Ribosomal DNA Transcription-Dependent Processes Interfere with Chromosome Segregation

Abstract

The ribosomal DNA (rDNA) is a specialized genomic region not only owing to its function as the nucleolar organizing region (NOR) but also because it is repetitive in nature and, at least in budding yeast, silenced for polymerase II (Pol II)-mediated transcription. Furthermore, cohesin-independent linkages hold the sister chromatids together at the rDNA loci, and their resolution requires the activity of the conserved protein phosphatase Cdc14. Here we show that rRNA transcription-dependent processes establish linkages at the rDNA, which affect segregation of this locus. Inactivation of Cfi1/Net1, a protein required for efficient rRNA transcription, or elimination of Pol I activity, which drives rRNA transcription, diminishes the need for CDC14 in rDNA segregation. Our results identify Pol I transcription-dependent processes as a novel means of establishing linkages between chromosomes.

Chromosome segregation is mediated by the removal of cohesin complexes, which hold sister chromatids together (24). A protease known as Separase removes cohesins from chromosomes by cleaving one of the cohesin subunits. Recently, it has become clear that segregation of the genes encoding rRNAs (ribosomal DNA [rDNA]), which are clustered in long tandem repeats on chromosome XII, require mechanisms in addition to cohesin removal for their segregation in budding yeast (6, 37, 40, 44). This observation is perhaps not surprising given that this genomic region is unique in several ways. First, owing to its repetitive nature, the rDNA is thought to be unstable (19). Second, the rDNA has the ability to organize an organelle, the nucleolus, around it (reviewed in reference 31). Third, 60% of all cellular transcription occurs at the rDNA, which is mediated by RNA polymerase I (Pol I) and Pol III (reviewed in reference 45). Fourth, DNA replication is largely unidirectional, coinciding with the direction of rRNA transcription (16). Fifth, the rDNA is silenced for Pol II-mediated transcription (5). Finally, at least in yeast, the rDNA segregates late during mitosis and meiosis (6, 13, 14).

The necessary mechanisms for rDNA segregation, in addition to cohesin removal, have been defined at some level. The protein phosphatase Cdc14, which is best known for its role in bringing about exit from mitosis, promotes rDNA segregation at least in part by targeting condensin, a protein complex required for chromosome condensation, to the rDNA (6, 37, 44). Cdc14 itself is regulated by an inhibitor Cfi1/Net1 that holds the protein inactive in the nucleolus during most of the cell cycle (reviewed in reference 35). During anaphase, two regulatory networks, the Cdc14 early anaphase release (FEAR) network and the mitotic exit network (MEN) promote the dissociation of Cdc14 from its inhibitor. Cdc14 activated by the FEAR network during early anaphase is primarily responsible for promoting rDNA and nucleolar segregation (6, 37, 40, 44). Why the rDNA requires Cdc14 for its segregation is not known, but it is clear that mechanisms in addition to cohesin-mediated cohesion enhance the cohesiveness of this genomic region, which are revealed when CDC14 is inactive. Throughout this study, we will refer to this cohesiveness at the rDNA that is observed in cdc14 and FEAR network mutants as “cohesin-independent linkages” or just as “linkages” at the rDNA.

We investigate here the nature of the linkages at the rDNA that require CDC14 and the FEAR network for their segregation. Our studies revealed that rDNA transcription imposes a need for CDC14 on rDNA segregation. By deleting factors involved in efficient rDNA transcription or eliminating RNA Pol I transcription, we show that rDNA segregation no longer relies on Cdc14 activity. The linkages at the rDNA caused by transcription are not Pol I specific, since rDNA segregation also requires CDC14 function when rRNA transcription is mediated by Pol II. We furthermore find that eliminating transcription also partially suppresses the rDNA segregation defect of cells defective in condensin function. Our results suggest that the production of rRNA and/or factors that assemble onto the rRNA represent a novel way of establishing linkages between chromosomes and impose a need for Cdc14 and condensin on rDNA and nucleolar segregation.

MATERIALS AND METHODS

Strains and plasmids.

