Axial contraction and short-range compaction of chromatin synergistically promote mitotic chromosome condensation

The segregation of eukaryotic chromosomes during mitosis requires their extensive folding into units of manageable size for the mitotic spindle. Here, we report on how phosphorylation at serine 10 of histone H3 (H3 S10) contributes to this process. Using a fluorescence-based assay to study local compaction of the chromatin fiber in living yeast cells, we show that chromosome condensation entails two temporally and mechanistically distinct processes. Initially, nucleosome-nucleosome interaction triggered by H3 S10 phosphorylation and deacetylation of histone H4 promote short-range compaction of chromatin during early anaphase. Independently, condensin mediates the axial contraction of chromosome arms, a process peaking later in anaphase. Whereas defects in chromatin compaction have no observable effect on axial contraction and condensin inactivation does not affect short-range chromatin compaction, inactivation of both pathways causes synergistic defects in chromosome segregation and cell viability. Furthermore, both pathways rely at least partially on the deacetylase Hst2, suggesting that this protein helps coordinating chromatin compaction and axial contraction to properly shape mitotic chromosomes. DOI: http://dx.doi.org/10.7554/eLife.10396.001


Introduction
The DNA molecule at the core of any eukaryotic chromosome is a hundred to million times longer than the average diameter of the cell that hosts it. Thus, cells need to fold their genetic material in order to fit it in the interphase nucleus; they need to pack it further during mitosis, in order to move sister-chromatids safely and symmetrically apart. Furthermore, chromatin folding must be dynamic to allow transcription and replication during interphase, and such that exceptionally large chromosomes can hyper-condense during anaphase in order to fit the size of the spindle and prevent chromosome missegregation (Neurohr et al., 2011;Titos et al., 2014). Moreover, mitotic condensation also facilitates the decatenation of sister chromatids during their separation (Charbin et al., 2014), and might help to 'cleanse' chromosomes from transcription, replication and cohesion factors (Yanagida, 2009). This is thought to 'reset' the transcriptional state of genes, and prevent displaced factors from interfering with chromosome segregation. However, despite their importance for chromosome segregation, the events ensuring the mitotic condensation of chromosomes are still only partially understood.
Early studies made evident that nucleosomes play a critical role in DNA packaging. In favor of the idea that they play specific roles in chromatin condensation, histone H3 is phosphorylated by aurora B throughout mitosis on a serine at position 10 in most, if not all, eukaryotes. Furthermore, aurora B inactivation leads to chromosome condensation and segregation defects in budding yeast (Lavoie et al., 2004), fission yeast (Petrova et al., 2013;Tada et al., 2011), HeLa cells (Tada et al., 2011) and roundworms (Hagstrom et al., 2002). However, the precise role of H3 S10 phosphorylation has remained unclear (Ajiro and Nishimoto, 1985). Recent data demonstrated that H3 S10 phosphorylation promotes the recruitment of the sirtuin-related deacetylase Hst2, which in turn deacetylates, at least, lysine 16 of histone H4 (Wilkins et al., 2014). This unmasks a basic patch, allowing H4 to interact with the acidic patch on H2A, most probably on an adjacent nucleosome (Robinson et al., 2008;Gordon et al., 2005). Thus, this cascade of events initiated by H3 phosphorylation is thought to tighten the interaction between neighboring nucleosomes. However, all studies carried out so far have failed to reveal strong phenotypes for H3 serine 10 to alanine mutations in a plethora of model organisms (de la Barre et al., 2001;Afonso et al., 2014;Ditchfield et al., 2003). Furthermore, mutation of this residue in budding yeast did not affect axial contraction of chromosomes and the condensation of the rDNA during regular mitoses (Neurohr et al., 2011;Lavoie et al., 2004;Lavoie et al., 2002). Indeed, the only phenotype identified upon replacement of H3 S10 with alanine in yeast so far is limited to the reduced ability to hyper-condense artificially long chromosomes in order to fit them in the spindle (Neurohr et al., 2011). Thus, it remains unclear whether H3 phosphorylation and H4 deacetylation play any general role in mitotic chromosome condensation.
The discovery that mitotic extracts of frog eggs lacking any one of the subunits of a protein complex called condensin largely failed to condense chromosomes (Hirano and Mitchison, 1994) opened new perspectives for understanding chromosome condensation (Piazza et al., 2013;Thadani et al., 2012). Condensin is a ring-shaped pentameric protein complex. The core of the ring is formed by two structural maintenance of chromosome (SMC) subunits, Smc2 and Smc4. Three non-SMC proteins (Brn1, Ycg1 and Ycs4 in budding yeast) close the ring. The mechanism of condensin loading on chromatin is not understood, but seems to depend on the activity of the kinase aurora eLife digest DNA in humans, yeast and other eukaryotic organisms is packaged in structures called chromosomes. When a cell divides these chromosomes are copied and then the matching pairs are separated so that each daughter cell has a full set of its genome. To enable these events to take place, the DNA must become more tightly packed so that the chromosomes become rigid units with projections called arms. Any failure in this chromosome "condensation" leads to the loss of chromosomes during cell division.
Within a chromosome, sections of DNA are wrapped around groups of proteins to make a series of linked units called nucleosomes, which resemble beads on a string. These units and other scaffold proteins together make a structure called chromatin and establish the overall shape of the chromosome. However, it is not exactly clear how the nucleosomes and scaffold proteins are rearranged during condensation.
