1. Chromosomes and Gene Expression
Download icon

High levels of histones promote whole-genome-duplications and trigger a Swe1WEE1-dependent phosphorylation of Cdc28CDK1

  1. Douglas Maya Miles  Is a corresponding author
  2. Xenia Peñate
  3. Trinidad Sanmartín Olmo
  4. Frederic Jourquin
  5. Maria Cruz Muñoz Centeno
  6. Manuel Mendoza
  7. Marie-Noelle Simon
  8. Sebastian Chavez
  9. Vincent Geli  Is a corresponding author
  1. Marseille Cancer Research Center (CRCM), U1068 Inserm, UMR7258 CNRS, Aix-Marseille Université, Institut Paoli-Calmettes, Equipe Labellisée Ligue, France
  2. Hospital Virgen del Rocío-CSIC-Universidad de Sevilla, Spain
  3. Barcelona Institute of Science and Technology, Spain
  4. Universitat Pompeu Fabra, Spain
Research Article
  • Cited 0
  • Views 792
  • Annotations
Cite as: eLife 2018;7:e35337 doi: 10.7554/eLife.35337

Abstract

Whole-genome duplications (WGDs) have played a central role in the evolution of genomes and constitute an important source of genome instability in cancer. Here, we show in Saccharomyces cerevisiae that abnormal accumulations of histones are sufficient to induce WGDs. Our results link these WGDs to a reduced incorporation of the histone variant H2A.Z to chromatin. Moreover, we show that high levels of histones promote Swe1WEE1 stabilisation thereby triggering the phosphorylation and inhibition of Cdc28CDK1 through a mechanism different of the canonical DNA damage response. Our results link high levels of histones to a specific type of genome instability that is quite frequently observed in cancer and uncovers a new mechanism that might be able to respond to high levels of histones.

https://doi.org/10.7554/eLife.35337.001

Introduction

Chromatin replication requires the synthesis and assembly of nucleosomes that wrap around DNA. Each time a cell divides several millions of histones, small basic proteins that conform nucleosomes, are synthesised and incorporated as the replication machinery copies the genome. Higher eukaryotes are unable to survive without histones and possess several alternative pathways to ensure enough histones during replication (Cook et al., 2011; Groth et al., 2005; Marzluff et al., 2008). Cells can also modulate cell cycle progression when histones become limiting to ensure the faithful replication of chromatin and avoid genome instability (Groth et al., 2007; Günesdogan et al., 2014; Murillo-Pineda et al., 2014). Histone excess has also been linked to genome instability and to a wide variety of processes in the cell including DNA repair and life span, which probably explains why all eukaryotes have several redundant pathways that ensure the absence of free histones beyond replication (Au et al., 2008; Castillo et al., 2007; Feser et al., 2010; Gunjan and Verreault, 2003; Singh et al., 2010; Takayama et al., 2010).

Accurate chromosome segregation is essential to prevent genome instability. Eukaryotic cells have different mechanisms or checkpoints able to specifically sense and respond to different types of errors. These checkpoints are able to modulate the length of the cell cycle and give the cells additional time to solve them ([Hartwell and Weinert, 1989]. Saccharomyces cerevisiae like most eukaryotic cells has two major checkpoints able to block cells prior to mitosis: the DNA damage response (DDR) (Ciccia and Elledge, 2010) and the spindle assembly checkpoint (SAC) (Musacchio and Salmon, 2007). Both of them are able to perform this block inhibiting the cleavage and degradation of the kleisin subunit of the cohesin complex Scc1RAD21. This inhibition takes place through a stabilisation of Pds1Securin either by phosphorylation (DDR) (Sanchez et al., 1999) or by preventing its degradation through the APC (DDR and SAC) (Agarwal et al., 2003; London and Biggins, 2014). The SAC can additionally respond to lack of tension keeping Shugosin at the pericentromere, which prevents cohesin cleavage through the inhibition of Esp1Separase (Clift et al., 2009; Nerusheva et al., 2014). The DDR can also block mitosis through the phosphorylation and inhibition of the G2/M cyclin-dependent kinase Cdc28CDK1 at Tyr19 (15 in humans) that plays a key role during mitosis (Agarwal and Cohen-Fix, 2002; Rahal and Amon, 2008; Zhang et al., 2016).

Besides these well-characterised checkpoints, studies in S. cerevisiae have revealed an additional checkpoint able to respond to actin cytoskeleton perturbations called the morphogenesis checkpoint, which delays cell-cycle progression when the actin cytoskeleton is perturbed (Lew, 2000; McMillan et al., 2002; Sakchaisri et al., 2004; Sia et al., 1998). This checkpoint is able to stabilise Swe1WEE1, a kinase able to promote a phosphorylation on Tyr19 of Cdc28CDK1 (Tyr15 in humans) that inhibits its activity and results in a delay on the metaphase to anaphase transition (Lianga et al., 2013). Swe1WEE1 is present in lower and higher eukaryotes during an unperturbed cell cycle. This kinase is expressed during replication and degraded before mitosis in a mechanism that involves the action of several kinases that promote Swe1WEE1 hyperphosphorylation and trigger its ubiquitination and subsequent destruction by the proteasome (Howell and Lew, 2012). Swe1WEE1 can also be stabilised upon DNA damage (Palou et al., 2015) and in response to certain types of stress (Chauhan et al., 2015; George et al., 2007; King et al., 2013). Interestingly, Swe1WEE1 was quite recently shown to be able to physically interact with histone H2B and promote its phosphorylation. This phosphorylation is conserved from yeasts to humans and seems to play an important role in the repression of histone transcription at the end of S-phase (Mahajan et al., 2012).

One crucial question that remains unsolved is whether cells are able to respond or sense high levels of histones as they do when they become limiting in order to prevent their undesirable effects on genome stability. To address this question, we have constructed a set of tools to test and analyse the cellular consequences of abnormal accumulations of histones beyond DNA replication in the budding yeast S. cerevisiae. Our results show that cells that fail to degrade histones after DNA replication or wild-type cells constitutively exposed to high levels of histones H2A and H2B have defects in chromosome segregation and are able to suffer aberrant cell divisions that result, in some ocassions in viable cells that have experienced a whole-genome-duplication event (WGD). We observe that high levels of histones promote changes in chromatin structure, increase nucleosome occupancy at centromeric and pericentromeric chromatin and decrease the incorporation of the histone variant Htz1H2A.Z to chromatin. This defect in Htz1H2A.Z incorporation is accompanied by a reduction of condensin recruitment to pericentromeric chromatin, a phenotype that could explain why cells exposed to high levels of histones experience WGDs (Kim et al., 2009; Oliveira et al., 2005; Woodward et al., 2016). We also show that Swe1WEE1 is stabilised in the presence of high levels of histones and promotes the phosphorylation of Cdc28CDK1 through a mechanism that does not require the activation of the DNA damage response. Our results, link for the first time high levels of histones with a cell cycle mark able to delay the cell cycle transition from G2 to mitosis and highlight histone levels as a potential and yet unexplored source of genome instability.

Results

Persistent generation of histones promotes WGDs

To determine the effects of histone overexpression beyond replication, we first decided to use a centromeric vector that expresses histones H2A and H2B under the control of a promoter that is not repressed outside of S-phase (CEN-HTA1-HTB1∆NEG referred as CEN∆NEG) (Osley et al., 1986). Deletion of the NEG regulatory site (∆NEG) in the HTA1-HTB1 cluster interferes with the normal repression of histones H2A and H2B by the HIR complex and leads to their persistent transcription beyond S-phase (Figure 1a) (Eriksson et al., 2012). This vector was introduced in wild-type cells and in cells that carried a mutation in RAD53 (rad53K227A) or a deletion of TOM1 (tom1∆), both of them required for the rapid degradation of histone excess (Gunjan and Verreault, 2003; Singh et al., 2009) (Figure 1a). When transformed in rad53K227A or tom1∆ cells, this construct generated a large number of small colonies that experienced severe growth defects and a small population of large colonies able to grow almost like wild-type cells (Figure 1b) (between 2–4 × 10−2, raw numbers are in the additional source data file for Figure 1b) (Figure 1—figure supplement 1a). FACS analysis of these two populations revealed that large colonies had experienced a WGD and became diploid (Figure 1b) [detailed information on methods, strain genotype and analyses can be found in the supplemental experimental procedures section]. This effect was not observed in wild-type cells (Figure 1—figure supplement 1b) and was not specific for the overexpression of histones H2A and H2B (Figure 1c). Growth analysis of small (n) and big colonies (2 n) in the rad53K227A mutant confirmed that diploids are able to tolerate better high levels of histones (Figure 1—figure supplement 1c). To rule out the possibility that WGDs occur independently of histone levels due to a selection of pre-existing rad53K227A diploids endowed with a growth advantage after transformation, we tested if increasing the number of copies of the HTA1-HTB1∆NEG construct would be sufficient to induce WGDs in a wild-type strain. Cells transformed with a high-copy vector that contains the HTA1-HTB1∆NEG construct (2µ∆NEG) vector gave rise to a significant number of cells (around 1 × 10−2, source data file for Figure 1b) that have experienced a WGD (Figure 1d). This construct increases by 50% the amount of histone H2B mRNA but does not induce a detectable change in the amounts of total histone H2B protein (Figure 1e and f). Diploids were perfectly viable, did not show any major chromosome reorganisation when compared to a haploid or diploid strain obtained by cross (Figure 1g) and were able to form triploids when crossed with a strain from the opposite mating type (Figure 1—figure supplement 1e). Diploids still had a small growth advantage when compared to haploids in wild-type conditions (Figure 1—figure supplement 1f). WGDs were never detected in wild-type cells transformed with an empty vector (more than 2.2 × 102 colonies analysed by FACS). These results support the fact that WGDs do not take place in the absence of histone deregulation.

Figure 1 with 1 supplement see all
Persistent transcription of histones promotes WGDs.

(a) Schematic representation of the three pathways that negatively regulate histone levels at a transcriptional (HIR complex), post-transcriptional (Lsm1-7 Pat1 complex), or post-translational (Rad53-Tom1-Ubc4/5 complex) level. (b) Left: small and large colonies observed after transformation of rad53K227A or tom1∆ cells with the centromeric vector pHTA1-HTB1∆NEG (pCEN∆NEG). Right: corresponding FACS profiles (c) FACS profiles from large and small colonies after transformation of rad53K227A cells with the indicated vectors. Histone expression is under the control of the GAL1-GAL10 galactose-inducible divergent promoter. 2µ GAL1pHTA2 GAL10pHTB2 simultaneously expresses histones H2A and H2B, while CEN GAL1pHHT2 GAL10pHHF2 expresses H3 and H4. Simultaneous expression of the four canonical histones is driven by two centromeric vectors. Samples were grown in galactose before and after transformation. (d) FACS profiles from small and big colonies obtained after the transformation of a wild type strain with a high-copy vector that contains the HTA1-HTB1∆NEG construct (2µ∆NEG). (e) H2B (HTB1) Q-PCR mRNA levels normalised to TUB1 mRNA in wild-type cells transformed with the 2µ∆NEG or an empty vector (p=0.029; t-test paired samples). Each point represents an independent experiment (f) Representative example of H2B protein levels in wild-type cells transformed with the 2µ∆NEG or an empty vector. Act1 was used as a loading control (g) Karyotypes from wild-type haploid cells (n), diploid strains (2 n) (obtained by cross) and big colonies obtained after transformation with the 2µ∆NEG (2 n*). Karyotypes were analysed by Pulse Field Gel Electrophoresis (PFGE) in five independent clones for each strain.

https://doi.org/10.7554/eLife.35337.002

Deregulation of histone levels delays chromosome segregation and promotes aberrant cell divisions in which daughter cells keep all the DNA content

To further understand how WGDs take place, we focused our study on the analysis of wild-type cells transformed with the 2µ∆NEG that still remained completely haploid. Time course experiments with these cells revealed that most of them usually turned into a diploid state in 48–96 hr after inoculation in liquid from a petri dish plate (colonies have already grown for 5 days in the plate) (Figure 2a). Cell cycle distribution of these haploids transformed with the 2µ∆NEG vector in a strain carrying several tags to follow kinetochores (Mtw1-mCHERRY), centrosomes (Spc42-CFP) and tubulin (TUB1-GFP)) revealed a large population of non-dividing cells as well as a significant decrease in the number of cells that have started to separate their chromosomes (Figure 2b). Interestingly, in some of these cells we observed that the whole spindle apparatus was able to traverse to the daughter cell before segregation (Figure 2c). In order to try to catch the exact moment in which some haploids become diploid, we constructed a strain carrying a tagged nuclear membrane protein and followed cell division using live microscopy in cells transformed with the 2µ∆NEG vector or an empty vector. This analysis confirmed a large population of non-dividing cells in the strain transformed with the 2µ∆NEG vector, consistent with previous observations (Morillo-Huesca et al., 2010), and a significant number of cells (18% vs 3% in the control strain) in which the nucleus remained undivided for several hours (Figure 2d, first panel and source data file for Figure 2d). This experiment also revealed two specific phenotypes for cells transformed with the 2µ∆NEG that were not observed in cells transformed with the empty vector. The first one was a small proportion of cells (8%) in which the nucleus migrated to the daughter cell before mitosis (Figure 2d, second panel). The second phenotype involved a combination of the two phenomena previously described and led on some occasions (1%) to an aberrant mitosis in which both nuclei remain trapped in the daughter, generating a binucleated cell (Figure 2d, third panel). To confirm that these phenotypes were specific for cells transformed with the 2µ∆NEG vector, we performed additional movies comparing cells transformed with this vector or an empty one. These movies confirmed that aberrant mitosis in which the daughter (never observed in the mother) keeps all the genetic material only take place in cells transformed with the 2µ∆NEG vector. In one of these events, we were able to follow a second cell division of one of these binucleated cells in which the two nuclear masses were split between the mother and the daughter (Figure 2e). We assume that these events may serve as a precursor for diploid cells that are probably being selected among haploids due to their growth advantage.

Figure 2 with 4 supplements see all
Deregulation of histone levels delays chromosome segregation and promotes aberrant cell divisions.

