A conditional mutation in Arabidopsis thaliana separase induces chromosome non-disjunction, aberrant morphogenesis and cyclin B1;1 stability

Plants build organs throughout their lives. Continual organogenesis endows the plant with plasticity in terms of ultimate size and fecundity, and allows the plant to acclimate to an altered environment by producing new organs with appropriate modifications. Plants develop, post-embryonically, from groups of pluripotent cells called meristems. A meristem generates cells that will acquire mature functions while simultaneously maintaining a population of proliferating cells. Conceptually, meristems allow the plant to keep making organs in much the same way as stem cells allow an animal organ to replace its cells.

Understanding the control of cell division is thus essential to an understanding of plant development. Generally, among eukaryotes, major features of the cell cycle and its regulation are strikingly conserved. However, when comparing a given protein among taxa, one often finds differences, sometimes fundamental ones.

The work reported here concerns the protein separase, which is a protease that is activated at the onset of anaphase to cleave the cohesin complex, which holds sister chromatids together (Onn et al., 2008). Separase targets the α-kleisin subunit of the cohesin complex. When separase activity is absent, chromosomes invariably fail to disjoin, causing chromosome breakage and multinucleate cells. Non-disjunction following separase inhibition has been reported in fungi [fission and budding yeasts and Aspergillus nidulans (Nasmyth et al., 2000)], Caenorhabditis elegans at both mitosis and meiosis (Siomos et al., 2001), Drosophila melanogaster (Jäger et al., 2001), various vertebrates (Papi et al., 2005; Kumada et al., 2006; Wirth et al., 2006; Shepard et al., 2007) and plant meiocytes (Liu and Makaroff, 2006). These results support the notion that separase activation at the onset of anaphase to disjoin chromatids is a widely conserved, if not ubiquitous, feature of the eukaryotic cell cycle.

Not surprisingly for a process as fundamental as the separation of chromosomes, separase is subject to a web of regulatory interactions. Some of these regulatory controls differ among organisms. Also, separase itself has been reported to have activities in addition to cleaving cohesin (Queralt and Uhlmann, 2005). For example, in budding yeast, separase plays a role in stabilizing microtubule dynamics of the anaphase spindle and in driving exit from M phase (Higuchi and Uhlmann, 2005; Baskerville et al., 2008). In C. elegans, separase is required for the secretion of the embryonic eggshell (Bembenek et al., 2007), and in D. melanogaster embryos separase deficiency has been associated with pseudostratification of the epithelium (Pandey et al., 2005). Finally, in human tissue culture cells, separase has been reported to be required for proper entry into M phase and spindle assembly (Giménez-Abián et al., 2005; Papi et al., 2005; Sak et al., 2008). Therefore, although all eukaryotic species examined use separase to cleave cohesin at anaphase onset, species appear to differ in how separase activity is regulated and to what extent the protein has acquired other functions.

Separase is an essential gene; therefore, studies of separase loss-of-function have, for the most part, been limited to tissue culture or early embryos. We know little about how separase acts in the cells of a mature organism or during development. This can be productively explored in plants because of their substantial, ongoing cell division. The root is particularly useful because it is accessible to observation, is organized linearly and grows indeterminately.

Here, we report that radially swollen 4 (rsw4), a temperature-sensitive mutant of A. thaliana isolated on the basis of aberrant root morphology (Wiedemeier et al., 2002), contains a mutation in the single-copy gene for separase. Consistently, rsw4 plants have defective disjunction in mitotic root cells; unexpectedly, they also have defective cortical microtubule organization and accumulate cyclin B1;1.

Plant material was Arabidopsis thaliana L. (Heynh), Columbia background. Plant growth was as described (Bannigan et al., 2006). The rsw4 line was described previously (Wiedemeier et al., 2002). The GFP-tubulin reporter line expresses GFP fused to the A. thaliana β-tubulin 6 (TUB6) gene (Bannigan et al., 2006). The GFP-histone lines express GFP fused to the A. thaliana histone H3.3 gene (At5g10980) and were made by Keith Earley and Craig Pikaard (Washington University of St Louis). For each reporter, two transgenic lines, representing independent transformants with the same construct, were crossed onto rsw4 and, from the F2, plants that were homozygous for rsw4 and brightly fluorescent were selected by visual inspection and bulked up. Cell production rates, and the spatial profile of elemental elongation rate, were measured as described (Rahman et al., 2007).

