CENP-A overexpression promotes distinct fates in human cells, depending on p53 status

Tumour evolution is driven by both genetic and epigenetic changes. CENP-A, the centromeric histone H3 variant, is an epigenetic mark that directly perturbs genetic stability and chromatin when overexpressed. Although CENP-A overexpression is a common feature of many cancers, how this impacts cell fate and response to therapy remains unclear. Here, we established a tunable system of inducible and reversible CENP-A overexpression combined with a switch in p53 status in human cell lines. Through clonogenic survival assays, single-cell RNA-sequencing and cell trajectory analysis, we uncover the tumour suppressor p53 as a key determinant of how CENP-A impacts cell state, cell identity and therapeutic response. If p53 is functional, CENP-A overexpression promotes senescence and radiosensitivity. Surprisingly, when we inactivate p53, CENP-A overexpression instead promotes epithelial-mesenchymal transition, an essential process in mammalian development but also a precursor for tumour cell invasion and metastasis. Thus, we uncover an unanticipated function of CENP-A overexpression to promote cell fate reprogramming, with important implications for development and tumour evolution.


Supplementary Figure 2 -Related to Figure 4 and Supplementary Table 1. RNA-seq non-inducible controls and gene cluster characterization
A) Principal component analysis (PCA) of gene expression profiles from bulk RNA-seq data, as described in Figure 4A, but including non-inducible (MCF10-2A WT) controls. Each dot represents a single sample. Legend indicated below. Red outline indicates Xirradiated samples (X-IR, 4 Gy). PC1 on X-axis shows that most of the variance can be attributed to increasing CENP-A levels across inducible cells. Non-inducible controls cluster separately and Dox treatment does not drive major differences in their expression. B) Effect of Dox on gene expression for inducible and non-inducible MCF10-2A cells.
Relative expression from RNA-seq data of the top 1000 upregulated genes (top) and top 1000 downregulated genes (bottom) by CENP-A overexpression in MCF10-2A TetOn-CENPA-FLAG-HA cells, plotted by increasing Dox concentrations for each cell line. Each line represents an individual gene, while the average is shown in red for each cell line. These genes are not significantly affected by Dox in the non-inducible control. C) Gene clusters 1-10 corresponding to heat map in Figure 4D. Box plots showing the distribution of relative expression levels of DEGs in each cluster, averaged by experimental condition. Each data point represents the mean expression level per condition, relative to average, for a particular DEG within the cluster (mean-centered counts, log2-transformed and TMM-normalized). Center lines show the medians; box limits indicate the 25th and 75th percentiles; whiskers extend 1.5 times the interquartile range from the 25th and 75th percentiles; outliers represented by dots. Main effects of experimental conditions are summarized at the bottom of each plot: up arrow indicates upregulation of genes in the cluster with CENP-A overexpression, X-irradiation, or p53-DN; down arrow indicates downregulation. Clusters 1 to 4 comprise genes that are upregulated as CENP-A is overexpressed, while genes in Clusters 5 to 10 are downregulated, each to varying degrees. Interestingly, genes in Clusters 6 to 8 also showed a coordinated response to radiation treatment or p53 inactivation. Importantly, the only cluster where p53 inactivation showed opposite effects to CENP-A overexpression was Cluster 8. Cluster 8 is also displayed in Figure 4E, duplicated here for direct comparison with other clusters. The top 10 enriched KEGG pathways for each DEG cluster are provided in Table S1. D) Reverse transcription quantitative PCR (RT-qPCR) for selected Cluster 8 genes, with PPIA as a reference gene. RNA extraction, DNase treatment and reverse transcriptase PCR of p53-WT and p53-DN MCF10-2A TetOn-CENPA-FLAG-HA cells, HCT116 TetOn-CENPA-FLAG-HA and DLD1 TetOn-CENPA-YFP-AID cells after 24h with 0, 10, or 100 ng/ml Dox (0X, 1X, 10X, respectively). Plots show gene expression for two biological replicates (dots) relative to PPIA, normalized to p53-WT 0X condition (MCF10-2A p53-WT 0X or HCT116 0X)(fold change ΔΔCt) on Log2 scale. were plated in culture dishes containing collagen/fibronectin-coated glass cover slips and then treated with 10 ng/ml Dox or no Dox control for 24h. Cells were then irradiated by Xray generator with 0-20 Gy and immediately fixed for immunofluorescence with γH2AX and CENP-A antibodies. Top: representative max intensity projection images from Zstack for cells with (+) or without (-) 10 ng/ml Dox after 0, 4 or 10 Gy X-irradiation. Scale bar = 20 µm. Bottom: quantification of mean γH2AX/DAPI intensity per nucleus for each condition (ImageJ). Center lines show the medians; box limits indicate the 25th and 75th percentiles as determined by R software; whiskers extend 1.5 times the interquartile range from the 25th and 75th percentiles. Light blue = No Dox. Dark blue = 1X (10 ng/ml) Dox. N = >90 for each condition B) Scheme of comet assays with relative CENP-A protein levels over time for MCF10-2A

