Micronuclei from misaligned chromosomes that satisfy the spindle assembly checkpoint in cancer cells

Abstract

Chromosome alignment to the spindle equator is a hallmark of mitosis thought to promote chromosome segregation fidelity in metazoans. Yet chromosome alignment is only indirectly supervised by the spindle assembly checkpoint (SAC) as a byproduct of chromosome bi-orientation, and the consequences of defective chromosome alignment remain unclear. Here, we investigated how human cells respond to chromosome alignment defects of distinct molecular nature by following the fate of live HeLa cells after RNAi-mediated depletion of 125 proteins previously implicated in chromosome alignment. We confirmed chromosome alignment defects upon depletion of 108/125 proteins. Surprisingly, in all confirmed cases, depleted cells frequently entered anaphase after a delay with misaligned chromosomes. Using depletion of prototype proteins resulting in defective chromosome alignment, we show that misaligned chromosomes often satisfy the SAC and directly missegregate without lagging behind in anaphase. In-depth analysis of specific molecular perturbations that prevent proper kinetochore-microtubule attachments revealed that misaligned chromosomes that missegregate frequently result in micronuclei. Higher-resolution live-cell imaging indicated that, contrary to most anaphase lagging chromosomes that correct and reintegrate the main nuclei, misaligned chromosomes are a strong predictor of micronuclei formation in a cancer cell model of chromosomal instability, but not in non-transformed near-diploid cells. We provide evidence supporting that intrinsic differences in kinetochore-microtubule attachment stability on misaligned chromosomes account for this distinct outcome. Thus, misaligned chromosomes that satisfy the SAC may represent a previously overlooked mechanism driving chromosomal/genomic instability during cancer cell division, and we unveil genetic conditions predisposing for these events.

Keywords: Mad2; aneuploidy; cancer; chromosomal instability; chromosome congression; cyclin B1; kinetochore; micronuclei; mitosis; spindle assembly checkpoint.

Graphical abstract

Figure 1

Schematic illustration of the high-content analysis of chromosome alignment defects Different steps between protocol optimization and automated live-cell imaging of 125 different RNAi conditions against genes previously implicated in chromosome congression.

Figure 2

A broad range of chromosome alignment defects directly lead to missegregation (A) Examples of time-lapse sequences illustrating the three main mitotic phenotypes observed. Arrows indicate chromosomes at the poles in cells exhibiting chromosome alignment defects. Pixels were saturated for optimal visualization of misaligned chromosomes. Scale bars, 5 μm. Time, h:min. (B) Quantification of congression phenotypes in control (siScramble) and siRNA-depleted cells. At least 2 independent experiments per condition were performed. The total number of cells analyzed for each condition is indicated in Data S1. ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001, ^∗∗∗∗p ≤ 0.0001; ns, not significantly different from control; Fisher’s exact two-tailed test; # highlights a possible off-target associated with siRNA oligo 1 against HURP.

Figure 3

Mild, yet penetrant, chromosome alignment defects are compatible with mitotic progression and cell viability (A) Examples of time-lapse sequences illustrating the fates exhibited by HeLa cells undergoing congression defects following siRNA knockdown. Time, h:min, from nuclear envelope breakdown (NEB) to each cellular outcome. Scale bars, 5 μm. (B) Frequency of cells that either died in mitosis (magenta) or died in interphase (green) in control and siRNA-depleted cells. The total number of cells analyzed for each condition is indicated in Data S1. ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001, ^∗∗∗∗p ≤ 0.0001; ns, not significantly different from control; Fisher’s exact two-tailed test; # highlights a possible off-target associated with siRNA oligo 1 against HURP. (C) Correlation between mitotic duration and cell death in mitosis for each condition. (D) Correlation between the severity of the congression phenotypes and the frequency of cell death in mitosis. (E) Correlation between the mitotic duration after siRNA treatment and cell death in interphase. (F) Correlation between congression severity and the frequency of cell death in interphase. Pearson’s correlation (r) and respective p values are indicated in the plots (two-tailed test).

Figure 4

Cells with misaligned chromosomes enter anaphase after satisfying the spindle assembly checkpoint (A) Selected time frames of representative HeLa cells stably expressing Mad2-GFP (green) and chromosomes labeled with SiR-DNA (magenta) in control and after CENP-E depletion. White arrowheads point to a misaligned chromosome during anaphase. Time, min:s. Time 00:00, anaphase onset. (B) Immunofluorescence of HeLa cells stained for DNA (blue), Mad1 (green), CENP-C (white), and β-tubulin (magenta). Insets show higher magnification of selected regions with misaligned chromosomes (grayscale for single channels of Mad1 and CENP-C). Images are maximum-intensity projections of deconvolved z stacks. Scale bars, 5 μm. (C) Quantification of the fluorescence intensity of Mad1 relative to CENP-C on misaligned chromosomes. Each dot represents an individual kinetochore. The horizontal line indicates the mean of all quantified kinetochores, and the error bars represent the standard deviation from a pool of two independent experiments (mock/prometaphase, n = 90 kinetochores, 9 cells; siCENP-E/prometaphase, n = 72 kinetochores, 17 cells; siCENP-E/anaphase, n = 19 kinetochores, 14 cells; ^∗∗∗∗p ≤ 0.0001 relative to control, Mann-Whitney test).