For mapping, DNA samples of 1152 plants from an F2 population created from crossing rsw4 and Landsberg erecta, and a set of 12 SSLP and CAPS markers (see Table S1 in the supplementary material) designed from information provided by The Arabidopsis Information Resource (http://www.arabidopsis.org), were used to map RSW4 between markers mpi120 and mpi130 (see Table S1 in the supplementary material) in a ~56 kb region of chromosome 4, according to the procedures described (Lukowitz et al., 2000).

For complementation, colony hybridization (following the manufacturer's manual; Roche, Indianapolis, IN, USA) with five sequence-specific digoxigenin-11-dUTP-labeled DNA probes (336 to 542 bp; see Fig. 1B, asterisks) was used to isolate 32 clones covering the 56 kb RSW4-containing interval from a genomic cosmid library constructed in the binary vector pBIC20 (Meyer et al., 1994). A set of seven clones covering the interval was identified by multiplex PCR using the same primer pairs that amplified the hybridization probes. Mutant rsw4 plants were transformed by Agrobacterium tumefaciens (GV3101) carrying cosmid clones according to Clough and Bent (Clough and Bent, 1998), and T1 transformants were selected on half-strength Murashige and Skoog plates containing kanamycin (50 μg/ml). Surface-sterilized T2 progeny of the transgenic T1 plants were grown on vertical agar plates supplemented with 1% sucrose for 7 days (5 days at 22°C followed by 2 days at 30°C) and scored for swollen root tips.

For sequencing, genomic DNA was prepared using a cetyl-trimethylammonium bromide-detergent extraction method (Lukowitz et al., 2000). Four DNA fragments (2209, 2491, 2814 and 3155 bp) encompassing the entire RSW4 (At4g22970) coding sequence were amplified independently by PCR, three times from wild type and rsw4 using specific primer pairs and a mixture of Taq and proofreading Pfu polymerases (Promega, Madison, WI, USA). Cycle sequencing of both strands of the PCR products was performed with a set of thirty primers and Big-Dye dideoxy terminator reaction mix (Applied Biosystems, Foster City, CA, USA), and products were resolved on an ABI310 sequencer. Likewise, a 1936 bp genomic fragment containing At4g22980 was amplified by PCR and sequenced with a set of six primers.

Protocols for real-time quantitative PCR were as described previously (Czechowski et al., 2004; Czechowski et al., 2005). Total RNA was extracted from excised root tips (~5 mm apical segments) using an extraction kit (Plant RNeasy; Qiagen, Valencia, CA, USA). Genomic DNA was eliminated by treatment with RNase-free DNase I according to the manufacturer's instructions (Invitrogen, Carlsbad, CA, USA). The concentration of total RNA was measured by spectrophotometer, and 1 μg of total RNA was reverse-transcribed with the Superscript II Reverse Transcription Kit according to the manufacturer's instructions (Invitrogen). Efficiency for all primers is above 1.8, and sequences are given in Table S2 in the supplementary material. These primers all have a T_m of ~58°C. ACTIN2 (At3g18780) or UBQ10 (At4g05320) were used as reference genes. The cDNA was amplified using an appropriate kit (SYBR Green Master, containing Hot-Start Taq polymerase; Applied Biosystems). The cycling program was as follows: 10 minutes at 95°C, followed by 40 cycles of 30 seconds at 95°C, 1 minute at 57°C and 30 seconds at 72°C. To check the specificity of annealing of the oligonucleotides, melting curves were established at the end of the run between 55°C and 95°C and a ramping speed of 2°C per minute. Fluorescence data were acquired at the 72°C step and during the melting curve program. Relative transcript levels of the gene of interest were expressed as 2^−ΔCt, where −ΔCt is the difference in the number of cycles to the threshold between the gene of interest and the reference. In each run, at least three technical replicates were performed for each sample. The results presented are based on three independent biological replicates.