All strains are isogenic with strain W303 (K699) and are listed in Table 1. Unless stated otherwise, gene deletions were constructed through one-step PCR as described previously (22). NOY1071 was used to make strain A14628 and NOY794 was used to make strain A14742, and A14743 (11, 27). rDNA-green fluorescent protein (GFP) dot strains were made as described previously (6). The MET-DHFRts-FLAG construct, which was used to generate a conditional depletion allele of RPA135, was made by cloning the MET promoter in front of the previously described DHFR-ts degron construct and by fusing the FLAG epitope to DHFR (8). This construct was then integrated upstream of the RPA135 open reading frame by using a PCR-based strategy. pGAL-URL-CDC14 was made by integrating pGAL-URL-3HA upstream of the CDC14 gene by a PCR-based strategy. The pGAL-URL-3HA fusion was constructed from the previously described pWS103 plasmid containing the UPL degron (32).

TABLE 1.

Strains used in this study

Strain	Relevant genotype
A1536	MATacfi1::URA3
A1665	MATacfi1::URA3 cdc14-3
A2587	MATa (wild type)
A2596	MATacdc15-2
A5321	MATacdc14-3
A9972	MATacdc15-2 PURA3::tetR::GFP::LEU2 RDN1::tetOx112::URA3
A10676	MATacdc14-3 fob1Δ::HIS3
A12629	MATacdc15-2 dn14Δ::TRP1
A12631	MATacdc14-3 dn14Δ::TRP1
A12712	MATadn14Δ::TRP1
A12732	MATacdc15-2 rad52Δ::TRP1
A12734	MATacdc15-2 rad52Δ::TRP1
A12894	MATacdc14-3 rad52Δ::TRP1
A13902	MATaset1Δ::kanMX6 dot1Δ::TRP1 hmlΔ::LEU2 cdc14-3
A13908	MATaset1Δ::kanMX6 dot1Δ::TRP1 hmlΔ::LEU2 cdc15-2
A14237	MATaPura3::tetR::GFP::LEU2 RDN1::tetOx112::URA3 cdc15-2 UBR1::GAL-Ubi-M-lacI-frag-Myc-UBR1::HIS3 CaURA3MX6-Degron-FLAG-RPA135
A14615	MATaPURA3::tetR::GFP::LEU2 RDN1::tetOx112::URA3
A14617	MATacdc14-3 PURA3::tetR::GFP::LEU2 RDN1::tetOx112::URA3
A14619	MATacdc14-3 cfi1Δ::HIS5 PURA3::tetR::GFP::LEU2 RDN1::tetOx112::URA3
A14620	MATacdc15-2 cfi1Δ::HIS5 PURA3::tetR::GFP::LEU2 RDN1::tetOx112::URA3
A14628	MATacdc14-3 fob1Δ::HIS3 25xRDN1
A14742	MATarpa135Δ rrn9Δ PURA3::tetR::GFP::LEU2 RDN1::tetOx112::URA3
A14743	MATarpa135Δ rrn9Δ cdc14-3 PURA3::tetR::GFP::LEU2 RDN1::tetOx112::URA3
A15651	MATacdc14-3 rad1Δ::kanMX
A15652	MATacdc15-2 rad1Δ::kanMX
A15653	MATarad1Δ::kanMX
A15663	MATacdc14-3 rad1Δ::kanMX rad52Δ::TRP1
A15665	MATacdc15-2 rad1Δ::kanMX rad52Δ::TRP1
A15667	MATarad1Δ::kanMX rad52Δ::TRP1
A15669	MATacdc14-3 rad1Δ::kanMX dn14Δ::TRP1
A15670	MATacdc15-2 rad1Δ::kanMX dn14Δ::TRP1
A15672	MATarad1Δ::kanMX dn14Δ::TRP1
A15673	MATα cdc15-2 dn14Δ::kanMX rad52Δ::TRP1
A15675	MATadn14Δ::kanMX rad52Δ::TRP1
A15677	MATacdc14-3 dn14Δ::kanMX rad52Δ::TRP1
A15679	MATacdc14-3 rad1Δ::kanMX dn14Δ::TRP1 rad52Δ::TRP1
A15680	MATα cdc15-2 rad1Δ::kanMX dn14Δ::TRP1 rad52Δ::TRP1
A15682	MATarad1Δ::kanMX dn14Δ::TRP1 rad52Δ::TRP1
A15691	MATaPura3::tetR::GFP::LEU2 RDN1::tetOx112::URA3 cdc14-3 UBR1::GAL-Ubi-M-lacI-frag-Myc-UBR1::HIS3 CaURA3MX6-Degron-FLAG-RPA135
A15692	MATaPura3::tetR::GFP::LEU2 RDN1::tetOx112::URA3 UBR1::GAL-Ubi-M-lacI-frag-Myc-UBR1::HIS3
A15693	MATα Pura3::tetR::GFP::LEU2 RDN1::tetOx112::URA3 UBR1::GAL-Ubi-M-lacI-frag-Myc-UBR1::HIS3
A15713	MATacdc14::GAL-URL-3HA-CDC14::kanMX PURA3::tetR::GFP::LEU2 RDN1::tetOx112::URA3
A15714	MATacfi1Δ::HIS5 PURA3::tetR::GFP::LEU2 RDN1::tetOx112::URA3
A15715	MATacdc14::GAL-URL-3HA-CDC14::kanMX cfi1Δ::HIS5 PURA3::tetR::GFP::LEU2 RDN1::tetOx112::URA3

