Cytological and genetic consequences for the progeny of a mitotic catastrophe provoked by Topoisomerase II deficiency

Seventy five percent of the progeny of a top2-5 mitotic catastrophe is inviable

We have recently reported that the top2-5 thermosensitive mutant undergoes timely progression through the cell cycle until a MC occurs in late anaphase [25]. Importantly, top2-5 gives a clear point-of-no-return in the MC phenotype because cytokinesis makes the top2-5 anaphase bridges collapse irreversibly. In many ways, this MC is similar to other previously studied top2 conditional alleles [13, 24], although top2-5 provides a better synchrony for the MC since a larger percentage of cells quickly sever the anaphase bridge [25].

We performed single-cell videomicroscopy on agar plates through long-range objectives and found that mother and daughter cells struggled to rebud (the most obvious yeast signal for a new cell cycle) without Top2 (Figure 1A) [25]. Whereas TOP2 unbudded (G₁/G₀) cells were able to form microcolonies of around 10 cell bodies after 6 h at 37 °C, top2-5 cells stopped dividing at either 2 (~65%) or 3 (~20%) cell bodies (Figure 1A). We hereafter refer to cell bodies rather than cells or buds since it is difficult to conclude whether a 3 cell-body is part of a single multi-budded cell, a budded mother with a daughter, or a mother with two daughters. This 2-3 cell-body pattern was an end-point phenotype upon continuous Top2 inactivation, since we observed the same proportions after 24 h at 37 °C (Figure 1B). Next, we investigated whether reactivation of Top2 by shifting the temperature down to 25 °C would allow any of these bodies to form a viable population. In order to have an overall picture of cell viability, we first determined clonogenic survival after different incubation periods at 37 °C. Because of the complexity of the budding patterns after the MC, we chose a solid medium-based clonogenic assay that allows to determine if at least one of the cell bodies was still viable by the time of the temperature shift, no matter how many cells are present in the progeny (Figure 1C). We found that top2-5 had a gradual loss of viability (50% survival after ~ 4 h), and less than 5% clonogenic survival was obtained after 24 h at 37 °C (Figure 1D); the TOP2 isogenic strain retained the expected 100% clonogenic survival in this assay (Supplementary Figure 1A).

Because in these clonogenic assays there is a mixture of budded (S/G₂/M) and unbudded (G₁/G₀) cells at the time of plate seeding, we repeated the clonogenic survival after 6 h at 37° C, but photomicrographing the plate surface at different time points. Through this analysis, we determined that at time 0 h the unbudded:budded ratio was 2:1; however, only half of the surviving macrocolonies came from unbudded cells (Figure 1E). This result implies that the chance to become a macrocolony is doubled if the original cell was budded at the time of the temperature upshift. The most likely explanation for this bias resides in the fact that the top2 MC is expected to be milder if cells are closer to anaphase onset when Top2 is inactivated. Indeed, budded cells appeared to better complete the first two cell cycles at the restrictive conditions (Supplementary Figure 2). A calculation based on the cell proportions at 0h (66% G₁/G₀; 33% S/G₂/M), overall macrocolony formation from the 37 °C for 6h regime (~25%), and the origin of those macrocolonies (~50% from G₁/G₀; ~50% from S/G₂/M) led us to conclude that around ~20% of the original G₁/G₀ cells gave rise to survivors after 6 h at 37 °C [0.25 × 0.5 / 0.66 = 0.19]. This proportion increases to ~40% for the S/G₂/M cells at the time of the temperature upshift [0.25 × 0.5 / 0.33 = 0.38].

Most of the inviable progeny of a top2-5 mitotic catastrophe immediately stops dividing

Next, we tried to correlate the long-term clonogenic survival of the G₁/G₀ cells with their ability to form microcolonies. We reasoned that it is possible that many MCs could render viable progeny in the short-term (microcolony) but could not raise a visible colony later (macrocolony). This difference could reflect a gradual loss of viability in the progeny as a consequence of genomic imbalances acquired after the MC. To get further insights into this possibility, we took pictures of the cells on the surface of Petri dishes at the time of seeding (0h), right after the 37 °C incubations (6h or 24h), and 16 h or 24 h after the plates were shifted back to 25 °C (Figure 2A). As expected from above, most of the original G₁/G₀ (unbudded) cells did not go beyond the 2-3 bodies stage during the 37 °C incubations (Figure 2A, 2B; inner circles in the sunburst plots). Restoring permissive conditions for Top2 activity allowed very few of the cell progeny to divide again (Figure 2A, 2B; outer circles in the sunburst plots). This finding was true not only during the long (24 h) incubation at 37 °C, but also for the 6 h incubation. Indeed, only ~15% of the 2-3 cell bodies observed after 6 h at 37 °C were able to re-bud again once or more after shifting them back to 25 °C. Incidentally, a low but significant proportion of G₁/G₀ cells did not bud during the 37 °C regimes. However, even in these non-MC cases, cells did not divide after the Top2 reactivation, suggesting that this G₁/G₀ subpopulation was already incapable of cell division following growth at 25 °C. Longer incubations at 25 °C after the MC resulted in microcolonies of >50 cells that eventually developed into macrocolonies. With our cell density settings, microcolonies of more than 20-30 cells hindered us from raising conclusions about the fate of adjacent cells. However, the position of the center in these microcolonies suggests that most, if not all, of the G₁/G₀ cells that ended up as macrocolonies had re-budded again within the first 24 h that followed the temperature downshift.