Flow cytometry was performed on a CyFlow flow cytometer running FloMax (Partec, Münster, Germany). Apical, 3-mm root-tip segments of 6-day-old seedlings were frozen and chopped with a razor blade and the homogenate filtered through 30-μm mesh as described previously (Boudolf et al., 2004), except that CyStain solution (Partec) replaced the extraction and staining solutions.

Roots were embedded in butyl-methyl methacrylate and stained for tubulin and DNA as described (Baskin and Wilson, 1997). For whole-mount observations, roots were fixed and treated as described for microtubule observation (Bannigan et al., 2006). For labeling DNA and microtubules, after the secondary antibody, seedlings were rinsed in PBS and treated with 1 μg/ml RNase A for 1 hour at 37°C, followed by a PBS rinse and incubation in 3 μM propidium iodide for 1 hour at room temperature. Seedlings were rinsed again in PBS, root tips were excised and mounted in Prolong Gold (Invitrogen) and observed by confocal fluorescence microscopy (Zeiss Meta 510C). To score mitotic index, roots were fixed as described above and stained in 5 μg/ml 4,6-diamino-2-phenylindole dihydrochloride (DAPI) for 15 minutes, rinsed, mounted, and imaged with the 405 nm laser line of a Nikon Eclipse C1 confocal microscope.

To observe microtubules or chromosomes in living roots, GFP reporter lines were handled as described (Bannigan et al., 2006). To observe root apical meristem histology, seedlings were stained in growth medium containing 3 μM FM 4-64 dye (Invitrogen) for 5 to 10 minutes, then washed in growth medium briefly and examined under the Nikon confocal microscope. To quantify cortical microtubule alignment, roots of GFP-tubulin-expressing seedlings were imaged through the Nikon confocal microscope and projection images of epidermal cells obtained. The angle formed by the long axis of the root and microtubules was measured in Image J (http://rsbweb.nih.gov/ij/). Angles were binned in 10° increments between 0° and 90°, with angles between 90° and 180° taken as the supplementary angle. For each treatment, at least 400 microtubules were measured from five cells per root and between eight and 12 roots. For transmission electron microscopy, the method of Eleftheriou et al. was used (Eleftheriou et al., 2005).

GUS staining was performed on transgenic seedlings expressing a cyclin B1;1-GUS reporter construct (Colón-Carmona et al., 1999). Seedlings were incubated in cold acetone for 15 minutes, and then incubated in staining solution (100 mM sodium phosphate buffer pH 7.0, 10 mM EDTA, 0.5 mM potassium ferricyanide, 0.5 mM potassium ferrocyanide, 0.5% Triton X-100 and 2 mM X-glucuronide) at 37°C for 3 hours. Seedlings were washed three times with the above solution without the substrate, and then cleared according to Malamy and Benfey (Malamy and Benfey, 1997).

For the terminal transferase-mediated DNA end-labeling (TUNEL) assay, seedlings were fixed in 4% paraformaldehyde in PBS for 1 hour. Seedlings were digested in 0.1% pectinase and 0.01% pectolyase for 15 minutes, and rinsed in 10% glycerol and 0.2% Triton X-100 for another 15 minutes. Wild-type seedlings treated with DNase I (200 U/ml) for 15 minutes were used as a positive control. TUNEL-positive nuclei were detected with the In Situ Cell Death Detection Kit (fluorescein) according to the manufacturer's protocol (Roche).

To identify RSW4, we undertook positional cloning (see Fig. S1 in the supplementary material). We crossed rsw4 onto the Landsberg erecta wild type and tested mutant seedlings in the F2 progeny for linkage to mapped polymorphic SSLP and CAPS markers. Based on 2304 chromosomes, RSW4 was localized to a ~350 kb region between ciw32 and ciw44 on the bottom arm of chromosome 4. Then, additional markers narrowed the location to a 56 kb genomic segment that contains 12 annotated genes. To cover this region, we isolated a non-redundant set of seven cosmid clones by colony hybridization and multiplex PCR. The clones were transformed individually into rsw4 and one of them (A1) complemented the mutant phenotype, with the F2 progeny showing a wild type to mutant segregation ratio of 3:1. The complementing cosmid contained two complete annotated gene sequences, At4g22970 and At4g22980, which were absent (or incomplete) in the non-complementing cosmids. Sequencing genomic DNA revealed that in rsw4, there was a specific point mutation in the coding region of At4g22970 that led to a missense mutation, whereas At4g22980 was identical in rsw4 and wild type. Therefore, we conclude that At4g22970 is RSW4.

