Abstract
Transcription is repressed in regions of the fission yeast genome close to centromeres, telomeres, or the silent mating-type cassettes mat2-P and mat3-M. The repression involves the chromo-domain proteins Swi6 and Clr4. We report that two other chromo-domain proteins, Chp1 and Chp2, are also important for these position effects. Chp1 showed a specificity for centromeric regions. Its essentiality for the transcriptional repression of centromeric markers correlates with its importance for chromosome stability. Chp2 appeared more pleiotropic. Its effects on centromeric silencing were less pronounced than those of Chp1, and it participated in telomeric position effects and transcriptional silencing in the mating-type region. We also found that PolII-transcribed genes were repressed when placed in one of the Schizosaccharomyces pombe rDNA clusters, a situation analogous to that in the budding yeast Saccharomyces cerevisiae. Chp2, Swi6, Clr4, and, to a lesser extent, Chp1 participated in that repression.
EUKARYOTIC chromosomes are organized in domains with distinctive properties, among which are the potential for transcription and for recombination. These two parameters can vary along chromosomes according to the DNA sequence, the proximity of specialized structures such as telomeres and centromeres, and the specific state of individual cells. Repetitive sequences and the proteins that interact with them often define regions where transcription and recombination are reduced. For example, heterochromatic areas, which are predominantly composed of repeated sequences, are poorly transcribed. In Drosophila they can silence euchromatic genes translocated nearby (reviewed by Weiler and Wakimoto 1995). In the yeast Saccharomyces cerevisiae, recombination and PolII transcription occur differently in the rDNA, a chromosomal region where repeats occur naturally in many eukaryotes, compared with other genomic regions (see references in Smithet al. 1999).
We are interested in position effects in the fission yeast Schizosaccharomyces pombe and in determining whether repeated sequences influence these effects. S. pombe has three chromosomes of respectively 5.7, 4.6, and 3.5 Mb (Fanet al. 1988; Hoheiselet al. 1993; Mizukamiet al. 1993). Each chromosome contains a 40- to 100-kb centromeric region comprising a 4- to 7-kb central core of unique sequence flanked by large inverted repeats and repeats placed in tandem (Takahashiet al. 1992; Steiner and Clarke 1994 and references therein). Because of their partially symmetrical arrangement relative to the central core, centromeric sequences have been proposed to fold into a hairpin. The rDNA forms two large clusters of ~500–1000 kb at both ends of chromosome 3 (Umesonoet al. 1983; Todaet al. 1984; Allshireet al. 1987; Hoheiselet al. 1993). Telomeres consist of degenerate repeats of the sequence TTACAGG. These repeats are ~300 bp long. They are adjacent to the rDNA genes at both ends of chromosome 3 and to telomere-associated sequences at the ends of chromosomes 1 and 2 (Sugawara 1989).
Position effects on fission yeast transcription have been observed near centromeres, telomeres, and in the vicinity of the mat2-P and mat3-M silent mating-type cassettes in the right arm of chromosome 2. The mat2-P and mat3-M cassettes are linked and 4.3 kb of the DNA that separates them displays striking sequence homology with a centromeric repeat (Grewal and Klar 1997). Marker genes introduced near centromeres, telomeres, or in the mating-type region are repressed, either stringently or in a variegated fashion, by mechanisms that utilize the trans-acting factors Clr1, Clr2, Clr3, Clr4, Swi6, and Rik1 (Lorentzet al. 1992; Thon and Klar 1992; Allshire et al. 1994, 1995; Ekwall and Ruusala 1994; Nimmoet al. 1994; Thonet al. 1994). In addition to repressing transcription, these six factors inhibit meiotic recombination in the mating-type region (Egelet al. 1989; Klar and Bonaduce 1991; Lorentzet al. 1992; Thonet al. 1994), indicating that position effects on transcription and recombination can involve related structures or mechanisms. Whether these silencing factors are also responsible for the reduced recombination observed in centromeric areas (Nakasekoet al. 1986; Chikashigeet al. 1989), or whether they repress tran-scription or recombination at other genomic locations, is not known. In addition to the position effects on transcription, position effects on recombination have been documented in S. pombe by a transplacement study of the ade6-M26 hot spot for recombination which showed that the ability of the M26 mutation to promote recombination in the ade6 gene depends on the chromosomal context (Virginet al. 1995).