(a) Representative FACS profiles of haploid wild type cells after transformation with the 2µ∆NEG vector grown during several days. Samples were taken every 24 hr and analysed by FACS to estimate DNA content. (b) Quantification of the different cell cycle stages observed in exponentially growing haploid cells after transformation with the 2µ∆NEG or an empty vector (11 Z-sections, 5µ total). The strain (DVY15) carries markers for the kinetochore (Mtw1-mCherry), the spindle pole body (Spc42-CFP), and tubulin (Tub1-GFP). Three independent samples were counted for the vector (n = 986) and the 2µ∆NEG (n = 1613). Significant p-values are indicated with an asterisk (0.022 and 0.021 from left to right; t-test, paired samples, two tails). (c) Example of a cell in which the whole mitotic machinery can be observed at the daughter cell before mitosis. (d) Live microscopy (5-min interval, 11 Z-sections, 3µ total) of strain DVY12 (Nup49-GFP) after transformation with the 2µ∆NEG vector. first panel: Representative example of cells that remain with an undivided nucleus for more than 2 hr (anaphase in control cells usually takes place in less than 20 min). second panel: Representative example of a cell in which the whole nucleus migrates to the daughter before anaphase. third panel: Representative example (10-min interval) of a cell in which both nuclei remain trapped in the daughter cell. Complete movies can be found in Figure 2d-videos 1–4. (e) Representative example of a cell in which both nuclei initially trapped in the daughter cell are able to enter a new round of division, and segregate (two nuclei at time 280). Percentages are estimated from the analysis of all movies performed during the first biological replicate (n = 228). Full data from this quantification can be observed in the additional Source Data file.

https://doi.org/10.7554/eLife.35337.005

Rad53 depletion allows a large and conditional overproduction of histones beyond S-phase in a lsm1∆ background

In the experiments described above, we have performed single-cell analysis of cells transformed with a vector that constitutively expresses histones H2A and H2B during the cell cycle. This construct constitutes a useful tool to test the effects of histones per se in genome stability in a mutant-free background. This tool, however, does not allow us to perform molecular studies in wild-type cells exposed to high levels of histones since most of the population becomes fully diploid before large volumes of culture are obtained. Synchronisation experiments in these cells were also problematic. Cells transformed with the 2µ∆NEG vector accumulate large amount of cells that do not exit G1 after an alpha factor release or have a very slow progression to S-phase, consistent with previous observations (Morillo-Huesca et al., 2010). To overcome these problems and further prove our initial observations, we designed an alternative experimental setting that would theoretically be able to generate large amounts of histones in the cell. We therefore created a strain that combined a deletion of LSM1, involved in the posttranscriptional degradation of histone mRNA (Herrero and Moreno, 2011) with an Auxin-Inducible Degron of Rad53 (rad53-AID) (see Eriksson et al., 2012) and Figure 1a for additional information about histone regulation in S. cerevisiae). This strategy allowed us to conditionally block at the same time the two major pathways involved in histone degradation at the end of S-phase which is otherwise lethal (Herrero and Moreno, 2011).

The rad53-AID degron was fully functional as indicated by the low level of Rad53 after one hour of treatment with 500 µM NAA (1-Naphthaleneacetic acid, NAA) (Figure 3—figure supplement 1a). rad53-AID mutant cells were sensitive to histone overexpression and deletion of LSM1 was lethal in these cells in the presence of Auxin (Figure 3—figure supplement 1b and c). Indeed, 4 hr of Auxin treatment was sufficient to reduce 90% of cell viability (Figure 3—figure supplement 1d). This lethality was much higher than the one observed in cells transformed with the 2µ∆NEG (Figure 3—figure supplement 1e) indicating that this conditional system was much more toxic for the cell.

We next measured histone levels in rad53-AID LSM1 or rad53-AID lsm1∆ cells. To this purpose, we purified core histones (Jourquin and Géli, 2017) and measured their amounts relative to a non-specific band in a Comassie Blue-stained acrylamide gel (Figure 3a). Quantification of histone levels indicated that Rad53 depletion per se had little or no effect on histone accumulation but did increase histone levels in an lsm1∆ background (Figure 3b). Western blot analysis with these samples confirmed a significant increase of canonical histones compared to the levels of histone H2A.Z, a histone variant that is not regulated by Lsm1 as canonical histones (Figure 3c, Figure 3—figure supplement 3f). To address if these changes in chromatin composition were able to affect chromatin structure, we performed a Microccocal Nuclease (MNase) digestion analysis of chromatin extracted from single and double mutants treated or not with NAA (Figure 3d). rad53-AID-lsm1∆-treated cells displayed a clear aberrant nucleosome pattern. While mono and di-nucleosomes were visible at concentrations ranging from 2.5 to 5 mU of Mnase in controls, chromatin from rad53-AID-lsm1-treated cells produced a high-molecular-weight smear that suggests that this chromatin is somehow less accesible to Mnase digestion. Overall, our results indicate that the rad53-AID lsm1∆ system constitutes a good tool to generate conditional large accumulations of histones in the cell and that high levels of histones are able to promote changes in chromatin structure.

Figure 3 with 3 supplements see all
Rad53 depletion in lsm1∆ cells allows large and conditional overproduction of histones beyond S-phase.

(a) Purified core histones from rad53-AID and rad53-AID lsm1∆ cells that were treated or not 2 hr with NAA and blocked with Nocodazole 2 additional hours. The image represents a Coomassie-Blue-stained gel of histone-purified samples. (b) Quantification of histones in the samples loaded in (a). Histone signal (indicated with a bracket) was normalised using the signal of a non-specific band (indicated with an asterisk). The results were normalised to lane 1. Quantification was done using image J. Two independent experiments were performed. *p=0.026 (t-test, paired, two tails) (c) Western Blots of canonical histones H3, H4 and H2A, and the histone variant H2A.Z. performed with purified core histones samples. Ratio H2A/H4 and H2AZ/H4 are shown in Figure 3—figure supplement 3f. (d) Micrococcal nuclease (Mnase) analysis of chromatin extracted from rad53-AID and rad53-AID lsm1∆ cells treated as described in (a). (e) Live microscopy (20-min interval, 14 Z-sections, 2,89µ total) in strain DVY36 (rad53-AID lsm1∆ NUP49-GFP) after 90 min of Auxin treatment (Auxin was maintained in the media). (f) Cell cycle distribution of rad53-AID, lsm1∆ and rad53-AID lsm1∆ cells after treatment with 500 µM NAA. (g) Pulse field gel electrophoresis of rad53-AID lsm1∆ cells that were treated with 200 mM Hydroxyurea (HU), 10 µg/ml Nocodazole (NOC), or 500 µM NAA (NAA). Control asynchronous growing cells were not treated (UT). HU-treated cells are blocked in S-phase preventing chromosomes from entering the gel. Nocodazole allows full replication but blocks mitosis. (h) Percentage of rad53-AID lsm1∆ cells that have experienced or not a WGD after a 4 hr treatment with 500 µM NAA. The DNA content of all survivors was analysed by FACS. two independent experiments (n = 100 or 75) p-value=0.038 (t-test, two tails, unequal variance).

https://doi.org/10.7554/eLife.35337.011

We next sought to investigate if this system phenocopied some of the cell cycle defects observed in cells transformed with the 2µ∆NEG vector. Live microscopy and DAPI staining analysis of rad53-AID-lsm1∆-treated cells reveals a large majority of cells that arrest as large budded cells in which the nucleus migrates between the mother and the daughter cell (Figure 3e and Figure 3—video 1 and 2), a behavior similar to the one observed in the 2µ∆NEG cells depicted in the third panel of Figure 2. These cells usually died after 6–8 hr of NAA treatment and were unable to enter or skip mitosis. FACS analysis suggested that most of them were able to complete replication (Figure 3f). To confirm this observation, we next performed pulse field gel electrophoresis (PFGE) allowing us to differentiate chromosomes in which DNA is fully replicated (chromosomes are able to enter the gel) from chromosomes in which some parts are not (DNA is unable to enter the gel) (Ide and Kobayashi, 2010). Cells treated for 4 hr with NAA exhibited a similar pattern to the one observed in cells treated with Nocodazole and clearly different from the one observed in cells treated with Hydroxyurea that leads to the accumulation of cells that have not fully replicated their DNA (Figure 3g). To confirm if this system was a useful tool to study the generation of WGDs, we estimated ploidy by FACS in a population of rad53-AID lsm1∆ that were platted in rich media plates after 4 hr of treatment with Auxin or DMSO (control). Treatment with Auxin specifically raised a population of cells that had doubled their DNA content (Figure 3h). This result reinforces the idea that transient accumulation of high levels of histones are sufficient to promote WGDs in the cell and suggest that the rad53-AID lsm1∆ double mutant is a useful tool to study how this process takes place at a molecular level.

High levels of histones decrease Htz1H2A.Z and condensin incorporation to pericentromeric chromatin

Cse4CENP-A and Htz1H2A.Z are two histone variants that can replace canonical histones H3 and H2A, respectively. Cse4CENP-A lies at the centromere and is essential to recruit the kinetochore (Stoler et al., 1995; Meluh et al., 1998; Fachinetti et al., 2013). Htz1H2A.Z is enriched at the centromere and pericentromeric regions and is required for efficient chromosome segregation (Albert et al., 2007; Krogan et al., 2004). Cse4CENP-A incorporation outside of the centromere region has been previously linked to WGDs in certain tumours (Tomonaga et al., 2003) and to aneuploidy in yeast (Castillo et al., 2007; Collins et al., 2007). Similarly, high-levels of Htz1H2A.Z incorporation at the pericentromeric region in yeast also affected ploidy (Chambers et al., 2012). Since histone overexpression is able to promote changes in centromeric chromatin (Au et al., 2008; Castillo et al., 2007; Salzler et al., 2009; Takayama et al., 2010) we checked if high levels of histones were able to promote WGDs by changing the stoichiometry of canonical histones versus non-canonical histones at centromeric chromatin. Chromatin immunoprecipitation (ChIP) of histones (H3, H4, H2B and H2A) and histone variants (Cse4-MycCENP-A and Htz1H2A.Z) were carried out in rad53-AID and rad53-AID lsm1∆ mutants that were treated or not with Auxin (Figures 4 and 5). These experiments were carried out in the presence of Nocodazole to confirm that all the changes observed are related to histone accumulation and not to changes in the composition of centromeric chromatin related to the cell cycle.

High levels of histones do not affect CENP-A recruitment to centromeres or the attachment of chromosomes to the spindle axis.

(a) Relative levels of histones H3, H4, Cse4 (CENP-A) measured by ChIP-Q-PCR at the indicated loci in rad53-AID (DVY22) and rad53-AID lsm1∆ (DVY23) cells. Samples were obtained from cultures treated (or not) with Auxin for 2 hr and incubated 2 additional hours with Nocodazole. CEN4 and CEN12 correspond to the centromeric region of Chromosomes IV and XII. INT IV and INT XII to intergenic regions in the same chromosomes. ALG9 corresponds to a small amplicon in the coding region of ALG9. p-Values were obtained from a Student T-test (paired samples, two tails) that compares rad53-AID-untreated cells versus rad53-AID-lsm1∆-treated cells *p<0.05. (b) MTW1 chromatin association (ChIP-qPCR) obtained in two independent experiments in which asynchronous growing rad53-AID and rad53-AID lsm1∆ cells were treated or not with NAA for 4 hr (c) Same as in b but using samples that were synchronised with NAA and Nocodazole as in (a). (d) Two representative examples of the normal distribution of kinetochores (MTW1-mCherry) along the spindle axis (Tub1-GFP) in rad53-AID lsm1∆ cells after 4 hr of Auxin treatment.

https://doi.org/10.7554/eLife.35337.016
Figure 5 with 1 supplement see all
High levels of histones decrease Htz1H2AZ and condensin incorporation into pericentromeric chromatin.

(a) Relative levels of histones H2B, H2A and H2A.Z at the indicated loci measured by ChIP-Q-PCR in rad53-AID (DVY22) and rad53-AID lsm1∆ (DVY23) cells. Samples were obtained from cultures treated (or not) with Auxin for 2 hr and incubated 2 additional hours with Nocodazole. This synchronisation was performed to avoid cell-cycle-related differences. CEN4 (or 12) L (left) and R (right) correspond to the pericentromeric regions of centromeres 4 and 12. p-Values were obtained from a Student T-test (paired samples, two tails) that compares rad53-AID-untreated cells versus rad53-AID-lsm1∆-treated cells *p<0.05; **p<0.01 (b) Representative example of H2A.Z protein levels obtained in wild-type cells transformed with an empty vector or the 2µ∆NEG vector (c) Quantification of protein levels of H2B and H2A.Z relative to Act1. p-Value for H2B differences (t-test, paired, one tail) 0.36; p-value for H2A.Z=0.033 (d) Representative example of H2A.Z protein levels obtained in wild-type cells transformed with either an empty vector or the 2µ∆NEG vector. These cells carry a second empty vector (prs425) or the multicopy vector prs425-HTZ1 driving the overexpression of HTZ1 (H2A.Z) (e) WGD events observed in cells transformed with the indicated plasmids. WGDs were estimated as in Figure 1 (described also in the Supplemental Experimental Procedures). p=0.045 (t-test, paired, one tail) four independent experiments (f) Percentage of cells with a haploid, mixed, or diploid DNA content during a 4 days (D1–D4) time-course in wild type or swr1∆ cells after transformation with the NEG vector. Percentages were obtained from the analysis of 15 clones for each background. (g) Relative levels of Brn1 associated to chromatin measured by ChIP-Q-PCR. Experiments and statistical tests were carried out as in (a) in strains DVY32 and DVY33 expressing HA-tagged Brn1. This experiment was performed five times with five independent biological replicates. (h) Plate growth assay of a wild-type strain and a condensin thermosensitive mutant (smc2-8) transformed with an empty vector or the centromeric version of the ∆NEG vector (CEN∆NEG).

https://doi.org/10.7554/eLife.35337.018

Analysis of several regions of chromosomes IV and XII revealed no changes in the amount of Cse4CENP-A at centromeres (Figure 4a, left panel). As expected, incorporation of Cse4 outside of the centromeres was low and differences in the Cse4 at these regions were not significant. Canonical histones in contrast did increase their occupancy along several regions and were highly enriched at centromeres (Figure 4a, middle and right panels). To check if kinetochores were still able to attach to centromeres, we measured by ChIP the recruitment of Mtw1, an essential component of the MIND complex that binds centromeric DNA through the association with inner and outer kinetochore elements (Hornung et al., 2014) in asynchronous growing cells (Figure 4b) and in cells synchronised with Nocodazole (Figure 4c). Mtw1 recruitment to centromeres IV and XII was similar and did not significantly increased in other regions of the genome. To confirm that kinetochores remained bound to the spindle, we studied the behaviour of chromosomes along the spindle axis in rad53-AID lsm1∆ arrested cells using a strain with tags to follow kinetochores (Mtw1-mCHERRY) and the spindle axis (TUB1-GFP) (Liu et al., 2008). As shown in Figure 4d, most of the arrested cells displayed one single mass of chromosomes in the daughter or the mother cell that always remained in line with the spindle axis. We concluded that histone overexpression is able to increase histone H3 and H4 occupancy at centromeres and several other regions but this incorporation does not seem to affect CENP-A recruitment, the attachment of kinetochores to chromosomes, or the attachment of both to the spindle.