Kruitwagen et al. used microscopy to study chromosome condensation in budding yeast. The experiments reveal that condensation involves two separate processes. First, modifications to the nucleosomes result in these units becoming more tightly packed in a process called short-range compaction. Second, a group of proteins called condensin is responsible for rearranging the compacted chromatin to enforce higher-order structure on the arms of the condensed chromosome (long-range contraction). Further experiments suggest that an enzyme called Hst2 may help to coordinate these processes to ensure that chromosomes adopt the right shape before the cell divides. For example, Hst2 ensures that longer chromosomes condense more than shorter ones.
A future challenge will be to find out whether chromosome condensation works in a similar way in humans and other large eukaryotes, which form much larger chromosomes with more complicated structures than yeast.
B (Lavoie et al., 2004;Tada et al., 2011). Furthermore, the non-SMC subunits were recently shown to directly bind DNA (Piazza et al., 2014), potentially followed by topological entrapment of chromatin inside the condensin ring . How condensin performs its functions in chromosome condensation is unclear, but it has been proposed that condensin's role might be structural, by inducing loops within the same DNA strand  or might be enzymatic by promoting positive DNA supercoiling (Baxter and Aragó n, 2012), both assisting in a decrease in length of mitotic chromatids.
Mitigating the central role of condensin in chromosome condensation, however, were observations in model organisms as diverse as fission yeast, fly, chicken and mammalian cells that indicate that chromosomes can still, at least partially, condense in the absence of condensin (Petrova et al., 2013;Coelho et al., 2003;Vagnarelli et al., 2006;Gerlich et al., 2006). Thus, although condensin was established as a key player in chromosome condensation, it cannot be the sole factor shaping mitotic chromosomes.
In order to gather insights into whether and how H2A-H4 interaction contributes to the organization of mitotic chromosomes, we sought for a method to assay the condensation state of chromatin in vivo. Here, we use a fluorescence-based assay to investigate short-range chromatin compaction and use it to study the relationships between condensin and histone modifications during chromosome condensation in mitotic cells.

Results
A microscopy-based assay to measure chromatin fiber compaction In order to develop a chromatin condensation assay, we reasoned that increased nucleosome-nucleosome interaction might render chromatin less accessible to DNA-binding proteins. To test this idea directly, we asked whether chromatin condensation restricted access for heterologous reporter proteins to their binding sites when those are introduced at a chosen chromosomal locus. Therefore, we used a yeast strain in which a set of Tet operator (TetO) repeats are inserted at the TRP1 locus on chromosome IV, 15 kb from CEN4, and constitutively expressing the TetR-mCherry fusion protein, which efficiently binds the TetO repeat. As a consequence, these cells exhibit a red dot in their nucleus throughout the cell cycle ( Figure 1A). To test whether the intensity of TetR-mCherry fluorescence possibly varied over the cell cycle, we measured the fluorescence intensity of this dot in G1 cells (unbudded), when the chromatid is decondensed, but not replicated yet, and in late anaphase mother cells, when the chromatid is separated from its sister and has reached full condensation (Figures 1B and 3C;(Neurohr et al., 2011;Sullivan et al., 2004;D'Amours et al., 2004) and see below). After subtracting background fluorescence, we noticed a highly significant (p<0.0001), 2-2.5-fold decrease in mCherry fluorescence intensity at the TetO repeats on the anaphase compared with the G1-phase chromosomes ( Figure 1A, B).
We next asked whether the variations of fluorescence intensity at the TetO repeats reflected changes in H2A/H4 interaction. Supporting this view, mutating key residues in the H3 phosphorylation and H4 deacetylation pathway established by (Wilkins et al., 2014) affected these variations ( Figure 1B,C). Strikingly, mutations that abrogate the mitotic interaction between H2A and H4, such as H3 S10A, hst2D and the H4 D9-16 mutations, all abolished the reduction in brightness of the TetR-mCherry focus normally observed in anaphase cells ( Figure 1B). In reverse, mutations that promote constitutive H2A/H4 interaction, such as H3 S10D and H4 K16R, caused the TetR-mCherry focus to constitutively show, that is, even in G1 cells, the low fluorescence intensity normally specific of anaphase cells. The effect of the H3 S10D mutation was indeed mediated by the recruitment of Hst2, since the hst2D mutation suppressed it; the H3 S10D hst2D double mutant cells showed constitutive high brightness, similar to hst2D single mutant cells. Interestingly, however, introducing the H4 K16R mutation in the hst2D mutant cells did not restore the intensity drop normally observed during anaphase, suggesting that H4 K16 is not the sole residue that Hst2 deacetylates to promote nucleosome-nucleosome interaction ( Figure 1B).
To test whether the observed fluctuations in fluorescence intensity were specific for the TRP1 locus or TetO/TetR-mCherry, we also measured the fluorescence intensity at the LYS4 locus, in the middle of the right arm of chromosome IV, where we integrated LacO repeats in cells expressing LacI fused to Green Fluorescent Protein (GFP). Although the effect was slightly less pronounced, we observed a similar, and significant, decrease in reporter brightness in anaphase compared with G1 cells at this locus ( Figure 1C). As for the TRP1 locus, mutants in the H3 S10 pathway also affected fluctuations in intensity at the CEN distal locus: hst2D, H3 S10A and H3S10D hst2D showed a continuously higher fluorescent intensity on the LacO repeats near the LYS4 gene and the H3 S10D mutation also resulted in a continuously lower fluorescent signal. These data indicated that changes in chromatin organization during mitosis indeed affected either the recruitment or the fluorescence intensity of TetR-mCherry and LacI-GFP on two distant chromatin loci, one close to the centromere and the second in the middle of the second longest yeast chromosome arm.
