|Home | About | Journals | Submit | Contact Us | Français|
An average of 200 copies of the rRNA gene (rDNA) is clustered in a long tandem array in Saccharomyces cerevisiae. FOB1 is known to be required for expansion/contraction of the repeats by stimulating recombination, thereby contributing to the maintenance of the average copy number. In Δfob1 cells, the repeats are still maintained without any fluctuation in the copy number, suggesting that another, unknown system acts to prevent repeat contraction. Here, we show that condensin acts together with FOB1 in a functionally complemented fashion to maintain the long tandem repeats. Six condensin mutants possessing severely contracted rDNA repeats were isolated in Δfob1 cells but not in FOB1+ cells. We also found that the condensin complex associated with the nontranscribed spacer region of rDNA with a major peak coincided with the replication fork barrier (RFB) site in a FOB1-dependent fashion. Surprisingly, condensin association with the RFB site was established during S phase and was maintained until anaphase. These results indicate that FOB1 plays a novel role in preventing repeat contraction by regulating condensin association and suggest a link between replication termination and chromosome condensation and segregation.
Repeated sequences and genes are extensively distributed throughout the genomes of many organisms. A common feature of these repeats is instability; the copy number of repeats frequently fluctuates, and they are exposed to the risk of contraction. However, the mechanisms responsible for the fluctuation of repeat copy number and for the maintenance of the repeats remain largely unknown. The eukaryotic rRNA gene (rDNA) family is a typical repeated structure. The genes are generally clustered in long tandem repeats on one or several loci in most eukaryote genomes. In the yeast Saccharomyces cerevisiae, 200 copies on average are tandemly arrayed in a central position on the longest chromosome, chromosome XII. A single rDNA unit (9.1 kb) comprises a 35S rRNA coding region which is transcribed by RNA polymerase I, the 5S rRNA coding region transcribed by RNA polymerase III, and two nontranscribed spacers (NTS), i.e., NTS1 and NTS2 (see Fig. Fig.5A).5A). The replication origin (ARS) and the replication fork barrier (RFB) are located within NTS2 and NTS1, respectively (5, 33, 42). The whole length of the tandem array extends ~1.9 Mb.
One system that maintains the average copy number of the rDNA long tandem array has been reported in recent years. High frequencies of recombination events and expansion/contraction of the repeats are observed for FOB1+ strains, whereas deletion of FOB1 (Δfob1) eliminates both events (14, 24, 27). Furthermore, DNA strand break formation around the RFB site was observed to be FOB1 dependent (7, 29, 47). FOB1 is therefore thought to be required for the maintenance of average copy number by increasing or decreasing copy number through the regulation of recombination events between the repeats. In Δfob1 cells, surprisingly, the rDNA tandem array is still maintained stably without any expansion/contraction. The static rDNA tandem array maintained in Δfob1 suggests that a second, unknown system, one that is different from the FOB1-stimulated recombinational system, acts to maintain the long tandem repeats. Originally, FOB1 was identified as the gene required for replication fork blocking in the direction opposite to 35S rRNA transcription at the RFB site (26). Recently, both in vivo and in vitro RFB site binding activity of Fob1p was demonstrated (28, 36). It is believed that the Fob1p/RFB system is required to avoid a head-to-head collision between replication and transcription of RNA polymerase I (5, 26, 33) and to contribute in the maintenance of chromosome integrity.
Condensin in S. cerevisiae is a protein complex consisting of five subunits, Smc2p, Smc4p, Brn1p, Ycs4p, and Ycs5p (Ycg1p), and plays an important role in condensation and segregation of chromosomes during mitosis (23, 31). All subunits are conserved across the eukaryotes. The two core subunits, Smc2p and Smc4p, belong to the “structural maintenance of chromosomes” family, whose members are involved in various chromosome dynamics; SMC1 and SMC3 encode subunits of cohesin that are required for connection of sister chromatids (35), and SMC5 and SMC6 are found to work for the segregation of repetitive chromosome regions (48). Condensin is essential for cell growth and has been reported to play a role in mitosis as well as in interphase (2). Cytological observations in S. cerevisiae demonstrated that condensin binds throughout the genome during interphase and is concentrated to the nucleolus in anaphase (4, 17, 34). Recently, a novel finding was reported. Although the majority of sister chromatid disjunctions are triggered by cleavage of cohesin during metaphase/anaphase transition (49), sister chromatids on the long arm of chromosome XII that contain the rDNA repeats are still held together despite cohesin destruction. The disjunction of sister chromatids of this locus occurs in mid-anaphase, long after cohesin destruction, and condensin is required for this step (13, 44). However, the mechanisms by which the sister chromatids at the rDNA locus are held together and the role of condensin in chromosome segregation and condensation remain largely unknown.