From previous analysis by videomicroscopy, we know that most top2-5 doublets (2 bodies) and all triplets (3 bodies) have passed anaphase and thus completed a MC after 6 h at 37 °C [25]. Nevertheless, we decided to complete the analysis of the immediate progeny by testing whether the observed cell bodies have accomplished cell separation. Our reasoning was that separation by micromanipulation would demonstrate that cell bodies have become individual daughter cells. Indeed, we could separate with the needle more than half of the doublets and triplets (Figure 2C, middle concentric circle in the sunburst chart). We next tried to correlate the ability of all these cells to form macrocolonies but found that they were largely unviable. The percentage of macrocolonies (~2%) was much lower than expected, indicating that that micromanipulation killed cells that otherwise would have retained viability.

We noticed from the microcolony experiments performed above that cells swelled after prolonged incubations at 37 °C (Figure 2A). After 24 h at 37 °C, the volume occupied by the original mother cell doubled (Figure 2D). Downshift of the temperature to 25 °C only modestly deflated these cells. In addition, a low proportion of cell bodies underwent lysis (“0”; 2 → 1 and 3 → 2 categories in Figure 2B sunburst charts). These findings, together with the hypersensitivity to micromanipulation, led us to consider osmotic stress (due to the continuous growth in size of the mother cell) as a possible cause underlying the long-term inviability and/or short-term inability to divide of the top2 progeny. However, addition of 1.2 M sorbitol, an osmotic stabilizer, neither prevented cells from swelling (Figure 2D) nor improved long-term viability (Figure 2E; Supplementary Figure 1B) or short-term division capability (Figure 2F; Supplementary Table 1).

Cell death after prolonged absence of Top2 activity occurs slowly, asynchronously, asymmetrically and is independent of the Yca1 metacaspase

Next, we analysed cell bodies in these microcolonies more closely, seeking other morphological patterns of cell disease aside from swelling; for example, lysis, darkening and loss of the rounded shape (Figure 3A). Most of these morphological features have been previously related to different forms of cell death. Because the top2-5 strain also carries a GFP-labeled H2A histone, we also monitored nuclear morphology and chromatin integrity (i.e., GFP intensity). Cells still looked fully healthy after 6 h at 37 °C (no difference with 0 h), despite the great loss in the ability of the progeny to divide. Only after prolonged 37 °C incubations did the cells start to look clearly sick. Still, more than 50% of cell bodies harboured a nuclear GFP signal even 24 h following the 37 °C upshift. This GFP signal co-existed in cell progenies that showed unhealthy patterns in the bright field such as swelled, darkened, or non-rounded cell bodies

In order to better study cell death after top2 MC, we employed other means that required experiments to be performed in liquid media instead of Petri dishes. We first quantified the rate of cell death and metabolic decline by staining with the vital dye methylene blue (MB) (Figure 3B). This dye stains dead cells blue, although it can also stain cells that are alive but metabolically attenuated [29]. A time course after the 37 °C upshift showed that there was not a major increase in MB positive cell bodies in the first 4 h (the equivalent in liquid cultures to 6 h on solid medium; [25]). In general, the increase in MB+ bodies was linear, but even 24 h after the 37 °C upshift ~40% of cell bodies were not stained by MB. These staining experiments uncovered two other properties of the top2 MC: i) it was common that only one cell body was MB+ in doublets and triplets (Supplementary Figure 3A); and ii) both unbudded and budded cells were equally stained (Supplementary Figure 3B). Regarding the first observation, the result suggests that loss of vitality is not coordinated between mother and daughter(s) cells. This asymmetry also confirms that many doublets must have completed cytokinesis despite remaining together. As for the second observation, the staining pattern suggests that loss of vitality occurs in an asynchronous fashion in terms of any preference for a cell cycle stage.