We next analysed the recruitment of Htz1H2A.Z to chromatin (Figure 5a). We observed a clear decrease in the incorporation of Htz1H2A.Z to pericentromeric chromatin in agreement with the global decrease of the H2AZ/H4 ratio shown in Figure 3c. This decrease was opposite to what was observed for histones H2B and H2A and indicated that Htz1H2A.Z molecules were being replaced by the canonical histone H2A in the rad53-AID lsm1∆ mutant treated with Auxin.

Interestingly, although wild-type cells transformed with the 2µ∆NEG vector do not seem to increase the protein levels of H2B (see above Figure 1f), the level of Htz1H2A.Z was decreased in these cells (Figure 5b and c). To investigate if changes in Htz1H2A.Z levels were directly related to WGDs, we decided to interfere with the incorporation of this histone variant in both, the 2µ∆NEG and the rad53-AID lsm1∆ systems. Htz1H2A.Z overexpression using a high copy vector largely increased the levels of Htz1H2A.Z observed in cells transformed with the 2µ∆NEG vector (Figure 5d) and was able to decrease the amount of WGDs observed in these cells (Figure 5e). Diploid cells were not affected by Htz1H2A.Z overexpression discarding an indirect effect due to a counterselection (Figure 5—figure supplement 1a). Consistent with the result of Figure 5e, deletion of SWR1, which decreases Htz1H2A.Z incorporation to chromatin (Mizuguchi et al., 2004) had an opposite effect to the one conferred by Htz1H2A.Z overexpression and increased the formation of WGDs in cells transformed with the 2µ∆NEG (Figure 5f). Overexpression of histone Htz1H2A.Z in rad53-AID lsm1∆ cells also increased the amounts of total histone Htz1H2A.Z (Figure 5—figure supplement 5b) but was unable to supress the defect in the incorporation of this histone variant to chromatin, the G2/M arrest or the lethality after NAA treatment (Figure 5—figure supplement 1c, d ane e).

In summary, our results with the 2µ∆NEG vector indicate that Htz1H2A.Z relative levels are important in the generation of WGDs in cells exposed to a persistent transcription of histones H2A and H2B. In contrast, overexpression of this histone variant is unable to suppress the defects observed in rad53-AID-lsm1∆-treated cells and suggests that this double mutant might have additional effects on chromatin structure that interfere with Htz1H2A.Z deposition, cell cycle, and cell viability.

Excessive incorporation of Htz1H2A.Z to pericentromeric chromatin has been previously linked to WGDs (Chambers et al., 2012). While the opposite has been linked to chromosome segregation defects, it has never been associated to WGDs. In S. pombe Htz1H2A.Z is required to stabilise the recruitment of condensin to pericentromeres, a complex able to embrace each of the daughter DNA strands and required to allow their proper segregation (Cuylen and Haering, 2011; Kim et al., 2009). Depletion of the CAP-H kleisin subunit (Brn1) of Condensin II in drosophila S2 cells has been shown to lead to chromosome segregation defects and produce a small number of abortive mitosis in which all the genetic material is retained in one cell (Oliveira et al., 2005). Mutations in the condensin II subunit Caph2 in mice have also been recently involved in the generation of cells with an increased ploidy content (Woodward et al., 2016). Based on these published results, we tested if Htz1H2A.Z defective recruitment observed in the rad53-AID lsm1∆ double mutant would affect condensin recruitment to pericentromeric chromatin (Figure 5g). Recruitment of Brn1 to pericentromeric chromatin decreased in three out of the four pericentromeres examined in five independent experiments suggesting that Htz1H2A.Z is important for condensin recruitment to pericentromeric regions in S. cerevisiae. We further tested the effects of histone overexpression in a strain that carries a thermosensitive version of the condensin subunit SMC2 (smc2-8). The smc2-8 mutant, was extremely sensitive to histone overexpression (Figure 5h) further linking histone levels to condensin function. Cell cycle kinetics to measure WGDs in these cells was not possible due to the extreme sensitivity of this mutant to histone overexpression.

Accumulation of histones triggers a DNA damage-independent phosphorylation of Cdc28CDK1

One of the initial goals of this work was to evaluate if cells have mechanisms able to modulate cell cycle progression in the presence of abnormal levels of histones as they do when histones become limiting during replication. This question is particularly relevant since persistent activation of the DDR, the SAC and Swe1 overexpression have all been linked to WGDs (Davoli et al., 2010; Sotillo et al., 2007; Kawasaki et al., 2003). We therefore investigated if high levels of histones were able to promote the activation of the DDR, the SAC or affect Swe1 stability using our two systems (rad53-AID lsm1∆ and 2µ∆NEG) able to deregulate histone levels during the cell cycle.

Auxin treatment in rad53-AID lsm1∆ cells did not promote Pds1Securin phosphorylation (Figure 6a), a marker for DDR activation, or enhanced the formation of Mad2 foci at kinetochores, a marker used to detect SAC activation (Figure 6b). To further confirm or discard the activation of the SAC, we decided to follow Pds1SECURIN degradation during a G2/M release in rad53-AID and rad53-AID lsm1∆ cells treated or not with NAA before the release. This experiment revealed a slight stabilisation of Pds1SECURIN in rad53-AID-lsm1∆-treated cells, higher levels of Scc1Rad21, the main substrate of Pds1, and Clb2 (Figure 6c and Figure 6—figure supplement 1a and b). Clb2 started to degrade but reaccumulated later suggesting that the activity of APC-Cdh1 is impaired in this mutant. We next looked at Swe1WEE1 levels and observed that this kinase was stabilised in a hypophosphorylated form in rad53-lsm1∆-treated cells (Figure 6d). This stabilisation led to a higher level of Cdc28CDK1 phosphorylation that depend on Swe1WEE1 (Figure 6e) and was independent of the DDR, since MEC1 deletion did not abolish it (Figure 6f). SWE1 deletion was able to suppress Cdc28CDK1 phosphorylation but unable to suppress Pds1 stabilisation suggesting that the mechanism by which Pds1 is stabilised does not depend on Cdc28CDK1 activity (Figure 6g and Figure 6—figure supplement 1c). The results obtained with the rad53-AID lsm1∆ system suggest that histone accumulation can stabilise Pds1 and delay cohesion degradation, and that it promotes a Swe1-dependent phosphorylation of Cdc28CDK1. The absence of Pds1SECURIN phosphorylation added to the fact that its stabilisation does not depend on Cdc28CDK1 phosphorylation favor the idea that Pds1 stabilisation could be due to an activation of the SAC.

Figure 6 with 1 supplement see all
Histone accumulation triggers a Swe1-dependent Cdc28CDK1 phosphorylation.

(a) Pds1 phosphorylation (Pds1-HA) in rad53-AID and rad53-AID lsm1∆ cells either after 4 hr treatment with NAA or bleomycin (Ble, 15 µg/ml). Pds1 appears as a doublet. Its phosphorylation can be visualised as a shifted band after bleomycin treatment. (b) Average number of Mad2 foci (Mad2-GFP) that co-localise with kinetochores (Mtw1-mCHERRY) in rad53-AID and rad53-AID lsm1∆ cells after 4 hr of treatment with either DMSO or Auxin. Two independent experiments were performed in which at least 100 cells were counted. (c) Cell-cycle progression from G2/M to the next G1 in rad53-AID lsm1∆ cells after treatment or not with Auxin. Exponentially growing cells were treated (or not) with Auxin for 2 hr and incubated for 2 additional hours with Nocodazole. Cells were then washed to eliminate Nocodazole and released into media containing Auxin and alpha-factor to arrest cells able to enter mitosis the next G1. This experiment was performed in cells carrying either a PDS1-HA tag or a SCC1-MYC tag. The presence of Clb2 and Rap1 was monitored with anti-Clb2 and anti-Rap1 antibodies. Rap1 was used in both as a loading control. A representative full example of the experiment for each strain is shown in the Figure 6—figure supplement 1a and b. This data file also includes the results obtained with the single mutant that are not shown in the this figure. Graphs below each panel represent the progression of cells to the next G1. These percentages were estimated from fixed cells that were analysed using DAPI staining. (d) Swe1 protein levels in rad53-AID and rad53-AID lsm1∆ cells treated as indicated. (e and f) Cdc28CDK1 phosphorylation levels in the indicated strains. Cdc28CDK1 phosphorylation is monitored with phospho-cdc2 (Tyr15) antibodies. Cells were treated as indicated (g) Cell-cycle progression from G2/M to the next G1 in rad53-AID lsm1∆ and rad53-AID lsm1∆ swe1∆ similar to the one described in (6d), using the strain that expresses Pds1-HA. The results obtained with the single mutant rad53-AID are shown in the Figure 6—figure supplement 1c. All the experiments observed in Figure 6 were performed twice and gave similar results.

https://doi.org/10.7554/eLife.35337.021

We then tested checkpoint activation in cells transformed with the 2µ∆NEG vector. Cells transformed with this vector did not trigger Rad53CHK2 phosphorylation or enhanced the formation of Ddc2ATRIP foci, two common markers of the DNA damage checkpoint (Figure 7a and b). The fact that Mad2 foci were slightly increased (Figure 7c) suggest that the SAC could be activated at least in some cells when histones are overexpressed. We failed to detect Cdc28CDK1 phosphorylation in asynchronous wild-type cells transformed with an empty vector or the 2µ∆NEG construct (data not shown). To confirm if Cdc28CDK1 phosphorylation was related to high levels of histones, we compared the phosphorylation kinetics of Cdc28CDK1 during one complete cell cycle (from G1 to the next G1) in wild type and rad53K227A cells in which histones H2A and H2B were overexpressed from a galactose inducible promoter (Figure 7d). Cdc28CDK1 phosphorylation in wild-type cells with no histone overexpression was normally enhanced as cells entered replication and decreased when they reached the G2/M transition (Figure 7d upper left panel). This phosphorylation was only modestly affected in wild-type cells overexpressing histones (Figure 7d upper right panel) but was maintained for a longer time when histones were expressed in the rad53K227A strain (Figure 7d compare lower left and right panel). Cells exposed to high levels of histones seem to be able to promote a Cdc28CDK1 phosphorylation, at least when Rad53 activity is impaired. This result suggest that the Cdc28CDK1 phosphorylation observed in rad53-AID-lsm1∆-treated cells is directly related to the presence of high levels of histones.

Plasmid-driven overexpresson of histones promotes Cdc28CDK1 phosphorylation.

(a) Rad53 phosphorylation in wild type cells transformed with an empty vector (UT), the centromeric, or the 2µ version of HTA1-HTB1∆NEG (CEN∆NEG and 2µ∆NEG respectively). Phosphorylated Rad53 appears as a shifted band clearly observed after treatment with 200 µM Hydroxyurea (HU) or 0.03% MMS (MMS). (b) Average and SEM (four independent experiments, 200 cells counted for each) of wild-type cells exhibiting Ddc2-Foci after transformation with an empty vector or the 2µ∆NEG vector. DNA content was measured for each sample to confirm that cells still remain haploid (not shown). (c) Average number of Mad2 foci (Mad2-GFP) that co-localise to kinetochores (Mtw1-mCHERRY) in wild-type cells after transformation with an empty vector or the 2µ∆NEG vector. DNA content was measured for each sample to confirm that cells still remain haploid (not shown). p-value=0.031 (t-test, paired, two tails) (d) Phosphorylation of Cdc28 upon H2A-H2B overexpression. Wild type and rad53K227A cells were transformed with either a control vector or a high-copy vector that expresses histones H2A and H2B from a galactose-inducible promoter. Cells were grown in raffinose, arrested in G1 with alpha-factor, incubated 2.5 hr with galactose to induce H2A-H2B expression, and released from alpha-factor block to follow cell cycle progression (FACS) and Cdc28 phosphorylation. Alpha factor was re-added 75 min after to re-arrest cells in G1. These experiment was performed twice and gave similar results.

https://doi.org/10.7554/eLife.35337.024

To address which is the contribution of these proteins to both, the arrest observed in rad53-AID-lsm1∆-treated cells and to the generation of WGDs in 2µ∆NEG transformed cells, we finally performed cell cycle kinetics in rad53-AID lsm1∆ cells and wild-type cells transformed with the 2µ∆NEG vector in cells lacking Pds1Securin, Mad2MAD2 and/or Swe1WEE1. PDS1 deletion had only a slight effect on the cell cycle arrest observed in rad53-AID-lsm1∆-treated cells and did not significantly change the kinetics of diploidisation observed in cells transformed with the 2µ∆NEG vector (Figure 8—figure supplement 1a–c). MAD2 and SWE1 deletions were able to accelerate the generation of WGDs in 2µ∆NEG transformed cells (Figure 8a) and did not change the ploidy content of cells transformed with an empty vector (Figure 8—figure supplement 1d). These deletions had almost no impact on the the cell cycle arrest observed in rad53-AID-lsm1∆-treated cells individually (Figure 8—figure supplement 1d) but significantly suppressed the arrest when combined (Figure 8b and Figure 8—figure supplement 1d). Overall, our results confirm that the WGD events observed upon histone overexpression are not generated by a persistent activation of the DDR, the SAC, or a stabilisation of Swe1WEE1. The results observed with Mad2 and Swe1WEE1 suggest that the SAC might instead cooperate with Swe1WEE1 and help to prevent them.

Figure 8 with 1 supplement see all
Deletion of SWE1 or MAD2 enhances WGDs.

(a) Percentage of cells with a haploid, mixed, or diploid DNA content during a 4 days (D1–D4) time-course in wild type, swe1∆, or mad2∆ cells after transformation with the NEG vector. These kinetics were performed as the ones shown in Figure 2a. Eight clones were analysed for each genetic background. DNA content of the indicated mutant cells is measured after transformation with 2µ∆NEG. Samples were taken every 24 hr and analysed by FACS to estimate DNA content. Eight independent clones were analysed for each genetic background. (b) Number of unbudded and budded cells with one or two well-differentiated nuclei observed in rad53-AID lsm1∆ and rad53-AID lsm1∆ mad2∆ swe1∆ cells after NAA treatment. Percentages were estimated by DAPI in fixed samples. At least 100 cells were counted. p-Values were obtained for each cell type with a one-way ANOVA test (**p<0.01 ***p<0.001) in which two independent experiments were compared.

https://doi.org/10.7554/eLife.35337.026

Discussion

In this study, we demonstrate that high levels of histones interfere with chromosome segregation and are able to promote WGDs. Collectively, our results indicate that these events are linked to a defect in the incorporation of the histone variant Htz1H2A.Z. We propose a competition model in which the canonical histone H2A competes with the histone variant H2A.Z. In this model, increasing the levels of canonical histone H2A will decrease the ratio H2A.Z/H2A in nucleosomes thereby decreasing the interaction of H2A.Z with the readers of this histone variant. Histone overexpression has been previously shown to saturate certain histone-modifying enzymes (Singh et al., 2010) and could also in theory, saturate chromatin remodelers able to bind and exchange H2A by H2A.Z. However, this explanation is not sufficient to explain how the overexpression of histones H2A and H2B trigger WGDs since H3 and H4 overexpression also triggers WGDs in a rad53K227A mutant. Two reports have shown that the incorporation of Htz1H2A.Z (Ranjan et al., 2013) and condensin (Toselli-Mollereau et al., 2016) requires nucleosome-free regions to be incorporated efficiently to chromatin (Ranjan et al., 2013). Our results show that the regions in which Htz1H2A.Z incorporation is decreased in rad53-AID-lsm1∆-treated cells exhibit higher nucleosome occupancy and are less accessible to Micrococcal Nuclease digestion. Accumulation of histones beyond replication might have a general effect on chromatin structure and reduce the amount of nucleosome-free regions. This general effect on chromatin structure may explain why the overexpression of Htz1H2A.Z is able to decrease the amounts of WGDs in cells overexpressing histones H2A and H2B but does not improve the incorporation of Htz1H2A.Z in rad53-AID-lsm1∆-treated cells. We infer from our results that histone expression may be restricted to S-phase to avoid an excessive incorporation of histones to chromatin and maintain nucleosome-free regions that would in particular promote the specific incorporation of Htz1H2A.Z (Boyarchuk et al., 2014; Nekrasov et al., 2012). Efficient incorporation of Htz1H2A.Z would regulate condensin recruitment throughout the cell cycle (Figure 9). Condensin depletion has been previously linked to chromosome segregation defects and WGDs (Oliveira et al., 2005; Woodward et al., 2016) and could explain why high levels of histones promote WGDs.