Growth conditions.

Strains were grown to log phase at 25°C in the medium specified in the figure legends and arrested in G₁ using a final concentration of 5 μg of α-factor/ml as described previously (6). Then, 2.5 μg α-factor was added to all cultures again 2 h later to prevent escape from the arrest. Strains were grown/ml to the log phase at 25°C in the medium specified in the figure legends and arrested in S phase by using 10 mg of hydroxyurea/ml as described previously (reviewed in reference 1). If hydroxyurea arrest was longer than 2 h, 5 mg of hydroxyurea/ml was re-added to all cultures to prevent escape from the arrest. Thiolutin (CMS Chemicals, Ltd.) was dissolved in dimethyl sulfoxide (DMSO; Fluka Chemicals) and used at a final concentration of 87 μM (18).

Other techniques.

Indirect immunofluorescence on whole cells was carried out as described previously (43). Immunofluorescence samples were visualized using a Zeiss Axioplan 2 microscope. Unless noted otherwise, 100 cells were scored for each time point. Rat anti-tubulin antibodies (Oxford Biotechnology) and fluorescein isothiocyanate-conjugated anti-rat antibodies (Jackson Immunoresearch) were both diluted at 1:500. Mouse anti-Nop1 (EnCor Biotechnology, Inc.) and cyanine Cy3-conjugated anti-mouse (Jackson Immunoresearch) antibodies were diluted at 1:750 and 1:1,000, respectively.

Anaphase cells were defined as cells with an elongated mitotic spindle and two distinct DAPI (4′,6′-diamidino-2-phenylindole) masses. Segregated nucleoli were defined as anaphase cells with two distinct and unconnected Nop1 masses of approximately equal size. Detection of rDNA-GFP dots was performed as described previously (6). To score rDNA-GFP segregation in anaphase cells, a cell was defined as being in anaphase if it was a dumb-bell shaped cell with fully segregated DAPI masses. These anaphase cells were scored for either unsegregated rDNA-GFP dots (one rDNA dot total) or segregated rDNA-GFP dots (two rDNA dots total; one dot per nuclear lobe). The percentage of anaphase cells with segregated rDNA-GFP dots was calculated by dividing the number of anaphase cells with segregated rDNA GFP dots by the number of anaphase cells with segregated and unsegregated GFP dots. The small percentage of anaphase cells with two distinct GFP dots that had not yet segregated to opposite lobes of the cell were not included in these calculations, because they fit neither into the fully segregated nor unsegregated category.

Western blot analysis was performed as previously described (42). Mouse anti-Flag antibodies (Sigma) were used at 1:625, mouse anti-Pgk1 antibodies (Molecular Probes) were used at 1:25,000, mouse anti-Vph1 antibodies (Molecular Probes) were used at 1:2,000, mouse anti-HA antibodies (HA-11; Covance) were used at 1:1,000, and anti-mouse horseradish peroxidase-conjugated (Amersham) antibodies were used at 1:2,000.

RESULTS

An rDNA array of only 25 repeats segregates as poorly as a 150-repeat array in cdc14-3 mutants.

In wild-type W303 yeast strains, the rDNA locus is composed of about 150 repeats (29). To determine whether the number of rDNA repeats influences the segregation behavior of this chromosomal region, we analyzed cdc14-3 mutants carrying either the wild-type rDNA array or 25 copies of the repeat (confirmed by Southern blot analysis, data not shown; note that cdc14-3 mutants carried a deletion in FOB1 to stably propagate 25 rDNA repeats (11, 27). Cells were synchronized in G₁ and then released into the cell cycle at 37°C to inactivate the cdc14-3 gene product. As cells progressed through the cell cycle, we examined when the nucleolus segregated by indirect immunofluorescence using an antibody against the nucleolar protein Nop1. As reported previously, cdc14-3 mutants arrested in anaphase with largely unsegregated nucleoli (Fig. 1A and B) (6, 14, 37, 40, 44). Reduction of the rDNA repeat number to 25 did not ameliorate the nucleolar segregation defect of cdc14-3 mutants (Fig. 1A and B). In fact, nucleolar segregation was less efficient than that of a wild-type array (Fig. 1A). A wild-type strain with 25 rDNA repeats segregated without any delays (data not shown). Thus, even a short rDNA locus requires CDC14 for its segregation.