At4g22970 encodes a protein with homology to separase. This enzyme has been widely studied in animals and fungi and is a protease that cleaves the cohesin complex at the start of anaphase, thereby allowing sister chromatids to disjoin (Nasmyth et al., 2000). Recently, a reverse-genetics approach was used to study separase in A. thaliana (Liu and Makaroff, 2006). These authors showed that evidently null alleles containing T-DNA insertions in At4g22970 give rise to embryo lethality. They also showed that decreasing the expression of this gene specifically in male meiotic cells, by means of a targeted RNAi construct, gave rise to the defective disjunction that is characteristic of a failure to remove cohesin. Because the function of separase was discovered in budding yeast, where the protein was named Extra Spindle Poles (Ciosk et al., 1998), we will refer to At4g22970 as AtESP.

A hallmark of diminished separase function is aberrant chromosome disjunction. To investigate the status of chromosomes, we observed mitotic cells in root tips. Roots were fixed, stained with DAPI and examined in whole-mounts using the 405 nm laser line of the confocal microscope (Fig. 1A-D). In the wild type exposed to 30°C for 24 hours, chromosomes separated cleanly at anaphase (Fig. 1A). However, in rsw4, chromosome disjunction was anomalous. In some cells, disjoined and non-disjoined chromosomes were found in the same spindle (Fig. 1B); in other cells, chromosome fragments were present in aberrant configurations (Fig. 1C), often apparently numbering more than ten (the 4C chromosome number of this species) (Fig. 1D). To image chromosomes along with the mitotic spindle, roots were embedded in methacrylate, sectioned and double stained to show DNA and microtubules (Fig. 1E-L). In roots exposed to 30°C for 14 hours, rsw4 cells in metaphase could not be distinguished from those of the wild type (not shown) or from rsw4 maintained at 19°C (Fig. 1E,I versus 1F,J). In anaphase or telophase, bridges linking separated chromatin masses or nuclei were not seen in wild-type cells at 30°C (Fig. 1G,K), whereas they were common in rsw4 at this temperature (Fig. 1H,L). These defects are typical of separase-deficient eukaryotes, and confirm the molecular identification of RSW4 as AtESP.

Because mitotic figures examined in sections can often be ambiguous, we used whole-mount preparations stained for tubulin and DNA (with propidium iodide) and observed by confocal fluorescence microscopy (Fig. 2). Again, roots were exposed to 30°C for 14 hours prior to fixation. In rsw4 cells at telophase, chromosome bridges were common, as were mitotic cells with apparently more than ten chromosomes (not shown). In some rsw4 cells, chromatin masses appeared to interrupt a forming cell wall or to be stuck to the forming cell wall (Fig. 2A-C). Interestingly, a pair of spindles sometimes formed adjacent to a common cell wall (Fig. 2D), as if daughter nuclei had remained anchored to the new cell wall until entering mitosis again. Neither type of formation was observed in the wild type. Although the arrangement of chromatin in rsw4 was commonly aberrant, microtubule organization in spindles and phragmoplasts appeared similar to that of the wild type, in both methacrylate sections (Fig. 1) and whole-mounts (Fig. 2).

To evaluate the penetrance of aberrant chromosome disjunction, roots were fixed and stained with DAPI and examined in whole-mounts. With optical sectioning, metaphase and anaphase division figures could be found throughout the root. In the wild type at either temperature, and in rsw4 at 19°C, almost half of such figures were in metaphase, and among cells in anaphase only a few were scored as abnormal (Table 1). With rsw4 at 30°C for only 2 hours, the percentage of metaphase division figures grew modestly, whereas that of abnormal anaphase figures grew substantially. By 24 hours at 30°C, nearly all of the anaphase figures in rsw4 roots appeared to be aberrant. These data show that transfer of rsw4 to the restrictive temperature rapidly and pervasively interrupts chromosome disjunction in the root.