Proposed model to explain how histone-stress impacts mitosis.

Histone dimers are depicted as small rectangles (H3-H4 in white, H2A-H2B in grey and H2A.Z-H2B in blue). During an unperturbed cell cycle (a, upper scheme), canonical histones and Swe1WEE1 increase during replication. Swe1WEE1 will phosphorylate Cdc28CDK1 and maintain it inactive. During G2, histone synthesis will be repressed and all histones (mRNAs and proteins) that are not incorporated to chromatin as well as Swe1WEE1 will be degraded. During mitosis, Htz1H2A.Z will stabilise condensin recruitment at pericentromeric regions allowing the proper function of this complex in chromosome segregation. When histone degradation is compromised (b, lower scheme), cells reach G2 with high levels of histones (histone-stress). This accumulation of histones will promote Cdc28CDK1 phosphorylation and inactivation. This inhibition will delay the entry into mitosis and presumably give time to the cell to lower histone levels. Since this phosphorylation depends on Swe1WEE1, we propose that histone-stress promotes Cdc28CDK1 phosphorylation through a stabilisation of Swe1WEE1. Cells unable to efficiently lower histone levels after replication will increase the amount of canonical nucleosomes incorporated at centromeres and pericentromeres. We propose that this increase in nucleosome density will decrease the efficient exchange of histone H2A by histone Htz1H2A.Z, and reduce Htz1H2A.Z incorporation. This defect in incorporation would consequently lead to a less stable association of condensin to pericentromeres and trigger chromosome segregation defects.

https://doi.org/10.7554/eLife.35337.029

We have also addressed in this study if high levels of histones modulate cell cycle progression to avoid their potential toxicity. Our results confirm that histone overexpression does not trigger the canonical DNA damage response. The Pds1 stabilisation observed in rad53-AID lsm1∆ treated cells added to the fact that MAD2 deletion enhances WGDs indicate that the SAC could be activated in cells that overexpress histones. This activation, however, does not seem to be the cause behind WGDs but rather a mechanism that helps to prevent them. One of the most interesting results of our checkpoint analysis is that cells that accumulate large amount of histones exhibit a DNA damage-independent phosphorylation of Cdc28CDK1 that likely results from the stabilisation of Swe1WEE1. As mentioned, this activation would not be the cause of WGDs but rather would help to prevent them. Swe1WEE1 was previously reported to interact with histone H2B in yeast and humans, phosphorylate H2B at Tyr37 (Tyr40 in yeast), and promote the efficient repression of histones transcription at the end of S-phase (Mahajan et al., 2012). The fact that this kinase is stabilised in the presence of high levels of histones unveils an unsuspected and general role of Swe1WEE1 in histone metabolism. Noteworthy, low levels of histone supply during S-phase have been shown to delay mitosis in Drosophila embryos through the increase of CDK1 phosphorylation levels mediated by the transcriptional downregulation of the Cdc25 phosphatase (Günesdogan et al., 2014). Our results added to this study suggest that the regulation of Cdc28CDK1 could be a mechanism by which cells can respond to both high and low levels of histones.

WGDs are quite frequently observed in cancer (Zack et al., 2013) and play an important role in tumorigenesis (Davoli and de Lange, 2012; Dewhurst et al., 2014; Gordon et al., 2012; Janssen and Medema, 2013; Santaguida and Amon, 2015). Given the different ways in which higher and lower eukaryotes regulate histone levels, one reasonable question that arises from our study is if our results can be extrapolated to higher eukaryotes. Overexpression of histone H2A has been linked to the transformation of normal liver to preneoplastic and neoplastic stages of Hepatocellular Carcinoma (Khare et al., 2011) where WGDs and aneuploidy are both frequent (Davoli and de Lange, 2011; Duncan et al., 2010). Two different lines of investigation, one in drosophila and one in bronchial cells, have shown that a partial depletion of the histone mRNA-binding factor SLBP leads to the accumulation of abnormal levels of polyadenylated histones mRNA beyond replication. These polyadenilated mRNAs can be translated and escape the normal cell cycle control of histones (Brocato et al., 2015). In the first study, this depletion correlated to the generation of a tetraploid population (Salzler et al., 2009). The second study demonstrated that arsenic, a common carcinogenic agent able to promote aneuploidy and WGDs, partially depletes SLBP, which in turn increases the tumorigenic potential of transformed cells (Brocato et al., 2015; States, 2015). Interestingly, arsenic-induced cellular transformation induces a more compact chromatin structure (Riedmann et al., 2015). Our work, combined with these reports, supports the idea that histone deregulation might constitute an important and yet unexplored potential source of genome instability in cells. It would therefore be interesting to revise upwards the amplification of the small arm of chromosome 6, which contains 55 out of the 65 genes that encode canonical histones. This amplification takes place in a wide variety of tumors and correlates with cancer aggressiveness (Santos et al., 2007).

Materials and methods

Strains, plasmids, growth conditions and cell-cycle synchronisation

All strains used in this study derive from the S288C background. Genotypes are listed in Supplementary file 1a, plasmids in Supplementary file 1b and oligonucleotidess in Supplementary file 1c.

Rad53 Auxin-Induced Degron (AID) was constructed as described in Nishimura et al., 2009. Rad53 was tagged by PCR using a carboxy terminal tag obtained from vector pMK43. The strain carrying the rad53-AID allele also bears the integrative vector pNHK36, which contains the OsTYR1 protein under the control of the GAL1-inducible promoter. Both vectors were obtained from Addgene. All rad53-AID strains were constructed in an sml1∆ background to compensate the essential function of Rad53 in promoting dNTP production. The functionality of tagged proteins was systematically checked.

Exponentially growing rad53K227A (or tom1∆) cells were transformed with a centromeric vector carrying the HTA1-HTB1∆NEG construct while wild-type cells were transformed with a 2-µ vector carrying the same cassette. After transformation, strains were plated on SC-Uracil selective media at 30°C. The frequency of WGDs in these strains was calculated based on the percentage of cells able to form a colony after 3 days of growth compared to the total amount of colonies visible after 5 days of growth. This method to estimate WGDs is based on the growth advantage that WGDs confer upon histone overexpression. Ploidy analysis of individual transformants by FACS indicates that roughly than 90% of cells able to grow at day 3 have experienced a WGD (8 or 9 out of 10, two independent repeats for each strain). As control, similar experiments were performed with strains transformed with an empty vector to show that generation of WGDs in linked to the presence of the ∆NEG construct. Time courses to determine the timing of WGDs were performed always from single small colonies that were grown and maintained in exponential growth conditions in liquid SC-uracil. As control, single colonies obtained from the corresponding strain transformed with an empty vector were analysed in parallel. Samples were taken every 24 hr to estimate the proportion of haploids and diploids by FACS.

For plate growth assays, yeast cultures were diluted to an O.D.600nm of 0.5 and serial 1:10 dilutions were spotted on the indicated plates. Doubling times were calculated from 12 hr time-courses in which cells were maintained in exponential growth conditions. Doubling times correspond to the growth rate of cells that still remain haploid during the time course (checked by FACS).

For experiments involving the rad53-AID construct, glucose was replaced with galactose to allow the expression of the OsTyr1 protein, required for AID-mediated degradation. These experiments were always carried out in yeast-peptone rich media (YP) except those performed for microscopy that were carried in SC media. Rad53 depletion was induced adding 1-Naphthaleneacetic acid (SIGMA N0640, referred as NAA or Auxin in the text). NAA was normally added to a final concentration of 500 µM (rich media) or 200 µM (SC) unless indicated. This difference in NAA concentration used between rich and minimal media is due to the fact that high concentrations of NAA in minimal media have deleterious effects on yeast growth.

Cell-cycle synchronisations in G1 were performed adding alpha factor to a final concentration of 0.01 µg/ml. All strains were bar1∆ to ensure a complete arrest that was checked by FACS and microscopy (90% or more of G1-arrested cells). Cell-cycle synchronizsations in G2/M were performed adding 10 µg/ml of Nocodazole. Cells were maintained in the presence of alpha-factor or Nocodazole for a duration corresponding of at least one whole cell cycle (estimated from duplication time). Cell-cycle arrest was confirmed by FACS and DAPI staining after fixation in 70% ethanol. After rehydratation in 1X PBS, samples for FACS were incubated for 3 hr with RNaseA (0.5 mg/ml) at 37°C. Cells were then resuspended in 0.5 µg/ml SYTOX green in PBS and incubated for at least 20 min at room temperature. After a 1X PBS wash, cells were briefly sonicated to remove cell clumps and DNA content was determined with a Becton Dickinson FACS-Calibur. For DAPI analysis, fixed cells were incubated with 0.2 µg/ml of DAPI for 30 min before washing twice with 1x PBS. Unless indicated, three independent repetitions were performed for all experiments.

Western blot, chromatin immunoprecipitation, MNase digestion, histone purification and pulse field gel electrophoresis

Cell lysates to perform Western Blot were prepared using a standard protocol of TCA precipitation. Laemmli-boiled crude extracts were run on a SDS-polyacrylamide gel (8, 10 or 12% depending on experiment) and transferred to a nitrocellulose membrane (Hybond-ECL). Membranes were blocked in PBS-T milk 5% and incubated with antibodies: 9E10 Myc (sc-40, Santa Cruz Biotechnology), anti-Rad53 (mouse monoclonal gift from Marco Foiani), HA (clone 3F10, Roche), H3 (ab1791, Abcam), H2B (ab1790, Abcam), H2A (39235), H4 (ab10158, Abcam), H2A.Z (39647 Actif Motif), phospho-cdc2 (Tyr15) (9111 Cell Signalling), cdc2-p34 (sc53, Santa Cruz Biotechnology), Clb2 (SC9071, Santa Cruz Biotechnology), Actin (ab8224, Abcam) Swe1 (rabbit polyclonal, gift from Doug Kellogg) and Rap1 (V. Géli Lab). Peroxidase-conjugated goat anti-mouse or anti-rabbit IgG (both from Bio-Rad) were used to detect proteins using a ChemiDoc Gel Imaging System. Protein quantification was performed with the ImageJ software. For Chromatin Immunoprecipitation (ChIP) and MNase digestion samples were previously fixed 15 min with 1% formaldehyde. Glycine was added to quench the reaction at a final concentration of 125 mM. Cells were sedimented, washed twice with cold TBS and stored at −80°C until use. ChIP samples were obtained breaking cells with a FastPrep-24 (116004500 MP biomedicals) (3 pulses of 30 s at 5.5 intensity) in lysis buffer (HEPES 50 mM NaCl 140 mM 1 mM EDTA 1%Triton 0.1% Sodium deoxycolate 1 mM PMSF) supplemented with protease cocktail inhibitors (ROCHE). Supernatant was transferred to new tubes by centrifugation piercing the bottom of the tube with a G25 needle and chromatin was concentrated by centrifugation. Chromatin was sheared via sonication to a size between 200 and 600 bp (Diagenode) in a bioruptor (Diagenode). The supernatant was divided in equal volumes, keeping 10 µl that were used as the input DNA control, incubated with specific antibodies (4°C overnight incubation, 1 µg of antibody) and immunoprecipitated with Protein G Dynabeads (Novex). After immunoprecipitation, samples were washed once with 1 ml of the following solutions: lysis buffer, lysis buffer 0.5M NaCl, wash buffer (0.25M LiCl 10 mM Tris HCl 1 mM EDTA 0.5% NP-40 0.5% Sodium deoxycolate) and TE. Samples were then eluted from magnetic beads with a 1% SDS TE solution, incubated overnight at 65°C to reverse crosslinking, treated with 0.15 mg of Proteinase K, and extracted using a DNA purification kit (Qiagen 28004). DNA was analysed by real-time qPCR using SYBR Green Premix Ex Taq (Takara) in a Rotor Gene 6000 (Corbett Research, Labgene, Archamps, France). Primers used are listed in Supplementary file 1c. Extracts for Mnase were resuspended in 1M sorbitol, digested 1 hr with 4.5 mg of Zymoliase 20T (AmsBio 120491–1) and treated 30 min with different concentrations of Mnase (SIGMA N3755) that was inactivated with SDS (0.4%) and EDTA (8.5 mM). Samples were incubated 1 hr and 30 min at 37°C with proteinase K and overnight at 65°C for reversal crosslinking. DNA was extracted from samples using a standard phenol-chloroform extraction, treated with RNAse and loaded in a 2% agarose gel. Histones were purified as described (Jourquin and Géli, 2017). Samples for PFGE were obtained as previously described (Dueñas-Sánchez et al., 2010). PFGE was performed on a BioRad CHEF DR-III system in a 0.8% agarose gel in 1 × TAE (40 mM Tris-acetate buffer, 2 mM Na2EDTA, pH8.3) at 14°C. After electrophoresis, DNA was stained and visualised with Ethidium Bromide.

RNA extraction and Q-PCR analysis

RNA was extracted using a standard protocol with hot acid phenol. RNA samples were treated with DNase I (USB) prior to use in RT-PCR experiments. Quantitative RT-PCR experiments were performed using one-step RT-PCR, in a LightCycler 480 II (Roche) using One Step SYBR PrimeScript RT-PCR Kit (Takara Bio Inc., Japan), 0.2 µM of each primer and 50 ng of RNA in a 10 µl reaction. Primers used: H2B (HTB1) HTB1-F: AGAGAAGCAAGGCTAGAAAGGA HTB1-R: GGAAATACCAGTGTCAGGGTG; TUB1 TUB1-F: TCTTGGTGGTGGTACTGGTT TUB1-R: TGGATTTCTTACCGTATTCAGCG.