FIG. 1.

The rDNA repeat number does not affect rDNA segregation in cdc14-3 mutants. (A) cdc14-3 (A5321), cdc14-3 fob1Δ (A10676), and cdc14-3 fob1Δ 25xRDN1 (A14628) cells were arrested in G₁ with α-factor (5 μg/ml), followed by release into fresh medium at 37°C. The percentage of cells in anaphase was determined by using tubulin staining at the indicated times after release (left panel). Nucleolar segregation in anaphase cells was determined as described in Materials and Methods using Nop1 and is shown in the right panel. At least 50 anaphase cells were analyzed. (B) Examples of unsegregated (top) and segregated (bottom) nucleolar masses in anaphase cells. Nop1 staining is shown in red, antitubulin staining is in green, and DNA staining is in blue. (C) Schematic representation of the location of the rDNA GFP dot on chromosome XII. (D) Wild-type cells (A14615; wild-type RDN1) cells carrying 400 copies of the rDNA (A14742; rpa135Δ rrn9Δ 400xRDN1), cdc14-3 mutants carrying a wild-type rDNA array (A14617; cdc14-3, wild-type RDN1), and cdc14-3 mutants carrying 400 copies of the rDNA array (A14743; cdc14-3 rpa135Δ rrn9Δ 400xRDN1) are all carrying rDNA GFP dots and were grown at 37°C for 140 min, and anaphase cells were scored for segregation of the rDNA-GFP dots as described in Materials and Methods.

We also examined cdc14-3 mutants bearing 400 copies of the rDNA repeat. Cells that lack the RNA Pol I subunit RPA135 and in addition carry a deletion of the upstream activation factor component RRN9 inefficiently transcribe the rRNA using RNA Pol II (27). As a result, these polymerase-switched cells carry an increased number of rDNA repeats (approximately 400 copies [27]) to compensate for this decrease in rRNA transcription. To compare the segregation of this expanded rDNA locus accurately with that of a wild-type array, we analyzed the segregation of an array of tet operator (tetO) sequences integrated next to the rDNA locus proximal to the telomere (Fig. 1C; henceforth, rDNA GFP dots [6]). The segregation behavior of this array can be examined by expressing a tet repressor-GFP fusion in cells (38). Because cells lacking RPA135 and RRN9 grow very slowly (27), we analyzed rDNA segregation only in anaphase cells rather than in cells progressing through the cell cycle in a synchronous manner. Wild-type cells carrying 400 copies of the rDNA (confirmed by Southern blot analysis [data not shown]) segregated rDNA GFP dots between mother and daughter cells as efficiently as cells carrying 150 copies of rDNA (Fig. 1D). Furthermore, an array of 400 rDNA repeats (confirmed by Southern blot analysis [data not shown]) did not further impair rDNA GFP-dot segregation in cdc14-3 mutants (Fig. 1D). These results do not exclude the possibility that rDNA array size does not influence the segregation of the rDNA at all, but our findings do suggest that repeat length within the range tested does not impose a need for CDC14 on rDNA segregation.

Recombination does not cause rDNA segregation defects in cdc14-3 mutants.

The rDNA locus can undergo contraction and expansion (reviewed in reference 15), which are mediated by recombination and, in yeast, involve the replication fork barrier protein Fob1 (7, 19-21, 23). Deletion of FOB1 does not improve rDNA segregation in cdc14-3 mutants and in fact slightly worsened the process (Fig. 1A) (37). Inactivation of the homologous recombination machinery (by deleting RAD52 [37]) or of the nonhomologous end-joining pathway (by deleting DNL4) or the single-strand annealing recombination pathway (by deleting RAD1) did not allow the nucleolus to segregate efficiently in cdc14-3 mutants either (see Fig. S1 in the supplemental material; also data not shown). Nor did double and triple mutant combinations affect rDNA segregation in cdc14-3 mutants (see Fig. S1C in the supplemental material). To control for the possibility that inactivation of recombination factors themselves causes rDNA segregation defects, we examined the consequences of deleting either RAD52 or DNL4 or RAD1 in cdc15-2 mutants, which like the cdc14-3 mutants arrest in anaphase but do not exhibit rDNA segregation defects (6). Inactivation of these recombination factors did not affect rDNA segregation in a cdc15-2 mutant (see Fig. S1 in the supplemental material; also, data not shown). Our results indicate that neither homologous recombination, nor nonhomologous end joining, nor single-strand annealing generates the linkages between sister chromatids at the rDNA that need resolving by Cdc14.