Since chromatin condensation is regulated by the kinase aurora B (Ipl1 in budding yeast), we last asked whether Ipl1 activity is required for the intensity decrease of TetR-mCherry at the TRP1 locus in mitotic cells. We arrested wild type yeast cells and cells containing the temperature sensitive ipl1-321 allele in G1 with alpha-factor, released them at the restrictive temperature of 35ºC and determined TetO/TetR-mCherry fluorescence intensity in the same G1 or following anaphase ( Figure 1D). Whereas wild type cells showed no significant difference in G1 and anaphase TetO/TetR-mCherry fluorescence intensity, the ipl1-321 strain showed a significantly brighter dot when undergoing Representative images of a cell in G1 and anaphase, containing a TetO array at the TRP1 locus and expressing TetR-mCherry (red). Fluorescence intensity of a focus is measured by determining the total fluorescence and subtracting the background, giving the corrected fluorescence intensity. Scale bar is 2 mm. (B) TetR-mCherry intensities for the indicated wild type (WT) and mutant strains in G1 and anaphase mother cells. One way Analysis Of Variance ( ANOVA) was performed to test significance. (C) Fluorescence intensity for a wild type strain containing LYS4:LacO and expressing LacI-GFP. Student's t-test was performed to determine significance. (D) Anaphase TetR-mCherry intensities for the indicated strains, synchronized in G1 by alpha-factor treatment and released at the indicated temperatures. Intensities for G1 were determined 5 min after release from alpha-factor induced arrest. All data are means and standard deviation for n>30 cells. **** p<0.0001 and n.s. not significant. DOI: 10.7554/eLife.10396.003 anaphase at the restrictive temperature. Compaction in G1 of ipl1-321 cells at the restrictive temperature was not affected, presumably due to the fact that this protein has no activity in G1, even in wild type cells (Buvelot et al., 2003). Thus, we conclude that the enhanced H2A-H4 interaction triggered by aurora B-dependent recruitment of the deacetylase Hst2 onto chromatin indeed affects the intensity of the TetR-mCherry signal on the chromosome.

Fluorophore concentration quenching causes fluctuations in brightness
We next wanted to better understand the molecular processes and structural changes of chromatin that were underlying the fluorescence variation at the TetO array over the cell cycle. Assuming enhanced nucleosome-nucleosome interaction promotes chromatin compaction, three models may explain the observed decrease of fluorescence in mitosis. First, chromatin compaction might reduce access of DNA-binding proteins, such as TetR-mCherry, to their binding site on DNA and cause their removal, as postulated by the chromosome cleansing hypothesis. Second, chromatin compaction might increase the local packing of TetR-mCherry, leading to quenching of the fluorophore (Lakowicz, 2013); these two first models are depicted in Figure 2A. Third, the changed local environment of mitotic chromatin might reduce the intrinsic fluorescence of mCherry and GFP.
In order to better distinguish between these models, we rationalized that coexpressing TetR-GFP with TetR-mCherry would not protect mCherry from a cleansing effect (model 1) or a change in local environment (model 3), but should strongly reduce any quenching, due to intercalation of a second fluorophore with a different excitation spectrum. Furthermore, in this context, quenching might be replaced by Fö rster Resonance Energy Transfer (FRET) between the TetR-GFP and TetR-mCherry molecules. Remarkably, unlike the cells expressing only TetR-mCherry, cells expressing both versions of TetR failed to show significant variation of the fluorescence signal for either mCherry or GFP at the TetO array between anaphase and G1 ( Figure 2B). Thus, cleansing and a general change in the local environment of the fluorophores are unlikely to explain the fluorescence drop observed at the TetO array during anaphase in the cells expressing solely TetR-mCherry. Supporting the idea that the intensity drop was due to a quenching effect, FRET was indeed observed upon exciting in the GFP and recording emission in the red channel in cells expressing both TetR-mCherry and TetR-GFP, but not in cells expressing TetR-mCherry alone ( Figure 2C). Moreover, FRET was significantly increased during anaphase compared with G1-phase ( Figure 2B,C), indicating that the fluorophores are indeed brought in closer proximity during anaphase compared with interphase. We conclude that increased H2A/H4 interaction results in a tighter packing of fluorophores and their quenching, establishing that H2A/H4 interaction leads to compaction of mitotic chromatin in vivo. Furthermore, cell cycle dependent changes of TetR-mCherry or TetR-GFP signals on TetO arrays is a reliable measure of short-range compaction of the underlying chromatin.