Figure 5

Cells with misaligned chromosomes enter anaphase after undergoing normal cyclin B1 degradation (A) Selected time frames from live-cell microscopy of HeLa cells stably expressing H2B-mCherry and cyclin B1-Venus in control, CENP-E, and TACC3 RNAi. Time, min:s. Time 00:00, anaphase onset. Scale bars, 5 μm. Black arrowheads point to misaligned chromosomes at anaphase onset. (B) Cyclin B1 degradation curves for control, CENP-E-, and TACC-3-depleted cells that properly align their chromosomes at the metaphase plate or exit mitosis with misaligned chromosomes and form micronuclei. The curves represent mean cyclin B1-Venus fluorescence intensity from all analyzed cells, and error bars represent the standard deviation from a pool of two independent experiments (siScramble n = 20; siCENP-E [misaligned + micronuclei] n = 22; siCENP-E [aligned] n = 15; siTACC3 [misaligned + micronuclei] n = 5; siTACC3 [aligned] n = 12). (C) Frequency of anaphase cells with aligned chromosomes, misaligned chromosomes, and misaligned chromosomes that result in micronuclei in control (black bars), CENP-E- (green bars), and TACC3-depleted cells (magenta bars).

Figure 6

Although most micronuclei originate from anaphase lagging chromosomes, misaligned chromosomes are a stronger predictor of micronuclei formation (A) Examples of time-lapse sequences illustrating the different origins of micronuclei. Time, min:s. Time 00:00, anaphase onset. White arrowheads track misaligned chromosomes, DNA bridges, or lagging chromosomes until they eventually form micronuclei. Pixels were saturated for optimal visualization of misaligned chromosomes, DNA bridges, and lagging chromosomes. Scale bars, 5 μm. (B) Frequency of daughter cells with micronuclei that derived either from lagging chromosomes (black bars), DNA bridges (green bars), or misaligned chromosomes (magenta bars) under the specified conditions (siScramble, n = 1,700; MonWO, n = 327; siAstrin, n = 423; siBub1, n = 457; siKif18a, n = 540; siCENP-N, n = 422; siSka1, n = 395; siTACC3, n = 485; siNsl1, n = 400; siSka3, n = 383; siZw10, n = 404; siNdc80, n = 440; siAurora A, n = 388; siCLERC, n = 263; siNuf2, n = 428; siCENP-I, n = 389; siAurora B, n = 499; siDsn1, n = 688; siCENP-E, n = 346; siSpc24, n = 418; siBubR1, n = 387; siSpc25, n = 425; siHURP_oligo1, n = 296; siHURP_oligo2, n = 200; siKNL1, n = 413; pool of 2 independent experiments for each siRNAi per condition, with the exception of Aurora A and CLERC in which only 1 experiment for the second siRNAi was performed. All independent experiments were pooled). ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001, ^∗∗∗∗p ≤ 0.0001; ns, not significantly different from control; Fisher’s exact two-tailed test; # highlight a possible off-target associated with siRNA oligo 1 against HURP. (C) Relative probability (sum of the 3 independent absolute probabilities normalized to 1) of micronuclei formation from a lagging chromosome (black bars), a DNA bridge (green bars), or a misaligned chromosome (magenta bars) under the specified conditions (^∗p ≤ 0.05, ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001, ^∗∗∗∗p ≤ 0.0001; ns, no significant difference from what would be expected if all missegregation events were equally likely to cause micronuclei in each experimental condition; chi-square test).

Figure 7

Micronuclei formation from misaligned chromosomes is a frequent outcome in a chromosomally unstable cancer cell model, but not in non-transformed cells (A and B) Examples of time-lapse sequences illustrating possible origins of micronuclei in RPE-1 (A) and U2OS (B) cells. Time, min:s. Time 00:00, anaphase onset. White arrowheads track misaligned chromosomes, DNA bridges, or lagging chromosomes until they eventually form micronuclei. Pixels were saturated for optimal visualization of misaligned chromosomes, DNA bridges, and lagging chromosomes. Scale bars, 5 μm. (C) Frequency of RPE-1 and U2OS daughter cells with micronuclei that derived either from lagging chromosomes (black bars), DNA bridges (green bars), or misaligned chromosomes (magenta bars) in control, siCENP-E, and after monastrol treatment/washout (MonWO). RPE-1 cells: control, n = 163; siCENP-E, n = 95; MonWO, n = 105. U2OS cells: control, n = 250; siCENP-E, n = 81; MonWO, n = 49 (Fisher’s exact two-tailed test). (D) Relative probability (sum of the 3 independent absolute probabilities normalized to 1) of micronuclei formation from a lagging chromosome (black bars), a DNA bridge (green bars), or a misaligned chromosome (magenta bars) in RPE-1 and U2OS cells in control and after CENP-E depletion or monastrol treatment/washout. ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001, ^∗∗∗∗p ≤ 0.0001; ns, no significant difference from what would be expected if all missegregation events were equally likely to cause micronuclei in each experimental condition; chi-square test.