Microscopy

All microscopy analyses except those corresponding to live-microscopy were performed in liquid SC synthetic media using a Nikon Eclipse Ti microscope with a 100x objective. Cell images were captured with a Neo sCMOS Camera (Andor). Images were analysed using ImageJ on 2D-maximum projections from 11-Z-stacks spaced 0.5µ each. For live microscopy, cells were plated in SC synthetic medium on concanavalin A–coated (C2012, Sigma) Lab-Tek chambers (Thermo Fisher Scientific). Imaging was performed using a spinning-disk confocal microscope (Revolution XD; Andor Technology) with a Plan Apochromat 100×, 1.45 NA objective equipped with a dual-mode electron-modifying charge-coupled device camera (iXon 897 E; Andor Technology). Time-lapse series with variable stacks (check figure legends) were acquired every 5 min. iQ Live Cell Imaging software (Andor Technology) was used for image acquisition. Images were analysed on 2D maximum projections and denoised with the Despeckle function in ImageJ 1.46b (National Institutes of Health). Graphs and statistical analysis were performed with Prism or Excel. Fun1 staining analysis was performed with a commercial kit (L-7009 from Molecular Probes) following the standard protocol described in the kit.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
  69. 69
  70. 70
  71. 71
  72. 72
  73. 73
  74. 74
  75. 75
  76. 76
  77. 77
  78. 78
  79. 79
    Overexpression and mistargeting of centromere protein-A in human primary colorectal cancer
    1. T Tomonaga
    2. K Matsushita
    3. S Yamaguchi
    4. T Oohashi
    5. H Shimada
    6. T Ochiai
    7. K Yoda
    8. F Nomura
    (2003)
    Cancer Research 63:3511–3516.
  80. 80
  81. 81
  82. 82
  83. 83

Decision letter

  1. Jerry L Workman
    Reviewing Editor; Stowers Institute for Medical Research, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

[Editors’ note: this article was originally rejected after discussions between the reviewers, but the authors were invited to resubmit after an appeal against the decision.]

Thank you for submitting your work entitled "A histone-sensing checkpoint prevents undesired endomitosis promoted by histone overexpression" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

The reviewers were excited by the initial observations in the manuscript. However, they felt that there were too many shortcomings in the data and unexplored mechanistic issues that the current manuscript is not sufficient for publication. They also felt that these issues could not be addressed in a timely manner and did not recommend revision of the manuscript for eLife. We hope that the detailed reviews below will be useful for moving the study forward.

Reviewer #1:

Miles et al. present multiple lines of evidence that histone overexpression causes mitotic defects in yeast cells. The DNA damage and spindle assembly checkpoints are not activated by such a stress, whereas Tyr15 of Cdk1 is phosphorylated, suggesting a novel Cdk1-dependent, but DDR- and SAC-independent regulatory function of mitosis. Whole genome duplication is an important phenomenon associated with development of many species and of human cancers. This work is thus of strong interest to many. However, this manuscript suffers from a slew of deficiencies, ranging from writing to experimental design to result interpretation. It'll take a very significant effort to resolve these issues satisfactorily.

1) One of the most interesting and valuable observations from this work is probably the visualization of how diploids might be generated in the wake of histone overdose. That is, the movement of both sets of duplicated genome to the daughter cell without cytokinesis. However, the authors fall short on testing the functionality of other key mitotic machinery such as spindle, spindle pole bodies, separation of centromeres, etc. of these cells. Moreover, there is no telling whether these cells can proceed successfully to the next round of mitosis, resulting in the formation of a homogenous diploid population.

2) It is good that the authors use different approaches to induce histone stress, however, the basis of which remains unclear. The authors cite a previous finding that Swe1 contacts H2B, but do not elaborate. They also suggest that reduced H2A.Z recruitment to chromatin might be responsible. However, there is no report of a similar mitotic defect in known htz1 alleles. It is also unclear how skewing histone expression would lead to WGD. For example, what is the minimal degree of overexpression needed? Figure 1C shows that GAL-controlled expression of all four core histones in the same rad53K227A background results in the retention of 1n population, suggesting that concerted expression of histones, not merely the number of histone molecules, is critical for mitotic integrity. Furthermore, the authors favor the idea of downregulation of H2A.Z loading to pericentromeres being the underpinning of the observed mitotic defects. If so, will co-expressing H2A.Z with H2A and H2B alleviate at least some of the stress? How is it that H3/H4 overexpression also confers the growth phenotype as does H2A/H2B ectopic expression?

3) The ChIP-qPCR data are weak. What is the resolution of the qPCR? What are the reaction conditions? The provision of oligo primers in Table 3 does not help much.

4) Figure 7 MNase assays suggest that the subject chromatin is more resistant to MNase, indicating higher nucleosome occupancy. However, this single-dose experiment does not afford sufficient quantitative power. There are many advanced, more sophisticated approaches available for MNase assays. The authors should go deeper than the current endeavor.

5) The experimental procedures provided are far from complete. Key experiments such as MNase digestion, cell lysate preparation for immunochemical analysis, and ChIP-qPCR are all left out.

6) There are many writing and grammatical errors throughout the text.

Reviewer #2:

Genome stability is critical for normal physiology of an organism. Genomic instability at the chromosome level results in aneuploidy, which can cause birth defects and cancer. As cancer cells typically gain copies of chromosomes, one popular theory is that aneuploidy follows whole-genome duplication and polyploidy. Polyploidy puts more burden on the mitotic chromosome segregation machinery, increases errors in that process, and allows cells to better tolerate the resulting aneuploidy. Thus, it is important for us to understand the potential causes and mechanisms of polyploidy. Using the budding yeast as a model organism, Miles et al. showed in the current study that histone overexpression (coupled with or without mutations in other histone-surveillance genes) causes polyploidy. They further probed the underlying mechanisms and made two interesting findings. First, histone overexpression promotes Wee1-dependent negative phosphorylation on Cdk1 and delays mitotic entry. This mechanism is not, however, required for polyploidization, but may impede it. Second, histone overexpression limits the deposition of the histone variant H2A.Z and its downstream effector condensin on chromosomes, particularly near centromeres. They believe that this mechanism underlies polyploidization caused by histone overexpression.

Overall, this study advances our understanding of ploidy regulation. The findings are novel and significant. The results are for the most part convincing and logically presented. Publication is recommended, provided that the authors address the following major points.

1) While it is clear that H2A.Z and condensin deposition is defective under conditions of histone overexpression, the evidence linking this effect to polyploidization needs to be further strengthened. The only strong evidence presented is that deletion of the H2A.Z depositor Swr1 exacerbates the effect of histone overexpression. A genetic suppression experiment will be much more convincing. The authors should test whether H2A.Z overexpression or forced targeting of condensin to peri-centromeres reduces polyploidy caused by histone overexpression.

2) Much of the results on DNA damage checkpoint, spindle assembly checkpoint, and Shugoshin are negative results. These sections should be shortened. This will leave more space for the authors to address point 1 above.

Reviewer #3:

This study builds on the notion that histone protein levels are tightly regulated by multiple mechanisms and that misregulation or histone overexpression causes toxicity, cell cycle arrest, sensitivity to DNA damaging agents, and mitotic chromosome loss.

One key observation in this study is that conditions that cause increased histone protein expression lead to whole genome duplication events in yeast, i.e. cells fail to divide the nuclear material between mother and daughter cell in a timely manner; these cells stop proliferating, or the resulting diploid cells as a result of endoreduplication have a growth advantage and rapidly take over the population. This is also apparent on plates where diploids form larger colonies. Upon presumed severe increased histone levels (by disrupting the histone RNA and protein degradation pathways, which is lethal) cells arrest in G2/M and fail to exit from a G2/M arrest. In this first part, the authors investigated whether there is a sensing mechanism, or a checkpoint, for high histone levels. The failure to exit from G2/M was accompanied by a delay in cohesin degradation and increased Cdc28 phosphorylation by Swe1. The authors propose that this event delays the entry into mitosis. However, deletion of Swe1 abolishes Cdc28 phosphorylation but does not suppress the G2/M arrest. Swe1 loss does enhance WGD events.

In the second part, investigating the cause of the WGD events, the authors show that presumed very high histone levels do not affect CENP-A recruitment to centromeres, but do impair H2A.Z and condensin recruitment to pericentric heterochromatin. A condensin mutation enhances the toxicity of histone overexpression and loss of the H2A.Z assembly factor Swr1 enhances the selection for diploid cells upon transformation with a histone H2A/B overexpression vector, together suggesting that loss of these two chromatin components can further exacerbate the effects of increased histone dosage. This leads the authors to suggest that the decrease in H2A.Z and condensin is at least one of the causes that trigger WGDs when excess free histones accumulate. However, another interpretation of these results (in essence negative genetic interactions) is that histone excess and loss of H2A.Z/condensin act by independent mechanisms because a loss of H2A.Z alone was not sufficient to cause WGD events. It is likely that increased histone dosage affects WGD also by other mechanisms.

This study presents a series of interesting observations that warrant further investigation. However, the claims made by the authors are not always justified and supported by data. Importantly, whether there is a CDC28/Swe1-mediated sensing mechanism has not been fully addressed. In addition, for many of the experiments, replicates are missing and the study would benefit from a more quantitative appreciation of the altered histone levels. Finally, the story is composed of two pieces that are currently not well connected.

- The claims made by the authors are not always justified and supported by data, e.g. "we have uncovered a new type of stress (histone-stress) able to trigger a stabilization of Swe1WEE1 that promotes Cdc28CDK1 phosphorylation and delays the entry into mitosis". Similar conclusions are drawn elsewhere (e.g. Discussion, third paragraph and Abstract). That histone overexpression causes stress was already known. That it stabilizes Swe1 is not shown in this manuscript. Furthermore, that Cdc28 phosphorylation delays entry into mitosis is not so clear because deletion of Swe1, the responsible kinase, does not affect the delay.

- For many key experiments replicates or quantification of replicate experiments are missing. For most of the western blots, FACS analyses, microscopy, and MNase sensitivity assays in the manuscript no replicates or quantification of replicates are provided. The authors should provide additional data on these experiments to demonstrate reproducibility, following the guidelines of eLife.

- The authors use various conditions to increase histone expression outside S-phase. One approach is overexpressing the histone genes from strong promoters. This is in principle straightforward. The other main approach is combining a Rad53 degron with an lsm1 mutation, which causes cell death. There are several issues with the approaches.

1) In the latter case (lsm1+rad53-aid) it cannot be excluded that the phenotypes seen are at least in part caused by other functions of the inactivated proteins. Although this is partially addressed in Figure 1D, the Cdc28 phenotype has only been observed in rad53 mutant backgrounds.

2) The authors use various conditions to induce histone overexpression but show in very few cases to what extent histone proteins are overexpressed. This is an important issue because the various methods and combinations of alleles and/or plasmids have different phenotypic consequences (e.g. 2-micron-δ NEG vs. lsm1+rad53-aid). The authors should examine histone protein levels for the different conditions used, for example not only lsm1 with or without rad53-aid (Figure 3), but also for the conditions used in Figure 1 and Figure 5D. For example, based on the phenotypes, one might assume that the δ NEG or GAL-H2A-H2B plasmids cause lower histone protein expression than the combined rad53+lsm1 mutation. However, this is not demonstrated by histone blots.

3) The assay first shown in Figure 2A involves transformation with a 2-micron-HTA-HTB1-δ-NEG plasmid. The authors monitor diploidization of the pool of cells over time starting at the time of transformation. This is a somewhat unconventional assay that would benefit from more explanation of the approach and experimental details. For example, at what time point was the vector control analyzed? It would be appropriate to take a vector control along for every time point and also take a t=0 point. In this assay several factors can influence the outcome or dynamics: First, after transformation only a small proportion of the cells will contain a plasmid and survive the selection, most cells will die. Therefore, the kinetics might depend on the transformation efficiency. Second, cells that become diploid will have a growth advantage and take over the population. How fast they do so will depend on the rate of diploidization as well as the growth rate. These different parameters should be explained more clearly and taken into account experimentally. Another point here concerns panel 2D: what is the time point after transformation, how did the authors assess whether the cells examined actually contained a histone plasmid, and what are the numbers for a wild-type cell or vector control?

Other Points:

Figure 1C. How long were the cells grown in galactose to induce histone expression?

The authors state that 'Rad53 depletion was lethal in the absence of Lsm1 and largely increased the amount of histones in the cell (Figure 3A).' However, the increase in H2B seems modest.

In Figure 3G, the colony formation data should be represented as absolute numbers instead of relative

In Figure 5A, the blots are confusing. In the top panel it is hard to judge the absence of the band in the last lane, in the bottom panel it is not clear why the intensity of the two bands changes?

In Figure 5D, the cell cycle progression is hard to judge from the images shown; the phenotype of the rad53 mutant seems independent of the histone plasmid (and this may also be the case for cdc28 phosphorylation). Furthermore, in the third paragraph of the subsection “Histone accumulation triggers a Cdc28CDK1 phosphorylation that delays the entry into mitosis” the authors conclude that the phosphorylation in the upper right panel was slightly increased. However, this cannot be concluded from the blots shown because the comparison is made between two separate blots.

For Figures 78 it is not clear how many biological replicates were performed.

The authors suggest that 'rad53 AID lsm1Δ treated cells were more resistant to MNase digestion' (Figure 7B, and subsection “High levels of histones decrease Htz1H2A.Z and condensin incorporation to pericentromeric chromatin”, second paragraph). This is not immediately obvious from the figure. The authors should provide additional analyses or discussion to support this conclusion.

Regarding the mechanism of WGD by increased histone dosage, a role for H2A.Z in condensin stability has previously been shown in S. pombe. Given the differences in pericentric chromatin between the models yeasts, it would be useful to establish whether there is also a connection in S. cerevisiae.

[Editors' notes: What follows is the decision after the authors appealed the previous decision.]

Thank you for choosing to send your work entitled "A histone-sensing checkpoint prevents undesired endomitosis promoted by histone overexpression" for consideration at eLife. Your article and your letter of appeal have been considered by a Senior Editor, and we regret to inform you that we are upholding our original decision.

All three reviewers think the work is of sufficient interest, in principle, for publication in eLife. However, the reviewers pointed out many problems, and Reviewers 1 and 3 in particular thought that addressing these problems would take too much time to merit a "Revise" decision. I should say that eLife policy is that "revise" decisions should be made when the amount of time needed to address the reviewers concerns takes about two months or less. If it will significantly longer than this, the policy is to reject the paper even if the potential is there for publication. Upon reconsideration, the original opinion about "too many problems" with "not enough time" holds; hence the rejected appeal. However, if you can address all the major concerns, eLife would reconsider a new submission on this topic. Technically this would be initially considered a new manuscript, but it would endeavor to have it considered by as many of the original reviewers as possible.