Chromatin compaction precedes the axial shortening of chromosomes Next, we investigated the dynamics of chromatin compaction during the cell cycle. To this end, we visualized both the TetO/TetR-mCherry (at TRP1) and LacO/LacI-GFP (at LYS4) loci simultaneously.This presence of two labeled loci on the same chromosome allowed measuring the physical distanceseparating them and hence the long-range contraction of the chromosome arm along its longitudinal axis during anaphase. Using this strain, we first recorded time-lapse movies ( Figure 3A) in which we measured the intensity of the LacI-GFP fluorescence at the LYS4 locus in cells progressing through mitosis ( Figure 3B). Upon averaging the signal of at least 15 (t = -18 minutes) to maximum 31 (t = 0 minutes) such traces, we observed that the intensity of the signal was indeed lowest during the first 12 minutes of anaphase, while starting to increase as soon as the cells started to exit mitosis ( Figure 3B, blue line indicates the formation of the first bud in the population). We also noticed that fluorescence intensity at the LacO locus was highly variable throughout every single movie, leading to high standard deviations. This variation was lowest during anaphase and started to increase as soon as chromatin was decondensing, consistent with the idea that chromatin is more constrained when it is most compacted and fluorophore quenching is highest. The source of this fluorescence variation is not known, but might reflect breathing movements of the underlying chromatin or complex photochemistry effects. In either case, this intrinsic cell-to-cell variability precludes drawing conclusions at the single cell level and emphasizes the fact that the quenching assay introduced here is statistical in nature.
Next, we wanted to determine whether the dynamics of chromatin compaction could be related to the contraction of the chromosome arm measured using the TRP1-LYS4 distance, as described previously by us and others (Neurohr et al., 2011;Petrova et al., 2013;Guacci et al., 1994;Vas et al., 2007). To avoid variations due to photobleaching, we used snapshot images of cells at precise and representative time points in mitosis ( Figure 3C): metaphase (large buds but neither the TRP1 nor the LYS4 loci were separated), early anaphase (sister TRP1 loci -in red -are separated, but the two LYS4 loci are not), mid-anaphase (both loci have undergone separation but the LYS4 locus still lags behind), late anaphase (all loci are separated and moved to the opposite poles of the cell) and G1 (unbudded cell). For each of these stages (>25 cells each), we measured both the intensity of the fluorescence on the two arrays and the distance between them. In this study, we focused specifically on the loci segregated to the mother cell, as we showed before that mother and bud are not directly comparable (Neurohr et al., 2011).
Analysis of this data set indicated that the two marked loci underwent compaction and decompaction with slightly different kinetics ( Figure 3C, upper panel). During early anaphase, both the TetO and LacO array seemed to be compacted to some extent already. As cells progressed to mid anaphase, the CEN4 proximal TetO arrays seemed slightly more compacted than the distal LacO arrays. In late anaphase, the TetO array was already starting to unpack, whereas the LacO remained compacted. Both dots had recovered their full intensity in the G1 cells, demonstrating that the intensity decrease in anaphase was not solely the consequence of sister-chromatid separation, which is expected to reduce fluorescence intensity by half, down to its G1 level until the next Sphase. In the same cells, measuring the TetO-LacO distance ( Figure 3C, lower panel) indicated that the chromosome first stretched out upon anaphase onset, to subsequently contract, reaching their shortest length in late anaphase, as reported (Guacci et al., 1994;Harrison et al., 2009). The TetO-LacO distance re-extended then to its steady state average in G1 cells.
The changes in distance between the two loci could be due to changes in nuclear diameter, which would constrain the maximal distance that the loci can move apart by random motion. However, we did not observe any significant changes in nuclear diameter (as determined by GFP-tagging the nucleoporin Nup170) when comparing late anaphase and the G1 phase in wild type and hst2D cells ( Figure 3D). Thus, the shortening of the TRP1-LYS4 distance in late anaphase cells truly reflects the effect of chromatin condensation by axial contraction of the chromosomes. Furthermore, our results established that short-range chromatin compaction was not strictly concomitant with longrange axial contraction of the chromosome, but rather preceded it.

Condensin is not involved in local chromatin compaction
These observations suggested that chromatin compaction and axial contraction of mitotic chromosomes might be distinct processes. Thus, we asked whether condensin, which is essential for axial chromosome contraction, contributed to short-range compaction of chromatin. We analyzed the brightness of the TetR-mCherry focus in yeast cells carrying the smc2-8 allele, a temperature sensitive mutation in the condensin subunit Smc2 ( Figure 4A). Remarkably, condensin inactivation for 90 min at the restrictive temperature had no effect on the changes in mCherry brightness between the anaphase and G1-phase of the cell cycle, whereas it indeed abrogated shortening of the TetO-LacO distance during anaphase (see below). We therefore conclude that condensin, unlike histone 3 phosphorylation and histone 4 deacetylation, does not promote nucleosome-nucleosome interaction. To test this idea further, we directly probed H2A/H4 interaction by using genetically encoded Ultraviolet (UV) inducible crosslinking (Wilkins et al., 2014). We arrested cells in G1 with alpha-factor and released them in the presence of nocadozole under wild type and smc2-8 conditions at 37ºC. Fluorescence-Activated Cell Sorting (FACS) analysis showed that the release and arrest was equally efficient in both cells (Figure 4-figure supplement 1). In wild type cells, as reported before, H4/H2A crosslinking is observed in mitosis and correlated strongly with H4 K16 deacetylation ( Figure 4B). In fitting with the microscopy data ( Figure 4A), the crosslinking between H4 and H2A showed no difference in kinetics in the condensin inactivated and in the wild type cells ( Figure 4B). Thus, condensin function is not required for proper, short-range compaction of mitotic chromatin.