In this study, we attempted to understand the putative second, unknown system by which the long rDNA arrays are maintained by isolating mutants carrying an unstable, shortened rDNA tandem array in a Δfob1 background. We found that condensin mutants showed severe contraction of rDNA tandem array and slow growth in Δfob1 cells but not in FOB1+ cells. Chromatin immunoprecipitation (ChIP) showed that condensin complex was loaded onto the RFB site in a FOB1-dependent fashion during S phase and remained associated until anaphase. These results indicate that there is a novel system to prevent contraction of long rDNA tandem array securely through FOB1-dependent condensin complex association to the rDNA.
Yeast strains and plasmids used in this study are summarized in Table S1 in the supplemental material.
UKJY225 carrying pKJY181 was mutagenized by use of ethyl methanesulfonate (3%) and plated on yeast extract-peptone-dextrose (YPD). Red colonies without any detectable white sector were screened. Candidates were transformed with pKJY053 to test whether the isolated clones still maintained sector formation ability. Subsequently, pKJY181 was removed by selecting for 5-fluoroorotic acid (FOA)-resistant (FOAR) colonies, and chromosome XII sizes of both FOA-sensitive (FOAS) and FOAR cells were analyzed. pKJY181 carried in the mutant was substituted with pKJY182 (LEU2 marker) to use a URA3-marked genomic library (YCp50 based).
Chromosomal DNA preparation in gel molds was performed as described in the instruction manual (clamped homogeneous electric field [CHEF]; Bio-Rad). Chromosomes in the gel mold were separated in 0.8% pulse field-certified agarose by using CHEF for 72 h at 14°C with 3 V/cm at a linear pulse of 5 to 15 min. In order to determine the rDNA tandem array length, the chromosomes prepared in the gel molds were digested with BamHI, and the DNA fragments were separated in 1.0% pulse field-certified agarose by using CHEF for 30 h at 14°C with 6 V/cm at a linear pulse of 5 to 25 seconds.
The tandem affinity purification-tagged BRNI (BRN1-TAP) and SMC4-TAP strains were used for the ChIP assay. Asynchronous cultures grown at 30°C or cells arrested with α-factor (1.5 μM), hydroxyurea (HU) (180 mM), or nocodazole (10 μg/ml) for 2 h were used for the ChIP assay. To analyze the SMC4 association in anaphase-arrested cells, the strains possessing the cdc15-2 mutation (3) were used to arrest cells in anaphase at 37°C for 2 h. ChIP assay was performed in accordance with the method of Claypool et al. (11) with some modifications. Protein concentrations of the whole-cell extracts (WCE) with and without cross-linking were adjusted to 5 mg/ml. Five hundred microliters of WCE was mixed with 35 μl (bed volume) of immunoglobulin G-agarose beads (Sigma), and immunoprecipitation (IP) was performed. Thirty microliters of WCE was stored for use as the Input sample. Input and IP samples were finally suspended in 100 ml Tris-EDTA after reverse cross-linking and phenol extraction. A 30-fold dilution of the Input sample was used for the PCR (17 cycles). The PCR primers in the rDNA and CUP1 regions were 26-mer oligonucleotides with 50% GC content. The PCR products were separated in 2.5% low-melting-temperature agarose gels (NuSieve GTG agarose; Cambrex) and stained with SYBR green (Cambrex). Signal intensities of the PCR products were measured using a LAS-1000 instrument (Fuji).
Fluorescence in situ hybridization (FISH) was carried out as described in the work of Guacci et al. (20) with some modifications as follows. Zymolyase treatment was performed after the cells had adhered to the polylysine-coated slide glass. The 8.5-kb fragment containing most of the rDNA unit was used for probe preparation. The probe was labeled with biotin by use of a biotin-nick translation kit (Roche) and detected with fluorescein isothiocyanate avidin D (Vector).
