Ribonuclease Activity of Dis3 Is Required for Mitotic Progression and Provides a Possible Link between Heterochromatin and Kinetochore Function

Background Cellular RNA metabolism has a broad range of functional aspects in cell growth and division, but its role in chromosome segregation during mitosis is only poorly understood. The Dis3 ribonuclease is a key component of the RNA-processing exosome complex. Previous isolation of the dis3-54 cold-sensitive mutant of fission yeast Schizosaccharomyces pombe suggested that Dis3 is also required for correct chromosome segregation. Methodology/Principal Findings We show here that the progression of mitosis is arrested in dis3-54, and that segregation of the chromosomes is blocked by activation of the mitotic checkpoint control. This block is dependent on the Mad2 checkpoint protein. Double mutant and inhibitor analyses revealed that Dis3 is required for correct kinetochore formation and function, and that this activity is monitored by the Mad2 checkpoint. Dis3 is a member of the highly conserved RNase II family and is known to be an essential subunit of the exosome complex. The dis3-54 mutation was found to alter the RNaseII domain of Dis3, which caused a reduction in ribonuclease activity in vitro. This was associated with loss of silencing of an ura4+ reporter gene inserted into the outer repeats (otr) and central core (cnt and imr) regions of the centromere. On the other hand, centromeric siRNA maturation and formation of the RITS RNAi effector complex was normal in the dis3-54 mutant. Micrococcal nuclease assay also suggested the overall chromatin structure of the centromere was not affected in dis3-54 mutant. Conclusions/Significance RNase activity of Dis3, a core subunit of exosome, was found to be required for proper kinetochore formation and establishment of kinetochore-microtubule interactions. Moreover, Dis3 was suggested to contribute to kinetochore formation through an involvement in heterochromatic silencing at both outer centromeric repeats and within the central core region. This activity is likely monitored by the mitotic checkpoint, and distinct from that of RNAi-mediated heterochromatin formation directly targeting outer centromeric repeats.


INTRODUCTION
Accurate chromosome segregation during mitosis is an essential event for proper cell multiplication. A large number of the genes required for different aspects of chromosome segregation have been identified. This includes genes encoding proteins implicated in centromere/kinetochore structure and function, sister chromatid cohesion, and chromosome condensation, together with proteins required for the spindle assembly checkpoint, activation of APC (anaphase promoting complex)/C (cyclosome) for ubiquitin-dependent proteolysis, and for anaphase promoting proteolysis that leads to the destruction of securin and the activation of separase [1]. In addition, various protein kinases such as Cdc2, polo and aurora and related protein phosphatases are implicated in mitotic control. These essential genes for chromosome segregation are conserved throughout eukaryotic organisms.
A series of fission yeast Schizosaccharomyces pombe cold sensitive mutants have been previously isolated that showed severe defects in sister chromatid separation at 20uC [2]. These mutants, termed dis1, dis2, and dis3 (defective in sister chromatid disjoining), also contain hyper-condensed chromosomes at the restrictive temperature. Dis1 is a microtubule binding protein that is required for the proper attachment of kinetochore microtubules to the kinetochores, and is localized at the kinetochore specifically during mitosis [3], [4]. Dis1 is also the founding member of the Dis1-XMAP215-chTOG family of microtubule-associating proteins that are universally present in eukaryotes. Dis2 is one of the two catalytic subunits (Dis2, Sds21) of Type 1 protein phosphatase PP1. The dis2-11 mutation is semi-dominant and blocks sister chromatid separation at the restrictive temperature [5]. Both Dis1 and Dis2 proteins are phosphorylated and their activities are regulated by Cdc2 kinase [6], [7]. Cdc2 down regulates the phosphatase activity of Dis2 and up regulates the kinetochore localization of Dis1 by direct phosphorylation during mitosis.
Unlike Dis1 and Dis2, the role of Dis3 in proper kinetochore function or mitotic progression is not clear. Dis3 is a relatively large protein (110 kD) that includes the characteristic RNase II (RNB) domain, which is highly conserved in bacteria, archaea and eukaryotes. RNase II contains a highly processive 39R59 exoribonuclease activity, and eukaryotic Dis3 homologs (Rrp44 in budding yeast) were identified as a component of the exosome that is required for RNA maturation and turnover in both the nucleus and cytoplasm [8]- [10]. Rrp44/Dis3 is one of more than ten protein components in the exosome, which can act upon different RNA substrates including mRNA, snRNA, snoRNA, rRNA and tRNA. The structure of E. coli RNase II was recently determined [11], [12] and revealed that the active site in the RNB domain is composed of four Asp residues and one Arg residue in two respective motifs referred to as Motif I and Motif IV. Mg 2+ ion is chelated by the first and fourth asp residues in Motif I.