Supplementary
TetOn-CENPA-FLAG-HA cell lines. To test the impact of p53 status and CENP-A overexpression on DNA damage directly, we performed alkaline comet assays after γirradiation on cells stably expressing empty vector (p53-WT) or p53-DN, with (+) or without (-) 24h 10 ng/ml Dox. C) Representative images of comets pertaining to scheme in B. D) Quantification of DNA damage measured as the product of the comet tail length and fraction of DNA in tail (olive tail moment) for each condition pertaining to scheme in B and representative images in C. Plots show mean with 95% confidence interval for the mean olive tail moment from three independent experiments (circles). # Same data displayed for 4 Gy 0h treatment shown in both plots for comparison. E) Quantification of chromosomal abnormalities from mFISH karyotypes in MCF10-2A TetOn New chromosome rearrangements per cell were determined as the total number of structural chromosomal anomalies observed per spread, excluding the ones that were common amongst the p53-WT no Dox spreads. F) Western blot of TCEs corresponding to Day 17 of growth curve in G for 0X and 1X Dox conditions. Load corresponds to ~50000 cells each. Primary antibodies are indicated on the right. # = high sensitivity ECL exposure. H4 used as loading control. G) Growth curve, as in Figure 4G Table 2. Single-cell RNA-seq epithelial cell clusters and EMT signatures A) UMAP plot of scRNA-seq data pertaining to scheme in Figure 5A, showing three main clusters of cells based on Leiden clustering after cell cycle correction. Cluster I (orange), Cluster II (light blue), and Cluster III (dark blue), are sub-divided as either Cycling (dark) or Non-cycling (pale), according to the previous analysis in Figure 5. See Table S2 for gene markers associated with each cluster, ranked by one-vs-all logistic regression. Cluster I is negative for epithelial markers (e.g. CDH1, EPCAM and KRT19) and shows high expression of mesenchymal genes (e.g. VIM, FN1 and CDH2). Clusters II and III are positive for epithelial markers but show differential expression of genes involved in cell metabolism (e.g. CTSV, AFMID, GLA, ACAT2, CBR4). B) Top: Cell-to-cell variability along the second principal component of the scRNA-seq data can be explained by differences in expression of cell metabolism genes. Principal component analysis (PCA) of scRNA-seq experiment. All conditions merged. Each dot represents a single cell on the first (PC1) and second (PC2) principal component, coloured by expression of genes involved in Fatty Acid (FA) Metabolism, Oxidative Phosphorylation (Phos.), and Glycolysis (Hallmark gene sets, MSigDB v6.2). Bottom: As above, except data plotted in UMAP space. Epithelial clusters II and III show low and high expression of genes involved in cell metabolism processes, respectively. C) Bar plot showing percentage of scRNA-seq Clusters II (cell metabolism low) and III (cell metabolism high) in A by experimental condition. Proportions are similar for untreated p53-WT and p53-DN cells. CENP-A overexpression (acute or chronic) reduces the proportion of cell metabolism high cells in both cases and the opposite for cell metabolism low cells. We speculate that this represents natural cell-to-cell variation in cell metabolism associated with nutrient availability in the cells, which may be hindered by increased production of CENP-A. D) UMAP of scRNA-seq samples across all conditions, coloured by the score for selected expression signatures including genes that are up-or down-regulated (arrows) after EMT or in luminal-and basal-like versus mesenchymal breast cancer cell lines (curated gene sets from MSigDB v6.2, see Table S2).

Supplementary Figure 6 -Related to Figure 6. Early detection of EMT in p53-DN MCF10-2A cells by brightfield microscopy and lack of EMT in HCC1954.
A) Brightfield images of cells after 6 and 10 days passed in parallel with either no Dox (-) or continuous CENP-A overexpression (++, 10 ng/ml Dox). White dashed lines in zoomed images outline edges of epithelial cell groups. Epithelial cells are characterized by strong cell-cell contacts, compared to cells with mesenchymal-like characteristics (isolated/showing reduced cell-cell contacts and elongated/distorted shape). Images taken of live cells by brightfield microscopy with a 4X objective. Scale bars = 40 µm B) Assessment of EMT by immunofluorescence of HCC1954 TetOn-CENPA-FLAG-HA cells after 34 days with (++) or without (-) continuous CENP-A overexpression (10 ng/ml Dox). Representative max intensity projections: DAPI (cyan), E-cadherin (yellow, epithelial marker), and Vimentin (magenta, mesenchymal marker). Scale bars = 40 µm.
No EMT was detectable in any of the images taken.

Supplementary Figure 8 -Uncropped Western blots from Figure 3
Pierce Universal Protein stain, colorimetric images and primary antibodies indicated on the right. Antibodies used on the same membrane stained in succession in the order shown. # = high sensitivity ECL.

Supplementary Figure 10 -Uncropped Western blots from Supplementary Figures 3 and 4
Pierce Universal Protein stain, colorimetric images and primary antibodies indicated on the right. Antibodies used on the same membrane stained in succession in the order shown. # = high sensitivity ECL. Dox used at 10ng/ml (1X).