (y-axis) for select targets (x-axis) at G1 (brown), S (gray) and G2/M (green) stages. Values are from one representative experiment. Typically, 200 to 500 cells were analyzed per gene per experiment.

Comparative analysis of cell lines confirms common molecular determinants of spatial centromere distribution.

a, 52 genes (black mesh) that are hits in both HCT116 and RPE1 cells. A total of 113 hits were selected in RPE1 cells for either Clustering Score (pink, 45) or spot count (blue, 87) and 111 hits in HCT116 cells for either Clustering Score (gold, 89) or spot count (black, 45). White non-shaded areas indicate unique hits in HCT116 (51) and RPE1 (53) cells. Values are from one representative experiment. Typically, 200 to 500 cells were analyzed for each target gene per experiment. b, Z-scores (x-axis) of spot count and Clustering Score for the 52 common hits (y-axis) in HCT116 and RPE1 cells. Genes are color coded based on their category as indicated in Fig. 2e.

Validation of protein depletion in siRNA knockdown a,

Number of cells per well at 72 hours after transfection of siDEATH and siScrambled control siRNAs in cycling (red), G1/S (green) or G2/M (blue) synchronized HCT116 cas9 cells as indicated. b, Western blots showing NCAPH2 protein levels after 72 hours of siRNA knockdown using siNCAPH2 and siScrambled in cycling, G1/S or G2/M synchronized HCT116 cas9 cells as indicated. NCAPH2 levels across samples were normalized using β-Actin and quantitated band intensities of NCAPH2 in siNCAPH2 samples were expressed as fraction of the corresponding siScrambled samples at cycling, G1/S and G2/M cells.

Construction and genotyping of FLAG-dTAG-SPC24 and NUF2-dTAG-FLAG cell lines.

a, b, CRISPR knock-in strategy for homozygous tagging of SPC24 (a) and NUF2 (b) with the dTAG-FLAG epitope. Horizontal black arrows indicate positions of primers used for PCR confirmation of the tagged allele. c, d, PCR genotyping of single cell clone for FLAG-dTAG-SPC24 (c) and NUF2-dTAG-FLAG based on the strategy explained in a and b.

Characterization of FLAG-dTAG-SPC24 and NUF2-dTAG-FLAG cell lines.

a, b, Representative images of knock-in cell lines expressing FLAG-dTAG-SPC24 (a), and NUF2-dTAG-FLAG (b) stained with DAPI (gray), CENP-C (green) and FLAG (red). Scale bar: 10 µm c, d, Westen blot images showing levels of FLAG-dTAG-SPC24 (c), and NUF2-dTAG-FLAG (d) at indicated time points after incubation with dTAG ligands and the relative ratios of dTAG-SPC24 or NUF2-dTAG to tubulin control compared to the level at the beginning of depletion (0 hr) are indicated below. e, f, western blots showing comparative levels of SPC24 and FLAG-dTAG-SPC24 (e) and NUF2 and NUF2-dTAG-FLAG (f) proteins in the indicated cell lines. Relative intensity ratio of FLAG-dTAG-SPC24 or NUF2-dTAG-FLAG proteins in the respective cell lines to the untagged SPC24 or NUF2 protein level in HCT116 cells is indicated.

Quantification of the cell cycle stage specific depletion of FLAG-dTAG-SPC24 and NUF2-dTAG-FLAG.

a, b, Western blots showing degron-based depletion of FLAG-dTAG-SPC24 in G1/S synchronized and cycling cells (a) and G2/M synchronized and cycling HCT116 cells (b). c. d, Western blots showing degron-based depletion of SPC24-dTAG-FLAG in G1/S synchronized and cycling cells (c) and G2/M synchronized and cycling HCT116 cells (d).

Quantification of cell cycle stages during G1 and mitotic release experiment.

a, Fraction of G1 (blue), S (green) and G2/M (orange) cells were quantified before (0h) and after (6h) release from double thymidine block in the presence or absence of indicated clustering factors. b, Fraction of G1 (blue), S (green) and G2/M (orange) cells were quantified before (0h) and after (6h) release from G2/M block in the presence or absence of indicated clustering factors. Values are from one representative experiment containing three technical replicates. Typically, 200 to 500 cells were analyzed per sample. c, and d, shows percent of G1, S and G2/M cells (x-axis) in presence (blue) and absence (red) of indicated mitotic factors before (0h) and after (6h) release from G1/S block (c) and G2/M block (d). Statistical significance of difference was tested using t-test and significantly (p< 0.05) different pairs are indicated by stars, where * indicates p ≤ 0.05, ** indicates p ≤ 0.01. Higher number of stars indicate lower p-value.

Mitotic defects in the absence of SPC24.

a-d, Representative images showing examples of mitotic defects including karyokinesis defect (a), mis-segregation (b), micronuclei formation (c), and aberrant metaphase alignment (d) observed at 6 hours after G2/M arrested HCT116 cells were released in the absence of SPC24. Cells were immunofluorescent labelled with CENP-C (green) and DAPI (gray). Scale bar: 10 µm