The human homologue of Dis3 was identified as an oncogene [13], consistent with a potential role in chromosome segregation and/or mitotic progression. Human Dis3 is tightly bound to Ran a small G protein [14]. Fission yeast Ned1, a homologue of human disease gene product lipin, directly interacts with Dis3, Pim1/Rcc1 (GEF for Ran) and nuclear pore proteins, and is implicated in complex cellular functions including chromosome organization and nucleo-cytoplasmic transport [15]. The Dis3 protein is located in both the nucleus and cytoplasm throughout the cell division cycle, and deletion of the dis3 + gene in fission yeast showed that it is essential for cell viability [16]. The mitotically arrested phenotype is thus specific to the dis3-54 mutant allele upon shifting to the restrictive temperature.
In order to determine how Dis3 contributes to mitotic progression, and more importantly, whether this represents a role for Dis3 ribonuclease in kinetochore formation/function, we have focused on the dis3-54 mutant of fission yeast. Here, we report that the dis3-54 mutation resides in the Dis3 RNB domain, and that the mutant protein contains a reduced RNase activity. We show that the spindlekinetochore interaction is abnormal in dis3-54 mutant cells and that this is associated with mitotic arrest at a pre-anaphase stage, which is dependent on the essential spindle assembly (mitotic) checkpoint protein, Mad2. We also show evidence that Dis3 is involved in a heterochromatic silencing process within the centromere. Mitotic spindles in wild-type (WT) and dis3-54 mutant cells at the restrictive temperature of 20uC. Cells grown in liquid culture were at 33uC were shifted to the restrictive temperature of 20uC for 4 hr and fixed with methanol. Anti-tubulin antibody TAT1 (green) and DAPI (red) were used for immunostaining and DNA staining, respectively. WT cells containing nuclei consistent with both metaphase and anaphase were observed as indicated, whereas dis3-54 cells appeared to have arrested during mitosis with abnormally long metaphase-like spindles. Scale bar = 10 mm. (B) Mitotic spindle length distributions in WT and dis3-54 mutant cells grown at restrictive temperature of 20uC for 4 hr. Cells were stained with anti-Sad1 (spindle pole body component) and anti-tubulin antibody after fixation with 3% paraformaldehyde and the pole-to-pole distance of the spindle microtubule in mitotic cells was measured. The inverted triangle represents the position of mode. (C) Time-lapse images of dis3-54 mutant cells expressing chromosomally integrated GFP-tagged a2-tubulin (Atb2) under the inducible nmt1 promoter. Images of cells growing at both 20uC (restrictive temperature, left panels) and 33uC (permissive temperature, right) were taken at 30 sec intervals for approx. 90 min with a confocal microscope. Numbers within each panel represent the time (min) after start of observation. The white arrow indicates the collapsed spindle microtubule, often observed after an extended period of mitotic arrest in dis3-54. (D) Time course plot of spindle length based on time-lapse images shown in Fig. 1C. Cells grown at 33uC (blue line) and 20uC (yellow line; ''cell 1'') were measured and the 0 min time point was adjusted to reflect the start of mitosis. Normal changes in spindle length representing phases I, II and III were observed during the time-course at 33uC [18]. doi:10.1371/journal.pone.0000317.g001

RESULTS Arrested dis3-54 mutant cells show an elongated metaphase-like spindle
Cold-sensitive dis3-54 mutant cells grown at 33uC were shifted to the restrictive temperature of 20uC and incubated for 4 hr to induce the nuclear division arrest phenotype. Arrested cells were fixed with methanol and the chromosomal DNA and microtubules were stained using DAPI and anti-tubulin antibodies, respectively. As expected, mutant cells displayed the dis-like phenotype [2], in that sister chromatids appeared not be separated and condensed chromosomes were scattered along the mitotic spindle (Fig. 1A). The average spindle length in dis3-54 mutant cells arrested at 20uC (4.5 mm, Fig. 1B) was longer than that of the wild-type metaphase spindle (2.5 mm), but much shorter than the fully extended anaphase B spindle in wild-type (approximately 10-13 mm).
In order to monitor spindle changes over a period of time in actual living cells, a chromosomally integrated GFP-tagged a2tubulin gene [17] was introduced into dis3-54. Mutant cells expressing GFP-tubulin both at restrictive (20uC) and permissive (33uC) temperatures were examined using a confocal microscope system equipped with a temperature control unit. Time-lapse images of cells cultured at 20uC and 33uC were obtained (representative cells shown in Fig. 1C; Movies S1, S2 and S3) and used for quantitative measurements of spindle lengths following initiation of mitosis ( Fig. 1D). At the restrictive temperature of 20uC, phase 1 (period of spindle formation; [18]) occurred normally in dis3-54 cells with an initial spindle length of 2.5-3.0 mm, however, the cells then appeared to remain in phase 2 (constant spindle length period; metaphase) as the spindle slowly reached a length of 4-6 mm within 90 min. Phase 3 (spindle extension period; anaphase B) did not take place and the spindle often collapsed after prolonged incubation at 20uC (indicated by white arrows in Fig. 1C). In contrast, dis3-54 cells grown at the permissive temperature (33uC) displayed normal spindle elongation and completed anaphase with a long anaphase B spindle within 30 min after entry into phase 1.