The chromatin compaction pathway does not contribute to axial contraction of chromosomes
In order to investigate in more detail how the phosphorylation of H3 S10 and the subsequent activation of Hst2 contributed to chromosome condensation, we next characterized if these events promoted axial contraction of chromosome IV, using the LacO-TetO distance as a readout (see Figure 3D). Confirming the role of aurora B/Ipl1 in chromosome condensation, the ipl1-321 mutant cells failed to undergo chromosome contraction when shifted to the restrictive temperature prior to mitosis, compared with wild type cells at these temperatures ( Figure 5A). As expected from previous studies (Neurohr et al., 2011;Lavoie et al., 2002), the function of Ipl1 in the contraction of regular chromosomes was unlikely to require H3 S10 phosphorylation and H2/H4 interaction, since the mutations H3 S10A and H4 D9-16 did not impair anaphase contraction ( Figure 5B,C ). Thus, Ipl1 promotes the axial contraction of chromosomes independently of phosphorylating H3 S10 and of promoting H4/H2A interaction, but possibly by promoting condensin function (see discussion).
In contrast, the hst2D mutation did abrogate the proper contraction of the chromosome during mitosis, implying that Hst2 acts in both axial chromosome contraction and short-range chromatin compaction ( Figures 5C and 1B). Even more remarkably, the H3 S10D phospho-mimicking allele caused chromosome IV to remain in a constitutive state of axial contraction throughout the cell cycle ( Figure 5B,C). This most probably reflected constitutive recruitment of Hst2 to nucleosomes, since inactivation of the HST2 gene in these cells abolished contraction ( Figure 5B,C). Thus, boosting the recruitment of Hst2 to chromatin through mimicking constitutive and ubiquitous phosphorylation of H3 promoted both chromatin compaction and axial chromosome contraction, despite the fact that the H3-dependent pathway of Hst2 recruitment is normally dispensable for Hst2 function in axial chromosome contraction during regular mitoses (though, it is essential for adaptive hyper-condensation of an artificially long chromosome). We next sought to determine the role of H4 K16 deacetylation in anaphase chromosome contraction. In accordance with the idea that H4 deacetylation solely plays a role in mitotic chromatin compaction, the H4 K16R allele did not show any significant differences in chromosome contraction compared with wild type cells ( Figure 5D). When we deleted HST2 on top of H4 K16R, we observed a significant decrease in chromosome contraction as compared with wild type and H4 K16R cells, but no significant difference with hst2D cells ( Figure 5D). Based on this we conclude that H4 deacetylation has no evident role in chromosome contraction. Furthermore, our data suggest that HST2, when deleted, does not have its defects on chromosome contraction through the chromatin compaction pathway, but rather by acting on another factor.
Together, in agreement with the fact that they follow different kinetics (Figure 3), our data reveal that axial contraction of chromosomes and compaction of chromatin are two independent processes. They depend on distinct molecular pathways that are at least in part coordinated by the same regulatory input, namely the activity of the kinase aurora B and the deacetylase Hst2. To inactivate Smc2, cells were shifted to 37ºC for 90 min. One way ANOVA was performed to test significance, **** p<0.0001 and n>40. (B) Yeast cells producing H2A Y58BPA were synchronized with alpha-factor at permissive temperature and then released into medium containing nocodazole at restrictive temperature. Samples were taken at indicated times, irradiated with UV and histones extracted with acid from isolated nuclei. Western blot against H4 detects the H2A-H4 crosslink (upper row), bulk H4 (lower row) and blotting against H4 K16Ac shows cell cycle progression. DOI: 10.7554/eLife.10396.006 The following figure supplement is available for figure 4:

Hst2 may promote condensin function
The data above demonstrated that hst2D mutant cells have strong defects in axial chromosome contraction, while other components of the chromatin compaction pathway (H3 S10 and H4) do not. Thus, we rationalized that Hst2 might not only promote chromosome compaction, but also the function of factors required for the axial contraction of chromatids, such as condensin. To Figure 5. Chromatin compaction does not influence axial chromosome contraction. (A) TRP1-LYS4 distances for the indicated strains, synchronized in G1 by alpha-factor treatment and released at the indicated temperatures. (B) TRP1-LYS4 distances were determined in the mother cell for the indicated strains and cell cycle stages. Box shows median value, whiskers all data points n>30 cells. One way ANOVA was performed to test significances, ** p<0.01, **** p<0.0001, n.s. not significant. (C) Example cells containing the indicated mutations and their impact on chromosome length as determined by the TRP1 (red) to LYS4 (green) distance. (D) TRP1-LYS4 distances were determined for the indicated strains in anaphase. Box shows median value, whiskers all data points n>45. One way ANOVA was performed to test significances, *** p<0.001, * p<0.05, n.s. not significant. DOI: 10.7554/eLife.10396.008 test this notion, we measured the effect of combining the smc2-8 mutation with H3 and hst2D mutations on axial chromosome contraction during anaphase. As expected, the smc2-8 single mutant cells grown at the restrictive temperature failed to axially contract their mitotic chromosomes to wild type levels ( Figure 6A). Deletion of Hst2 in these cells did not exacerbate their contraction phenotype ( Figure 6A), suggesting that condensin and Hst2 might act in the same genetic pathway. In contrast, contraction was restored in the H3 S10D smc2-8 double mutant cells. This effect was likely due to boosting Hst2 recruitment and activity since it disappeared in the H3 S10D hst2D smc2-8 triple mutant cells. The effect of H3 most likely depended on the phosphorylation state of S10, since the S10A mutation did not promote contraction. Thus, enhanced activation of Hst2 (H3 S10D) suppressed the effect of the smc2-8 mutation on chromatin contraction. Hst2 might mediate the suppression of the contraction defect due to the smc2-8 mutation either through activation of a still unknown, alternative contraction pathway or through stimulation of condensin activity such as to restore at least part of its function. Axial contraction and chromatin compaction synergistically facilitate chromosome segregation In order to establish the physiological relevance of our phenotypic observations, we next asked whether combining compaction and contraction defects cell growth and viability. Consistent with condensin and H3 S10 phosphorylation pathways acting synergistically to promote proper chromosome condensation, the smc2-8 H3-S10A double mutant cells were growing slower than the smc2-8 single mutant cells, and were unable to grow at 33˚C, unlike the single mutant cells ( Figure 6B and Figure 6-figure supplement 1). Accordingly, the smc2-8 hst2D double mutants showed the same phenotype. Thus, when the condensin complex was only partially active, cells depended on Hst2 and phosphorylation of histone H3 on S10 for growth. Remarkably, however, the smc2-8 H3 S10D double mutant did not show any altered or improved growth or viability phenotype compared with the smc2-8 single mutant cells. It grew reasonably well at 33˚C, while being non-viable at 37˚C. Thus, restoring axial contraction of the chromosomes using the H3 S10D mutation was not sufficient to restore the viability of the cells lacking condensin function. These results are consistent with condensin playing more roles than simply promoting the axial contraction of chromosomes.