To gain further insight into the defective chromosome separation we used GFP reporters. We crossed rsw4 onto a line expressing tubulin-GFP and onto a line expressing histone H3.3-GFP. In both metaphase and anaphase, mitotic spindles appeared indistinguishable in the two genotypes and at the two temperatures (see Movies 1, 2 in the supplementary material). This confirms the inference from the fixed cells that the rsw4 phenotype does not include defective mitotic spindles. In contrast to microtubules, chromosome movement in rsw4 was aberrant. In the wild type, metaphase cells commonly entered anaphase, which took ~2 minutes to complete (Fig. 3A; see Movie 3 in the supplementary material). By contrast, in rsw4 seedlings exposed to 30°C for only 3 hours, anaphase was prolonged. In some cells, chromosomes on a metaphase plate initiated anaphase but did not separate completely (Fig. 3B). However, in most cells, chromosome masses were found that appeared to comprise incompletely segregated chromosomes, which remained so during the period of observation (Fig. 3C,D; see Movies 4, 5 in the supplementary material).

At the permissive temperature, although rsw4 roots do not swell, they elongate more slowly than wild-type roots (Wiedemeier et al., 2002). However, the chromosome disjunction anomalies reported above were rare at the permissive temperature, suggesting additional effects of the rsw4 mutation. To investigate cell division, we quantified cell production rate, which represents the total output of cells per file of a given tissue with time (Rahman et al., 2007). Note that this assay for cell production rate assumes steady state behavior and hence cannot be applied to roots at the restrictive temperature because their growth changes steeply over time. The slowed elongation at 19°C involved no significant decrease in cell length and hence a substantial decrease in cell production rate (Fig. 4). This implies that the decreased elongation rate of rsw4 at 19°C results from defects in cell division rather than in expansion. Consistently, direct measurement of maximal expansion rate within the elongation zone revealed no difference between rsw4 and the wild type (data not shown).

To determine whether the cell cycle becomes arrested in rsw4, we first used flow cytometry (Fig. 5). Exposure of the wild type to 30°C for 24 hours caused no detectable change in the distribution of DNA content among root tip nuclei. In rsw4 at 19°C, ~2% of nuclei had reached 32C, whereas in the wild type, nuclei with 32C levels were undetectable. For rsw4 at 30°C, ploidy tended to increase further, which might reflect a contribution from nuclei formed following non-disjunction. However, the number of nuclei with 2C or 4C DNA content tended to decrease, a trend that is inconsistent with cell cycle arrest at a specific checkpoint.

To assess the cell cycle further, we used quantitative real-time RT-PCR (qRT-PCR) to measure the steady state transcript level for genes known to be regulated transcriptionally during the cell cycle (Menges et al., 2005). To enrich for proliferating cells, RNA was obtained from 5-mm apical root-tip segments. RNA was extracted 12 hours after transfer to 30°C, which is after a period of significant disturbance to mitosis but precedes detectable root swelling. With just one exception, marker expression was little affected (Fig. 6), confirming the lack of cell cycle arrest inferred from flow cytometry.

The exception was a histone H4 transcript, which decreased by ~60%. However, out of more than 40 histone genes subsequently assayed (see Table S3 in the supplementary material), the expression of only two changed more than twofold (one H1 decreased by 50% and one H3.2 doubled). Furthermore, these three histone transcripts changed in rsw7, a mutant with defective mitosis caused by a mutation in a kinesin (Bannigan et al., 2007). Insofar as histone genes are all transcriptionally upregulated in S phase (Marzluff and Duronio, 2002), these results substantiate our conclusion that the cell cycle is not arrested in rsw4 and indicate that the H4 marker originally chosen (At5g59970) behaved anomalously.

Along with radial swelling, the orientation of cell division is altered in rsw4, perturbing histology (Wiedemeier et al., 2002). Although phragmoplasts and preprophase bands formed in the mutant with evidently wild-type structure, aberrant chromosome masses interfered with cell plate formation (see Fig. S2 in the supplementary material). Cell walls were often completely or partially interrupted by a nucleus. When cells had multiple chromatin masses after anaphase, several phragmoplasts formed, leading to wandering cell walls at unusual orientations. It remains to be determined whether these hindrances to cell wall construction are sufficient to explain the altered histology seen in the mutant or if there are additional disturbances to organ polarity.