To determine whether dis3-54 cells cultured at 20uC were arrested at a pre-anaphase stage or following the switch to anaphase, the dis3-54 mutation was combined with a chromosomally integrated copy of a GFP-tagged Cut2/securin gene under the control of its native promoter. Cut2/securin acts to prevent sister chromatid separation during metaphase and is degraded upon the triggering of anaphase [19]. In wild-type cells, Cut2-GFP signals were associated with condensed chromosomes during metaphase and disappeared in all anaphase and post-anaphase cells. In contrast, dis3-54 cells cultured at 20uC constantly showed intense Cut2-GFP signals that remained with the condensed chromosomes and did not disappear (Fig. 2). This confirmed that anaphase (sister-chromatid separation) did not occur and dis3-54 cells were indeed arrested at a pre-anaphase stage. Taken together, the above data indicates that a pre-anaphase spindle is produced in arrested dis3-54 cells grown at the restrictive temperature, which remains present for 60-80 min.

Mitotic arrest of dis3-54 involves the checkpoint protein Mad2
Correct progression through mitosis is essential for cell survival and it is therefore carefully monitored by several proteins that prevent completion of the mitotic cycle should a potentially lethal problem be encountered [20], [21]. Mad2 is known to restrain the mitotic progression by preventing the activation of ubiquitin-dependent anaphase promoting proteolysis. Although dis3-54 cells were found to be arrested at a pre-anaphase stage, it was unclear whether this was due to the dis3-54 mutation directly preventing sister chromatid separation, or due to dis3-54 resulting in activation of the mitotic checkpoint to halt mitotic progression at the pre-anaphase stage.
To distinguish between these two possibilities, we first investigated the localization of two spindle assemble checkpoint proteins, Mad2 and Bub1, in wild-type and mutant cells using chromosomally integrated GFP-tagged versions of these genes driven by their native promoter. Intense kinetochore dot signals corresponding to Mad2-GFP were observed in most dis3-54 cells arrested at 20uC, whereas such signals were only normally seen in a small number of wild-type mitotic cells (Fig. 3A). Similar results were also obtained from the analysis of Bub1-GFP. In wild-type cells, Bub1-GFP was only observed in early mitotic stages, however. Intense kinetochore localization of Bub1-GFP was observed in most of the arrested dis3-54 cells (Fig. 3B).
To confirm that the observed presence and kinetochore localization of Mad2 was functionally contributing to the mitotic arrest in dis3-54, the mutant cells were crossed with a Mad2 deletion mutant (Dmad2). In striking contrast to that of single dis3-54 mutants, the spindle was fully extended in the double mutant cultured at 20uC (Fig. 3C) and similar to that observed in wild-type (see Fig. 1A). Analysis of GFP-tagged Cut2/securin localization in the double mutant also revealed loss of Cut2 associated with the appearance of daughter nuclei following pre-anaphase (Fig. 3D). This was similar to that observed in wild-type (see Fig. 2) and demonstrated that in the absence of Mad2, dis3-54 mutant cells are able to overcome the arrest at pre-anaphase and proceed to anaphase. Two daughter nuclei that were asymmetrical in size or lagging chromosomes were often observed in the dis3-54 Dmad2 double mutant cells, suggesting that although the spindle was WT and dis3-54 mutant strains that carried a chromosomally integrated Cut2/securin-GFP fusion gene under control of the native promoter were cultured at 20uC for 4 hr in synthetic EMM2 medium. Cells were stained with Hoechst H33342 for visualizing DNA together with GFP signals, and were not subject to fixation. WT cells in both pre-anaphase (p) and anaphase (a) were observed, whereas persistent Cut2 signals in dis3-54 (three representative cells shown) confirmed they were in a preanaphase stage. Scale bar = 10 mm. doi:10.1371/journal.pone.0000317.g002 extending normally, chromosome segregation was unequal. In both fission yeast and animal cells, the spindle checkpoint proteins Mad2 and Bub1 have been shown to be recruited to kinetochores that are not attached by kinetochore microtubules or in which the stable bipolar attachment is not established [22]- [24]. The above data strongly indicates that normal bipolar attachment between the kinetochore and spindle is impaired in dis3-54 mutant cells, and the resulting spindle checkpoint inhibits the metaphase/ anaphase transition.

Sister chromatid separation and abnormal segregation in dis3-54 Dmad2
To investigate further whether sister chromatid separation could occur when the spindle checkpoint was compromised in dis3-54 cells, the nuclear phenotype of both single dis3-54 and double dis3-54 Dmad2 mutants was examined over an extended time period following shift to the restrictive 20uC temperature. As shown in Fig. 4A, almost all single mutant cells contained the typical dis-like pre-anaphase nuclei (see Fig. 1A). In contrast, the proportion of cells with the ''asymmetrical daughter nuclei'' phenotype (see Fig. 3C) greatly increased in the double mutant cells and approximately 60% of the double mutant cells showed the asymmetrical nuclei phenotype by 4 hr after the shift to 20uC.