Deletion of CFI1/NET1 partially suppresses the rDNA segregation defect of cdc14-3 mutants.

Cfi1/Net1 binds to the nontranscribed spacer (NTS) regions 1 and 2 found in each rDNA repeat (17). There, the protein is required not only to hold the protein phosphatase Cdc14 in an inactive state (33, 43) but also for assembling rDNA silencing complexes, as well as for efficient Pol I transcription (17, 32, 36). Because deletion of CFI1/NET1 disrupts nucleolar organization (32, 36), we examined the segregation behavior of rDNA GFP dots. Deletion of CFI1/NET1 significantly ameliorated the segregation defect of the rDNA GFP dots in cdc14-3 mutants (Fig. 2A and B). This rescue was not due to the inactivation of CFI1/NET1 suppressing the temperature sensitive cdc14-3 allele (Fig. 2C). It was, however, possible that rDNA segregation required lower levels of Cdc14 activity than viability at 37°C. We therefore also examined the effects of deleting CFI1/NET1 in cells depleted for Cdc14. The open reading frame of CDC14 was fused to a ubiquitin-arginine-lacZ fusion (32) and cloned under the control of the glucose-repressible GAL1-10 promoter (GAL-URL-CDC14). The fusion was efficiently depleted within 1 h in the presence of glucose (Fig. 2D) and led to the accumulation of anaphase cells with unsegregated rDNA loci (Fig. 2E and F). Deletion of CFI1/NET1 allowed rDNA segregation to occur more efficiently in Cdc14-depleted cells (Fig. 2F). Together, these results suggest that Cfi1/Net1's role in nucleolar organization, chromatin structure, and/or rRNA transcription rather than its role in regulating Cdc14 activity was responsible for efficient rDNA segregation in cdc14-3 cfi1/net1Δ mutants.

FIG. 2.

Deletion of CFI1/NET1 rescues the rDNA segregation defect in cells lacking CDC14. (A) Examples of unsegregated (top) and segregated (bottom) rDNA GFP dots in anaphase cells. GFP dots are shown in green and DNA in blue. (B) cdc14-3 (A14617), cdc15-2 (A9972), cdc14-3 cfi1Δ (A14619), and cdc15-2 cfi1Δ (A14620) cells all carrying rDNA GFP dots were incubated at 37°C for 140 min in YEPD medium, and anaphase cells were scored for segregation of the rDNA-GFP dots as described in Materials and Methods. (C) Serial dilutions of wild-type (A2587), cdc14-3 (5321), cfi1Δ (A1536), and cdc14-3 cfi1Δ (A1665) strains grown on yeast extract-peptone-dextrose (YEPD) plates at 35°C for 2 days. (D to F) Wild-type (A14615; ▪), cfi1Δ (A15714, □), GAL-URL-3HA-CDC14 (A15713, •), and GAL-URL-3HA-CDC14 cfi1Δ (A15715, ○) cells all carrying rDNA GFP dots were incubated in YEPD medium at 30°C to repress production of Cdc14. The efficiency of the Cdc14 depletion was determined by Western blot analysis with protein samples from GAL-URL-3HA-CDC14 (A15713) (D) and GAL-URL-3HA-CDC14 cfi1Δ (A15715) (D). Vph1 was used as a loading control in the Western blot analysis. (E) At least 50 cells at each indicated time point were counted to determine the percentage of anaphase cells, which were defined as dumb-bell-shaped cells with fully segregated DAPI masses, as shown in Fig. 2A. (F) At least 50 of these anaphase cells were counted to determine the percentage of segregated GFP dots in anaphase cells at the designated times.

Inhibition of transcription allows rDNA segregation to occur in cdc14-3 mutants.