Among the proteins that associate with chromatin and determine its properties are proteins containing a chromatin organization modifier (chromo) domain (for review see Cavalli and Paro 1998). This motif was first identified in the Drosophila HP1 and Pc proteins (Paro and Hogness 1991) and subsequently in >40 proteins from various organisms (Aasland and Stewart 1995; Kooninet al. 1995; current sequence databases). It is found, either alone or in combination with other domains, in both repressors and activators of transcription and in several proteins that localize to pericentric heterochromatin. The chromo-domain and the related chromo-shadow domain mediate the formation of protein complexes and their association with chromatin (Messmeret al. 1992; Powers and Eissenberg 1993; Saunderset al. 1993; Plateroet al. 1995; Strutt and Paro 1997). Many potential partners have been proposed to interact with chromo-domain proteins, including components of the nuclear membrane (Ye and Worman 1996; Yeet al. 1997) and components of the origin recognition complex (Paket al. 1997).
Two chromo-domain proteins, Swi6 and Clr4, influence position effects in S. pombe (Lorentzet al. 1992; Ekwall and Ruusala 1994; Thonet al. 1994; Allshireet al. 1995). Swi6 contains a chromo- and chromo-shadow domain (Lorentzet al. 1994; Aasland and Stewart 1995). Clr4 contains a chromo- and SET domain [Su(var)3-9, Enhancer of zeste, trithorax; Tschierschet al. 1994; Ivanovaet al. 1998]. When either swi6 or clr4 is mutated or deleted, genes normally subject to transcriptional silencing near centromeres, telomeres, or in the mating-type region are derepressed. Consistent with these phenotypes, Swi6 localizes at centromeres, telomeres, and in the mating-type region (Ekwallet al. 1995). Clr4 is important for the localization of Swi6 (Ekwallet al. 1996) and, when expressed from a plasmid, is itself seen in the rDNA (Sawin and Nurse 1996). We tested whether two other chromo-domain proteins of S. pombe, Chp1 (Doeet al. 1998) and Chp2, were involved in the same processes as Swi6 and Clr4. In addition to assaying the effects of Chp1 and Chp2 in the regions where silencing was previously observed, we investigated whether Swi6, Clr4, Chp1, or Chp2 modified the efficiency of PolII transcription in the rDNA.
MATERIALS AND METHODS
Sequence of chp1 and chp2: Chp1 (GenBank accession no. Q10103) is encoded in the cosmid SPAC18G6 (Z68198) and Chp2 (CAA16917) in SPBC16C6 (AL021767).
Bacterial strains used for cloning: Cloning of plasmid DNA was performed in the Escherichia coli strains DH5 (Hanahan 1983) or S1754 (Thonet al. 1999).
Cloning of chp1: chp1 is contained within a 4.3-kb BamHI-Ecl136I fragment of genomic DNA. That fragment was purified from cosmid ICRFc60G0618 (Reference Library, ICRF; Lehrach 1990) and cloned into Bluescribe (Stratagene, La Jolla, CA) digested with BamHI and HincII, to create pGT121. The presence of chp1 in pGT121 was ascertained by restriction mapping and partial sequencing.
Partial replacement of chp1 with LEU2 or ura4+: EcoRV cleaves pGT121 75 bp downstream of the predicted initiating ATG of the chp1 open reading frame (ORF) and BglII cleaves pGT121 1355 bp upstream of the predicted stop codon. A total of 1450 bp of pGT121 DNA between those EcoRV and BglII sites were replaced with the 2.2-kb EheI-BamHI fragment of pJJ250 (Jones and Prakash 1990), which contains the S. cerevisiae LEU2 gene, to create pGT122, or with the 1.8-kb HincII-BamHI fragment of pON94 (a gift from Olaf Nielsen), which contains the S. pombe ura4+ gene, to create pGT162. pON94 was constructed by filling in the ends of the 1.8-kb HindIII fragment, which contains ura4+ (Grimmet al. 1988), adding SphI linkers, and ligating the product into the SphI site of a modified version of pUC19 (Yannish-Perronet al. 1985) where the HindIII site was converted to a BamHI site. The insert of pGT122 was released by digesting with BamHI and ScaI and used to transform a diploid strain produced by the mating of PG9 and PG1141. chp1+/chp1Δ::LEU2 diploids were identified among the Leu+ transformants by Southern blot of genomic DNA restricted with SphI or SphI and Nsi and hybridized with the chp1 probe described below (Southern and Northern blot analyses). Two such diploids were subjected to tetrad dissection. The insert of pGT162 was released with BamHI and FspI and used to transform SP837. Ura+ transformants with the chp1Δ::ura4+ allele were identified by Southern blot of genomic DNA digested with NsiI or SphI and SacI and probed with the chp1 probe.