To directly observe how the sister centromeres were segregating in individual dis3-54 and double dis3-54 Dmad2 mutant cells, we crossed these mutants with a strain that expresses LacI-GFP and contains lacO repeats integrated at the lys1 locus that is close to the centromere of chromosome 1 (cen1-GFP; [18]). All wild-type cells contained a single cen1-GFP spot at opposite ends of the cell corresponding to each copy of chromosome 1 (Fig. 4B), confirming normal segregation during anaphase. On the other hand, one or two cen1-GFP spots on the single pair of condensed chromosomes could be observed in the single dis3-54 mutant cells arrested at 20uC, consistent with the notion that they were arrested at a metaphase-like stage. In the double mutant dis3-54 Dmad2 cells, however, approximately 55% of anaphase cells showed the two segregated spots as seen in wild-type. The remaining double mutant cells contained two closely situated spots within the same daughter nucleus, suggesting that the sister chromatids had separated but did not move to opposite nuclei. This data indicates that although sister chromatid separation occurred in the double mutant upon loss of the mitotic checkpoint, chromosome segregation towards the opposite poles became abnormal, possibly due to lack of bi-orientation in paired sister centromeres that then moved to the poles in an apparently random fashion and caused the high frequency of unequal segregation.

Hypersensitivity to a microtubule poison
Fission yeast mutants defective in kinetochore-microtubule attachment are usually hypersensitive to microtubule destabilizing drugs such as thiabendazole, TBZ [22], [25]- [28]. We examined whether this was also the case for the single dis3-54 and double dis3-54 Dmad2 mutants. Serial dilution of mutant cells were spotted onto YPD plates containing 0 and 10 mg/ml concentrations of TBZ, together with the control hypersensitive a1-tubulin mutant, nda2 [29]. Both dis3-54 and dis3-54 Dmad2 were hypersensitive to TBZ even at the semi-permissive temperature of 30uC (Fig. 4C). The double mutant dis3-54 Dmad2 also showed a recognizable synthetic effect, as the cells grew slowly in comparison with single dis3-54 or Dmad2 mutant cells.

S. pombe Dis3 contains an RNase activity that is reduced by the dis3-54 mutation
In order to further understand the role for Dis3 in mitosis, the causative mutation in dis3-54 was first determined by sequencing the dis3 gene in dis3-54 mutant cells. A single nucleotide substitution was identified (C1690RT1690) that resulted in an amino acid substitution from Pro509 to Leu509 (Fig. 5A). Pro509 is located within the central region of Dis3 that contains a highly conserved RNB domain [30]. The RNB domain has been shown to be a catalytic module in RNaseII family proteins [11], [31] and Pro509 is conserved in all eukaryotic homologues, although the prokaryotic E. coli RNase II contains a serine residue at the equivalent position.
We next attempted to confirm that Dis3 does indeed contain ribonuclease activity, and if so, whether this is affected by the dis3-54 mutation. Plasmids were constructed that contained either wild-type Dis3 (WT) or dis3-54 mutant Dis3 (P509L) fused to a GST tag under control of the nmt1 inducible promoter. Another residue in the RNB domain, Thr808, has been shown to be located in the catalytic region of RNB by crystallographic structure analyses [11], [12] and differs from Pro509 in that it is also highly conserved in prokaryotes, although its contribution to RNase activity was yet to be determined. We therefore targeted this residue as a control and constructed a third plasmid with RNB mutant Dis3 (T808A) fused to GST. After overproduction in fission yeast cells, GST fusion proteins were purified and in vitro RNase assays were performed at 37uC and 20uC using total RNA isolated from fission yeast as the substrate [8].
When the substrate RNA was incubated with control GST protein only, the RNA remained stable throughout the reaction periods. In contrast, the substrate RNA was rapidly degraded in a time-course dependent manner when incubated with wild-type Dis3-GST protein (Fig. 5B). This demonstrates that fission yeast Dis3 is indeed a ribonuclease that exhibits activity towards at least rRNAs in vitro. The RNase activity of wild-type Dis3 was  Fig. 1A. Sister chromatids were able to separate in the double mutant, but chromosome segregation was abnormal. (C) dis3-54 and dis3-54 Dmad2 mutants are hypersensitive to TBZ. Serial dilutions (1:5) of each strain were spotted onto YPD solid media containing 0 and 10 mg/ml TBZ and incubated at 30uC. The cold-and TBZ-sensitive nda2 mutant strain defective in a1-tubulin was used as control [25]. doi:10.1371/journal.pone.0000317.g004 somewhat slower at 20uC compared to that at 37uC, as most RNA was degraded within 15 min at 37uC compared to substrate RNA still remaining by 60 min at 20uC.