To further test the requirement of condensin and H3 S10 pathways on chromosome segregation, we imaged anaphase cells stained with 4', 6-diamidino-2-phenylindole (DAPI) to visualize the frequency with which the single and double mutant strains produced lagging chromatin. The cells were grown at 25˚C, the permissive temperature for the smc2-8 mutation or shifted for 90 min to 33˚C, a semi-permissive temperature for smc2-8 mutant cells. At 25˚C, the smc2-8 single and the smc2-8 H3 S10A, smc2-8 hst2D and smc2-8 H3 S10D double mutants cells showed no strong differences (6.3-8.0%) in the frequency at which lagging chromatin was observed in the center of the spindle ( Figure 6C). In contrast, the smc2-8 H3 S10A and the smc2-8 hst2D double mutants grown at 33˚C showed a marked increase in the frequency (18.9% and 16.9%, respectively) of lagging chromatin compared with the smc2-8 single mutant cells. The smc2-8 H3 S10D mutant cells did not show such an additive phenotype (10.0%; Figure 6C). These data indicate that the pathways promoting the short-range compaction of chromatin and the axial contraction of chromosomes contribute synergistically to shaping mitotic chromosomes in order to ensure their correct segregation prior to cell division.

Discussion
Mitotic chromosome condensation is the process through which a relatively relaxed interphase chromosome condenses into two relatively short, compact sister chromatids that the spindle can symmetrically segregate. The kinase aurora B has long been implicated in this process (Lavoie et al., 2004;Tada et al., 2011;Lavoie et al., 2002). Through activation of condensin and phosphorylation of S10 on histone H3 it is thought to promote the formation of a mitotic chromosome. However, how these pathways interacted with each other to shape mitotic chromosomes was poorly understood.
Here, we monitored chromosome condensation in living yeast cells by using two microscopy assays, allowing us to distinguish long-range contraction of chromosomes along their longitudinal axis from short-range chromatin compaction. The first assay has been used before (Petrova et al., 2013;Guacci et al., 1994;Vas et al., 2007), but did not detect a role for histone phosphorylation during regular mitoses (Neurohr et al., 2011). The second assay, introduced here, uses fluorophore quenching to show that nucleosome-nucleosome interaction via H3 phosphorylation and in an Hst2dependent manner promotes short-range compaction of mitotic chromatin in vivo. This enhanced packing is evidenced by the increased quenching of fluorophores when the chromosome locus is decorated by multiple copies of a fluorescent reporter. The quenching effect was observed similarly well whether we used GFP or mCherry as a fluorophore, and independently of whether it was fused to TetR or to LacI. Though we cannot exclude any local effects of the repetitive nature of the TetO and LacO repeats on chromatin compaction, this novel fluorescence-based assay allows for a simple detection of chromatin compaction states. Beyond promoting chromosome segregation, the exact function of this compaction process will be interesting to address. Our studies establish that chromatin compaction does not displace DNA-associated proteins such as TetR or LacI, as a strong cleansing model would predict. Although they do not exclude that other DNA-binding factors might be removed from the DNA during this process, these data suggest that compaction might serve other functions, such as to change the biophysical properties of the chromatin fiber during segregation. Chromatin compaction might for example affect the local stiffness, elasticity and mechanical resistance of chromosomes to facilitate their decatenation, protection and movement during anaphase.