Deletion of chp1 with no substituted marker: pGT121 was digested with BsaAI and AccI, filled in at the ends with the Klenow fragment of E. coli DNA polymerase I, and religated, to create pGT187. The entire chp1 ORF is deleted in pGT187, as well as 33 nucleotides 5′ and 77 nucleotides 3′ to the chp1 ORF. The insert of pGT187 was released with BamHI and SphI and used to transform PG1544. The DNA of FOAR transformants was digested with NsiI or BamHI and Ecl136III and hybridized with the chp1 probe, which allowed us to identify transformants in which the chp1Δ::ura4+ allele of PG1544 had been replaced with the chp1Δ allele.
Replacement of chp2 with ura4+: The 1.8-kb HindIII fragment containing the S. pombe ura4+ gene (Grimmet al. 1988) was cloned in Bluescript KSII(+) (Statagene) with the EcoRV site in the ura4+ gene close to the EcoRV site in the polylinker, to create pGT188. The oligonucleotides GTO-150 (5′ TCCCCCCGGGAGCTCAGATCGTTATACACTTTACGTATCTAGG 3′) and GTO-151 (5′ CGGGATCCGTCGACTATTAAGACTTTCCAGATATACCAAC 3′) were used to amplify 777 bp of genomic DNA on one side of the chp2 ORF. GTO-152 (5′ AACTGCAGTCGACTTGATCTTTGGACTTAATAATTAGAATTACG 3′) and GTO-153 (5′ TCCCCCGGGAAGCTTCTCGAGATGGCGCTTGAAGGGCTTAGTGCGC 3′) were used to amplify 776 bp of genomic DNA on the other side of the chp2 ORF. The oligonucleotides were purchased from Operon Technologies (Alameda, CA). The PCR were performed with Native Pfu DNA polymerase (Stratagene) in 80 μl containing the buffer provided by the manufacturer, 1 μg S. pombe genomic DNA, 1 μm of each primer, and 0.5 mm dNTPs. The amplification cycles were as follows: 3 min 94°; 15 times [1 min 94°, 1 min 60°, 1 min 72°]; 7 min 72°. The PCR products were gel purified. The product amplified with GTO-150 and GTO-151 was digested with SacI and BamHI. The product amplified with GTO-152 and GTO-153 was digested with SalI and SmaI. These two fragments were ligated in a four-way ligation with Bluescript SKII(−) (Stratagene) digested with SacI and SmaI and with the 1.8-kb BamHI-SalI fragment of pGT188, which contains ura4+. A clone containing the ura4+ gene flanked by sequences normally flanking the chp2 ORF was obtained from this ligation and named pGT190. The insert of pGT190 was released with SacI and XhoI and used to transform SP837. The DNA of Ura+ transformants was digested with HindIII or BamHI and PstI and hybridized to the chp2 probe described below (Southern and Northern analyses). This identified a strain with the chp2Δ::ura4+ allele, PG1744.
Replacement of chp2 with LEU2: The 3.7-kb BamHI-PstI fragment and the 0.8-kb PstI-SalI fragment of pGT190 were ligated in a three-way ligation with the 2-kb BamHI-SalI of pJJ250, which contains the S. cerevisiae LEU2 gene (Jones and Prakash 1990). A clone similar to pGT190, but containing LEU2 instead of ura4+, was obtained from this ligation, pGT193. The insert of pGT193 was released by digesting with SacI and XhoI and used to transform a diploid strain produced by the mating of PG1636 and PG1637. chp2+/chp2Δ::LEU2 diploids were identified among the Leu+ transformants by a Southern blot analysis similar to that described above for the chp2Δ::ura4+ allele and two such diploids were subjected to tetrad dissection.
Deletion of chp2 with no substituted marker: pGT190 was digested with SalI. The 3.7-kb SalI fragment, which contains Bluescript and 777 bp of DNA flanking chp2 on its centromere-proximal side, was ligated with the 0.8-kb SalI fragment, which contains 776 bp of DNA flanking chp2 on its centromere-distal side. A clone with a precise deletion of the chp2 ORF was obtained from this ligation, pGT191. The insert of pGT191 was released by digesting with SacI and XhoI and used to replace the chp2Δ::ura4+ allele of PG1753. The DNA of FOAR transformants was digested with ClaI or BamHI and PstI and hybridized to the chp2 probe. This identified a strain with the chp2Δ allele, PG1776.
Introduction of ura4+ in the rDNA repeats: The 1.8-kb HindIII fragment containing ura4+ (Grimmet al. 1988) was cloned into pUC8 (Viera and Messing 1982). The resulting construct was linearized with BsrFI and used to transform PG1634. Ura+ transformants were analyzed by Southern blots of genomic DNA restricted with EcoRV, HindIII, or BamHI and hybridized with a LEU2 or ura4 probe. A transformant with pUC8 and ura4 integrated in the pBR322 sequence present in the rDNA repeat of PG1634 was saved as PG1661.