The dis3-54 (P509L) and RNB (T808A) mutant Dis3 proteins were assayed in a similar manner to determine whether these mutations alter Dis3 RNase activity. Time-course analysis of substrate RNA degradation at 37uC showed that the P509L substitution result in reduced RNase activity of Dis3. In reactions containing wild-type Dis3, 50% of substrate RNA was degraded within 4 min at 37uC, whereas approximately 10 min was required for the same level of degradation with P509L and T808A mutant Dis3 (Fig. 5C). At 20uC, the effect of P509L and T808A mutations on RNase activity was more severe. Approximately 110 min was required for 50% degradation of substrate RNA by P509L mutant Dis3, in contrast to only 26 min for wildtype Dis3. The T808A mutation severely disrupted Dis3 activity as more than 50% of the target RNA still remained after 240 minutes. The Dis3 protein contains the PINc, RNB and S1 domains. PINc belongs to a large family of protein domains associated with exonuclease or nucleotide binding activity. S1 is an RNA binding domain [65]. The RNaseII RNB domain contains four high consensus regions, Motif I, II, III and IV [30]. Amino acid sequences from Motifs I and IV of E. coli The dis3-54 mutation can be complemented with a plasmid containing a wild-type copy of the dis3 + gene. To confirm that the reduction in RNase activity of Dis3 is directly linked to the dis3 phenotype, we constructed equivalent Dis3 expression plasmid containing either the P509L or T808A mutations. As expected, P509L mutant Dis3 was unable to complement the dis3-54 cold-sensitive (cs) phenotype. Expression of RNB mutant (T808A) Dis3 also failed to complement the cs phenotype, confirming the importance of Dis3 RNase activity for its kinetochore function (Fig. 5D).
Expression of the integrated ura4 + gene was assayed by spotting serial dilutions of each strain on either non-selective media (N/S) or counter-selective media containing FOA (inhibitor of growth of Ura4 + -expressing cells). The otr::ura4 + construct was normally silenced in the wild-type strain such that cells could grow in the presence of FOA, whereas growth of the control non-silenced strain Ddcr1 otr::ura4 + was inhibited by FOA (Fig. 6). In dis3-54 mutant cells at 26uC (the semi-permissive temperature), silencing at the outer otr repeat and inner imr1 regions was clearly reduced as FOA inhibited colony formation of dis3-54 otr::ura4 + and dis3-54 imr1::ura4 + . The ura4 + gene inserted at cnt1 central core region was also desilenced in the dis3-54 mutant at the semi-permissive 26uC. This silencing defect was more severe than that at the imr1 and otr regions, and could be observed even at the permissive 33uC (Fig. 6). The Dis3 protein therefore has a role in silencing in all three major centromeric regions. In addition to the centromeric silencing defect, we found that silencing at the heterochromatic mating type locus (mat) was also reduced in the dis3-54 mutant (Fig. 6).
Native transcripts from outer centromeric repeats accumulate to low levels in the dis3-54 mutant RT-PCR was employed to directly detect transcripts from the otr region in the dis3-54 mutant. Specific primers were used to distinguish between both forward and reverse transcripts derived from the dh sequence of the otr repeats (see Fig. 6). In control RNAi mutant strains Ddcr1, Dago1 and Drdp1, strong accumulation of transcripts from both the forward and reverse strands were observed, whereas only the reverse strand was just detectable in wild-type ( Fig. 7A and [32]). In dis3-54 cells grown at the restrictive temperature, accumulation of both the forward and reverse strands was observed, although this was weak and much less than that for RNAi mutants. Loss of centromeric silencing in dis3-54 was nonetheless significant as no transcripts could be detected in wild-type cells grown under the same conditions.

Centromeric siRNA production and incorporation into RITS is not affected in the dis3-54 mutant
RNAi-mediated centromeric heterochromatin formation in S. pombe involves the formation of 22 , 26 nucleotide (nt) small interfering RNAs (siRNAs) by Dcr1-dependent cleavage of centromeric dsRNAs derived from the otr repeats [32], [40]. These siRNAs are also incorporated into the effector complex called RITS, which is required for the downstream histone modification within the otr region [33]. In order to examine the role, if any, of Dis3 protein in siRNA production or incorporation Figure 6. The dis3-54 mutant shows defects in centromeric silencing. Centromeric silencing of four S. pombe strains, wild-type (WT), Dicer deletion RNAi mutant (Ddcr1), Mis6 kinetochore mutant (mis6) and single dis3-54 mutant, was examined in a genetic background containing the ura4 + reporter gene integrated into different regions of the centromere (otr::ura4, imr1::ura4, cnt1::ura4). Schematic diagram of S. pombe centromere I (upper panel) shows the ura4 + insertion sites. A strain without the otr::ura4 reporter was included as a control (ura4 2 ). Silencing of a reporter gene integrated into the mating type locus (mat1) was also examined for dis3-54. Serial dilutions (1:5) of each strain were plated on YPD media without selection (N/S) and on YPD media containing 5-FOA (inhibitor of Ura+ strains), and incubated at permissive (33uC) and restrictive (26uC) temperatures as indicated. Note that mis6 is a heat-sensitive mutant and that the permissive and restrictive temperatures for mis6 are 26uC and 33uC, respectively. doi:10.1371/journal.pone.0000317.g006 into RITS, accumulation of centromeric siRNAs was first determined in dis3-54 mutant cells.