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for resubmitting your work entitled "High levels of histones promote Whole-Genome-Duplications and trigger a Swe1WEE1 dependent phosphorylation of Cdc28CDK1" for further consideration at eLife. Your revised article has been favorably evaluated by Kevin Struhl (Senior editor), a Reviewing editor, and two reviewers.

The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance, as outlined below:

Genome stability is critical for normal physiology of an organism. Genomic instability at the chromosome level results in aneuploidy, which can cause birth defects and cancer. As cancer cells typically have chromosome numbers larger than the normal 46, one popular theory is that aneuploidy follows whole-genome duplication and polyploidy. Polyploidy puts more burden on the mitotic chromosome segregation machinery, increases errors in that process, and allows cells to better tolerate the resulting aneuploidy. Thus, it is important to study the potential causes and mechanisms of polyploidy.

Using the budding yeast as a model organism, Geli et al. showed in the current study that histone overexpression (coupled with or without mutations in other histone-surveillance genes) causes polyploidy. They further probed the underlying mechanisms and made two interesting findings. First, histone overexpression limits the deposition of the histone variant H2A.Z and its downstream effector condensin on chromosomes, particularly near centromeres. They believe that this mechanism underlies polyploidization caused by histone overexpression. Published results in the mouse and fly also link condensin to ploidy regulation, suggesting that the connection between condensin defects and polyploidization might be evolutionarily conserved. Second, histone overexpression stabilizes Wee1, promotes Wee1-dependent negative phosphorylation on Cdk1 and delays mitotic progression. Furthermore, histone overexpression activates the spindle checkpoint. These mechanisms are not, however, required for polyploidization, but actually impede it.

Overall, this study advances our understanding of ploidy regulation. The findings are novel and significant. The results are for the most part convincing and logically presented. Publication is recommended, provided that the authors address the following point.

Major point

1) The authors argue that histone overexpression activates the spindle checkpoint independently of the kinetochore. The major evidence supporting this claim is the lack of strong Mad2 foci formation. However, data in Figure 7C does show an increase of Mad2 foci formation. It thus remains possible that Mad2 is activated through the canonical kinetochore pathway. The authors should include positive controls in Figure 6B and 7C, such as cells treated with nocodazole. If the authors can't do additional experiments, then they should revise the text and discussion to reflect this caveat.

https://doi.org/10.7554/eLife.35337.034

Author response

Reviewer #1:

Miles et al. present multiple lines of evidence that histone overexpression causes mitotic defects in yeast cells. The DNA damage and spindle assembly checkpoints are not activated by such a stress, whereas Tyr15 of Cdk1 is phosphorylated, suggesting a novel Cdk1-dependent, but DDR- and SAC-independent regulatory function of mitosis. Whole genome duplication is an important phenomenon associated with development of many species and of human cancers. This work is thus of strong interest to many. However, this manuscript suffers from a slew of deficiencies, ranging from writing to experimental design to result interpretation. It'll take a very significant effort to resolve these issues satisfactorily.

1) One of the most interesting and valuable observations from this work is probably the visualization of how diploids might be generated in the wake of histone overdose. That is, the movement of both sets of duplicated genome to the daughter cell without cytokinesis. However, the authors fall short on testing the functionality of other key mitotic machinery such as spindle, spindle pole bodies, separation of centromeres, etc. of these cells. Moreover, there is no telling whether these cells can proceed successfully to the next round of mitosis, resulting in the formation of a homogenous diploid population.

We are more than willing to try to perform live microscopy in the strain used in Figure 2B (that contains a tag to follow kinetochores, tubulin and the SPB) in a wild type strain carrying an empty vector or the 2μΔNEG construct. This experiment is feasible and within our capabilities but probably complex to analyze. Regarding the second question, we can try longer films and check if are able to catch a second division in cells that have experienced a WGD event. However, the facts that the observation of a WGD event is something that takes place at a low frequency added to the long time required for the cell to resume mitosis in the presence of histone overexpression may make this task rather difficult.

2) It is good that the authors use different approaches to induce histone stress, however, the basis of which remains unclear.

During this study we have tried several strategies to try to increase the amount of histones in the cell. There are two main reasons that lead us to finally decide to carry out our experiments with the HTA1-HTB1ΔNEG system. The first one is that this system allows us to overexpress histones when histones are not normally required (beyond replication) but should not interfere with the normal supply of histones required for DNA replication. The second one is that this construct was the only one among those tested (including all the plasmids in which histones are overexpressed using galactose inducible promoters) in which most of the phenotypes observed in the rad53 and tom1 mutants were reproduced in wild type cells. Moreover, cells -in which histone overexpression is driven by the GAL promoter- are quite difficult to release from G1 arrest. The reason to use a conditional strain (rad53-AID lsm1Δ) able to cut off all histone degradation is explained in the text (Results section). We are aware of the fact that lsm1 and rad53 play key roles in other processes in the cell but to our knowledge there is no direct evidence of a common function of these two proteins other than the regulation of histone levels during the cell cycle.

The authors cite a previous finding that Swe1 contacts H2B, but do not elaborate.

We do. Please check the second paragraph of the Discussion section.

They also suggest that reduced H2A.Z recruitment to chromatin might be responsible. However, there is no report of a similar mitotic defect in known htz1 alleles.

Chromosome segregation defects have been associated with htz1 mutants (Keogh et al. Genes and Dev. 2006; Krogan et al. PNAs 2004) and polyploidy with H2A.Z overexpression (Chambers et al. Genes and Dev 2012). We agree that there is no direct link between the absence of H2A.Z and the generation of polyploid cells but our results do indicate that this phenomenon takes place at a low frequency and could have therefore passed unnoticed. For instance, polyploidy induced by telomere dysfunction was described just a few years ago (Davoli and de Lange, Cell 2010) despite numerous studies in the field of telomere.

It is also unclear how skewing histone expression would lead to WGD. For example, what is the minimal degree of overexpression needed? Figure 1C shows that GAL-controlled expression of all four core histones in the same rad53K227A background results in the retention of 1n population, suggesting that concerted expression of histones, not merely the number of histone molecules, is critical for mitotic integrity.

We agree, overall if we consider that Meeks-Wagner and Hartwell (Cell 1986) observed differences in the frequency of chromosome loss when histones were overexpressed in pairs (H2A-H2B or H3-H4) or simultaneously. This said, our results do allow us to conclude that both pairs of core histones expressed alone or in combination are sufficient to promote WGDs in a strain deficient for histone degradation. H2A-H2B overexpression is able to generate this phenotype even in wild type cells.

Furthermore, the authors favor the idea of downregulation of H2A.Z loading to pericentromeres being the underpinning of the observed mitotic defects. If so, will co-expressing H2A.Z with H2A and H2B alleviate at least some of the stress? How is it that H3/H4 overexpression also confers the growth phenotype as does H2A/H2B ectopic expression?

We were also initially puzzled by this result and propose a possible explanation based on our results in the Discussion section.

3) The ChIP-qPCR data are weak. What is the resolution of the qPCR? What are the reaction conditions? The provision of oligo primers in Table 3 does not help much.

We disagree. All our ChIP-qPCR experiments include a control sample (no antibody or no tag depending on the experiment) that has a signal at least one order of magnitude lower than the signal obtained for the proteins analyzed and is usually included in all figures. All comparisons mentioned in the text as significant normally include a paired t-test that is something not always present in many recently published papers. Regarding the resolution, we include an image at the end of this paragraph in which we can observe chromatin fragmentation from the first biological replicate. The DNA smear in this biological replicate is similar for all of them and seems to be more enriched in a region slightly smaller than 200 bp (last band of the ladder). We also include a first test that was performed with the rad53-AID CSE4-MYC (CENP-A) strain to determine the efficiency of the H2A.Z antibody and the MYC antibody (CENP-A). In this test, we can see that CENP-A is specifically enriched at both centromeres (CEN4 and CEN12) and that this enrichment is significantly decreased when the PCR set of primers is displaced less than 200 bp to the left or the right side of centromeres (CEN4 or 12 left and right). This test also shows that H2A.Z is highly enriched at intergenic regions and pericentromeric chromatin, two regions in which H2A.Z is normally present. Each primer used for the qPCR reactions includes a number with its exact position on the chromosome (start or end of the amplicon depending if it’s the forward or reverse primer). We can include a supplemental figure that maps the position of all set of primers.

4) Figure 7 MNase assays suggest that the subject chromatin is more resistant to MNase, indicating higher nucleosome occupancy. However, this single-dose experiment does not afford sufficient quantitative power. There are many advanced, more sophisticated approaches available for MNase assays. The authors should go deeper than the current endeavor.

We agree and we are willing to perform Mnase-Seq with the conditional strains if the paper is considered for review.

5) The experimental procedures provided are far from complete. Key experiments such as MNase digestion, cell lysate preparation for immunochemical analysis, and ChIP-qPCR are all left out.

We will include a detailed protocol for each of these techniques in the revised version if the paper is further considered.

6) There are many writing and grammatical errors throughout the text.

We sincerely apologize for this. We will pay more attention in the final version.

Reviewer #2:

Genome stability is critical for normal physiology of an organism. Genomic instability at the chromosome level results in aneuploidy, which can cause birth defects and cancer. As cancer cells typically gain copies of chromosomes, one popular theory is that aneuploidy follows whole-genome duplication and polyploidy. Polyploidy puts more burden on the mitotic chromosome segregation machinery, increases errors in that process, and allows cells to better tolerate the resulting aneuploidy. Thus, it is important for us to understand the potential causes and mechanisms of polyploidy. Using the budding yeast as a model organism, Miles et al. showed in the current study that histone overexpression (coupled with or without mutations in other histone-surveillance genes) causes polyploidy. They further probed the underlying mechanisms and made two interesting findings. First, histone overexpression promotes Wee1-dependent negative phosphorylation on Cdk1 and delays mitotic entry. This mechanism is not, however, required for polyploidization, but may impede it. Second, histone overexpression limits the deposition of the histone variant H2A.Z and its downstream effector condensin on chromosomes, particularly near centromeres. They believe that this mechanism underlies polyploidization caused by histone overexpression.

Overall, this study advances our understanding of ploidy regulation. The findings are novel and significant. The results are for the most part convincing and logically presented. Publication is recommended, provided that the authors address the following major points.

1) While it is clear that H2A.Z and condensin deposition is defective under conditions of histone overexpression, the evidence linking this effect to polyploidization needs to be further strengthened. The only strong evidence presented is that deletion of the H2A.Z depositor Swr1 exacerbates the effect of histone overexpression. A genetic suppression experiment will be much more convincing. The authors should test whether H2A.Z overexpression or forced targeting of condensin to peri-centromeres reduces polyploidy caused by histone overexpression.

We are more than willing to test if H2A.Z overexpression is able to suppress WGDs in wild type cells transformed with the 2μΔNEG vector. The second experiment may be more complex to achieve. Indeed artificial targeting of condensin to pericentromeres may be detrimental for the cells.

2) Much of the results on DNA damage checkpoint, spindle assembly checkpoint, and Shugoshin are negative results. These sections should be shortened. This will leave more space for the authors to address point 1 above.

We can easily re-adapt the paper once all the experiments are finished in order to clarify the message.

Reviewer #3:

This study builds on the notion that histone protein levels are tightly regulated by multiple mechanisms and that misregulation or histone overexpression causes toxicity, cell cycle arrest, sensitivity to DNA damaging agents, and mitotic chromosome loss.

One key observation in this study is that conditions that cause increased histone protein expression lead to whole genome duplication events in yeast, i.e. cells fail to divide the nuclear material between mother and daughter cell in a timely manner; these cells stop proliferating, or the resulting diploid cells as a result of endoreduplication have a growth advantage and rapidly take over the population. This is also apparent on plates where diploids form larger colonies. Upon presumed severe increased histone levels (by disrupting the histone RNA and protein degradation pathways, which is lethal) cells arrest in G2/M and fail to exit from a G2/M arrest. In this first part, the authors investigated whether there is a sensing mechanism, or a checkpoint, for high histone levels. The failure to exit from G2/M was accompanied by a delay in cohesin degradation and increased Cdc28 phosphorylation by Swe1. The authors propose that this event delays the entry into mitosis. However, deletion of Swe1 abolishes Cdc28 phosphorylation but does not suppress the G2/M arrest. Swe1 loss does enhance WGD events.

In the second part, investigating the cause of the WGD events, the authors show that presumed very high histone levels do not affect CENP-A recruitment to centromeres, but do impair H2A.Z and condensin recruitment to pericentric heterochromatin. A condensin mutation enhances the toxicity of histone overexpression and loss of the H2A.Z assembly factor Swr1 enhances the selection for diploid cells upon transformation with a histone H2A/B overexpression vector, together suggesting that loss of these two chromatin components can further exacerbate the effects of increased histone dosage. This leads the authors to suggest that the decrease in H2A.Z and condensin is at least one of the causes that trigger WGDs when excess free histones accumulate. However, another interpretation of these results (in essence negative genetic interactions) is that histone excess and loss of H2A.Z/condensin act by independent mechanisms because a loss of H2A.Z alone was not sufficient to cause WGD events. It is likely that increased histone dosage affects WGD also by other mechanisms.

This study presents a series of interesting observations that warrant further investigation. However, the claims made by the authors are not always justified and supported by data. Importantly, whether there is a CDC28/Swe1-mediated sensing mechanism has not been fully addressed. In addition, for many of the experiments, replicates are missing and the study would benefit from a more quantitative appreciation of the altered histone levels. Finally, the story is composed of two pieces that are currently not well connected.

- The claims made by the authors are not always justified and supported by data, e.g. "we have uncovered a new type of stress (histone-stress) able to trigger a stabilization of Swe1WEE1 that promotes Cdc28CDK1 phosphorylation and delays the entry into mitosis". Similar conclusions are drawn elsewhere (e.g. Discussion, third paragraph and Abstract). That histone overexpression causes stress was already known. That it stabilizes Swe1 is not shown in this manuscript. Furthermore, that Cdc28 phosphorylation delays entry into mitosis is not so clear because deletion of Swe1, the responsible kinase, does not affect the delay.

Of course, we are aware that we are not the first ones reporting that large accumulation of histones causes stress in cells or the first ones that demonstrate that Swe1 is a stress responsive kinase able. What we want to point out with this sentence is that we have uncovered a type of stress able to trigger a Cdc28CDK1 phosphorylation that depends on the stress-responsive kinase Swe1WEE1. We agree that Swe1 stability is not directly addressed in the paper. However, we do show that Cdc28 phosphorylation in rad53-AID lsm1Δ treated cells depends on the presence of this kinase and does not take place when Swe1 is lacking. We have recently found a specific antibody for Swe1 that we could order and use it to address this issue in Figures 5A and 5D. The conclusion of the delay is based on the results obtained in Figure 5D and not in Figure 6A (this result is discussed in the second paragraph of the Discussion section). In Figure 5D we do observe that rad53 cells increase the levels of phosphorylated Cdc28 and the time required to reach the next G1 (compare the FACS profile at 105 and 120 minutes) specifically when these cells are exposed to high levels of histones H2A and H2B.