Because MB does not distinguish whether cell bodies are dead or simply metabolically stressed, we next sought other more informative vital stains. First, we employed the fluorescent vitality probe FUN1©. This probe stains metabolically active live cells with red vacuolar aggregates [30]. FUN1© is considered more informative and reliable than MB. With this probe we confirmed that only ~10% of cell bodies have lost vitality after just 4 h at 37 °C (left chart of Figure 3C; Supplementary Figure 4). It was also surprising that vitality decline still affected no more than 40% of all cell bodies after 24 h at 37 °C.

We also used Propidium Iodide (PI) to monitor cell viability. Loss of plasma membrane impermeability is considered a bona fide marker of cell death [27]. PI is only able to fluorescently stain cells that have lost such impermeability. Anticipating some sort of RCD after the top2 MC, we decided to accompany PI staining in red with reporters for either reactive oxygen species (ROS) or externalization of phosphatidylserine (PS) at the plasma membrane; both in green (DCFH-DA and annexinV-FITC, respectively). Intrinsic ROS production has been observed during RCD in all eukaryotes, including yeast, and is considered one of most reliable RCD markers [31]. After overnight growth at 25 °C (0h), the top2-5 strain had neither dead cells nor cells with ROS (right part of Figure 3C, Supplementary Figure 5). Four hours after the 37 °C temperature shift, there was only a slight increase in dead cell bodies (~6%) and almost no signs of ROS in the rest (~3%). Only after 24 h of incubating the cells at 37 °C, we found a significant proportion of both ROS (~20%) and dead cells (~20%). It is noteworthy that the percentage of PI+ cells was still relatively low after this 24 h incubation, and 60% of all cell bodies were still resistant to PI and free of ROS. Similarly, PS externalization is considered a conserved bona fide marker of RCD [32]. Double staining with PI and annexin V-FITC in cells where the cell wall has been digested (a necessary step for annexin V to reach externalized PS) showed less than 10% of cell bodies with the staining expected for early apoptosis, even 24 h after the temperature upshift (Supplementary Figure 6; annexin V positive plus PI negative). It is noteworthy that there were more dead cell bodies (PI positive) in this assay than in the ROS/PI assay (~60% vs. ~20%). This observation is likely a consequence of digesting the cell wall, since the plasma membrane of swollen cells at 24 h might collapse during the treatment.

We next examined if the observed pattern of death after the top2 MC was altered by a mutation in caspase. We reasoned that this approach might shed more light on the RCD/ACD nature of the observed death, considering the low presence of intrinsic ROS and the technical caveats of the annexin V assay. Hence, we deleted the only caspase-like gene in yeast, YCA1/MCA1. Yca1 is required for RCD in yeast in response to several environmental stresses [33–35]. In addition to cell death, we checked whether Yca1 modulated the other behaviours seen in the top2 MC progeny (for example, the inability to divide and the slow decline in cell vitality). The conclusions derived from comparing top2-5 YCA1 and top2-5 yca1Δ were that Yca1/Mca1: (i) had no influence in the percentage or rate of cells that end up dying (Figure 3B, 3C); (ii) neither accelerated nor slowed down the vitality decline (Figure 3B, 3C); (iii) did not modify the profile of clonogenic survival after the top2 MC (Figure 3D; Supplementary Figure 1C); and (iv) its absence did not improve the ability of the immediate cell progeny to divide (Figure 3E; Supplementary Table 1).

The overall conclusion from these experiments is that the immediate progeny of the top2 MC enter a senescent-like state as they retain vitality but loses their ability to re-bud. Senescence is only a transient state that lasts several hours or days, until cells eventually die. The pattern of cell death (morphologically diverse, slow, asynchronous and asymmetrical), together with the lack of effect of Yca1, suggest that loss of Top2 leads to ACD.

Chromosome ploidy modulates the ability of the progeny to divide

All the experiments described above were carried out in haploid yeast cells. In haploids, MC associated with the presence of partly unresolved sister chromatids, as in the top2-5 mutant, is expected to be highly deleterious. Severing of these anaphase bridges result in daughter cells that may lack several chromosome arms [11]. Based on this consideration, one might expect that diploid cells would be more resistant to the consequences of top2 MC than haploid cells. Thus, we studied the top2 MC in an isogenic homozygous top2-5/top2-5 diploid (2N) strain. Unlike its haploid counterparts, diploid cells were more often able to re-bud at least once. In fact, ~50% of all 2-3 cell bodies originated from just after 6h at 37 °C were able to do so after the 25 °C downshift (Figure 4A; Supplementary Table 1). This percentage dropped considerably if the progeny was incubated 24 h at 37 °C. Strikingly, however, the increase in the ability to divide again after the MC did not yield better clonogenic survival (Figure 4B; Supplementary Figure 1D), indicating that most of this viable progeny was competent to re-bud only in the first generations. We also performed a microdissection analysis of the diploid top2 progeny. Unlike haploid top2-5 cells, which was rather sensitive to micromanipulation, 13% of the diploid progeny raised a macrocolony after the separation attempt (Figure 4C). Altogether, we conclude that a top2 MC in diploids results in better short-term ability to divide.