In order to elucidate the second, unknown system involved in rDNA repeat maintenance, we attempted to isolate mutants in which the rDNA copy number is drastically reduced under FOB1-defective conditions. However, such cells are expected to exhibit very slow growth or lethality under fob1− conditions because of an insufficient rDNA copy number. To avoid the cessation of cell growth in such mutants, the FOB1 gene was introduced into the parental strain via a plasmid; this allowed rDNA amplification to occur and thus supplied the rDNA copies. Thus, mutants carrying the shortened rDNA array in fob1− conditions are expected to exhibit FOB1 plasmid-dependent faster growth or survive. Therefore, a synthetic lethality screen using a colony color-sectoring assay was set up with the Δfob1 strain (Fig. (Fig.1A).1A). In brief, the UKJY225 strain carrying the pFOB1 plasmid (pKJY181) was mutagenized. Since the pFOB1 plasmid is not essential for growth in the parental cells, segregation of the plasmid can be observed by colony color sectoring in the parental cells, whereas no sector is expected to be observed for the mutants exhibiting pFOB1-dependent growth. The red colonies without any detectable sectors were screened at 30°C. Among ~150,000 colonies, 88 such clones were isolated. pFOB1 was then removed from each mutant by selecting for FOAR cells, and all 88 FOAR clones were grown. Subsequently, the sizes of chromosome XII under the FOB1+ (FOAS) and Δfob1 (FOAR) conditions were determined by CHEF electrophoresis. Although most cells carried a long chromosome XII under both conditions, we found six clones (clones #31, #75, #99, #157, #193, and #200) that carried an extremely short chromosome XII in the Δfob1 background but carried one of almost normal size in the FOB1+ background (Fig. (Fig.1B).1B). The parental FOB1+ (YMW1) (54) and Δfob1 (UKJY225) cells carried a 3.1-Mb-long chromosome XII that was estimated to contain approximately 220 copies of the rDNA (the size of the sum of non-rDNA regions of chromosome XII is approximately 1.1 Mb, and the size of a single rDNA unit is 9.155 kb); however, under the Δfob1 condition, the size of the chromosome XII of five mutants (#31, #99, #157, #193, and #200) was reduced to ~1.2 Mb, and that of mutant #75 was reduced to ~1.7 Mb. To confirm the shortening of chromosome XII under Δfob1 conditions caused by the contraction of the rDNA tandem array, chromosome DNA samples were digested with BamHI, which does not cut within the rDNA (Fig. (Fig.1C),1C), and their fragment sizes were determined by CHEF (Fig. (Fig.1D).1D). The copy number of the rDNA repeats of the five mutants was dramatically reduced (7 to 16 copies; lanes 2, 3, 4, 6, and 7) and that of mutant #75 (lane 5) was reduced to approximately 60 copies (based on the results shown in Fig. Fig.1B).1B). Therefore, the shortening of chromosome XII was due to the supercontraction of the rDNA tandem array. The growth rate of the six mutants carrying pFOB1 (doubling time = 90 to 100 min) was approximately equal to or slightly lower than that of the parental strain (90 min), whereas the mutants in the Δfob1 background showed a severe slow-growth phenotype (doubling time = 170 to 260 min).
In order to identify the genes responsible for the shortening of the rDNA array in the six mutants, we screened the genomic library and identified the gene that was capable of complementing the lack of sector formation in mutants #193 and #31. In the case of mutant #193, the BRN1 gene—a gene encoding a subunit of the condensin complex—was identified as the gene complementing sector formation at the wild-type level. We then examined whether the BRN1 gene was able to complement the supercontraction of the rDNA array in mutant #193 after the removal of pFOB1. As shown in Fig. Fig.2A,2A, the loss of pFOB1 in mutant #193 led to the shortening of its chromosome XII (compare lanes 2 and 3). However, in the presence of the BRN1 gene, the length of chromosome XII remained unchanged even after pFOB1 loss (compare lanes 4 and 5), confirming that BRN1 complemented the supercontraction of the rDNA array in mutant #193. Further, the subsequent removal of the BRN1 plasmid led to the shortening of chromosome XII (lane 6), indicating that the BRN1 gene was required to prevent repeat contraction in Δfob1 cells. In the case of mutant #31, SMC4, whose product is another subunit of the condensin complex, was identified.
Since all six mutants exhibited similar phenotypes, we asked whether condensin could complement the defect in the other four mutants as well. Five condensin genes, SMC2, SMC4, BRN1, YCS4, and YCS5, were cloned on a centromere plasmid (LEU2 marker) and introduced into the remaining four mutants carrying pFOB1; sector formation was then examined. The results of this complementation test revealed that two condensin genes were able to complement the defects in all four mutants; SMC2 and SMC4 complemented in mutants #157 and #99 and mutants #75 and #200, respectively (data not shown). Therefore, all six mutants were identified as condensin mutants, suggesting that condensin plays an extremely important role in preventing the contraction of the long rDNA tandem repeats, at least under Δfob1 conditions. Each mutation was identified as a single base substitution within the coding region, leading to a single amino acid substitution (summarized in Fig. Fig.2B).2B). We refer to these mutations as indicated in Fig. Fig.2B;2B; for instance, we refer to mutation in the mutant #193 as brn1-193. Four mutations in the SMC genes (smc2-99, smc2-157, smc4-31, and smc4-75) are located near the ATP binding Walker B motif. The brn1-193 mutation is located within the N-terminal domain conserved among the kleisin superfamily.
