Contribution of CENP-F to FOXM1-Mediated Discordant Centromere and Kinetochore Transcriptional Regulation

Abstract

Proper chromosome segregation is required to ensure chromosomal stability. The centromere (CEN) is a unique chromatin domain defined by CENP-A and is responsible for recruiting the kinetochore (KT) during mitosis, ultimately regulating microtubule spindle attachment and mitotic checkpoint function. Upregulation of many CEN/KT genes is commonly observed in cancer. Here, we show that although FOXM1 occupies promoters of many CEN/KT genes with MYBL2, FOXM1 overexpression alone is insufficient to drive the FOXM1-correlated transcriptional program. CENP-F is canonically an outer kinetochore component; however, it functions with FOXM1 to coregulate G2/M transcription and proper chromosome segregation. Loss of CENP-F results in altered chromatin accessibility at G2/M genes and reduced FOXM1-MBB complex formation. We show that coordinated CENP-FFOXM1 transcriptional regulation is a cancer-specific function. We observe a small subset of CEN/KT genes including CENP-C, that are not regulated by FOXM1. Upregulation of CENP-C in the context of CENP-A overexpression leads to increased chromosome missegregation and cell death suggesting that escape of CENP-C from FOXM1 regulation is a cancer survival mechanism. Together, we show that FOXM1 and CENP-F coordinately regulate G2/M genes, and this coordination is specific to a subset of genes to allow for maintenance of chromosome instability levels and subsequent cell survival.

Keywords: Centromere; cell cycle; chromosome segregation; kinetochore; transcriptional regulation.

Figure 1.

CENP-F and FOXM1 systematically regulate chromatin and centromeric function. (A) Immunoblot analysis showing protein knockdown in response to 20 nM siRNA treatment of CENP-F, FOXM1 or double KD for 48 h. Wedge represents loading titration: 1× > 0.75× > 0.5× > 0.25× > 0.125×. All other lanes are loaded at 1× concentration. N = 2. (B) Quantification of CENP-F and FOXM1 protein level fold changes compared to siNeg from immunoblot shown in (A). P values calculated using two-tailed paired Student’s t test. N = 3. (C) Representative images of hTERT-RPE1 cell treated with siCENP-F, siFOXM1 or both. Immunofluorescence was conducted to visualize CENP-F (green), FOXM1 (red), and DAPI (blue). (D) Representative mitotic images of hTERT-RPE1 cells treated with siCENP-F, siFOXM1 or both. Immunofluorescence was conducted to visualize centromeres by ACA (red), α-tubulin (green), and DNA by DAPI (blue). Arrows indicated the presence of lagging chromosomes. (E) Quantification of micronuclei frequency from (D). P values calculated using one-tailed Student’s t test. N = 3. (F) Quantification of lagging anaphase events from (D). P values calculated using one-tailed Student’s t test. N = 3.

Figure 2.

CENP-F and FOXM1 coordinately co-regulate G2/M genes. (A) Fold change of CENP-F and FOXM1 transcript levels normalized to GAPDH, in response to 20 nM siRNA treatment for 48 h in hTERT-RPE1 cells. P values calculated using two-tailed paired Student’s t test. N = 3. (B) Scatterplot of genes from RNA-seq plotted as log₂FC in siCENP-F vs siFOXM1 treated cells. Correlation coefficient and P value was calculated using Pearson’s method. N = 3. (C) Heatmap showing Log₂FC of top 500 upregulated and downregulated genes in siCENP-F and siFOXM1 conditions, clustered by unbiased euclidean method. (D) GO-term analysis showing top five enriched GO-terms in top 500 downregulated genes in siCENP-F, siFOXM1 and siDouble cells. GO-terms enriched in multiple conditions are plotted on a single line with different colored circles. (E) Heatmap showing overlapping genes between top 500 downregulated genes in the RNA-seq datasets and the mitotic cell cycle gene set (enriched in both siFOXM1 and siDouble). Log₂FC of overlapping genes are shown in all conditions.

Figure 3.

CENP-F influences the chromatin landscape of G2/M genes. (A) Log₂FC of ATAC-seq signal in hTERT-RPE1 cells treated with siCENP-F and siFOXM1 for 48 h, mapped against FOXM1 binding sites in MCF7 and hTERT-RPE1 cells, with k-mean clustering = 3 (bottom). N = 2. (B) Graphical depiction of ATAC-seq profiles showing changes in accessibility under siCENP-F or siFOXM1 conditions in each cluster. (C) GO-term enrichment analysis of each cluster. Terms enriched in multiple groups are represented on the same line. Negative log₁₀ P value of enrichment is represented on the y-axis. (D) Heatmap of the subset of genes in cluster 1 and 2 from ATAC-seq that have log₂FC>|0.5| in labelled RNA-seq data set. Cell cycle associated genes are highlighted with red arrows.