Genome instability footprints in the surviving progeny from the top2-mediated mitotic catastrophe

Above, we have just shown that ~25% of the top2-5 diploid cells still gave rise a macrocolony after the 6h at 37°C regime. We next examined the genomes of these survivors in search for specific genomic footprints of the top2 MC. To accomplish this goal, instead of using the homozygous isogenic diploid employed in the previous chapter, we generated a highly heterozygous hybrid top2-5 diploid. The hybrid diploid was generated by crosses of two sequence-diverged top2-5 haploids, derivatives of W303-1A and YJM789, which are heterozygous for more than 55,000 single-nucleotide polymorphisms (SNPs) distributed throughout the yeast genome (the yeast genome is 15 Mb). These heterozygous SNPs allow the analysis of various types of genomic alterations by using SNP-specific microarrays [36, 37]. The hybrid diploid is also engineered to select various types of chromosome alterations on chromosome V (Figure 5A). It is homozygous for the ade2-1 mutation on chromosome XV (an ochre-suppressible allele) and heterozygous for the SUP4-o suppressor gene in chromosome V [38]. Strains with the ade2-1 mutation form red colonies in YPD in the absence of the SUP4-o suppressor. Diploid strains with one or two copies of SUP4-o form pink and white colonies, respectively [39]. Thus, loss of the SUP4-o gene by either mitotic recombination or chromosome loss results in a red colony instead of the pink colonies characteristic of the original strain. Colony colour changes are also expected if the copy number of the ade2-1 allele varies by loss or duplication of chromosome XV. In addition, aneuploidy for other chromosomes sometimes alters the color of the colony.

Microcolony and clonogenic experiments performed in the top2-5 hybrid diploid FM1873 showed that this strain lost viability quicker than the isogenic homozygous diploid in the S288C background (Figure 5B, 5C; Supplementary Figure 1E). There was a steady rise in red and/or red/white sectored colonies among the survivors during the 37 °C incubation, as expected if the top2 MC increases genome instability (Figure 5D). We used microarrays to examine genomic alterations in the control FM1873 isolate (no exposure to 37 °C) and in isolates exposed to 37 °C for 6 h, and then allowed to form colonies at 25 °C (Figure 5E; see also Supplementary Information for interpretation on the genomic alterations picked up by SNP array). It is important to stress that the SNP microarrays allow analysis of genomic alterations throughout the genome in addition to those changes that occur on chromosome V [36, 37]. When the control FM1873 strain was examined before exposure to the restrictive temperature, surprisingly, we found that it was altered relative to an isogenic TOP2/TOP2 hybrid [40]. More specifically, we realized that FM1873 carried a terminal loss of heterozygosity (T-LOH) on the right arm of chromosome XII (the longest chromosome arm in yeast). In addition, all isolates had three to four copies of chromosome XIV. We tried several times to recreate the hybrid top2-5 diploid but were unable to isolate a derivative that had only two copies of XIV. Since this chromosome is the location of top2-5, it is likely that chromosome XIV trisomes and tetrasomes have a selective growth advantage at the permissive temperature over the diploids that have only two top2-5 copies. We generated an isogenic derivative of FM1873 (MD684) that lacked the T-LOH event on XII, although it still had extra copies of XIV. For our subsequent genomic analyses, we studied both FM1873 and MD684.