CFI1/NET1 is responsible for establishing a specialized silent chromatin structure at the rDNA. However, the silenced nature of this region is not likely to impose a need for Cdc14 on its segregation because neither deletion of the silencing factor SIR2 (3, 12, 34) nor deletion of Set1 and Dot1, components of histone H3 methyltransferases, which methylate histone H3 on lysine 4 and lysine 79, respectively, affected rDNA segregation in cdc14-3 mutants (2-4, 25) (see Fig. S2 in the supplemental material). We therefore considered the possibility that rRNA transcription, which occurs throughout the cell cycle (10) and which requires CFI1/NET1 to occur efficiently (32), affected rDNA segregation.

Thiolutin is an efficient inhibitor of all three RNA polymerases in yeast, since transcription is strongly inhibited 20 min after the addition of the drug (18). Treatment of wild-type cells with thiolutin for 2 h did not affect the segregation of rDNA GFP dots (Fig. 3A), but it significantly ameliorated the segregation of these GFP dots in cdc14-3 mutants. The segregation of rDNA GFP dots occurred almost to the same extent as in wild-type cells (Fig. 3A), indicating that transcription-dependent processes induce linkages at the rDNA that impose a need for CDC14 on their segregation.

FIG. 3.

Inhibition of transcription by thiolutin treatment rescues the rDNA segregation defects. (A) Wild-type (A14615) or cdc14-3 (A14617) carrying rDNA-GFP dots were grown in YEPD or YEPD containing DMSO (+DMSO) or YEPD containing thiolutin (87 μM; +Thiol) at 37°C for 140 min. At least 50 anaphase cells were scored for segregation of the rDNA-GFP dots as described in Materials and Methods. (B) ycs4-1 cells with rDNA-GFP dots (A15096) were treated as described in panel A, and at least 50 anaphase cells were scored for rDNA GFP dot segregation. Note that the ycs4-1 strain was analyzed at the same time as cells shown in panel A.

CDC14 has been shown to mediate rDNA segregation in a condensin-dependent manner (6, 37, 44), suggesting that these factors function in the same pathway to bring about rDNA segregation. If this were true, the rDNA segregation defects observed in condensin mutants should, like those of cdc14-3 mutants, be ameliorated by thiolutin treatment. Indeed, the addition of thiolutin to temperature-sensitive condensin mutants (ycs4-1) led to the segregation of rDNA GFP dots in 60% of the cells (Fig. 3B). In contrast to cdc14-3 mutants, the rescue of the rDNA segregation defect of condensin mutants was incomplete. This is likely due to the fact that chromosome segregation as a whole is impaired in condensin mutants. Our results indicate that the rDNA segregation defects observed in cdc14-3 and ycs4-1 mutants are suppressed by the inactivation of transcription.

Inactivation of Pol I transcription suppresses the rDNA segregation defect of cdc14-3 mutants.

Transcription of the 18S, 5.8S, and 28S encoding rRNA transcript is mediated by RNA Pol I, whereas that of the 5S transcript is mediated by RNA Pol III. To determine whether it was Pol I-mediated transcription of rRNA rather than transcription at other chromosomal loci that inhibited rDNA segregation in cdc14-3 mutants, we created a conditional allele of the Pol I subunit RPA135. The open reading frame of RPA135 was fused to a ubiquitin-DHFR fusion known as “degron,” which leads to degradation of the fusion protein via the N-end rule pathway at 37°C (reviewed in reference 8). The fusion was then placed under the control of the methionine-repressible MET3 promoter, allowing for the rapid depletion of the fusion in the presence of methionine at 35°C (Fig. 4A). To examine the effects of depleting Rpa135 on rDNA segregation in cdc14-3 mutants, we arrested cells in S phase using the DNA replication inhibitor hydroxyurea and then released cells into the cdc14-3 block under Pol I-depleting conditions (in the presence of methionine at 35°C). Cells also carried overexpressed ubiquitin ligase Ubr1 responsible for N-end rule mediated protein degradation (reviewed in reference 41) to more efficiently deplete cells of the Pol I subunit. Rpa135 was depleted upon inhibition of RPA135 transcription and transient overexpression of the ubiquitin ligase (Fig. 4A). Upon release from the HU block, cells progressed through metaphase and entered anaphase irrespective of whether Rpa135 was present or not (Fig. 4B). We did, however, notice that anaphase spindles appeared fragile and frequently broken in both strains at later time points. Importantly, segregation of rDNA GFP dots occurred much more efficiently in cdc14-3 mutants lacking Rpa135 than in cdc14-3 mutants with an intact polymerase I (Fig. 4C). Similar results were obtained when we analyzed the segregation of the entire nucleolus using Nop1 staining as a means of visualizing the nucleolus (Fig. 4C). We did note, however, that the fraction of cells that had segregated the nucleolus as observed using an anti-Nop1 antibody was lower than that that had segregated rDNA GFP dots. This is due to the fact that the segregation of Nop1 masses was only scored when two clearly distinct Nop1 positive masses of equal size were present in cells with intact anaphase spindles. In contrast, rDNA GFP dots were considered segregated when the two GFP dots were separated to opposite ends of the bilobed nucleus, which also occurs in cells where the nucleolus has not yet completely partitioned. Our findings indicate that rRNA transcription antagonizes rDNA segregation in cdc14-3 mutants. The finding that Nop1 staining remained intact during the anaphase after the depletion of RPA135 furthermore not only indicates that inhibiting rRNA transcription does not immediately lead to disassembly of all nucleolar structures but also implies that the nucleolus itself, at least as defined by Nop1 staining, is not the reason why this chromosomal region requires CDC14 for its segregation.