Yeast media and culture conditions: YES (Thon and Friis 1997) was used as rich medium; MSA (Egelet al. 1994) supplemented with 100 mg adenine, 100 mg uracil, and 200 mg l-leucine per liter was used as sporulation medium; drop-out media (Roseet al. 1990), FOA medium (Thonet al. 1999), and YE (Morenoet al. 1991) were used to test auxotrophies. Plates were incubated at 33°.
Yeast transformation and strain construction: A lithium acetate protocol adapted from Heyer et al. (1986) and Moreno et al. (1991) was used for transforming S. pombe. Strains transformed with constructs containing LEU2 or ura4+ were selected on drop-out media lacking leucine or uracil. FOA medium was used to select for loss of ura4+. The strains produced by transformation and their progeny from subsequent crosses are listed in Table 1.
Southern and Northern blot analyses: S. pombe DNA was prepared according to Moreno et al. (1991) and RNA according to Schmitt et al. (1990) from liquid YES cultures incubated at 30°. Blotting and hybridization conditions were as described previously (Thonet al. 1999). The probes used for Southern hybridization were made with a random priming kit (Promega, Madison, WI), 3000 Ci/mmol [α-32P]dCTP from Amersham (Piscataway, NJ), and the following templates: the BamHI-SphI fragment of pGT121 for chp1; a purified PCR product amplified from genomic DNA with GTO-150 and GTO-153 for chp2; the 1.8-kb HindIII fragment for ura4 (Grimmet al. 1988); and the 2-kb BamHI-SalI of pJJ250 for LEU2 (Jones and Prakash 1990). Antisense ura4 riboprobes were made for Northern hybridization with a Riboprobe II core system (Promega), 3000 Ci/mmol [α-32P]UTP (Amersham), T7 RNA polymerase (Promega), and pON140 (a gift from Olaf Nielsen) linearized with BamHI as template. pON-140 is a clone of the ura4+ gene in pGEM4 (Promega).
Stability of Ura− phenotype: FOAR colonies were purified by two rounds of streaking on FOA-containing plates and patched on YES plates. Following growth under nonselective conditions on YES for ~15 generations, cells were suspended in water and used for spot tests on selective media or for DNA preparation.
Chromosome stability: Mitotic loss of the minichromosome Ch16m23::ura4+-Tel[72] was assayed as that of the minichromosome Ch16 (Allshireet al. 1995): cells containing the ade6-M210 allele on chromosome 3, the complementing allele ade6-M216 on Ch16m23::ura4+-Tel[72], and mutations of interest were propagated in medium lacking adenine (AA-ade) at 33°, plated on medium containing a small amount of adenine (MSA supplemented with 100 mg uracil, 200 mg l-leucine, and 15 mg adenine per liter), and incubated at 33°. The loss rate of Ch16m23::ura4+-Tel[72] was determined as the number of colonies with a red sector equal to or greater than half the colony divided by the sum of white and sectored colonies.
RESULTS
Deletion of two chromo-domain protein genes, chp1 and chp2, in S. pombe: A gene capable of encoding a 109-kD protein with an N-terminal chromo-domain is present in the left arm of S. pombe chromosome 1. That gene was the subject of a previous study and named chp1 (chromo-domain protein in S. pombe; Doeet al. 1998). Another gene, which we will refer to as chp2, is located in the right arm of chromosome 2 and encodes a 43-kD protein displaying extensive sequence similarity with the S. pombe Swi6 protein (Figure 1). Like Swi6 and a family of proteins related to Drosophila HP1, Chp2 contains both a chromo- and chromo-shadow domain. The chp1 and chp2 ORFs were both discovered in the S. pombe genome sequencing project (Sanger Center).
We created several deletion alleles of chp1 and chp2: part of the chp1 or the entire chp2 ORF was replaced with the S. cerevisiae LEU2 gene, a functional homolog of the S. pombe leu1+ gene, or with the S. pombe ura4+ gene. Deletion of the entire ORF with no substituted markers was also obtained. The deletions were originally introduced in the chromosome of diploid cells. The chp1 and chp2 deletions both proved viable in the haploid progeny of the transformed diploids. This had been observed independently for chp1 (Doeet al. 1998). chp1-deleted cells formed colonies significantly smaller than the wild type, whereas chp2-deleted cells formed colonies as large as the wild type.
Deleting chp1 or chp2 alleviates centromeric position effects: S. pombe centromeres are organized in large inverted repeats (Takahashiet al. 1992; Steiner and Clarke 1994 and references therein). The ura4+ gene was placed at several locations within the centromere of chromosome 1 by Allshire et al. (1994, 1995). Insertion at most locations results in a level of expression inferior to that of the ura4+ gene at its wild-type location and in a variegated phenotype where clonally derived fractions of the cell population express ura4+, whereas the rest of the cells keep it repressed. Swi6 and Clr4 are required for this centromeric repression (Allshireet al. 1995).