A sensitive indicator of organ polarity are the cortical microtubules, which in most of the A. thaliana root growth zone are strictly transverse to the long axis of the root. Root diameter in rsw4 diverged from that of the wild type, starting 15 hours after exposure to the restrictive temperature (Fig. 7A). In living root epidermal cells expressing GFP-tubulin, cortical microtubules in the wild type retained their principally transverse orientation, but those of the mutant were often disorganized (Fig. 7B-D). The disorganization seen in these images was confirmed by measuring microtubule angles, which revealed a progressive disorganization with time (Fig. 7E). Microtubule disorganization was concomitant with swelling and underscores the defective polarity of the mutant root, at least after 15 hours of the loss of separase function.

Several mutants have been identified in which disruption of organ polarity is associated with ectopic accumulation of cyclin B1;1 (Suzuki et al., 2005; Inagaki et al., 2006; Zhu et al., 2006). We therefore crossed rsw4 into a line that harbors a widely used cyclin B1;1 reporter (Colón-Carmona et al., 1999). In this reporter, GUS is driven from the native promoter and is translationally fused to an N-terminal portion of the protein that contains the signal for degradation (the D-box). In wild-type cells, GUS expression is punctate, reflecting the cells that happened to have been in G2/M at the time of staining (Fig. 8A). A similar pattern was seen in wild-type seedlings transferred to 30°C for 24 hours and in rsw4 at 19°C (Fig. 8B,C). However, the staining dramatically increased in rsw4 as a function of time at 30°C (Fig. 8D-F). Staining was constant for rsw4 expressing the auxin reporter DR5-GUS (Sabatini et al., 1999) (not shown), arguing against a change in oxidation state that enhances the reaction rate of the glucuronidase. Because the mRNA level for CYCB1;1 was little affected at 12 hours (Fig. 6), it appears that the pronounced increase in GUS staining reflects a failure to degrade the protein.

Ectopic accumulation of cyclin B1;1, as well as root swelling, can be caused by DNA damage (Culligan et al., 2004). We therefore determined to what extent DNA damage, or a response thereto, occurred in rsw4. First, we compared dose-response curves between wild type and rsw4 for hydroxyurea and mutagenic ultraviolet (250 nm) and found no significant differences between the genotypes (data not shown), including experiments at the permissive temperature, where the decrease in cell production (Fig. 4) indicates partial loss of separase function. Separase has been implicated in DNA repair (Nagao et al., 2004; Sak et el., 2008) and non-disjunction itself damages DNA. To assay double-stranded DNA breakage directly, we used the TUNEL assay. Although TUNEL-positive nuclei were easily detected in the DNase I control, nuclei in rsw4 were not labeled above background (Fig. 9A-C), suggesting that widespread DNA damage was absent. Finally, we examined the expression of three genes, WEE1, RAD51 and PARP2, the expression of which is known to be strongly increased in response to DNA damage (De Schutter et al., 2007). Expression of these genes was scarcely affected in rsw4 (Fig. 9B). Taken together, it appears that rsw4 has an essentially undisturbed response to DNA damage.

This work identifies RSW4 as AtESP, the A. thaliana homolog of separase. This identification is based on molecular analysis, as well as on the loss-of-function phenotype of rsw4 having defective chromosome disjunction, as invariably reported for separase mutants, including in A. thaliana when RNAi was used to downregulate AtESP in meiocytes (Liu and Makaroff, 2006; Yang et al., 2009). Indeed, meiocytes of rsw4 plants transferred to 30°C closely resemble those in which separase expression is decreased with RNAi (data not shown). A genomic fragment containing AtESP complemented the growth and division phenotypes of rsw4. An ambiguity exists for the latter result insofar as the complementing clone contained a small open reading frame upstream of the separase. However, this gene in rsw4 had a wild-type sequence, and its expression was neither altered in rsw4 nor, based on community resources, associated with dividing tissues. Taken together, these results indicate that RSW4 is AtESP.