FIG. 4.

Depletion of Rpa135 allows rDNA segregation in cdc14-3 mutants. cdc15-2 GAL-UBR1 MET-Degron-FLAG-RPA135 (A14237; ▪), cdc14-3 GAL-UBR1 MET-Degron-FLAG-RPA135 (A15691; ⋄), GAL-UBR1 (A15692; □), and cdc14-3 GAL-UBR1 (A15693; ⧫) cells all carrying rDNA GFP dots were arrested for 3 h in S phase by using 10 mg of hydroxyurea/ml (HU arrest) in synthetic medium lacking methionine containing 2% raffinose and 0.5% glucose at 25°C. These cells were then transferred into YEP medium containing 2% raffinose, 2% galactose, and 8 mM methionine at 35°C with 10 mg of HU/ml for an additional 2 h to deplete Rpa135. Cells were then released into prewarmed YEPD medium containing 8 mM methionine at 35°C (zero time point), and samples were taken at the indicated time points. (A) The depletion of Rpa135 was determined by Western blot analysis with protein samples from cdc15-2 GAL-UBR1 MET-Degron-FLAG-RPA135 (A14237; left panel) and cdc14-3 GAL-UBR1 MET-Degron-FLAG-RPA135 (A15691; right panel) cells. Pgk1 was used as a loading control in Western blot analyses (lower panels). (B) The percentage of cells with metaphase (left panel) and anaphase (right panel) spindles was determined by using tubulin staining at indicated times after the release from the second HU arrest. (C) rDNA GFP dots (left panel) and Nop1 (right panel) were scored for segregation in anaphase cells at the designated time after release from the second HU arrest.

DISCUSSION

In budding yeast, the rDNA and the genomic regions distal to the rDNA locus on chromosome XII segregate late during mitosis and require the protein phosphatase CDC14 to do so (6, 37, 40, 44). This requirement is not simply due to the length of the chromosomal arm carrying the rDNA array (1 to 2 Mb) because a chromosome XII carrying a short array of rDNA repeats (25 copies) also requires CDC14 for its segregation. Conversely, a dramatic increase in repeat number did not worsen rDNA segregation in wild-type or cdc14-3 cells. We therefore conclude that rDNA transcription-dependent processes induce a novel type of linkage that holds this specialized genomic region together. Furthermore, resolution of these transcription-based linkages need specialized pathways, such as the FEAR network-Cdc14-condensin pathway.

Previous studies and the findings presented here indicate that several types of linkages exist at the rDNA. Cohesins hold the rDNA together until their removal at the metaphase-anaphase transition. The inactivation of cohesins, however, does not eliminate the need of CDC14 in rDNA segregation, indicating that additional mechanisms exist that hold the duplicated rDNA together (6). Recombination or rDNA silencing do not appear to create linkages at the rDNA (6, 37). Furthermore, the repetitiveness of the rDNA locus per se does not appear to affect segregation, at least in the range examined here. A 25-repeat rDNA array segregates as poorly as, if not worse than, a 150-repeat array. We note, however, that this result does not exclude the possibility that a certain minimal number of repeats is necessary for rDNA linkages to be established that need CDC14 for their segregation, but our results nevertheless argue against repetitiveness being mainly responsible for establishing rDNA linkages.