Strains and their genotypes
Sequence comparison of Chp2 and Swi6. The predicted protein sequences of Chp2 and Swi6 were aligned using the CLUSTAL W program (Thompsonet al. 1994).
We tested the effect of Chp1 and Chp2 on the expression of ura4+ at six centromeric integration sites (Figure 2). In the first experiment, we compared the ability of wild-type, chp1-, or chp2-deleted cells containing a centromeric ura4+ gene to form colonies on medium lacking uracil or medium containing FOA, a toxigenic substrate of the Ura4 protein (Grimmet al. 1988; Figure 2A). In the second experiment, we examined the amount of ura4+ transcript present in cells of the same strains by Northern blot analysis (Figure 2B). Transcripts originating from the ura4DS/E allele (Allshireet al. 1994), a truncated ura4 gene occupying its natural chromosomal location, were used as an internal control.
In wild-type cells, the tightest repression of ura4+ was observed in the outer centromeric repeats (otr1L) within the dg1 region. Insertion within the central area (cnt) or the inner most repeats (imr1L) allowed a higher expression of ura4+. These phenotypes agree with those reported by Allshire et al. (1994, 1995). Some strains (FY939, FY967, FY501, FY336) grew more poorly on FOA medium in our experiments than originally observed (compare Figure 2A with Figure 1C and Figure 2B in Allshireet al. 1995). We believe that this is due to a difference in the FOA-containing media used. The amount of ura4+ transcript relative to the control transcript from ura4-DS/E was similar in both studies. As noted before (Allshireet al. 1995), an amount of ura4+ transcript inferior to that found in the wild type could support effective growth on medium lacking uracil.
Deletion of chp1 completely derepressed five centromeric loci and caused a small derepression at cnt1, the most central centromeric insertion site (Figure 2, A and B). Deletion of chp2 partially derepressed ura4+ at the six centromeric sites tested. It increased growth on medium lacking uracil and reduced growth on medium containing FOA (Figure 2A). It also increased the amount of ura4+ transcript detected in Northern blot relative to the amount arising from the same sites in wild-type cells (Figure 2B). However, the derepression was strikingly small compared with that caused by deletion of chp1. This was true for all sites examined except possibly cnt1.
The FOA resistance of strains containing a ura4+ centromeric marker was generally reduced by deletion of chp1 or chp2, but we noted that approximately one cell in a thousand was able to generate a full-sized colony on FOA-containing medium; the frequency of such cells varied from culture to culture. chp+ strains gave rise to a similar kind of papillation (Figure 2A). We investigated the nature of the FOA resistance by propagating cells from well-growing FOAR colonies under nonselective conditions and by measuring their ability to form colonies in the absence of uracil. We also performed Southern blots to test for possible alterations of the ura4 gene. Two independent FOAR colonies from each strain displayed in Figure 2, 1–6, were tested in this manner. We found that the FOA resistance was due to a loss of the ura4+ gene in all strains examined except for FY986 and FY648. In FY986 and FY648, the FOA-resistant state returned efficiently to Ura+ and the ura4+ gene appeared unchanged in Southern blots. This epigenetically controlled resistance is likely to have masked deletion events such as those occurring in the other strains tested.
Effect of Chp1 and Chp2 on gene expression near centromere 1. (A) Expression of the ura4+ gene assayed by plating efficiency on selective media. Serial 10-fold dilutions of cell suspensions were spotted on the indicated media. The cells contained centromeric insertions of ura4+ that are depicted in B (1–6), a random integration of ura4+ (R.I.), or no full-length ura4 gene (−). The strains were: 1: +, FY939; chp1Δ, PG1699; chp-2Δ, PG1767; 2: +, FY967; chp1Δ, PG1703; chp2Δ, PG1769; 3: +, FY986; chp1Δ, PG1709; chp2Δ, PG1771; 4: +, FY648; chp1Δ, PG1717; chp2Δ, PG1764; 5: +, FY501; chp1Δ, PG1691; chp2Δ, PG1765; 6: +, FY336; chp1Δ, PG1721; chp2Δ, PG1763; R.I., FY340; −, FY489. (B) Northern blot analysis of the ura4+ transcript. RNA was prepared from the strains displayed in A and the amount of ura4+ transcript was estimated by hybridization to a ura4 riboprobe.