This work adds mitosis in a seed plant to the list of eukaryotic cell types in which separase has been shown to be responsible for allowing sister chromatids to disjoin. Although this might be expected, the result is not a foregone conclusion. Plants have diverged from animals and fungi for hundreds of millions of years and have accumulated differences in their machinery for cell division. Dynein plays a leading role in the mitosis of animals and fungi but has been lost entirely from flowering plants. A regulator of separase called securin is essential in yeast (Onn et al., 2008), but is almost, although not quite, dispensable in mice (Kumada et al., 2006; Wirth et al., 2006) and, based on sequence similarity, is absent from plant genomes. The forms of chromosome disjunction in rsw4 closely resemble those of other eukaryotes with defective separase and substantiate our conclusion of conserved function.

The cell cycle bristles with regulatory spines to block untimely proliferation; in view of this, one might find the lack of cell cycle arrest in rsw4 at the restrictive temperature surprising. Nevertheless, cells appear to lack a checkpoint that monitors disjunction per se. In general, defects in disjunction, as caused by separase deficiency and examined in a wide variety of taxa, occur with continued proliferation, allowing highly polyploid cells to accumulate, often with two (or four or more) replicated chromosomes remaining bound together at the centromeres (Jäger et al., 2001; Siomos et al., 2001; Kumada et al., 2006; Wirth et al., 2006). This phenotype is inconsistent with cell cycle arrest and resembles that shown here for rsw4 (Fig. 1).

In rsw4, the accumulation of cyclin B1;1 could be interpreted as evidence for a G2/M arrest; however, there was no concomitant increase in the CYCB1;1 mRNA, which would be expected for a checkpoint-driven delay. Furthermore, when overexpressed on its own, cyclin B does not invariably block the cell cycle (Amon et al., 1994; Cross et al., 1999; Criqui et al., 2000).

The clearest example of separase deficiency disturbing the cell cycle is found in yeast, in which separase helps to regulate exit from M phase through its interactions with a regulatory phosphatase (Stegmeier et al., 2002; Sullivan and Uhlmann, 2003). Separase also helps regulate exit from M phase in oocytes, although using a different pathway (Terret and Jallepalli, 2006), but usually not in mitotic cells, with the exception of zebrafish embryos (Shepard et al., 2007). By contrast, delays entering mitosis occur in separase-deficient human cell lines (Sak et al., 2008), and these delays are associated with defective spindle assembly and chromosome alignment (Giménez-Abián et al., 2005; Papi et al., 2005). However, neither undue delays at G2/M nor defective spindles accompany separase deficiencies in whole animals and in certain other cell lines (Jäger et al., 2001; Siomos et al., 2001; Padney et al., 2005; Kumada et al., 2006; Wirth et al., 2006). For somatic animal cells, these observations argue against a substantial role for separase in regulating cell cycle progression.

Indications that AtESP is involved in cell cycle regulation, although not at a specific checkpoint, come from rsw4 phenotypes at the permissive temperature. In rsw4 at 19°C, there was little or no cytological evidence of chromosome non-disjunction, indicating that the phenotypes seen reflect other activities of the separase. First, cell production rate at 19°C was ~65% of that of the wild type (Fig. 4), a reduction that is consistent with reduced numbers of division figures found in DAPI-stained roots (data not shown). Second, rsw4 roots had a population of nuclei with a 32C DNA level (Fig. 5), which is a level rarely, if ever, attained in wild-type roots (Boudolf et al., 2004; Ramirez-Parra et al., 2004; Imai et al., 2006; Anonymous, 2009). Together, these phenotypes imply a role for separase in the cell cycle transition from proliferation to endoreduplication.