- For many key experiments replicates or quantification of replicate experiments are missing. For most of the western blots, FACS analyses, microscopy, and MNase sensitivity assays in the manuscript no replicates or quantification of replicates are provided. The authors should provide additional data on these experiments to demonstrate reproducibility, following the guidelines of eLife.

We would like to apologize for this issue. The new revised version will include an additional source data file that contains all the raw and processed data and a pdf file with original images for all the replicates performed in all key experiments. We are also happy to share raw data from live microscopy videos if there is a server in which we can upload them.

- The authors use various conditions to increase histone expression outside S-phase. One approach is overexpressing the histone genes from strong promoters. This is in principle straightforward. The other main approach is combining a Rad53 degron with an lsm1 mutation, which causes cell death. There are several issues with the approaches.

1) In the latter case (lsm1+rad53-aid) it cannot be excluded that the phenotypes seen are at least in part caused by other functions of the inactivated proteins. Although this is partially addressed in Figure 1D, the Cdc28 phenotype has only been observed in rad53 mutant backgrounds.

Please check the answer to reviewer 1 (Point 2 major issues).

2) The authors use various conditions to induce histone overexpression but show in very few cases to what extent histone proteins are overexpressed. This is an important issue because the various methods and combinations of alleles and/or plasmids have different phenotypic consequences (e.g. 2-micron-δ NEG vs. lsm1+rad53-aid). The authors should examine histone protein levels for the different conditions used, for example not only lsm1 with or without rad53-aid (Figure 3), but also for the conditions used in Figure 1 and Figure 5D. For example, based on the phenotypes, one might assume that the δ NEG or GAL-H2A-H2B plasmids cause lower histone protein expression than the combined rad53+lsm1 mutation. However, this is not demonstrated by histone blots.

We have now examined histone levels in the 2μΔNEG vector by RT-PCR and western blot. We will include this data in the revised version.

3) The assay first shown in Figure 2A involves transformation with a 2-micron-HTA-HTB1-δ-NEG plasmid. The authors monitor diploidization of the pool of cells over time starting at the time of transformation. This is a somewhat unconventional assay that would benefit from more explanation of the approach and experimental details. For example, at what time point was the vector control analyzed? It would be appropriate to take a vector control along for every time point and also take a t=0 point. In this assay several factors can influence the outcome or dynamics: First, after transformation only a small proportion of the cells will contain a plasmid and survive the selection, most cells will die. Therefore, the kinetics might depend on the transformation efficiency. Second, cells that become diploid will have a growth advantage and take over the population. How fast they do so will depend on the rate of diploidization as well as the growth rate. These different parameters should be explained more clearly and taken into account experimentally. Another point here concerns panel 2D: what is the time point after transformation, how did the authors assess whether the cells examined actually contained a histone plasmid, and what are the numbers for a wild-type cell or vector control?

All experiments involving the ΔNEG cassette started from exponentially growing cells transformed with a centromeric (rad53K227A and tom1Δ) or a 2-μ vector (wild type) carrying the HTA1-HTB1ΔNEG construct. After transformation, strains were plated on SC-Uracil selective media at 30ºC. Time courses to determine the timing of WGDs were performed always from single small colonies coming usually from a 5-days growth plate. These cultures were grown and maintained in exponential growth conditions in liquid SC-uracil minimal media and were only considered for further analysis when they had a full haploid DNA content the first day (day 2 or 1) of the kinetic. Number of days in these experiments reflects the number of days that cells were grown in liquid media culture and not the number of days after transformation. Each experiment performed carried at least one colony of the strain tested transformed with an empty vector. This control cell was treated essentially as cells transformed with the histone-overexpressing vector. Samples were taken daily to estimate the proportion of haploids and diploids by FACS.

Other Points:

Figure 1C. How long were the cells grown in galactose to induce histone expression?

They were maintained in galactose before and after transformation.

The authors state that 'Rad53 depletion was lethal in the absence of Lsm1 and largely increased the amount of histones in the cell (Figure 3A).' However, the increase in H2B seems modest.

We agree but it is something that seems to be quite repetitive in our four biological replicates. It is important to consider that histones are very abundant proteins in the cell and that even a small increase can represent a very large amount of additional histone molecules in the cell.

In Figure 3G, the colony formation data should be represented as absolute numbers instead of relative

We can change this graph by a new one that represents all the different cell types observed for each mutant and each condition.

In Figure 5A, the blots are confusing. In the top panel it is hard to judge the absence of the band in the last lane, in the bottom panel it is not clear why the intensity of the two bands changes?

We agree and we are willing to repeat this experiment to obtain images with a better quality.

In Figure 5D, the cell cycle progression is hard to judge from the images shown; the phenotype of the rad53 mutant seems independent of the histone plasmid (and this may also be the case for cdc28 phosphorylation).

We disagree. The FACS profiles and the levels of Cdc28 phosphorylation are clearly different at time 105 and 120 in the rad53 mutant when histones are overexpressed.

Furthermore, in the third paragraph of the subsection “Histone accumulation triggers a Cdc28CDK1 phosphorylation that delays the entry into mitosis” the authors conclude that the phosphorylation in the upper right panel was slightly increased. However, this cannot be concluded from the blots shown because the comparison is made between two separate blots.

We agree and will eliminate this comment.

For Figures 78 it is not clear how many biological replicates were performed.

All experiments were at least three times (unless indicated in the figure legend) using three independent biological replicates. We can correct this in the figure legend.

The authors suggest that 'rad53 AID lsm1Δ treated cells were more resistant to MNase digestion' (Figure 7B, and subsection “High levels of histones decrease Htz1H2A.Z and condensin incorporation to pericentromeric chromatin”, second paragraph). This is not immediately obvious from the figure. The authors should provide additional analyses or discussion to support this conclusion.

We agree and we are willing to perform Mnase-Seq with the conditional strains if the paper is considered for review.

Regarding the mechanism of WGD by increased histone dosage, a role for H2A.Z in condensin stability has previously been shown in S. pombe. Given the differences in pericentric chromatin between the models yeasts, it would be useful to establish whether there is also a connection in S. cerevisiae.

H2A.Z and condensin recruitment to pericentromeric chromatin is conserved from yeasts (S.cerevisiae and S.pombe) to higher eukaryotes (mouse and humans). Both of them play an important role during chromosome segregation and lead to chromosome segregation defects when they are absent. There is no direct evidence to our knowledge of a physical interaction between H2A.Z and condensin in S. cerevisiae.

[Editors' notes: What follows is the response to the previous round of reviews included in the authors' resubmission]

We are sending you a second revised version of our manuscript now entitled “High levels of histones promote Whole-Genome-Duplications and trigger a Swe1WEE1 dependent phosphorylation of Cdc28CDK1”. The previous version was entitled “A histone-sensing response prevents undesired endomitosis promoted by histone overexpression”. We have changed the title according to a comment raised by one of the reviewers.

In our previous submission, you explained us that all three reviewers thought that the work was of sufficient interest for eLife and that they would be willing to consider a new/revised submission that addresses all of the issues raised by the reviewers. You mentioned “Technically this would be initially considered a new manuscript, but we would endeavor to have it considered by as many of the original reviewers as possible”.

Reviewer #1:

The authors are praised for their endeavors to probe from different angles for the very interesting histone overdose-related cell division phenotypes in the budding yeast. The discoveries of the potential involvement of H2A.Z and condensin are very promising.

However, there are insufficiencies need to be addressed.

a) One of the fundamental issues that I have for this manuscript is the use of "endomitosis" for the observed phenotype. There are well-documented and -defined examples, normal or pathological, for endomitosis and endoreplication. The authors' initial observation might be better described as an escape from failed segregation, so that a small number of daughter cells inherit the full set of duplicated genomes, at the cost of losing the mother cell. Oddly, the "diploidization" stops at a 2n DNA content, and does not progress into 4n, 8n, or even higher ploidy that is the case for typical endocycles.

The term endomitosis was used to describe the phenomenon observed in Figure 2E in which the nucleus divides inside the daughter cell without cytokinesis. To avoid confusion, we have replaced the term “aberrant endomitosis” by “aberrant cell division”. We have detected 4n cells in rare occasions while we were analyzing the DNA content of rad53K227A and tom1∆ cells transformed with the CEN∆NEG vector.

We have never observed them in wild type cells transformed with the 2µ∆NEG vector. We think that this issue could be related to the fact that a 2N DNA content in wild type cells is sufficient to buffer the toxic effect generated by high levels of histones in haploid cells.

b) Another concern is that overexpressing histones H2A and H2B is sufficient to trigger the 2n phenotype in a wildtype background. The use of the rad53-AID lsm1 null strain does not add significant values to the study but might further complicate the matter (e.g., rad53 loss-of-function alleles can cause problems in cell cycle progression). Therefore, mechanistic in sights from the use of these strains may not be directly applicable to WGD caused by histone overexpression.

Rad53 and Lsm1 have no other function in common than the regulation of histone levels to our knowledge. However, it is certainly important to consider the possibility that some of the phenotypes observed in the double mutant may be indirect and not related to histone metabolism. To prove if the defect in H2A.Z incorporation observed in the double mutant was relevant for WGDs we have measured H2A.Z levels, and checked the effects of H2A.Z overexpression and SWR1 deletion in cells transformed with the 2µ∆NEG (new Figure 5). All the results are consistent with the hypothesis that H2A/H2A.Z balance is affected by histone overexpression and critical in the generation of WGDs.

To prove that Cdc28 phosphorylation is also directly related to histone levels, we have followed Cdc28 phosphorylation during a whole cell cycle in cells that were overexpressing histones H2A and H2B in a wild type and in rad53K227A cells in which histones are not efficiently degraded after histone overexpression (Gunjan et al., Cell 2003). Our results in the rad53K227A indicate that Cdc28 phosphorylation is maintained for a longer time in cells in which histones are being overexpressed.

c) Technical issues need to be addressed as well. Firstly, the characterization of histone dosages and chromatin distribution is less than ideal. The authors cherry-picked the methods and histones for analysis. Why not just purify all core histones and resolve them by SDS-PAGE for CBR staining?

As asked by reviewer 1, we have purified histones (Jourquin and Géli, 2017; Histones Methods and Protocols) in rad53-AID and rad53-AID lsm1∆ mutants treated or not with Auxin and replaced the results from the previous version (Western Blot and Chromatin Fractionation). Histones were resolved by SDS-PAGE and stained with CBR. Histone amounts were normalized to a non-specific band present in the purification. We observe a relative increase of histones compared to a non-specific band in the rad53-AID lsm1∆ double mutant treated with Auxin either compared to the non treated sample or to the rad53-AID single mutant treated or not (New Figure 3A).

d) The authors argued that H2A.Z is the key player for WGDs, and yet the characterization of H2A, the counterpart of H2A.Z, is lacking (except Figure 8C). Did they see the expected inverse correlation between H2A and H2A.Z?

As asked by reviewer 1, in the revised version of the manuscript we have performed ChIPqPCR with an H2A antibody (new Figure 5A). We observe that the level of H2A increases at those regions exhibiting a decrease of H2A.Z (new Figure 5A).

e) Secondly, Figure 7A shows that the centromeric Cse4 level does not change, yet histone H3 detection at centromeres shoots up by a factor of about 10, and H4 level is up at the similar level at centromeres. What is the explanation? Why does H3 not evict Cse4?

S. cerevisiae has just one centromeric nucleosome able to incorporate either one (Dalal et al., Plos Biology 2007) or two Cse4 molecule/s (Mizuguchi et al., Cell 2007; Wisniewski et al., eLife 2014). Our Cse4 ChIP indicates that the amount of Cse4 at centromeres does not change when histones are overexpressed (Figure 4A) and therefore disfavors a model in which H3 competes with Cse4 incorporation. The fact that Mtw1, a subunit of the MIND complex required to build a functional kinetochore, is recruited with a similar efficiency and always remains in line with tubulin filaments (Figure 4B,C and D), further suggests the formation of a functional kinetochore that is able to attach chromosomes to microtubules. We agree however with reviewer one that the large increase in H3 and H4 at centromeres and other regions of the genome is intriguing. We discussed thoroughly this point and propose two different explanations.

The first one is that histone overexpression favors the incorporation of additional nucleosomes at both sides of the centromeric nucleosome. The chromatin that surrounds the centromere is far from perfectly positioned (Cole et al., PNAS 2011) and may allow the incorporation of additional nucleosomes that will increase the signal at centromeres. This hypothesis is simple but has to face the fact that there is a space constraint. Indeed, only a limited amount of nucleosomes can be incorporated in 200600 bp fragments, the average size of DNA fragments that we obtain after sonication in our ChIP experiments. Therefore, the large increase in histone ChIP levels that we observe would be difficult to explain with the addition of 1 or 2 nucleosomes.

The second explanation that could explain this increase would be a change in chromatin compaction. Nucleosomes can physically interact with neighbor nucleosomes and promote chromatin condensation (Wilkins et al., Science 2014). Changes in chromatin condensation allow the interaction of nucleosomes with other nucleosomes that are not necessarily assembled at the same region. These nucleosomes should, in principle, retain these interactions after formaldehyde crosslinking, and should be able to increase the ChIP signal even if these nucleosomes are not directly interacting with centromeric chromatin.

f) Thirdly, overexpressing H2A.Z reduces the percentage of 2n* cells. Is this genuine suppression or that cells suffer from lower viability, hence the percentage of viable 2n* colonies decreases? Are the overall chromatin structure and condensin localization improved?

As demanded by reviewer 1, in the revised version we show the growth of haploids and 2n* cells transformed with the vector allowing the overexpression of histone H2A.Z or an empty control vector (Figure 5—figure supplement 1A). We observe no obvious differences in growth between n and 2n* cells overexpressing H2A.Z or carrying the empty vector control. In cells transformed with the 2µ∆NEG vector, H2A.Z overexpression enhances the growth of both haploids and 2n* cells. We do not observe a counter-selection of diploids.

We have also performed ChIP experiments to see if H2A.Z overexpression is able to suppress the defect in the incorporation of H2A.Z observed in rad53-AID lsm1∆ treated cells. We observe that H2A.Z overexpression is not sufficient to restore the levels of H2A.Z at pericentromeric chromatin in this double mutant. The rad53-AID lsm1∆ double mutant must have other defects that are able to maintain lower levels of H2A.Z incorporation to chromatin. We propose a possible explanation in the second paragraph of the Discussion.

g) Lastly, with all the supplemental figures, why did authors choose to not show the mating characterization results of the 2n* cells?

We did not include it because we considered it something anecdotic and not really important for the overall message of the paper. As asked by reviewer 1, we have included it in the new version (Figure 1—figure supplement 1E).