Small RNAs were purified from S. pombe wild-type, Ddcr1 deletion and dis3-54 mutant cells grown at 33 or 20uC, and subject to Northern blot analysis using a probe specific to the dh region. As shown in Fig. 7B, siRNA production in the dis3-54 mutant grown at the restrictive temperature was similar to that in wild-type and was not impaired. As expected, no siRNA was detected in the negative control Ddcr1 mutant strain.
We next examined whether the siRNAs were normally incorporated into the RITS complex in dis3-54 mutant cells. The dis3-54 mutant was crossed with an S. pombe strain that carried Figure 7. Centromeric siRNA production and and overall chromatin structure in dis3-54 is similar to wild-type. (A) Accumulation of RNA transcripts derived from native centromeric repeats was determined by semi-quantitative RT-PCR using strand-specific primers. Total RNA isolated from WT and S. pombe RNAi deletion mutants, Drdp1, Dago1 and Ddcr1, grown at 33uC, and from WT and dis3-54 cultured at the restrictive temperature (20uC) were assayed. Actin was also assayed as a loading control, and reactions without reverse transcriptase (-RTase) were included as a control against DNA contamination. (B) Total small RNAs were extracted from S. pombe wild-type (WT) and dis3-54 mutant cells cultured at the restrictive temperature (20uC). WT and Ddcr1 deletion mutant strains cultured at 33uC were included as a positive control. Small RNA northern blots were probed with a specific probe to detect siRNAs derived from the centromeric dh repeat. Position of RNA size marker is shown to the left. Blots were reprobed with U6 snRNA for loading control. (C) The RITS complex was isolated from wild-type (WT), Ddcr1 and dis3-54 mutant cells by immunoprecipitation of the FLAG-tagged Chp1 component. Small RNAs that co-purified with FLAG-Chp1 were extracted by phenol-chloroform and separated on a denaturing polyacrylamide gel after labeling with [59-32 P] pCp using RNA ligase. Position of the RNA size marker is shown to the right. (D) Localization of Swi6 protein at centromeric otr region in the dis3-54 mutant. ChIP analysis was performed using antibody against Swi6 protein [58]. Co-precipitated DNA was purified and PCR amplified using otr (dg1 and dg3) region specific primers. (E) Chromatin structure at the three centromere regions, cnt1, imr1 and otr (dg1) in WT, mis12 and dis3-54 mutants. Nuclear chromatin fractions were prepared from WT and dis3-54 mutant cells grown at 20uC for 8 hr, and mis12 at 36uC (restrictive) for 8 hr, and digested with MNase for 0, 1, 2, 4 and 8 min. Southern Hybridization was then performed using specific probes for cnt1, imr1 and dg1. doi:10.1371/journal.pone.0000317.g007 the chromosomally integrated Flag-tagged Chp1 protein, which is a chromo-domain protein and a component of the RITS complex [33], [41]. Chp1-Flag was purified from cells grown at either 33uC or 20uC by immunoprecipitation, and analyzed for co-precipitated siRNAs after labeling with [59-32 P] pCp and RNA ligase [33]. In dis3-54 mutant cells, small RNAs coprecipitated with Chp1-FLAG at both 20uC and 33uC similar to that in wild-type cells, although the association may be slightly weaker at the restrictive temperature (Fig. 7C). Consistent with previous published data, no small RNAs were associated with RITS in control Ddcr1 cells [33]. The above data demonstrates that centromeric siRNAs are produced normally and are bound to RITS in dis3-54 cells grown at the restrictive temperature.
Swi6 protein remains associated with the otr region at reduced levels in dis3-54 Heterochromatin structure at the otr region in S. pombe centromeres is characterized by specific histone modifications and the subsequent binding of Swi6, which is a homolog of the mammalian heterochromatin protein 1 (HP1) and involved in heterochromatin spreading [42]- [45]. The presence of Swi6 at the outer repeats in the dis3-54 mutant at the restrictive temperature was therefore examined using the chromatin immunoprecipitation (ChIP) assay. As previously reported, strong binding of Swi6 to the otr region was detected in WT cells, whereas no binding was observed in control Ddcr1 RNAi mutant cells and a Dswi6 mutant. In contrast, Swi6 remained associated with the otr region in the dis3-54 mutant, although at a reduced level compared to that in wild-type (Fig. 7D). This result is consistent with the slight accumulation of transcripts from dh repeats in dis3-54. The fact that Swi6 was still localized at the outer repeats also supports the above finding that the mechanism for RNAi-mediated heterochromatin formation is not affected in the dis3-54 mutant.