Related to this, it is remarkable that the H3 S10D mutation is able to establish an Hst2-dependent and constitutive state of compaction and contraction, as evidenced by both assays, without affecting much growth and viability. We propose that this mutation strongly boosts the recruitment and activation of Hst2 all along chromosomes by mimicking the phosphorylated state of histone H3. Two models may explain how H3 S10D elicits its effects. In the first, H3 S10D might lead to hyper-activation of chromatin compaction by over-recruitment of Hst2, such that it might shorten the chromosome axis, independently of the contraction machinery. However, the following observations speak against this idea: I. crosslinking studies do not indicate that H3 S10D increases nucleosome-nucleosome interaction (Wilkins et al., 2014), II. fluorescence quenching of TetR-mCherry and LacI-GFP is not enhanced in the H3 S10D mutant ( Figure 1B,C) compared with WT anaphase cells, and III. phosphorylation of H3 S10 does promote the condensin-dependent and Hst2-dependent contraction of artificially long chromosomes (Neurohr et al., 2011;Wilkins et al., 2014). Thus, these data support an alternative model: Hst2 may stimulate condensin function, and hyper-activation of Hst2 by H3 S10D may promote this effect and thereby chromosome arm contraction. To fully distinguish between these two models it will be important to investigate the effect of the H3 S10D mutation on condensin loading and activity. Furthermore, it is important to note that the phenotype of the H3 S10D mutant cells does not seem to reflect physiological conditions taking place during regular mitoses, indicating that the fraction of Hst2 driving wild type chromosome contraction does not depend on H3 S10 phosphorylation (dashed arrow Figure 7, left panel). Rather, it might reflect what happens when both chromosome condensation pathways are hyper-activated, for example under the presence of an artificially long chromosome or in small cells, such as cells grown on a poor carbon source (see below). In any case, it is striking that the H3 S10D strain does not show any obvious defects in growth and viability despite causing chromatin compaction and contraction throughout the cell cycle. Thus, this constitutively condensed state does not impair access and remodeling of chromatin by the transcription and replication machineries during interphase in any major manner.
Our kinetic data suggest that chromatin undergoes a number of successive structural changes during anaphase and that compaction precedes contraction. In this respect, two models could explain the extended TRP1-LYS4 distance observed in early anaphase cells. First, this might simply be due to the fact that the anaphase nucleus is more elongated, allowing for the chromosome arm to reach its real extension without being confined in the normally spherical morphology of late anaphase or G1 nuclei. In this scenario, the apparent stretching of the chromosome in early anaphase does not correspond to any rearrangement of the chromatin or changes in contraction, and the impression that contraction follows compaction is an illusion. However, previous work has established that residual cohesion between sister chromatids persists during early anaphase and that this actually causes chromosome stretching during early segregation stages (Renshaw et al., 2010;Figure 7. Model of chromosome condensation. H3 S10 phosphorylation leads to chromatin fiber closure, whereas condensin is required for axial shortening. Hst2 is needed for both of these levels of condensation. DOI: 10.7554/eLife.10396.011 Harrison et al., 2009). Thus, a second model would be that maximum contraction is only reached when cohesion is fully resolved, such as to not interfere with cohesion removal. This scenario suggests that chromosome contraction indeed needs to be a late anaphase event to complete chromosome segregation, and that perhaps the compaction events that precede play an important role in facilitating cohesion resolution. This model is attractive because it might more thoroughly account for the molecular events that are actually taking place during anaphase. Understanding how these processes are regulated during early anaphase will be necessary in order to distinguish between the two models. However, one definite conclusion that we can already draw is that arm contraction and chromatin compaction undergo relaxation between late anaphase and G1, since these nuclei are not significantly different in size, and hence the effects observed are not due to confinement alone.
Based on our data on chromosome contraction and compaction in Figure 3C, we propose that mitotic chromosome condensation entails at least three processes (Figure 7).
First, a histone 3-dependent and histone 4-dependent process, which we term chromatin compaction, ensures the short-range tightening of DNA into a smaller volume via nucleosome-nucleosome interaction (Wilkins et al., 2014). This process strictly depends on the phosphorylation of histone H3 on serine 10, and the subsequent deacetylation of lysine 16 of histone H4. Because this event has little impact on the length of condensed chromosomes, it has escaped attention until now. However, our data indicate that it is an important event for proper chromosome architecture and is an early event of every yeast mitosis.
Second, a condensin-dependent process, termed here axial chromosome contraction, ensures the long-range contraction of the chromosome, probably by facilitating the formation of higher order chromatin structures along the chromosome axis Baxter and Aragó n, 2012). This process is completed after chromatin compaction, in late anaphase, but does not strictly require it in order to proceed. Certainly, it will be interesting to clarify this order of event, to dissect how it is controlled, and to determine whether it facilitates the ordered condensation of wellseparated chromosome arms. Moreover, the observation that Hst2 contributes to both processes suggests that this enzyme might be pivotal for coordinating compaction and contraction with each other. In this respect, it will be interesting to determine how Hst2 promotes the shortening of the chromosome axis, and whether it does so by directly modulating condensin function.
Third, our data also indicate the existence of an additional, also condensin-dependent process devoted to some other aspect of chromosome organization, beyond compaction and contraction. Indeed, restoring the contraction of the chromosome along its axis in the smc2-8 condensin mutant cells by mimicking constitutive H3 phosphorylation did not alleviate the temperature-dependent lethality caused by the smc2-8 mutation. Furthermore, although inactivation of the deacetylase Hst2 largely abolished the axial contraction of chromosomes, HST2 is not an essential gene. Hence, we suggest that the essential function of condensin is not to mediate long-range contraction. Rather, the essential function of condensin most likely relates to its roles in orchestrating the decatenation of intertwined sister-chromatids (Charbin et al., 2014;Baxter and Aragó n, 2012;Baxter et al., 2011) and the re-annealing of single stranded DNAs (Sakai et al., 2003).