Figure 4.

CENP-F regulates FOXM1-MMB complex formation. (A) TCGA pan cancer analysis of gene expression represented as z-score of cancer vs normal of FOXM1, MYBL2 and CENP-F and components of the MuvB complex. (B) Immunoblot of FLAG immunoprecipitation in FLAG-FOXM1-hTERT-RPE1 cells. Cells were treated with doxycycline for 48 h to induce FOXM1 expression and simultaneously treated with 20 nM siNeg or siCENP-F. (C) Quantification of immunoprecipitation of LIN54 or MYBL2 shown in (B). P values calculated using two-tailed Student’s t test. N = 3.

Figure 5.

FOXM1 and MYBL2 co-occupy a specific subset of CEN/KT proteins. (A) Metaplot of FOXM1 and MYBL2 log₂FC ChIP signal vs IgG at FOXM1-MYBL2 intersected consensus peaks with k-means clustering = 3. N = 3. (B) Venn diagram showing unique and overlapping peaks in FOXM1 and MYBL2 ChIP-seq. (C) Transcript levels represented by normalized counts per million from RNA-seq in hTERT-RPE1 cells for each ChIP-seq cluster. (D) Annotation analysis of genes in each cluster, y-axis shows percentage of peaks within each annotation category. (E) GO-term enrichment analysis for each cluster. (F) TCGA pan cancer analysis of gene expression represented as z-score of cancer vs normal of CEN/KT subcomplexes. (G) Correlations of CEN/KT components with FOXM1 from TCGA data. Scatterplot of select genes shows degree of correlations between FOXM1 and CENPA or CENPC of the CCAN complex and FOXM1 and CENPF or MIS12 of the kinetochore complex. Pearson’s correlation coefficient was calculated from z-scores of cancer versus normal. (H) FOXM1 and MYBL2 ChIP profiles in hTERT-RPE1 cells showing peaks at selected genes. N = 3.

Figure 6.

FOXM1-CENP-F cancer-specific regulation excludes CENP-C to promote survival in cancer cells. (A) Immunoblotting of hTERT-RPE1, MCF7 and PC-3 cells treated with 20 nM siNeg or siFOXM1 for 48 h. N = 3. (B) RT-qPCR of CENP-F transcript levels, normalized to GAPDH, in FOXM1 knockdown conditions across normal and cancer cell lines. P values calculated using two-tailed Student’s t test. N = 3. (C) RT-qPCR of CENP-A transcript levels, normalized to GAPDH, in knockdown across normal breast epithelial (MCF10A) and breast cancer (MCF7) cell lines. P values calculated using two-tailed Student’s t test. N = 3. (D) RT-qPCR of CENP-C transcript levels, normalized to GAPDH, in knockdown conditions across normal breast epithelial (MCF10A) and breast cancer (MCF7) cell lines. P values calculated using two-tailed Student’s t test. N = 3.(E) RT-qPCR of CENP-A transcript levels, normalized to GAPDH, in knockdown across normal prostate epithelial (RWPE1) and prostate cancer (PC-3) cell lines. P values calculated using two-tailed Student’s t test. N = 3. (F) RT-qPCR of CENP-C transcript levels, normalized to GAPDH, in knockdown conditions across normal prostate epithelial (RWPE1) and prostate cancer (PC-3) cell lines. P values calculated using two-tailed Student’s t test. N = 3.

Figure 7.

CENP-C escapes CENP-F—FOXM1 regulation as a cancer survival mechanism. (A) Representative images of Tet-CENP-A-mCherry HeLa-Trex cells induced for 48 h and transfected with either CENP-C LAP or LAP empty vector (EV). Imaging was conducted for mCherry (CENPA), eGFP (LAP) and DAPI. (B) Quantification of percentage of cells with micronuclei in each condition shown in (A). P values calculated using two-tailed paired Student’s t test. N = 3. (C) Percent cell viability calculated by growth curves over 72 h post transfection in each condition. P values calculated using two-tailed paired Student’s t test. N = 3. (D) Illustrated model showing differential CENP-F-FOXM1 regulation in the context of cancer and how escape of CENP-C from this regulation protects cancer cells from fatal CIN.