To confirm and further analyze the effect of condensin mutation on rDNA array size, we introduced the most severe mutation (smc2-157) of mutant #157 into the wild-type strain. Mutant #157 under Δfob1 conditions carries the least number of rDNA repeats (seven copies; Fig. Fig.1D,1D, lane 3). The smc2-157 mutation was fused with a kanMX marker and introduced into the diploid strain (N2KJY038). The resulting smc2-157 heterozygous diploid strain (smc2-157/SMC2 Δfob1/Δfob1; N2KJY091) carried long rDNA repeats. In addition, the diploid strain carrying pFOB1 (N2KJY073) produced a wild-type level of sectored colonies, indicating that the smc2-157 mutation was recessive. A haploid Δfob1 smc2-157 strain carrying pFOB1 (N2KJY075), which is created by dissection of N2KJY073, produced sectorless colonies similar to those seen for the original mutant. The rDNA array size of the smc2-157 mutant was then analyzed by monitoring the chromosome XII size in the FOB1+ and Δfob1 cells. While chromosome XII in the FOB1+ condition (N2KJY075) was observed as a broad band with a major intense band of 2.35 Mb, corresponding to ~130 rDNA copies, it was dramatically shrunk to ~1.2 Mb in the Δfob1 condition (N2KJY173) (see Fig. S1 in the supplemental material), as observed in the original mutant #157. The BamHI digestion to analyze the rDNA array length in the Δfob1 smc2-157 double mutant showed two discrete bands of approximately ~140 kb, corresponding to 11 and 12 copies of rDNA (Fig. (Fig.1D,1D, lane 1). Therefore, it was confirmed that the smc2-157 mutation itself caused destabilization of the rDNA array even under the FOB1+ condition and supercontraction of the rDNA array under the Δfob1 condition.
The destabilized rDNA repeat observed in the mutant led us to analyze the recombination event within the repeats. In the FOB1+ smc2-157 strain, the recombination event measured by URA3 marker loss frequency increased to 9.5-fold that in the FOB1+ SMC2+ strain (82.5 ± 6.3 × 10−4 versus 8.8 ± 1.3 × 10−4). We also analyzed the production extrachromosomal rDNA circle (ERC) molecules. However, the accumulation of ERC molecules in the FOB1+ smc2-157 strain (N2KJY075) was only twofold that in the wild type (YMW1); in the Δfob1 smc2-157 double mutant (N2KJY173), ERC molecules were barely detected (unpublished results). Thus, the total copy number (chromosomal repeats and ERC molecules) of the rDNA in the Δfob1 smc2-157 double mutant is extremely low.
As described above, the growth rate of the six condensin mutants was almost at the wild-type level in the FOB1+ cells; however, the growth rate in the Δfob1 cells showed a severe decrease. A plausible reason may be that the copy number of rDNA was insufficient to ensure maximal growth. In order to verify this, the full length of the rDNA unit cloned on a high-copy-number plasmid, pRDN-wt (9), was introduced into the Δfob1 smc2-157 double mutant to supply sufficient copy numbers, and the growth rate was examined. Unexpectedly, as shown in Fig. Fig.3,3, pRDN-wt hardly improves the slow growth of the double mutant (N2KJY173*1), indicating that the dramatic reduction in rDNA copy number is not the primary cause of slow growth. On the other hand, growth rate improvement was observed by the introduction of the FOB1 gene (N2KJY173*2). The rDNA copy number of N2KJY173*2 cells gradually increased and possessed approximately 30 to 40 copies in this assay (data not shown). Despite the fact that N2KJY173*2 had a copy number lower than that of N2KJY173*1, which possesses the rDNA on the 2μm plasmid (an average of 60 copies per haploid cell) (10), growth rate recovery was observed only in the N2KJY173*2 transformant. These results suggest that FOB1 suppresses the slow growth defect of condensin mutation by a reason other than insufficient rDNA copy number. The slightly slower growth of N2KJY173*2 in comparison to that of N2KJY075, which has harbored pFOB1 for multiple generations, is probably due to the lower rDNA copy number (30 to 40 versus 150).