Deletion of chp2, but not of chp1, alleviates telomeric position effects: Ch16 is a 530-kb S. pombe minichromosome derived from chromosome 3 by radiation-induced breakage (Matsumotoet al. 1987). It contains the ade6-M216 allele, which allows one to monitor its presence in cells bearing the ade6-M210 allele due to the intragenic complementation between the ade6-M216 and ade6-M210 alleles (Gutz 1963). A truncated version of Ch16, Ch16m23::ura4+-Tel[72], where the ura4+ gene is ~1 kb from the telomere, was constructed in two steps (Niwaet al. 1989; Nimmoet al. 1994). ura4+ is expressed in a variegated manner in Ch16m23::ura4+-Tel[72] (Nimmoet al. 1994). Clr4 and, to a lesser extent, Swi6 contribute to the repression of ura4+ at that location (Allshireet al. 1995). We crossed the Ch16m23::ura4+-Tel[72] minichromosome into, respectively, chp1- and chp2-deleted strains. Crosses with swi6-115 and clr4-681 were performed at the same time for comparison. We assayed the expression of ura4+ in three mutant and three wild-type strains issued from each cross by spot tests on selective media. Strains with identical genotypes proved to have very similar plating efficiencies on the various selective plates (data not shown). Spot tests were repeated with one strain of each genotype and the amount of ura4+ transcript was estimated by Northern blot. As can be seen in Figure 3, deleting chp1 had no influence on telomeric position effect, but deleting chp2 an increased expression of ura4+. The increased expression caused by the chp2 deletion was similar to that caused by a mutation in clr4.
Role of S. pombe chromo-domain proteins in chromosome segregation: We measured the rate of loss of the minichromosome Ch16m23::ura4+-Tel[72] in wildtype, chp1Δ::LEU2, chp2Δ::LEU2, swi6-115, and clr4-681 cells. The results are presented in Table 2. swi6-115 and chp1Δ::LEU2 increased loss of Ch16m23::ura4+-Tel[72], in accordance with previous reports (Allshireet al. 1995; Doeet al. 1998). In contrast, deletion of chp2 or the clr4-681 mutation had a modest and possibly not significant effect on the rate of loss compared with the wild type. We constructed strains containing pairwise combinations of the chp1Δ::LEU2, chp2Δ::LEU2, swi6-115, and clr4-681 alleles. We found that all double mutants were viable and that the rate of loss of Ch16m23:: ura4+-Tel[72] was greater in the chp1Δ::LEU2 swi6-115 and chp1Δ::LEU2 clr4-681 double mutants than it was in any of the single mutants (Table 2). For all other combinations tested, the rate of loss was similar to that caused by the stronger mutation in the pair. The chp2 deletion did not increase chromosome loss in any of the backgrounds tested. The minichromosome loss assay we used relies on measuring changes from a white Ade+ phenotype displayed by cells containing the minichromosome to a red Ade− phenotype displayed after chromosome loss (materials and methods; Allshireet al. 1995). In all mutant strains tested, we noticed the presence of colonies that were mostly red with a narrow white sector (Table 2). This phenotype is indicative of multiple consecutive missegregation events and it is expected to occur very infrequently according to our calculated rates of chromosome loss. The existence of such a phenotype suggests that the mutations tested destabilize chromosomes in a heritable manner in a fraction of the cell population. This property could also account for the relatively large variations in loss rates observed in independent cultures of the same strain (Table 2). Alternatively, red colonies with a narrow white sector could be produced if Ade+ cells temporarily stopped dividing after a missegregation event.
Deletion of chp2, but not of chp1, derepresses transcription in the mating-type region: The mating-type region of S. pombe comprises three linked cassettes, mat1, which is transcribed, and mat2-P and mat3-M, which are silenced (Egel and Gutz 1981; Beach 1983). Silencing extends to the regions flanking mat2-P and mat3-M and to the 10.9-kb region that separates the two cassettes (Thon and Klar 1992; Thonet al. 1994; Grewal and Klar 1997). It affects the ura4+ gene when placed at an XbaI site located ~400 bp centromere-distal to the mat2-P cassette [mat2-P(XbaI)::ura4+ allele; Thonet al. 1994; Figure 4A] or at an EcoRV site located ~150 bp centromere-distal to the mat3-M cassette [mat3-M(Eco-RV)::ura4+ allele; Thon and Klar 1992; Figure 4B]. The products of clr1, clr2, clr3, clr4, clr6, swi6, rik1, esp1, esp2, and esp3 are essential for full repression. These factors seem to act in partially redundant pathways, such that double mutations are required to completely abolish silencing within the mating-type region. Double mutations with strong synergistic effects can either reside in two trans-acting factors not belonging to the same class or consist in a combination of cis- and trans-acting mutations (Thon et al. 1994, 1999; Thon and Friis 1997; Grewalet al. 1998). We examined whether Chp1 or Chp2 contributed to the repression of ura4+ placed near mat2-P or mat3-M, or to the repression of the mating-type genes at mat2-P or mat3-M. The tests were performed in strains with or without deletion of cis-acting silencing elements.