In an earlier study we demonstrated that yeast cells challenged by the presence of an exceptionally long chromosome or by being themselves exceptionally small (for example, whi3 mutant cells and cells growing on a poor carbon source) possess the ability to condense chromosomes beyond wild type levels to adjust the axial contraction of long chromosomes and adapt their length to the length of the spindle, a process we termed adaptive hyper-condensation (Neurohr et al., 2011), and which we would like to rename adaptive hyper-contraction. This process depended on condensin and phosphorylation of H3 S10, this last requirement being in contrast to what is observed for the contraction of most wild type chromosomes during regular mitoses. Accordingly, single smc2-8 and H3 S10A mutants failed to hyper-contract an exceptionally long chromosome (Neurohr et al., 2011). Since the H3 S10A mutation had no defect on the axial contraction of wild type chromosomes, these results indicated that adaptive hyper-contraction depended more heavily on Ipl1dependent phosphorylation of S10 on histone H3 than regular chromosome condensation does. Thus, we propose that H3 S10 is hyper-phosphorylated on long chromosomes by Ipl1 located in the center of the spindle (midzone), boosting Hst2 and condensin activity (dotted arrow, Figure 7). Our observation that mimicking S10 phosphorylation, using the H3 S10D allele, promotes axial shortening in an Hst2-dependent manner is indeed consistent with high levels of S10 phosphorylation promoting the crosstalk between compaction and contraction, via Hst2.
It will be interesting to test how the principles laid out here function in eukaryotes with a more complex chromosome structure, such as humans. A good starting point for such investigations would be to test the effects of H3 S10 mutation and/or removal of the Hst2 homologues on such cells. How important is a tight regulation between short-range chromatin compaction and long-range chromosome contraction? How does that impact chromosome structure? And how do such mutations affect the segregation of the genetic material? Together, our work demonstrates that yeast is a powerful system to dissect the mechanisms underlying chromosome condensation and how the disruption of such mechanisms affects chromosome segregation and cell viability.

Yeast strains
All yeast strains used in this study were derived from the S288c background and are described in Table 1. Histone mutants were amplified from the synthetic non-essential histone collection from Thermo Scientific and transformed in S288c-derived strains. HST2 was deleted by using standard methods (Janke et al., 2004). smc2-8 temperature sensitive strains were obtained by crossing and tetrad dissection. Spotting assay in Figure 5B was done by aligning cells at OD600 = 0.1 and diluting 1:5 in subsequent steps. 2.5 ml drops were plated on yeast extract peptone dextrose (YPD) medium and grown for 2.5 days.

Fluorescent microscopy
All strains were grown at 30ºC on YPD medium. Condensin temperature-sensitive strains were grown in liquid YPD until OD 600 = 0.4 and then transferred to 25ºC or 37ºC for 90 min prior to imaging. For imaging, cells were resuspended in non-fluorescent medium (Synthetic Defined (SD) minus tryptophan (-TRP)) and put on an agar pad. All microscopy was done using a Deltavision microscope (Applied Precision) with a CCD HQ2 camera (Roper), 250W Xenon lamps, Softworx software (Applied Precision), and a temperature chamber set to the desired temperature. For both chromosome contraction and chromatin compaction assays, still images were taken using 750 ms exposure in the Tetramethylrhodamine (TRITC (red)) channel and 500 ms exposure time for the Fluorescein (FITC (green)) over ten Z-slices that were 0.5 mm apart. Transmission images were taken as a reference. Analysis was done using Fiji image processing software. We defined G1 cells as round cells with no bud and anaphase cells as a budded cell where the red focus leads and the green focus followed. All measurements were done in mother cells. Sum intensity projections were used and a line was drawn between the green and red foci to determine chromosome contraction. For fluorescence intensity, a Region Of Interest (ROI) was drawn around a focus and the integrated density was determined. An identically sized ROI was put in the nucleus to determine the background signal. Background intensity was subtracted from the focus intensity to yield the fluorescent intensity for a given focus. For the example cell in Figure 3A, 1-hr long movies were taken using slightly modified conditions: 300 ms FITC exposure, 500ms TRITC exposure and eight Z-stacks that were 0.5mm apart. FRET was done using the same microscope described above. First the FRET channel was recorded using 500 ms exposure in the FITC channel while capturing emission in the TRITC channel. Then, excitation and emission in TRITC was recorded using 500 ms of exposure time. Analysis was done by drawing a ROI around the fluorescent focus in both channels, subtracting the background and determining the ratio between them. At least 50 cells were measured for all conditions. For Ipl1 inactivation, exponentially growing cells were arrested with alpha-factor according to standard protocols and released at 25˚C or 35˚C. To obtain G1 condensation, cells were released for 5 min, to obtain anaphase condensation cells were released for approximately 2 hr at the indicated temperatures before imaging.
Nuclear diameters were determined by imaging Nup170-GFP as described above and by using Fiji image processing software.
DAPI staining was performed by growing cells of a given strain to OD 600 < 1.0 in YPD at 25˚C, then two new diluted cultures were made at OD 600 = 0.2. One of them was again put at 25˚C, while the other was kept at 33˚C for 90 min. Cells were harvested by 90 s centrifugation in a 1.5 ml Eppendorf tube at 650 RCF. Cells were resuspended in 1 ml 70% ethanol, left to fix for 7 min, spun down again and washed in 1 ml sterile water, before being resuspended in 3-5 ml