The experimental strains were exposed to the restrictive temperature of 37 °C by incubating the cells for six hours either on plates or in liquid (details in Materials and Methods); these two protocols resulted in similar levels of instability. A total of 27 isolates were examined for FM1873 (13 experimental, 14 control), and 19 isolates of MD684 (10 experimental, 9 control). Somewhat surprisingly, the control single-colony isolates (cells not exposed to 37 °C) also had high rates of instability (Table 1; “C” samples), indicating that the Top2p encoded by top2-5 does not have wild-type activity even at the permissive temperature. Indeed, a previous biochemical study reported that the Top2-5 activity at 25 °C is 33% of that of wild type Top2 [41]. Among all isolates examined, we found 76 T-LOH events, 3 interstitial LOH (I-LOH) events, 31 trisomies, 2 monosomies, and 6 uniparental disomies (UPDs) (Figure 5E; Table 1). The average number of genetic changes per strain (including all the data of Table 1) was 3.3 alterations/isolate. Since the strains were grown approximately 40 generations before microarray analysis, the rate of alterations/cell division/isolate is about 8.3 x 10^-2. In a previous microarray analysis of a wild-type diploid isogenic with FM1873 and MD684, we found a rate of alterations of about 2 x 10^-3/division, a rate about 40-fold less than for the top2 strains. The most common alteration in the top2 strains was a T-LOH event (64% of the total events). These events likely reflect the repair of a DSB by either a crossover or a break-induced replication (BIR) event (Figure 5E). The chromosomal distribution of the events is striking. The right arms of chromosome IV and XII (the two longest arms in the yeast genome) had over 80% of the terminal LOH events (60/76), although these arms represent less than 30% of the yeast genome. Other chromosomes with T-LOH events (the number of events shown in parentheses) are: XIII (5), XV (4), VII (3), XIV (2), VII (1), XI (1), and V (1). Chromosomes XIII, XV and XIV have the third, fourth, and fifth longest chromosome arms in the genome, respectively; all of the mapped events in these chromosomes are on the longest arms. Strikingly, the very large (about 1.2 Mb) ribosomal RNA gene cluster (rDNA) on the right arm of chromosome XII was not a preferred site for T-LOH events. Although the rDNA is about 60% of the right arm of XII, only 30% (3 of 11) of the LOH events had a breakpoint in or near the rRNA genes. We should point out that T-LOH events on XII could only be followed in the MD684 strains (FM1873 already had a cXIIr T-LOH); hence, our estimate of the frequency of T-LOH events on the right arm of XII is a minimal estimate.

In addition to LOH events, we observed 33 changes in chromosome number (31 trisomies and 2 monosomies) (Table 1). The frequencies of trisomies are not simply related to the size of the chromosomes. Chromosomes V, VIII, and XV were the most frequently-observed trisomes; chromosomes V and VIII are medium-sized chromosomes (both about 550 kb), whereas chromosome XV is large (1092 kb). Only one trisomy was observed involving one of the three smallest chromosomes (I, VI, and III). Only three of the trisomies involved the two largest chromosomes IV and XII.

We also observed six UPD events (Figure 5E). In strains with these events, the homolog is present in two copies, but both copies are derived from one of the original parental homologs. There are two plausible pathways to generate UPD (Supplementary Figure 7): two non-disjunction events in different cell cycles or reciprocal UPD (RUPD) in which one pair of homologs segregates into one cell and the other pair segregates into the other cell. Although both pathways probably contribute to UPD in yeast, at least some of the events in wild-type strains are RUPD [42]. To determine whether RUPD events occurred frequently in the top2 cells, we used a protocol in which both daughter cells produced as a result of RUPD or a reciprocal crossover (RCO) in chromosome V could be recovered in different sectors of a sectored colony (Supplementary Figure 8). The sectored colonies were derived from the same strains (FM1873 and MD684) used for our single-colony analysis. A crossover or RUPD can produce a red/white sectored colony. However, to select for such events, both FM1873 and MD684 contained a heterozygous can1-100 marker located allelically to the SUP4-o insertion and a gene encoding hygromycin resistance (hph) located distal to the can1-100 insertion (Figure 5A). The can1-100 mutation is a nonsense mutation and is suppressed by SUP4-o. Strains that lack the suppressor are resistant to the drug canavanine and those with the suppressor are sensitive. A crossover between can1-100 and the centromere results in formation of a canavanine-resistant red/white sectored colony (Supplementary Figure 8A); a sectored colony with the same phenotype can also result from RUPD (Supplementary Figure 8B). RCO and RUPD events can be distinguished by microarray analysis (bottom panels of Supplementary Figure 8).

Following exposure of FM1873 and MD684 to 37 °C, we found 242 red/white Can^R sector colonies, 43 of which had the sectoring pattern for the hph marker suggestive of RCO or RUPD events. We found both RCO and RUPD events in two-thirds of these colonies (Supplementary Table 2). The rate of RUPDs in cells of FM1873 and MD684 treated for 6 h at 37 °C was the same, 1.1 x 10^-5/division; the rate of RUPD in cells of FM1873 that were not exposed to 37 °C was 1 x 10^-6. The rate of RCO in FM1873 cells treated at 37 °C was 2.7 x 10^-6; no RCO events were observed in MD684. In an isogenic wild-type strain, the rates of RUPD and RCO for chromosome V were 10^-7 and 1.6 x 10^-6, respectively [42]. These results support the conclusion that the top2 mutation results in a substantially elevated rate of RUPD (about 100-fold) and has little effect on the rate of RCO.