Overall centromeric chromatin structure in the dis3-54 mutant is similar to that in wild-type The central core region of the centromere is known to have a specific unique chromatin structure, which is observed as a smeared pattern following micrococcal nuclease (MNase) digestion as opposed to a regular nucleosomal ladder pattern observed for the outer repeat region [36], [46]. Several mutants defective in silencing at core centromere, such as mis6, mis12, cnp1, mal2 and sim4, have also been shown to be defective in maintenance of the specific central core chromatin structure [38], [39], [47]- [49]. We therefore used the MNase assay to investigate overall chromatin structure at the three major centromeric regions in the dis3-54 mutant. As previously reported, the smeared pattern for cnt1 and imr1 regions in wild-type was lost in the mis12 kinetochore mutant and changed to a regular pattern (Fig. 7E), consistent with the disruption of chromatin structure at core regions [48]. On the other hand, smeared MNase digestion for cnt1 and imr1 regions was maintained in the dis3-54 mutant. At the outer repeats, the regular pattern observed in wild-type was also maintained in dis3-54. This data suggests that although Dis3 has a role in centromeric silencing, it is not required for overall centromeric chromatin structure.
Synthetic lethality of dis3-54 mutation with Dswi6 or Ddcr1 Due to the fact that silencing of centromeric heterochromatin was defective in the dis3-54 mutant, particularly in the otr:: ura4 + region that is silenced by a spreading mechanism from neighbouring centromeric repeats, we examined whether the dis3-54 mutation genetically interacts with heterochromatin mutations such as Dswi6 and Ddcr1. Swi6 is a fission yeast homolog of mammalian heterochromatin protein HP1 that is involved in heterochromatin spreading [42]- [45] downstream of RNAi-mediated heterochromatin formation involving Dcr1. As shown in Fig. 8A, both the double mutants dis3-54 Dswi6 and dis3-54 Ddcr1 were synthetic lethal at 30 and 33uC, respectively, whereas each of the individual single mutants could produce colonies at the same temperature. The result suggested that the Dis3 protein might share an essential function with Swi6 and Dcr1 in heterochromatin formation or heterochromatin function during mitosis.

DISCUSSION
In this study, we show that the S. pombe cold-sensitive dis3-54 mutation results in a single amino acid change (Pro509RLeu; P509L) within the characteristic RNase II RNB domain of Dis3 and causes a Mad2-dependent mitotic arrest at the restrictive temperature. Arrested cells show hyper-condensed mitotic chromosomes that are scattered along a metaphase-like spindle, which is longer in length than that normally observed during metaphase in wild-type. Continuous localization of the GFP-tagged spindle checkpoint proteins Mad2 and Bub1 at the kinetochores of arrested dis3-54 cells suggested activation of the mitotic checkpoint. Consistent with this, deletion of the mad2 + gene in dis3-54 led to a release of the mitotic arrest and revealed unequal separation of sister chromatids and centromeres, strongly suggesting that the mitotic checkpoint was induced by a failure in proper kinetochore-spindle interaction. The segregation phenotype of the dis3-54 Dmad2 double mutant resembles that of other centromere/ kinetochore mutants in S. pombe [38], [47]- [49], and the elongated metaphase spindle in arrested dis3-54 cells is a common kinetochore mutant phenotype in fission yeast [48], [50], [51]. Taken together with the hypersensitivity of dis3-54 to the tubulin poison TBZ, the above data indicates that the kinetochore structure and function is impaired in dis3-54 at the restrictive temperature.
In other eukaryotes, Dis3 has been identified as a component of the exosome that is involved in 39R59 RNA processing. However, to the best of our knowledge, no reports to date have identified a relationship between the RNase activity of Dis3 and proper mitotic progression. It is possible that Dis3 in fission yeast had evolved a kinetochore-specific function and that the RNase activity is either lost or distinct from this role. Our in vitro analyses confirmed that WT S. pombe Dis3 protein does indeed retain RNase activity. Moreover, the amino acid change caused by the dis3-54 mutation, Pro509RLeu509, was found to cause reduced Dis3 RNase activity. The possibility that Pro509 is a specific residue required for a mitotic function independent of RNase activity was addressed by examining a second mutation in the RNB domain that is located almost 300 residues C-terminal of dis3-54 and also reduced RNase activity (Thr808RAla; T808A). Dis3 constructs containing mutations located outside the RNB domain are able to complement the cold-sensitive phenotype of dis3-54 (H.M., M.Y., unpublished data), similar to the wild-type Dis3 gene. In contrast, Dis3 constructs containing the T808A RNB mutation and the wild-type Pro509 residue could not complement the cold-sensitive dis3-54 phenotype. This strongly suggests that it is the correct RNase activity, rather than the actual Pro509 residue, that is required for the kinetochore function of Dis3.
Mitotic and chromosome segregation defects have been reported for three fission yeast mutants, dhp1-1, pfs2-3169 and Dcid14, that are also defective in RNA metabolism. Dhp1 is a 59R39 exoribonuclease required for the processing and export of poly(A) RNA in the nucleus, whereas Pfs2 and Cid14 are involved in mRNA 39-end processing and polyadenylation [52]- [54]. However, unlike for Dis3, these proteins are not directly related to the exosome. Nonetheless, the mitotic phenotypes of dis3-54 and the above mutants do indicate significant roles for RNA processing in correct kinetochore function and chromosome segregation.