First, we assayed ura4+ expression from the mating-type region in cells lacking chp1 or chp2 by spot tests on selective media (Figure 4). Wild-type, swi6-115, and clr4-681 cells were spotted on the same plates for comparison. Cells lacking Chp1 grew as well on FOA and as poorly on medium lacking uracil as cells containing a functional protein, indicating that Chp1 had no effect on transcription of the ura4+ gene placed in the mating-type region. In contrast, lack of Chp2 allowed increased growth in the absence of uracil, indicating that Chp2 was required for the repression of ura4+ near mat2-P and mat3-M. In addition to forming colonies in the absence of uracil, cells deleted for chp2 and containing ura4+ near mat2-P could also form colonies on FOA-containing medium, indicating that deletion of chp2 did not derepress ura4+ near mat2-P as efficiently as mutations in swi6 or clr4 (Figure 4A, top). A stronger derepression of ura4+ was obtained by simultaneously deleting chp2 and a mat2-P cis-acting element (Figure 4A, bottom). The derepression seen at mat3-M following deletion of chp2 appeared equivalent to that in swi6 and clr4 mutants (Figure 4B).
Effect of Chp1, Chp2, Swi6, and Clr4 on gene expression near a telomere. (A) Spot test analysis. Suspensions of cells with the minichromosome Ch16m23::ura4+-Tel[72] and the indicated mutations were spotted as in Figure 2 to estimate the level of expression of ura4+. m23-TEL[72] strains: +, FY520; chp1Δ, PG1732; chp2Δ, PG1773; swi6-115, FY611; clr4-681, PG1731; m23, FY489; −, FY521. (B) Northern blot analysis. RNA was prepared from the strains plated in A and hybridized to a ura4 riboprobe.
Wild-type S. pombe cells efficiently switch the content of their mat1 cassette between P and M by copying the silent information present at mat2-P and mat3-M. Following these switches, cells can mate and form zygotic asci within isolated colonies. Strains unable to switch their mat1 allele usually form colonies that do not contain spores. Under particular circumstances, such as when both mating types are coexpressed in single cells, haploid cells can sporulate without mating in a process referred to as haploid meiosis. Therefore, sporulation in unswitchable mat1-M cells can be used to monitor mat2-P expression and sporulation in unswitchable mat1-P cells can be used to monitor mat3-M expression. We examined various strains for the presence of haploid sporulation by microscopic examination (data not shown) and by exposing colonies to iodine vapors, which selectively stains spores black (Breschet al. 1968; Figure 5). According to this assay, deletion of neither chp1 nor chp2 significantly derepressed the silent cassettes. Deletion of chp2, however, had a synergistic effect with deletion of either a mat2-P or a mat3-M centromere-proximal silencing element. A qualitatively similar effect was observed in swi6 or clr4 mutant strains (Thon et al. 1994, 1999; Figure 5), but not following deletion of chp1 (Figure 5). Hence, Chp2, but not Chp1, appears to play a role similar to the other chromo-domain proteins Swi6 and Clr4 in silencing transcription of the mating-type region.
Effect of Chp1, Chp2, Swi6, and Clr4 on the stability of a minichromosome a
Effects of Chp1, Chp2, Swi6, and Clr4 on the expression of ura4+ from the mating-type region. (A) Effects near mat2-P. Cells with ura4+ at the XbaI site centromere-distal to mat2-P (Xb) were spotted on the indicated media as in Figure 2. (Top) Strains have intact mat2-P silencing elements. (Bottom) Strains have a mat2-P centromere-proximal deletion of ~1.5 kb between a BglII (Bg) and a BssHII (Bs) site. mat2-P(XbaI)::ura4 strains: +, SP1124; chp1Δ, PG1605; chp2Δ, PG1790; swi6-115, SP1126; clr4-681, PG1032; Δ(B-B)mat2-P(XbaI)::ura4 strains: +, SP1151; chp1Δ, PG1639; chp2Δ, PG1787; swi6-115, SP1640; clr4-681, SP1165. (B) Effects near mat3-M. Cells with the ura4+ gene at the EcoRV site centromere-distal to mat3-M were spotted as in Figure 2. +, PG9; chp1Δ, PG1454; chp2Δ, PG1783; swi6-115, PG596; clr4-681, PG912.