On the short-term cytological consequences of the top2 mitotic catastrophe

In this study we have adopted the term mitotic catastrophe (MC) in its broadest cytological sense, referring to aberrant mitoses that are expected to be deleterious for the progeny based on the degree of observed abnormalities. It is worth mentioning that other authors, especially those working with metazoans, restrict the MC term to death occurring in mitosis after a mitotic insult [44, 45]. With this restriction in mind, MC in metazoans is a sort of mitotic RCD. In our yeast experimental model with the top2-5 allele, death before anaphase is insignificant since cells go through G₂/M and anaphase as quickly as their TOP2 counterparts [25]. In nocodazole-treated cells, which arrest cells in G₂/M, 40% cell death was reported after 10 h in clonogenic assays [46]. This cell death was described to occur through an RCD apoptosis-like mechanism. By contrast, our clonogenic assays showed that a sudden drop in viability occurs between 3-6 h after Top2 inactivation (Figure 1D), about the time needed for cells to complete telophase and cytokinesis on solid medium [25]. From the microcolony experiments, we concluded that the ensuing top2-5 progeny are largely impaired in entering a second cell cycle (Figures 1A, 1B and 2). This impairment can be partly alleviated in diploids, yet only in the short-term (Figure 4). This observation leads us to propose that gross genomic imbalances prevent the immediate progeny from cell cycle progression. Taking into account previous reports on the formation of anaphase bridges in top2-ts mutants [13, 25] and high levels of chromosome missegregation [23], it appears logical that many of the haploid progeny lose entire chromosomes containing essential genes. In addition, loss of essential genes may reflect breakage of chromosomes at the bridge followed by loss of the distal chromosome regions [11].

Another possibility is that daughter cells immediately die upon the top2 MC through an RCD program. This would imply that either anaphase/telophase cells or their immediate progeny sense the MC and execute an RCD. Our results, however, argue against this possibility. Firstly, many cells stained negative for death (PI) and apoptotic (annexin V) markers, whereas they stained positive for metabolic activity (FUN1©), even after 24 h at 37 °C. Secondly, the decline in cell vitality occurred slowly, linearly (asynchronously) and asymmetrically (when comparing daughter cells that remained together after the MC). Thirdly, Yca1/Mca1, the main RCD player in S. cerevisiae, does not regulate the vitality decline. Even though there are other RCD proteins aside from Yca1/Mca1 [47, 48], we point out that that the pattern of cell death after the top2 MC is better explained through ACD. Although we found intrinsic ROS production in a minor subset of the top2 MC progeny, and this finding has been considered a marker of RCD [31, 49], we hypothesize that, in our case, ROS accumulation is a consequence of the steady decline of cell homeostasis. For instance, ROS might arise from loss of nuclear genes involved in eliminating ROS in metabolically active cells.

Comparing with previous studies, the events that lead to death after the top2 MC resemble those observed after prolonged G₂/M arrest in the cdc13-1 mutant, which results in irreversible DNA damage at chromosome ends. In arrested cdc13 cells, there are cell markers of RCD such as ROS production [50], although further biochemical assays and genetic manipulation suggest ACD rather than RCD [47]. In cdc13-1, because cells get blocked in G₂/M, there are no genetic imbalances prior to cell death. It was proposed that cell lysis, resulting from cell growth without cell division, was the ultimate cause of death, a hypothesis confirmed because sorbitol (an osmotic stabilizer) improved cell viability [47]. Although we also observed oversized cells one day after the top2 MC (Figure 2A and 2D), cell lysis was a rare event and sorbitol did not improve cell viability (Figure 2D–2F). Therefore, we propose that the secondary ACD observed after the top2 MC is the consequence of the steady decline of cell homeostasis resulting from loss of essential genes.

On the long-term genetic consequences of the top2 mitotic catastrophe

We have also compared the genomes of surviving diploids after 6h of Top2 inactivation. Even though we did not prove that all survivors came from a MC (i.e., they went through anaphase), the results shown in Figures 1–5, together with previous findings [13, 24, 25] lead us to conclude that most of these survivors likely came from a top2 MC. Indeed, many of the observed chromosome variations and rearrangements can be best explained if cells go through anaphase in the absence of proper sister chromatid disjunction [11]. One interpretation of the strong bias for T-LOH events relative to I-LOH events (76 terminal and 3 interstitial) is that DSBs are repaired primarily through BIR (Figure 5E) [11, 51]. In many previous studies, I-LOH represented a very significant fraction of the total LOH events. For example, in G₁-synchronized cells treated with ultraviolet light, we observed a 1:3 ratio between T-LOH and I-LOH [40]. I-LOH requires both ends of the DSB to invade the other homolog during repair through homologous recombination, a condition difficult to fulfil in DSBs generated by cytokinesis [11].