In order to examine the effect of mutation in the other condensin subunit, the rDNA array size was analyzed by monitoring the length of chromosome XII in the previously isolated ycs4-1 temperature-sensitive mutant (NBY521) (4) under both permissive (24°C) and semipermissive (30°C) conditions (Fig. (Fig.4).4). In the parental YCS4+ strain (NBY522), the length of chromosome XII was normal, whereas in the ycs4-1 mutant, its length decreased to ~2 Mb under the permissive condition and it was either too dispersed to detect or stacked in a well under the semipermissive condition. Thus, the mutation of another condensin subunit also resulted in shortened rDNA repeats, even in FOB1+ cells. We then asked whether the supercontraction of the array occurs in a Δfob1 ycs4-1 double mutant. We first attempted to delete FOB1 in the ycs4-1 strain (NBY521); however, we failed to disrupt the FOB1 gene. Subsequently, the diploid strain was dissected; this strain was constructed by crossing the Δfob1 strain (N2KJY002) with the ycs4-1 strain (NBY521). Surprisingly, none of the Δfob1 ycs4-1 double mutant progeny grew. We concluded that ycs4-1 is synthetically lethal with Δfob1, and the six mutants isolated in this study have properties similar to but less severe than those of ycs4-1. While a 9.5-fold increase in the frequency of marker loss was observed in the smc2-157 mutant, the 63-fold increase observed in the ycs4-1 mutants in comparison with the parental YCS4+ strain (4) is consistent with the above conclusion.
It was reported that the condensin complex plays an important role in sister chromatid segregation of the rDNA region (4, 17, 34). By using the ChIP assay, it was demonstrated that condensin physically binds to the rDNA in vivo (17, 34, 51). The results of the suppression in condensin mutants by FOB1 raise the possibility that FOB1 may affect the association of condensin with rDNA. We examined this possibility to analyze the profile of condensin complex association across the rDNA by a ChIP assay using TAP-tagged (41) BRN1 or SMC4, in which the endogenous copy of each gene was modified at its C terminus by TAP (BRN1-TAP and SMC4-TAP). Both the strains exhibited normal growth rates and did not exhibit temperature sensitivity, suggesting that the TAP-tagged condensin genes are fully functional. As shown in Fig. Fig.5B,5B, both BRN1-TAP and SMC4-TAP showed similar association patterns in asynchronous cultures of FOB1+ cells. The major peak in the association profiles of both subunits coincided with the RFB site, and the base of the peak was widely extended within the NTS region. In contrast, both subunits did not associate strongly with the 35S rRNA region; the binding was even throughout the 35S rRNA coding region and did not show any characteristic peaks. The major peak of association with the RFB site was consistent with results from a previous report (51), but no other significant peaks were observed within the 35S rRNA coding region. The virtually identical profiles of BRN1-TAP and SMC4-TAP association across the rDNA strongly suggest that both components associate as a condensin complex. In the Δfob1 cells, the peak of association at the RFB site was almost lost, but a small increase in the association profile within the NTS region was observed (Fig. (Fig.5B).5B). Therefore, we concluded that there were three modes of condensin complex association within the rDNA. The first mode was identified based on the peak of FOB1-dependent association with the RFB site, the second by the small increase in association observed within the NTS region in the Δfob1 cells, and the third by a low basal level of association observed in the 35S rRNA coding region without any characteristic peaks in both FOB1+ and Δfob1 cells. The preferential association of condensin with the RFB site in a FOB1-dependent fashion was also confirmed by monitoring the relative severalfold increases in the intensities of the PCR fragments around the RFB site from that of a fragment in the 35S rRNA coding region (see Fig. S2 in the supplemental material).
We also observed the BRN1-TAP association in an asynchronous culture of the smc2-157 mutant. As shown in Fig. Fig.5D,5D, while a peak at the RFB site was still observed in the smc2-157 mutant in FOB1+ cells, the association with the RFB site as well as the base of the peak extended within the NTS region was severely decreased; weak associations in the 35S rRNA coding region were also decreased. These results indicate that the smc2-157 mutation decreases the rDNA binding ability of the condensin complex.
Cytological observations demonstrated that condensin was dispersed within the nucleus during interphase and was concentrated in the nucleolus during mitosis (4, 17, 34). In order to determine whether the FOB1-dependent condensin association peak at the RFB site was cell cycle regulated, the SMC4-TAP strain was arrested in G1 phase (α-factor) or in metaphase (nocodazole), and a ChIP assay was carried out. As shown in Fig. Fig.5E5E (the topmost graph and the third graph from the top), the FOB1-dependent condensin association with the RFB site is not observed in G1-phase cells but is observed in metaphase cells. Thus, condensin association with the RFB site fluctuates during the cell cycle. The small peak at the RFB site observed in G1-phase-arrested cells may be the result of leaky arrest by the α-factor, which was observed in the fluorescence-activated cell sorter (FACS) profiles.