Reviewer #2:

This manuscript appears to be the re-submission of a previously submitted manuscript. I have attached my original review along with comments on the newly added data.

Genome stability is critical for normal physiology of an organism. Genomic instability at the chromosome level results in aneuploidy, which can cause birth defects and cancer. As cancer cells typically gain copies of chromosomes, one popular theory is that aneuploidy follows whole-genome duplication and polyploidy. Polyploidy puts more burden on the mitotic chromosome segregation machinery, increases errors in that process, and allows cells to better tolerate the resulting aneuploidy. Thus, it is important for us to understand the potential causes and mechanisms of polyploidy. Using the budding yeast as a model organism, Miles et al. showed in the current study that histone overexpression (coupled with or without mutations in other histone-surveillance genes) causes polyploidy. They further probed the underlying mechanisms and made two interesting findings. First, histone overexpression promotes Wee1-dependent negative phosphorylation on Cdk1 and delays mitotic entry. This mechanism is not, however, required for polyploidization, but may impede it. Second, histone overexpression limits the deposition of the histone variant H2A.Z and its downstream effector condensin on chromosomes, particularly near centromeres. They believe that this mechanism underlies polyploidization caused by histone overexpression.

Overall, this study advances our understanding of ploidy regulation. The findings are novel and significant. The results are for the most part convincing and logically presented. During the review of the initial draft of this paper, I raised the point that the connection between polyploidization and H2A.Z/condensin was weak, and asked for genetic suppression data. The authors have now provided convincing evidence to show that H2A.Z overexpression reduces the extent of polyploidization caused by canonical histone overexpression. This new piece of evidence strengthens their argument for a link between H2A.Z and polyploidization. In a second development, I recently heard at a meeting that condensin II mutations cause tissue-specific polyploidization in the mouse. The finding was actually published near the end of 2016 (Woodward et al. (2016) Genes Dev, 30:2173-2186), but apparently escaped my notice (and was not referenced in this paper). In the Woodward et al. study, Andrew Wood and coworkers showed that mutation of condensin II caused lymphoma and polyploidizition of tumorinitiating cells in the mouse. These findings, along with results in the fly (which the authors discuss), I am now convinced that there is an evolutionarily conserved connection between condensin defects and polyploidization. For these reasons, I recommend the publication of this excellent study.

We really appreciate the comments from reviewer 2. We have included this reference in the new version.

Reviewer #3:

This study presents several interesting findings. First, increased histone dosage in budding yeast leads to genome segregation errors and whole-genome duplication. Second, increased histone dosage triggers a series of events that can be described as a sensing mechanism. However, the function of this sensing mechanism in relation to WGDs remains elusive. The sensing mechanism seems to be independent of known checkpoint mechanisms. Finally, increased histone dosage leads to reduced H2A.Z levels in the cell and reduced condensin loading in pericentric regions; these events may contribute to the genome instability caused by histone stress. Although many questions remain to be answered, together, the results of this study will be of interest to a broad audience.

In this revised and improved version of the previously submitted manuscript the authors now show histone expression levels for some of the experiments and show Swe1 expression data to confirm its stabilization. In addition, more information is provided about the replicates and experimental conditions. Finally, the manuscript has been rewritten and generally, the claims made by the authors have a better match with the results than in the previous version. However, several concerns remain.

In the summary, the statement 'uncovers a mechanism able to sense histone levels before mitosis in order to avoid their undesired consequences' is not fully supported by the data. The function of the sensing mechanism still remains unclear.

In the Discussion, it is mentioned that 'Swe1 delays mitosis in the presence of high levels of histones'. The authors do not provide evidence to support this claim.

We have rewritten the sections related to the role of checkpoints to make a new version that is more descriptive and less conclusive. The statement that Swe1 is able to sense histone levels before mitosis in order to avoid their undesired consequences has been tuned down in the Summary, Results and Discussion. Possible consequences of Cdc28 phosphorylation on the cell cycle are mentioned in the discussion.

In Figure 5B, the authors show Swe1 blots to support the notion that Swe1 is stabilized upon histone stress. However, a loading control is missing. Cdc28 blots are shown but this protein is related to Swe1 function and it changes due to phosphorylation. Panel A also misses a loading control.

We have included Act1 as a loading control in the three figures, which are now shown in the new Figure 6 (E-G) (previous Figures 5A-C). See also our response to referee 1 (point 6). Cdc28 levels remain constant throughout the cell cycle in S. cerevisiae (Mendenhall and Hodge, 1998).

In Figure 8C, the authors show on blots that H2A.Z and H2.A levels change in opposite directions upon histone stress. However, why is the level of H2A.Z already low in rad53-AID lsm1 without treatment? Similarly, why are H4 levels lower in rad53-Aid treated vs untreated?

We believe that they represent loading differences rather than real differences in the levels of H2A.Z. In the new version of the manuscript, western blots are performed with histonepurified extracts. In the new blots performed with these samples (New Figure 3B) this difference is no longer observed.

The authors now explain more clearly in the text on page 7 how the assay for Figure 2A (and subsequent figures) was done: Cells that remained completely haploid were inoculated from plates into liquid media and then monitored over time. Here it would be appropriate to show that at t=0, the cells indeed started off as haploids. The vector control does not provide this information. And what is time point used for the vector control, why is only one time point shown for the vector alone? What was the time point used for Figure 2B?

In this study, FACS profiles are always made from exponentially growing cells in liquid cultures and not from cells directly collected from petri dishes. This decision was based on the fact that plated cells after 4 or 5 days tend to accumulate in G1/G0. As a result, diploid populations are difficult to differentiate from haploid cells at a G2/M stage. Obtaining a sample in exponential growth conditions usually takes two days in cells transformed with the 2µ∆NEG vector due to their slow growth. Day 2 therefore constitutes the first time-point for which we can harvest enough cells to perform a FACS experiment. At day 2, a large proportion of wild type cells transformed with the 2µ∆NEG vector remain haploid. However, in some cases, some of the transformed colonies have already started to change to a mixed population of haploids and diploids.

The vector control is used to prove that the strain is completely haploid before transformation with the 2µ∆NEG vector and also to prove that this strain does not go to a 2n state in the absence of the 2µ∆NEG vector. We have always included at least one in each kinetic and taken samples the first and the last day. The one shown in Figure 2A corresponds to the one taken the last day. We have never observed a diploidisation in any genetic context in the absence of histone overexpression. That is why we only represent one sample and not two. We believe that is important to point out that S. cerevisiae cells are very stable unicellular organisms that do not usually experience changes in their ploidy content after subcultivation.

In Figure 1—figure supplement 1E, the authors show doubling time data for haploids transformed with the multi-copy deltaNEG plasmid. However, in Figure 6—figure supplement 1D, the authors show that these cells rapidly diploidize. Please explain.

There is a certain variation in the time taken by each transformant to become fully diploid, something that can be appreciated in the kinetics shown in the new Figures 5F and 8A. Duplication times of transformants were estimated from a twelve-hour time course performed during day 2. FACS samples were taken at the beginning and at the end of the time course to check the ploidy. Samples that were already diploid or had started to diploidize during the kinetic were not considered for the estimation of the duplication time.

For many figures, the authors show the average of two experiments and a standard deviation. Although two replicates may sometimes suffice, including more replicates gives more confidence in the results. For small sample size, showing the two data points or the spread is more informative than standard deviation.

In this new version we have corrected the representation of our results according to the suggestion of reviewer 3. We have also extended the number of repeats of two of our key experiments, the condensin ChIP experiment, that now shows the average of five biological replicates and the suppression of WGDs by H2A.Z overexpression, that is done with 4 independent biological replicates. Most of the other results for which only two repeats are shown, are measured in more than one experiment (Mtw1 in the new Figure 4B and C; Pds1 in the new Figure 6D and H; Scc1 in the new Figure 6D and old Figure 4A; Cdc28 phosphorylation in the new Figures 6E-H), or demonstrated using different approaches (accumulation of histones in the rad53-AID lsm1∆ treated cells was shown by western blot and chromatin fractionation in the old version and is confirmed with histone purification in the new version and its incorporation to chromatin was shown with ChIP and Mnase).

For Figure 5—figure supplement 1C, why was the plating done on Gal media, did these strains express an inducible copy of histone genes?

The AID system includes an adaptor protein required for the degradation that is under the control of a galactose inducible promoter. This information is included in the Supplemental Experimental Procedures.

Figure 6—figure supplement 1C: it is not clear what the error bars refer to and what statistical test was performed.

It was included in the additional source data file but we forgot to include it in the figure legend. We have corrected this and included the statistical test in the new version.

In Figure 8E HTZ1 overexpression is used. Please provide the details of the HTZ1 vector and the vector control.

The details have been included in the new version.

In Figure 5A, the blots are confusing. In the top panel it is hard to judge the absence of the band in the last lane, in the bottom panel it is not clear why the intensity of the two bands changes?

We have performed new blots that include Act1 as a loading control.

[Editors’ note: the author responses to the re-review follow.]

Major point

1) The authors argue that histone overexpression activates the spindle checkpoint independently of the kinetochore. The major evidence supporting this claim is the lack of strong Mad2 foci formation. However, data in Figure 7C does show an increase of Mad2 foci formation. It thus remains possible that Mad2 is activated through the canonical kinetochore pathway. The authors should include positive controls in Figure 6B and 7C, such as cells treated with nocodazole. If the authors can't do additional experiments, then they should revise the text and discussion to reflect this caveat.

We have revised the text accordingly. We acknowledge the increase of Mad2 foci

formation shown in Figure 7C:

“The fact that Mad2 foci were slightly increased (Figure 7C) suggest that the SAC could be activated at least in some cells when histones are overexpressed.”

“The Pds1 stabilisation observed in rad53-AID lsm1Δ treated cells added to the fact that MAD2 deletion enhances WGDs indicate that the SAC could be activated in cells that overexpress histones.”

https://doi.org/10.7554/eLife.35337.035

Article and author information

Author details

  1. Douglas Maya Miles

    Marseille Cancer Research Center (CRCM), U1068 Inserm, UMR7258 CNRS, Aix-Marseille Université, Institut Paoli-Calmettes, Equipe Labellisée Ligue, Marseille, France
    Contribution
    Conceptualization, Investigation, Supervision, Validation, Writing—original draft
    Contributed equally with
    Vincent Geli
    For correspondence
    douglas.maya@cabimer.es
    Competing interests
    No competing interests declared
    ORCID icon 0000-0002-0669-6526
  2. Xenia Peñate

    Instituto de Biomedicina de Sevilla, Hospital Virgen del Rocío-CSIC-Universidad de Sevilla, Sevilla, Spain
    Contribution
    Investigation, Validation
    Competing interests
    No competing interests declared
    ORCID icon 0000-0002-4117-888X
  3. Trinidad Sanmartín Olmo

    1. Centre for Genomic Regulation, Barcelona Institute of Science and Technology, Barcelona, Spain
    2. Universitat Pompeu Fabra, Barcelona, Spain
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared
  4. Frederic Jourquin

    Marseille Cancer Research Center (CRCM), U1068 Inserm, UMR7258 CNRS, Aix-Marseille Université, Institut Paoli-Calmettes, Equipe Labellisée Ligue, Marseille, France
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  5. Maria Cruz Muñoz Centeno

    Instituto de Biomedicina de Sevilla, Hospital Virgen del Rocío-CSIC-Universidad de Sevilla, Sevilla, Spain
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  6. Manuel Mendoza

    1. Centre for Genomic Regulation, Barcelona Institute of Science and Technology, Barcelona, Spain
    2. Universitat Pompeu Fabra, Barcelona, Spain
    Contribution
    Investigation, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
  7. Marie-Noelle Simon

    Marseille Cancer Research Center (CRCM), U1068 Inserm, UMR7258 CNRS, Aix-Marseille Université, Institut Paoli-Calmettes, Equipe Labellisée Ligue, Marseille, France
    Contribution
    Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  8. Sebastian Chavez

    Instituto de Biomedicina de Sevilla, Hospital Virgen del Rocío-CSIC-Universidad de Sevilla, Sevilla, Spain
    Contribution
    Conceptualization, Writing—review and editing
    Competing interests
    No competing interests declared
  9. Vincent Geli

    Marseille Cancer Research Center (CRCM), U1068 Inserm, UMR7258 CNRS, Aix-Marseille Université, Institut Paoli-Calmettes, Equipe Labellisée Ligue, Marseille, France
    Contribution
    Conceptualization, Investigation, Writing—review and editing
    Contributed equally with
    Douglas Maya Miles
    For correspondence
    vincent.geli@inserm.fr
    Competing interests
    No competing interests declared
    ORCID icon 0000-0002-4103-7462

Funding

Ligue Contre le Cancer (Equipe Labéllisée 2017)

  • Vincent Geli

Fondation ARC pour la Recherche sur le Cancer (Aide Posdtdoctorale)

  • Douglas Maya Miles

Ministerio de Economía y Competitividad (BFU2012-37162)

  • Manuel Mendoza

European Research Council (2010-St-20091118)

  • Manuel Mendoza

Regional Andalusian Government (P12-BIO1938MO)

  • Sebastian Chavez

Ministerio de Economía y Competitividad (BFU2013-48643-C3-1-P)

  • Sebastian Chavez

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank A Gunjan, J Tyler, S Biggins, F Prado, Doug Kellogg, and S Mahajan for sharing strains, plasmids and/or reagents; We would also like to thank C Machu for imaging expertise and JH Guervilly, Felix Prado, E Bailly and V Geli team members for discussions. DM was supported by a postdoctoral fellowship from the Association pour la Recherche sur le Cancer (Fondation ARC). Work in VG laboratory is supported by the”Ligue contre le Cancer’ (Equipe Labéllisée 2017). Work in the laboratory of M M is supported of the European Research Council (ERC) Starting Grant 2010-St-20091118, and the Spanish Ministry of Economy and Competitiveness BFU2012-37162 to MM, and ‘Centro de Excelencia Severo Ochoa 2013–2017’, SEV-2012–0208 to the CRG. S C is supported by grants BFU2013-48643-C3-1-P from the Spanish MiNECO, and P12-BIO1938MO from the Regional Andalusian Government, both incluidng European Union funds (FEDER).

Reviewing Editor

  1. Jerry L Workman, Reviewing Editor, Stowers Institute for Medical Research, United States

Publication history

  1. Received: January 24, 2018
  2. Accepted: March 5, 2018
  3. Version of Record published: March 27, 2018 (version 1)

Copyright

© 2018, Maya Miles et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 792
    Page views
  • 96
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Chromosomes and Gene Expression
    2. Developmental Biology and Stem Cells
    Jan J Zylicz et al.
    Research Advance Updated
    1. Cell Biology
    2. Chromosomes and Gene Expression
    Madhav Jagannathan et al.
    Research Article Updated