According to a previous study [52], the frequency of DSBs in top2 mutants is higher for long chromosome arms than short chromosome arms. Spell and Holm [50] explained this observation by suggesting that intertwinings on short chromosome arms could be resolved by passive diffusion off the end, whereas such intertwinings on long chromosome arms required Top2. Our results are also consistent with this interpretation. Chromosome IV and XII right arms are the longest in the yeast genome and are overrepresented in the T-LOH events (normalizing for the size of the arm). An unusual feature of the T-LOH data is the rDNA is under-represented as a breakpoint in the cXIIr T-LOH events. It was previously shown, using an assay that detects loss of inserted marker within the rDNA, that top2 strains have substantially elevated rates of mitotic recombination in the rDNA [53]. One possible explanation of this discrepancy is that DSBs within the rDNA may be repaired by single-strand annealing between flanking copies of the rDNA genes [54], an event that would result in loss of an inserted marker without an interaction with the other homolog. Such an event could not be detected by the microarray analysis.

Another genetic alteration that is consistent with sister chromatid non-disjunction at the top2 MC is the elevated presence of trisomies. A straightforward explanation for this is that the intertwining of sister chromatids in the top2 strains often results in their co-segregation into one of the daughter cells (Figure 5E). Although this type of non-disjunction would be expected to create equal numbers of monosomic and trisomic strains, it is possible that the monosomic strains have a competitive growth disadvantage and are, therefore, selected against. In tel1 mec1 diploids, expected to enter anaphase with broken and/or underreplicated chromosomes, trisomies were five times more common than monosomies [55]. A similar bias towards trisomies was observed in cdc14-1 diploids [51]; cdc14 results in elevated levels of anaphase bridges because Cdc14 regulates condensin and Top2 actions in anaphase [56–58].

Lastly, the 100-fold enrichment in RUPD after the top2 MC is also in agreement with models of genomic instability generated by sister chromatid non-disjunction. Two models are proposed for the generation of RUPD. One model involves two independent missegregation events occurring in successive divisions. In the other model, segregation of the chromosomes occurs in a manner similar to meiosis-I (Supplementary Figures 7 and 8). In wild-type cells, we demonstrated that the second model is correct [42]. From our data in the current study, we cannot distinguish between these models. The rate of a single non-disjunction event of chromosome V in a top2 mutant is very high, about 3 x 10^-3/division [23]. Thus, the likelihood of two independent non-disjunction events in a single division is (3 x 10^-3)² or 10^-5 which is close to our observed rate of RUPD in the top2-5 diploid. It is possible that both mechanisms contribute to the high rate of RUPD observed in the top2-5 mutants. Another difference with our earlier study [42] is the low frequency (18%) of sectors reflecting RUPD and RCO recovered from the red/white Can^R sector assay. In our previous study, almost all sectored colonies reflected RCOs or RUPD events. It is likely that the high levels of aneuploidy observed in the top2 mutants affects colony color by mechanisms unrelated to RCO or RUPD events on chromosome V.

In conclusion, we have characterized the consequences for the progeny of depleting yeast cells of Top2. We determined that the top2 mitotic catastrophe leads to the sudden loss of the capability to divide again. Nevertheless, restrictions for cell division are not a consequence of immediate cell death as the progeny remain alive for several hours. In addition, survivors of the top2 mitotic catastrophe carry genomic footprints that point towards sister chromatid non-disjunction and breakage of anaphase bridges as the source of the top2-driven genome instability. Overall, the top2-mediated mitotic catastrophe is highly deleterious for the cell progeny but it might also bring about highly unstable surviving clones.

Strain construction

All strains used in this work are listed in Supplementary Table 3. Gene deletions were achieved using standard PCR and transformation methods [25, 59]. To obtain the transformation products, genomic DNA of the corresponding strain in the Euroscarf yeast haploid MATa knockout collection was used as the PCR template. Primers used in the PCR bind 100-400 bps upstream and downstream the deleted gene ORF (Supplementary Table 4).

The top2-5 isogenic homozygous diploid derivative was generated through a one-step marker-free transformation approach that takes advantage of the α-factor hypersensitivity in the haploid MATa bar1Δ genotype (a proof of concept is provided in Supplementary Figure 9). Further details in Supplementary Information.

The top2-5 heterozygous diploids FM1873 and MD684 were obtained by crossing of haploid strain top2-5 derivatives of PSL2 and PSL5 (see Supplementary Information). These two strains are isogenic with W303-1A and YJM789, respectively, and have been engineered to select and visually detect chromosome V rearrangements (Figure 5A) [38]. W303-1A and YJM789 differ by about 55,000 SNPs. Thus, these hybrid heterozygous diploids were also used for genome-scale detection of chromosome rearrangements by SNP arrays (see below).