Our data not only exclude silencing, recombination, and—to some extent—repetitiveness as the source of linkages at the rDNA that need CDC14 for their resolution but also implicate rRNA transcription and/or processes dependent on it. Reducing the efficiency of rRNA transcription by the deletion of CFI1/NET1 or by complete elimination of rRNA transcription by thiolutin treatment or by the inactivation of RNA Pol I allows cdc14-3 mutants to segregate their rDNA with an efficiency close to that seen in cdc15-2 mutants, which arrest in anaphase with their rDNA segregated. The rescue was nearly complete when cells were treated with thiolutin and significant when Rpa135 was depleted from cells. Although the fact that the rescue was less efficient in Rpa135-depleted cells than in thiolutin-treated cells is likely due to the incomplete depletion of Rpa135, we cannot exclude the possibility that either Pol III-mediated transcription of the 5S RNA contributes to the cohesiveness of the rDNA or that Cdc14 has additional roles in promoting rDNA segregation. The fact that rDNA repeat length does not significantly affect rDNA segregation in cdc14-3 mutants is also consistent with rRNA transcription preventing rDNA segregation. rRNA transcription at each individual repeat is increased when the number of repeats is reduced (11). Conversely, when the number of rDNA repeats is increased, it is possible that transcription of each individual rDNA repeat is decreased (27).

What do we know about the nature of these rDNA transcription-dependent linkages? They appear to be specific to the rDNA locus. The presence of a heavily transcribed gene (a gene transcribed from the strong GAL1-10 promoter) elsewhere in the genome does not affect the segregation of a neighboring tetO array (B. N. Tomson, unpublished observations). However, they are not specific to Pol I-mediated transcription. Cells in which rRNA is transcribed by Pol II also require CDC14 for the segregation of the rDNA. rRNA transcription could induce a local increase in catenation, which needs special pathways for resolution. In this scenario, the segregation defects of telomeric sequences observed in cdc14-3 mutants (6) could be a consequence of transcription-induced catenation events pushed toward the ends of chromosomes during chromosome segregation. We favor the idea that rRNA transcripts themselves and/or factors that assemble onto them (26) establish linkages at the rDNA. Electron microscopic studies of spreads of yeast rDNA genes (also known as Miller spreads) provide a glimpse as to the amount of rRNA and rRNA-binding proteins being concentrated at the rDNA (11, 28, 30). Protein complexes such as the large U3 snoRNP-containing complex called the SSU processome, which is required for the cleavage and maturation of the 18S rRNA (9), assembles in a stepwise manner onto rRNA cotranscriptionally (28). This result also implies that rRNA modifications, which are generally thought to occur prior to cleavage, also occur cotranscriptionally. Although our results imply transcription-dependent processes in establishing linkages between the sister rDNA loci, they argue against the nucleolus itself, whose assembly depends on rRNA transcription (26), imposing a need for CDC14 on rDNA segregation. Upon inactivation of RPA135, the nucleolus did not disassemble, at least as judged by Nop1 staining, but rDNA segregation occurred nevertheless efficiently in cdc14-3 mutants. A model in which the simple presence of many proteins assembled around the rDNA affect its segregation is thus unlikely.

rRNA synthesis is continuous throughout the cell cycle in yeast (10). It is thus not surprising that specialized mechanisms exist that dissolve these linkages. At least one such mechanism involves the protein phosphatase Cdc14 and condensin. Inhibition of rRNA transcription suppressed the rDNA segregation defect of cdc14-3 and condensin mutants, indicating that rRNA-induced linkages either act in parallel to those targeted by the phosphatase and condensin or are the linkages resolved by Cdc14-mediated processes. We favor the latter possibility because the rDNA segregation defect of cdc14-3 mutants is substantially, if not completely, rescued by the elimination of rDNA transcription, arguing for a linear relationship between these two events. How could condensin and CDC14 dissolve transcription-mediated connections between sister rDNA loci? Perhaps condensin targeted to the rDNA by Cdc14 individualizes DNA strands through its condensation function or organizes the rDNA repeats in a way that rRNA transcripts and factors that assemble onto them no longer interfere with segregation. It is also possible that Cdc14 and condensin induce changes in the nucleolar structures that facilitate its segregation and that of the DNA it embeds. Animal cells appear to use a different strategy to resolve transcription-induced rDNA linkages. In these systems, rRNA transcription and ribosome biogenesis are shut down, and the nucleolus is disassembled prior to the onset of mitosis (reviewed in reference 39). It thus appears that the different eukaryotes have evolved different strategies to resolve transcription induced linkages.