How might the exosome activity of Dis3 contribute to kinetochore formation, such that when this activity is compromised the mitotic checkpoint is activated? One possible explanation is that RNA metabolism of unknown target genes whose products are essential for kinetochore function is defective in dis3-54 mutant. In this case, identification of accumulated RNAs in dis3-54 mutant may be crucial for understanding the mechanism of mitotic arrest in dis3-54 mutant. Another plausible explanation suggested by our data is through a possible link with heterochromatin silencing. In the dis3-54 mutant, silencing was defective at reporter genes inserted into the outer repeats and central core of centromere. Both these regions of the centromere, especially the central core, are known to be essential for faithful segregation of chromosomes in mitosis [55].
At the central core region of the centromere, the dis3-54 mutant showed a more severe silencing defect without a disruption of the overall specific chromatin structure itself. The conserved kinetochore proteins Mis6 and Mis12 were also normally localized at core regions in the dis3-54 mutant (data not shown). This phenotype is similar to that in alp5-1134, a histone acetyltransferase component mutant, which was reported to be required for silencing at the central core [56]. The alp5-1134 mutant also causes a chromosome segregation defect that results in activation of spindle checkpoint in mitosis [56]. Like the Alp5 protein, Dis3 may have an important function in silencing at core centromere to ensure the faithful kinetochore function, but not in the maintenance of the structure itself.
Centromeric transcripts derived from outer centromeric repeats are known to be processed by the RNAi machinery to mediate heterochromatin formation [32]- [34]. Recently, it was shown that heterochromatin spreading and silencing of reporter genes neighboring outer centromeric repeats requires an endonucleolytic cleavage (slicing) activity by the Ago1 RNAi component [57]. Significantly, the cleavage products resulting from wild-type Ago1 slicing activity accumulated in mutants that lacked the Rrp6 nuclear-specific exosome subunit, indicating that the nuclear exosome is involved in processing these transcripts. In the absence of Ago1 slicing activity, heterochromatin structure was only slightly reduced at the outer centromeric repeats, whereas silencing of the otr::ura4 + reporter gene was lost. This is similar to the phenotype observed here for dis3-54. The Dis3 subunit was also recently shown to be responsible for core exosome activity and to have a nuclear function that partially overlaps with Rrp6 [58]. It is thus highly likely that Dis3 also processes centromeric transcripts, possibly even those cleaved by the Ago1 protein.
Although further study is required, the above data support a hypothesis that the Dis3 activity is involved in a heterochromatin silencing process that may be linked to its role in the formation of a function kinetochore (Fig. 8B)

Immunofluorescence microscopy
Cell expressing GFP-tagged genes integrated on the chromosome were fixed with methanol and imaged using immunofluorescence microscopy as previously described [61]. The TAT1 antibody was used for tubulin immunostaining [62].

In vitro RNase assays
The procedure for in vitro RNase assays was basically as according to [8]. Assays were performed using affinity-purified recombinant protein estimated by SDS-PAGE. Purity of the purified GST and Dicer (RNAi component; Ddcr1). Single and double mutants were plated on solid media and replica plates incubated at a range of temperatures from permissive to restrictive (36uC, 33uC, 30uC, 20uC). Synthetic defects were observed at semi-permissive temperatures for the double mutants as indicated. (B) One model for the role of Dis3 in correct kinetochore formation and function. Dis3 exosome activity is involved in heterochromatin silencing at regions neighbouring centromeric repeats, possibly processing transcripts cleaved by Ago1. This processing mechanism would also be important for establishment of a functional kinetochore and microtubule attachment. Dis3 activity (either directly or indirectly) is then monitored as a key step by the mitotic checkpoint, preventing serious errors in the absence of correct kinetochore formation. doi:10.1371/journal.pone.0000317.g008 fusion proteins were confirmed by Coomassie blue staining ( Figure  S1). Assay were performed in 10 mM Tris-HCl at pH7.6, 50 mM KCl, 5 mM MgCl 2 , 10 mM DTT, 100 mg/ml BSA, 5.0 mg Yeast RNA and 0.8 U/ml RNasin; Promega. Reactions were stopped by addition of 1/10 vol of 8% SDS and 100 mM EDTA, and heated at 65uC for 15 min after adding 1 vol of 50% formamide and loading dye. Aliquots of 20 ml were resolved by agarose gel (1.2% agarose, 20 mM MOPS, 1 mM EDTA, 50 mM sodium acetate and 2.2 M formaldehyde) electrophoresis, and stained by ethidium bromide (EtBr). The quantity of EtBr stained RNA was assessed by Molecular Imager GS-700 (Bio-Rad).

RNA analyses
Small RNA was prepared from cells growing exponentially in appropriate medium, and RT-PCR and Northern blots were performed as described [63]. Analysis of RITS complex containing siRNA was described previously [64].

MNase nuclease digestion
The MNase digestion assay was performed as described previously [47].