Despite the low population of metaphase cells in asynchronous cultures, the peak value of condensin association of asynchronous cells is relatively high compared to that for metaphase-arrested cells. This raises the possibility that condensin association occurs in a cell cycle phase earlier than metaphase. We examined the SMC4-TAP association in S phase by arresting the cells for 2 h using HU, which arrested the cell cycle in early S phase. As shown in Fig. Fig.5E5E (second graph from the top), the FOB1-dependent condensin association with the RFB site is observed in S-phase-arrested cells, and the peak value is almost the same as that observed in metaphase-arrested cells. A FOB1-dependent condensin association peak at the RFB site was also observed in the cells synchronized in S phase (the cells incubated for 30 minutes after being released from α-factor arrest) (data not shown). These results indicated that condensin was loaded onto the RFB site during the early S phase.
Recently, it was reported that condensin was required for the proper separation of sister chromatids in the rDNA region during anaphase (13, 44). We investigated whether the FOB1-dependent condensin association with the RFB site was maintained until the anaphase, in which cells were arrested by the cdc15-2 mutation. This mutation arrests the cell cycle in the anaphase at 37°C, and nucleolar localization of Ycs4p was observed in the cells arrested in anaphase (13). As shown in Fig. Fig.5E5E (bottom graph), the FOB1-dependent condensin association peak at the RFB site was also observed in the anaphase-arrested cells. Therefore, we concluded that the FOB1-dependent condensin association with the RFB site was established during the early S phase and maintained until anaphase.
What causes the slow growth rate of the Δfob1 smc2-157 double mutant? Microscopic observations showed that dumbbell-shaped cells of the Δfob1 smc2-157 double mutant were accumulated at a high level in an asynchronous culture. FACS analysis confirmed that a single peak corresponding to 2C cells was observed in the double mutant strain (data not shown); this indicates that the slow growth of the double mutant was caused mainly by slow progression during M phase. By FISH, we analyzed the segregation of the nucleolus and nucleus in these dumbbell-shaped cells, which had accumulated in the asynchronous culture of the Δfob1 smc2-157 double mutant. As shown in Fig. Fig.6A,6A, the rDNA signals decreased in size and intensity in the Δfob1 smc2-157 double mutant; this is consistent with the drastic reduction of the number of rDNA repeats (11 or 12 copies). In addition, despite our use of asynchronous cells, we observed a higher population of mitotic cells in which nuclei were completely separated but the nucleoli were still in the process of separation (Fig. (Fig.6A).6A). In contrast, most of the cells were in interphase in the cases of the wild-type, Δfob1, and smc2-157 strains. Accumulation of the cells in the process of rDNA segregation suggests that the slow growth of the Δfob1 condensin double mutant was caused primarily by the delay of rDNA segregation.
If this is true, then the mitotic delay in the double mutant is due to the presence of an rDNA repeat; thus, the complete removal of the rDNA array from chromosome XII must restore the growth rate of the Δfob1 condensin double mutant. We examined this possibility by constructing a Δfob1 smc2-157 double mutant lacking a chromosomal rDNA array by crossing the double mutant with the rdnΔΔ strain (50). The strains lacking chromosomal rDNA array constructed in this study carry a helper plasmid that can supply 35S rRNA by transcription of a galactose-inducible RNA polymerase II promoter. In addition, those strains also carry multiple copies of ERC molecules in which the 35S rRNA is transcribed by the native RNA polymerase I promoter, thus allowing cell growth even in a glucose medium (50). As shown in Fig. Fig.6C,6C, the growth rate of the Δfob1 smc2-157 double mutant lacking chromosomal rDNA array was almost same as those of the smc2-157 single mutant (N2KJY317) and the SMC2+ strains (N2KJY319 and N2KJY316). Thus, we confirmed that the slow growth observed in the Δfob1 smc2-157 double mutant was due to the defect in rDNA segregation. In addition, although the pFOB1 segregation frequency in the Δfob1 smc2-157 double mutant was particularly low, deletion of the chromosomal rDNA array restored the segregation frequency to the wild-type level (see Fig. S3 in the supplemental material). We therefore concluded that FOB1 plays an important role in rDNA segregation, especially in the condensin mutant.
A long rDNA tandem array is important for cell growth, but it is vulnerable to the risk of copy number decrease. Here, we found a condensin complex-based system that prevents the contraction of long rDNA array in S. cerevisiae. The condensin mutants isolated in this study exhibited a severely slowed growth and a supercontraction of the rDNA tandem array under Δfob1 conditions, whereas both phenotypes were nearly suppressed under FOB1+ conditions. Surprisingly, a sufficient number of rDNA copies supplied by a plasmid could not suppress the slow-growth phenotype of the Δfob1 condensin double mutant, whereas the FOB1 gene could suppress the slow-growth phenotype. Because the tandem array was stably maintained in Δfob1 cells, we initially assumed the existence of a system that maintains the long tandem array independent from FOB1 function. However, our results indicated that the condensin complex functions in concert with FOB1 to prevent contraction of the long tandem array. The FOB1-dependent condensin complex association peak at the RFB site is the basis of the genetic relationship between FOB1 and condensin.