Unless stated otherwise, all strains were grown overnight in air orbital incubators at 25 °C in rich YPD media (1% w/v of yeast extract, 2% w/v peptone and 2% w/v dextrose) before every experiment.

Assays to assess clone survivability and capability of single cells to divide

A modified clonogenic assay performed on agar plates was used to assess survivability of the progeny after the mitotic catastrophe (details in Supplementary Information). The purpose of this assay was to determine if at least one of the resulting cells in the progeny was still able to raise a new cell population, irrespective of how many times the cell has divided and how many cells in the progeny are viable. Half-life values (t ½), the time in which the clone survival drops to 50%, were calculated adjusting the data to a four-parameter model using Graphpad Prism 7.

The time-lapse microcolony analyses were performed with a 40x long-range objective mounted on a Leica LMD6000 direct microscope (details in Supplementary Information). The cell density on the agar plate surface was set to 25 cells per 10,000 μm². The cell volume of the original mother cell in all three frames was calculated assuming a perfect sphere and taking the cell diameter for calculations.

The single cell analysis by micromanipulation was performed on YPD plates with a Singer Sporeplay tetrad microdissector (details in Supplementary Information). Just 12 cells were harvested per plate in order to avoid prolonged incubations at 25° C at the beginning of the experiment.

The segregation and morphology of the histone-labelled nucleus (H2A-GFP) was analyzed by wide-field fluorescence videomicroscopy as described before [25]. Further details in Supplementary Information.

Assays to assess cell vitality and cell death

All the colorimetric and fluorometric assays to assess metabolic competence, ROS production, and plasma membrane permeability were carried out in asynchronous logarithmic cultures grown overnight at 25 °C, adjusting the OD₆₀₀ to 0.2 and incubated them for 24 h at 37 °C. Samples were taken at 0, 4 and 24 h and directly observed under the microscope. Fluorescence microscopy was used instead of flow cytometry because strains were already fluorescent for H2A-GFP. ROS were visualized with 10 μg/ml of 2′,7′-dichlorofluorescin diacetate (DCFH-DA; Sigma-Aldrich; D6883); and 3 μg/ml PI (Fluka; #81845) was used to count dead cells. Both dyes were directly mixed with a 200 μl aliquot of the culture and incubated 15 min at 37 °C in the dark. Cell bodies were considered ROS positive when the cytoplasm stained green in the absence of staining for PI. FUN® 1 (Invitrogen; F7030) staining was done washing 200 μl of each sample with water containing 2% D-(+)-glucose and 10 mM Na-HEPES, pH 7.2, resuspending the cells in the same buffer with 10 μM FUN® 1 and incubating 30 min at 37 °C in the dark. For MB (Sigma-Aldrich; M9140) staining, 1 μl of each sample was mixed with 1 μl of a 0.04% w/v MB solution in water onto a microscope slide and directly visualized (bright field). Staining with FITC-labeled annexin V was undertaken in spheroplasts as previously described [32]. Minor modifications to this protocol were employed; namely, we used the TACS Annexin V-FITC Apoptosis Detection Kit (R&Dsystems; 4830-01-K) as well as 60 units/ml of zymolyase (ZymoResearch; E1005) for cell was digestion.

Analysis of genomic rearrangements using microarrays

Two similar protocols were used to expose heterozygous diploids to the top2-mediated MC. For both protocols, cells were grown in YPD to an optical density of 0.5-1. For one set of experiments (marked as E1 in the tables in the text), the cells were struck for single colonies on plates containing solid YPD medium, incubated for 6 hours at 37 °C, then incubated at room temperature until colonies had formed. For the second protocol, the cells grown at room temperature were harvested by centrifugation, and resuspended in 37 °C liquid medium, followed by incubation for 6 hours at 37 °C. Following this incubation, they were struck on YPD plates and incubated at room temperature until colonies were formed. For both protocols, the control cells were struck to room temperature YPD plates without incubating them at 37 °C. In experiments to detect sectored colonies, the YPD medium was replaced with solid omission medium lacking arginine and containing 120 μg/ml canavanine [39].

We detected LOH events, aneuploidy and UPD using SNP-specific microarrays similar to those used by [60]. The details of this procedure can be found in the Supplementary Information and have been described before [36, 37]. In analysis of single-colony isolates, whole-genome arrays were used (Gene Expression Omnibus [GEO] #GPL20144). For analysis of sectored colonies, we used microarrays specific for the right arm of V, and chromosomes I, III, and VIII (GEO #GPL21274).