The rsw4 mutant was identified through its swollen root, but the identification of RSW4 as separase (AtESP) offered no explanation for this altered morphology. To account for swelling, one must examine elements that regulate cell wall anisotropy, such as the cortical array of microtubules. In the original characterization of rsw4 (Wiedemeier et al., 2002), cortical microtubules were reported to be unaffected, but we show here that they become disorganized, concomitant with root swelling. The reason for the discrepancy is not clear; however, Wiedemeier et al. used plastic sections, which sample the cortical array only in small, irregular patches, and looked for large disruptions, as occur with inhibitors. This was recently argued by Bannigan et al. (Bannigan et al., 2007), who found disorganized cortical microtubules in rsw7, a mutant that Wiedemeier et al. also reported to have unaffected microtubules. Disorganized microtubules in rsw4 account for root swelling. The question now becomes: what links microtubules to separase function.

We suggest that the link is cyclin B1;1. This cyclin accumulates in several backgrounds that are defective in genome structure and chromatin maintenance (Suzuki et al., 2005; Inagaki et al., 2006; Zhu et al., 2006) or in response to DNA damage (De Schutter et al., 2007), in some cases only in the presence of DNA-damaging agents (Culligan et al., 2004; Hefner et al., 2006). In all these backgrounds, the superabundant cyclin is associated with some degree of altered histology and often root swelling. In transgenic tobacco expressing a non-degradable version of cyclin B1;1 (Weingartner et al., 2004), as well as in the hobbit mutant of A. thaliana, which reduces the function of the complex responsible for degrading cyclin B1;1 (Serralbo et al., 2006), cortical microtubules are disorganized and to a similar extent as reported here for rsw4. Although the original study overexpressing cyclin B1 in tobacco BY-2 cells reported wild-type morphology (Criqui et al., 2000), a subsequent study found that higher expression levels are needed to provoke microtubule disorganization and altered cell shape (Weingartner et al., 2004). This suggests that among mutants with ectopic cyclin B1;1 accumulation, those with root swelling have higher levels of cyclin B1;1 expression than those without, and that in rsw4, ~15 hours are needed for levels to accumulate sufficiently to alter microtubules and hence morphology.

Thus, many observations fit with the idea that accumulating cyclin B1;1 disrupts cortical microtubules. But what is the connection between the cyclin and separase? The mutants cited above all involve chromatin maintenance, which is a separase function, and separase loss-of-function can damage DNA, as non-separated chromosomes are pulled apart on the mitotic spindle. Also, separase has been reported to be active in DNA repair (Nagao et al., 2004). Hence, in rsw4, DNA damage could be responsible for driving up cyclin B1;1 levels. However, we failed to obtain evidence that DNA damage responses are impaired in rsw4, finding neither cell cycle arrest, TUNEL-positive nuclei, altered sensitivity to DNA-damaging agents, nor alteration of message levels of three pivotal regulators.

Instead, separase might be linked to cyclin B1;1 directly. In animals, one way that separase is regulated is through binding cyclin B1, as part of the CDK complex (Gorr et al., 2005; Holland and Taylor, 2006; Boos et al., 2008). This interaction has been viewed in terms of cyclin B negatively regulating separase. At the onset of anaphase, separase is activated via the proteolysis of its inhibitors, including cyclin B1, a proteolysis that is initiated by the anaphase-promoting complex. It is tempting to speculate that once active, separase itself helps to ensure the swift destruction of cyclin B. Separase mutants of yeast (Tinker-Kulberg and Morgan, 1999) and Drosophila embryos (Padney et al., 2005) accumulate excess cyclin B at metaphase. This accumulation, whatever its mechanism, points to a novel function for separase, and one of importance given how profoundly too much cyclin B1;1 disturbs plant morphogenesis.

We thank Abidur Rahman (now at Iwate University) for establishing the TUNEL assay; Alex Bannigan (now at James Madison University) for help with confocal fluorescence microscopy; Keith Earley and Craig Pikaard (Indiana University) for the GFP-histone line; Chris Makaroff (Miami University of Ohio) for sharing unpublished data on rsw4 meiocytes; and Frederik Coppens [Flanders Institute for Biotechnology (V.I.B.) Ghent University] for unpublished data on HMG2. L.D.V. is a postdoctoral fellow of the Research Foundation-Flanders. This work was supported in part by the Max-Planck Society (W.-R.S. and D.S.), and by a grant from the US Department of Energy (DE-FG02 03ER15421 to T.I.B.), which does not constitute endorsement by that department of views expressed herein.