Condensin is required for chromosome condensation and segregation during the mitotic stage. Hence, it is believed that condensin is loaded onto chromatin during mitosis. However, our results clearly indicated that condensin complex loading onto the RFB site occurred in the early S phase at the rDNA locus. This suggests that condensin also plays some role during the S phase. In the fission yeast Schizosaccharomyces pombe, it was reported that condensin played a dual role during both mitosis and interphase (2). It was demonstrated that the condensin mutant exhibited a higher sensitivity to hydroxyurea and failed to activate the replication checkpoint. Thus, our results are consistent with the results demonstrated for S. pombe and suggest that condensin also plays an important role in the S phase of S. cerevisiae.
Cytological observations showed that the condensin complex is enriched at nucleolus during anaphase (4, 17, 34). Observation of condensin association by the ChIP method used in this study showed that there was almost no difference in the amounts of condensin bound to the rDNA from S phase until anaphase. Probably, high condensation of rDNA might produce an apparent increase in signal intensity in cytological observations, despite the fact that the total amount of condensin bound to rDNA could remain the same. Alternatively, condensin distribution during interphase is uniform on multiple chromosomal loci, but condensin remains bound only to the rDNA locus in anaphase. This might also produce an intense signal only in the nucleolus.
The simple explanation for the relationship between FOB1 and the condensin complex is that condensin is recruited to the RFB site by direct interaction with Fob1p. However, if that is the case, it is difficult to explain why the small amount of condensin association is observed within the NTS region in Δfob1 cells. In addition, the specific interaction between Fob1p and the condensin complex is not observed by use of the immunoprecipitation assay (unpublished data). Alternatively, the FOB1-dependent condensin loading onto the RFB site during the early S phase suggests that there is an intimate relationship between condensin loading and replication fork termination. In fact, most of ARS in the rDNA repeats fire in the early S phase (40), and replicating bubbles of an average size of 7 ± 4 kb were observed in the cells arrested for 2 h by HU treatment (43). At the RFB site, the progression of the replication fork in the direction opposite to that of 35S rRNA transcription is stopped in a FOB1-dependent manner during the early S phase. Subsequently, the fork moving in the same direction as that of 35S rRNA transcription approaches, and the two forks converge at the RFB site. Thus, the site of convergence of the replication fork is restricted to the RFB site in a FOB1-dependent manner. Completion of replication by the convergence of the two forks produces multiple intertwinings between sister chromatids (15, 45). Therefore, in the rDNA locus, the localization of the intertwinings must be restricted to the RFB site by FOB1. It is likely that condensin recognizes a certain DNA structure that is created at the RFB site, such as an entangled DNA structure formed by two converging replication forks or a Holliday-like structure formed by replication fork regression. In fact, biochemical analyses demonstrated that the human 13S condensin complex preferentially bound to cruciform DNA and DNA molecules with higher-order structures (25). Further study is necessary to elucidate the mechanism by which condensin is loaded onto the RFB site during S phase.
Multiple intertwinings between sister chromatids produced by converging replication forks must be untangled completely before their separation during mitosis. Condensin is loaded onto the RFB site in the S phase when the intertwinings occur, and it remains associated with the RFB site until the M phase, when the catenanes must be untangled. It can be stated that condensin, along with topoisomerase II (Top2p), plays a role in managing the untangling of the catenanes. In fact, it was reported that condensin was required to localize Top2p onto the chromatin (4). In addition, it was recently reported that although the disjunction of the majority of chromosomal regions occurred through cohesin destruction during the metaphase/anaphase transition, the rDNA locus on sister chromatids remained connected even after the destruction of cohesin, and both condensin and Top2p were required for proper disjunction of this region during mid-anaphase (13, 44). Efficient untangling by condensin and Top2p may be required for the proper segregation of the rDNA locus. Although it is still unclear why cohesin-independent connection remains until anaphase, it can be interpreted that catenation connects the sister chromatids of the rDNA region even after cohesin destruction.
