Depletion of separase does not block cohesion fatigue. (A) Upper Panels: Live cell imaging confirms loss of separase activity mediated by RNAi; cells fail to appropriately segregate sister chromatids during anaphase and cytokinesis results in a cut phenotype. Lower Panels: Separase depletion does not prevent scattering in cells arrested at metaphase by proteasome inhibition (MG132). (B) Chromosome spreads and cell extracts were prepared from arrested mitotic cells that had been transfected with non-targeting, separase-targeting, or SKA3-targeting siRNA duplexes. These cells had arrested in mitosis due to proteasome inhibition and/or SKA3 depletion. Graphically represented are the percentages of chromosome spreads classified as separated. Chromosome spreads reveal that separase depletion does not prevent cohesion fatigue in cells arrested by proteasome inhibition or SKA3 depletion. The number of spreads tallied for each condition ranged between 164 and 270. Western blotting indicates successful depletion of separase and SKA3 by RNAi. (C) Separase is required for early but not late chromatid separation induced by overexpression of degradation-resistant cyclin B. Upper panels: HeLa H2B-mCherry/R42A-CyclinB-GFP cells delay at metaphase for durations less than 1.6 hours, separate sister chromatids, and remain arrested in mitosis. Lower panels: HeLa H2B-mCherry/R42A-CyclinB-GFP cells lacking separase delay for several hours at metaphase and subsequently experience cohesion fatigue. HeLa H2B-mCherry/R42A-CyclinB-GFP cells separated sister chromatids with an average metaphase to scatter duration of 1.6 ± 1.1 hours. In contrast, 46% of HeLa H2B-mCherry/R42A-CyclinB-GFP cells lacking separase maintained metaphase alignment of chromosomes; the remaining 54% separated sister chromatids after an extended metaphase arrest. Bars = 10 μm, Time = hr:min, Error bars = standard deviation (see also Figure S2B, S2C and Movies S2, S3, and S4)

Manipulating cohesin levels on chromosomes affects the timing of chromosome scattering. (A) Upper panels: HeLa H2B-GFP cells depleted of SGO1 to reduce cohesin levels on mitotic chromosomes achieve metaphase alignment and then separate chromatids with an average duration from nuclear envelope breakdown (NEB) to scatter of 1.3 ± 0.3 hours. Lower panels: HeLa H2B-GFP cells depleted of both SGO1 and SKA3 separate chromatids in an average duration from NEB to scatter of 1.6 ± 0.6 hours. For comparison the graph also shows that cells depleted only of SKA3 separate chromatids after metaphase arrest with an average duration from NEB to scatter of 6.1 ± 2.9 hours. (B) Upper panels: HeLa H2B-GFP cells depleted of Wapl proceed through mitosis. Middle and lower panels: Depletion of Wapl inhibits chromatid scattering in SKA3-depleted or MG132-treated cells. In Ska3-depleted cells 94% underwent chromatid scattering. The percentage was reduced to 45% by co-depletion of Wapl, and depletion of Wapl reduced chromatid scattering to just 15% in MG132-treated cells. 6% of cells depleted only of SKA3 maintain metaphase alignment. For Wapl-depleted cells that did undergo scattering, the average time from metaphase to scattering was also increased. Bars = 10 μm, Time = hr:min, Error bars = standard deviation

Metaphase arrest of HeLa H2B-GFP cells results in chromosome scattering that is a consequence of unscheduled chromatid separation. (A) Live cell imaging shows that a cell arrested at metaphase by proteasome inhibition (MG132) scatters chromosomes. (B) Analysis of chromosome movements from 8 live cell imaging experiments indicates that scattering after metaphase arrest induced by proteasome inhibition is asynchronous within the culture. (C) Chromosome spreads reveal that scattered chromosomes are individual chromatids. The appearance after MG132-mediated metaphase arrest of individual chromatids within chromosome spreads correlate with the time-dependent scattering of chromosomes observed by live cell imaging (3 experiments, n > 167 for each time point within each experiment). Bars = 10 μm, Time = hr:min, Error bars = standard deviation (see also Figure S1, S2A and Movie S1)

Inhibition of Plk1 in metaphase-arrested cells does not block cohesion fatigue. (A) Live cell imaging indicated that HeLa H2B-GFP cells arrested in mitosis by proteasome inhibition (MG132) and treated with the Plk1 inhibitor, BI2536 at 200 or 500 nM, either formed monopolar spindles, or they maintained metaphase arrest that typically led to cohesion fatigue. (B) Quantification of HeLa H2B-GFP cells treated with MG132 and BI2536: Cells treated with only MG132 maintained bipolar spindles; 80% of these cells experienced cohesion fatigue with an average metaphase to scatter duration of 4.7 ± 1.4 hours. While the majority of cells treated with MG132 and BI2536 formed monopolar spindles, some maintained bipolar spindles, most of which experienced cohesion fatigue with an average metaphase-to-scatter duration similar to that of cells in MG132 alone. Bars = 10 μm, Time = hr:min, Error bars = standard deviation (see also Figure S3 and Movie S5)

Spindle microtubules are required for cohesion fatigue and cohesin is retained on separated chromatids. (A) Live cell imaging of LLC-PK cells stably expressing GFP-topoisomerase II: Sister centromeres (arrows) remain close together at metaphase within untreated cells. In contrast within cells experiencing cohesion fatigue, sister centromeres are widely separated while regions of the chromosome arms away from centromeres remain joined. Bar = 10 μm. (B) Quantified results of chromosome spreads prepared from mitotic HeLa cells treated for 17 hours with MG132 or MG132 + Nocodazole. Cultures were transfected with non-targeting siRNA duplexes, siRNA duplexes targeting SKA3, or siRNA duplexes targeting SGO1. Under all RNAi conditions, nocodazole-free conditions resulted in greater than 97% of spreads containing separated chromatids. Nocodazole reduced chromatid separation to 3% or less in MG132-treated or MG132-treated/SKA3-depleted cells. In contrast, nocodazole caused only a slight reduction in chromatid separation for MG132-treated/SGO1-depleted cells. (C) Chromosome spreads and extracts were prepared from mitotic cells transfected with non-targeting or separase-targeting siRNA duplexes and held in mitosis by proteasome inhibition (MG132) in the presence or absence of intact microtubules (± nocodazole). The graphs shows the percentages of chromosome spreads classified as separated at t = 0 hr and t = 8 hr post proteasome inhibition. The number of spreads tallied for each condition ranged between 207 and 312. Greater than 95% of cells with intact spindle microtubules underwent cohesion fatigue regardless of separase depletion. Isolated chromosomes or chromatids were blotted for SMC1 and RAD21/SCC1; histone H3 serves as a loading control. Chromatid separation due to cohesion fatigue (t = 8 hr, MG132 only, lane 5) does not result in reduction of chromatin-associated SMC1 and RAD21/SCC1 in comparison to t = 0 (lane 1) or in comparison to chromosomes that remain paired (t = 8 hr, MG132 + nocodazole, lane 6). Depletion of separase did not affect the levels of chromatin-associated SMC1 or RAD21/SCC1. Western blotting of cytoplasmic extracts indicated successful depletion of separase by RNAi; β-actin serves as a loading control. (see also Figure S4 and Movie S6)