In the Δfob1 condensin double mutant, there is some defect in rDNA segregation. As a result, segregation of sister chromatids is expected to occur without complete untangling of the catenanes accumulated in the rDNA region. This may lead to fatal damage and result in the supercontraction of the tandem array. Although the rad52 Δfob1 smc2-157 triple mutant was able to survive, the segregation frequency of the pFOB1 plasmid in the triple mutant was lower than that in RDA52+ cells (unpublished results). It appears that the supercontraction may require the double-strand break repair pathway; however, the single-strand annealing pathway is probably responsible for the rescue of the rad52 mutant. However, when supercontraction occurs and it results in carrying short tandem array in the Δfob1 condensin double mutant, the amounts of mutant condensin available for single rDNA copy are expected to increase. This would make the disjunction of short tandem array possible, although difficulties in disjunction still remain. This explanation is supported by the accumulation of dumbbell-shaped cells with still segregating rDNA regions in the asynchronous culture of the Δfob1 smc2-157 strain.
Why was the supercontraction observed in the condensin mutant under the Δfob1 condition but not under the FOB1+ condition? In Δfob1 cells, as the replication fork proceeds in the direction opposite to that of RNA polymerase I transcription, it moves through the RFB site and reaches the 35S rRNA coding region (46); therefore, converging replication forks are not restricted to the RFB site but are distributed randomly throughout the rDNA. This results in the random distribution of intertwinings throughout the rDNA. Approximately 72% (6.6 kb of a single 9.1-kb copy) of these positions are expected to be located within the 35S rRNA coding region. Further, more than half of the copies of rDNA repeats are actively transcribed (18), and active replication origins are clustered (40) and are localized downstream of the transcriptionally active copies (40, 38). These findings imply that most replication fork convergence sites must be localized in the active 35S rRNA coding region. Within the active copies, strong and ceaseless RNA polymerase I transcription that can dissociate even histone-DNA associations (12) continues throughout the cell cycle (16). Such strong transcription may have an inhibitory effect on condensin association with the rDNA. Furthermore, the condensin mutant isolated in this study exhibited a reduced ability to associate with the rDNA (Fig. (Fig.5D).5D). Thus, in the Δfob1 condensin double mutant, strong RNA polymerase I transcription may dramatically inhibit the association of the mutant condensin complex with the rDNA. It was reported that the distribution of a similar SMC family complex, cohesin, is also affected by RNA polymerase II transcription (30, 32).
In the case of FOB1+ cells, FOB1 restricts the position of replication fork convergence to the RFB site; therefore, the inhibitory effect of RNA polymerase I transcription on the reduced associative ability of the mutant condensin complex can be suppressed by keeping the condensin interaction sites out of the RNA polymerase I transcription zone. This would enable a level of untangling sufficient to allow the maintenance of a long tandem array even in the case of the condensin mutants. The importance of the FOB1-dependent condensin association system observed only in the condensin mutants is that it appears to ensure prevention of the contraction of the long repeats. In fact, in the case of wild-type condensin, the long tandem array is stably maintained even under the Δfob1 condition. The FOB1-independent condensin association observed within the NTS region under the Δfob1 condition may be sufficient to manage 200 copies, if the condensin is wild type, because the degree of condensin association with the rDNA in Δfob1 SMC2+ cells is higher than that in FOB1+ smc2-157 cells, in which ~150 copies are maintained by the mutant condensin.
Previous studies have demonstrated that cells possess an average copy number maintenance system, in which the FOB1 stimulates recombination, thus allowing expansion/contraction of the repeats (27). Here, we demonstrate that another task performed by FOB1 is prevention of the contraction of the tandem array via the regulation of condensin association with the RFB site. In performing both of these roles, the cells seem to acquire a higher potential for stably maintaining long repeat structures; this is because replication fork blocking activity and condensin are conserved in various organisms. From this point of view, it is very interesting that condensin in human cells is concentrated in the nucleolus during the interphase (8). Further, in bacteria, both SMC-like proteins (e.g., that encoded by mukB in Escherichia coli) and replication fork blocking systems (Ter site and Tus protein) are found, and chromosome segregation defects were observed in mutants of SMC-related genes (6, 19, 21, 22, 37, 39, 53). Therefore, the relationship between replication forks converging at specific sites and condensin association may be important not only for repeat but also for nonrepeat regions. In fact, it was reported very recently that condensin localized in the zone of converging DNA replication (52).
We thank Kunihiro Matsumoto, Andrew W. Murray, Masahiro Ajimura, Takehiko Kobayashi, Akio Toh-e, and Yoshikazu Ohya for providing the yeast strains, plasmids, and genomic library. We are grateful to Austen Ganley for critically reading the manuscript.
This work was supported in part by Grants-in-Aid for Scientific Research 13480234 and 13141205 from the Ministry of Education, Culture, Sports, Science and Technology.
†Supplemental material for this article may be found at http://mcb.asm.org/.