Effects of Chp1, Chp2, Swi6, and Clr4 on the expression of mat2-P and mat3-M. (A) Effects on mat2-P. Sporulated colonies of stable mat1-M strains grown on MSA plates supplemented with leucine, uracil, and adenine were exposed to iodine vapors and photographed. Strains in the top row have intact mat2-P cis-acting elements. Strains in the bottom row have the 1.5-kb deletion depicted in Figure 4A. mat2-P strains: +, SP1124; chp1Δ, PG1605; chp2Δ, PG1790; swi6-115, SP1126; clr4-681, PG1032; Δ(B-B)mat2-P strains: +, SP1151; chp1Δ, PG1639; chp2Δ, PG1787; swi6-115; PG1640; clr4-681, SP1165. (B) Effects on mat3-M. Sporulated colonies of stable mat1-P strains were photographed as in A. Strains in the top row have intact mat3-M cis-acting elements. Strains in the bottom row have the 482-bp deletion depicted in Figure 4B. mat3-M strains: +, PG445; chp1Δ, PG1796; chp2Δ, PG1799; swi6Δ, PG1797; clr4-681, PG1594; Δ(482)mat3-M strains: +, PG1402; chp1Δ, PG1779; chp2Δ, PG1778; swi6Δ, PG1781; clr4-681, PG1419.
We constructed six mat1-Msmt-0 strains containing all pairwise combinations of the chp1Δ::LEU2, chp2Δ::LEU2, swi6-115, and clr4-681 alleles. These strains were used to monitor the levels of expression of mat2-P, as revealed by haploid meiosis after growth on limiting nitrogen. The strains were examined by iodine staining and microscopic examination. None of the combinations led to an easily observable derepression of mat2-P (data not shown). This is consistent with our previous conclusion that Chp1 does not affect transcriptional silencing in the mating-type region and indicates that Chp2, Swi6, and Clr4 act in a single pathway.
Effects on mating-type switching: Switching to the opposite mating type is a regulated and extremely efficient process in homothallic strains with a mat1 mat2-P mat3-M mating-type region, also designated h90. Swi6 and Clr4 are required for that process (Egelet al. 1984; Gutz and Schmidt 1985; Thon and Klar 1993; G. Thon and A. J. S. Klar, unpublished results; Ivanovaet al. 1998). More specifically, Swi6 and Clr4 seem to be important for the choice of the silent cassette that converts mat1 since mutations in swi6 or clr4 not only reduce switching in h90 cells, but increase switching in cells with an engineered mat1 mat2-M mat3-P mating-type region (h09 cells; Thon and Klar 1993). Iodine staining and microscopic examination of the type of asci produced within a colony allow for estimating the efficiency of mating-type switching. High switching rates lead to homogenous mixtures of P and M cells within colonies and hence to a high efficiency of zygotic ascus formation and to a dark iodine staining phenotype. On the other hand, low switching rates lead to a reduced frequency of zygotic asci and to speckled iodine staining phenotypes. Because of the resemblance between Chp2 and Swi6, we investigated whether deleting the chp2 gene affected mating-type switching in h90 or h09 cells, in two different backgrounds known to affect switching, swi6-mod+ and swi6-mod− (Thon and Klar 1993). We found that the iodine staining and sporulation pheno-types of wild-type and chp2-deleted strains were similar (data not shown), an indication that mating-type switching can occur efficiently in the absence of Chp2. Mating-type switching also occurs normally in the absence of Chp1 as judged by the iodine staining phenotype of chp1-deleted h90 colonies (Doeet al. 1998; data not shown).
Effect of swi6, clr4, chp1, and chp2 on the transcription of PolII genes in the rDNA repeats: Genes for the 25S, 18S, and 5.8S ribosomal RNAs are present in 70–100 copies in the genome of S. pombe (Hoheiselet al. 1993). Most of these genes are organized in two large clusters located at both ends of chromosome 3 and consisting in tandem repeats of a 10.4-kb unit (Schaaket al. 1982; Todaet al. 1984; Hoheiselet al. 1993). The S. cerevisiae LEU2 gene and plasmid DNA were integrated in the rDNA by Toda et al. (1984) for the purpose of mapping the rRNA genes. We observed that the LEU2 marker was expressed rather poorly in a strain with such an insertion, YIp2.4-1. When replicated from complete medium to medium lacking leucine, colonies with the YIp2.4 insertion did not grow evenly (Figure 6A). Growth was limited to sectors of the colonies, giving rise to wheel-like patterns. This phenotype was suppressed by deleting chp1, chp2, swi6, or by a mutation in clr4 (Figure 6A; data not shown).
We introduced the S. pombe ura4+ gene near the LEU2 gene of YIp2.4 (Figure 6B). The integration was obtained in a swi6-115 background and tested by Southern blots (data not shown). As expected, the integrated ura4+ gene displayed tight linkage to LEU2 in tetrad dissections (0 recombinants/45 viable spores examined). We compared its expression with the expression of ura4+ integrated at a random site (Allshireet al. 1994; Figure 6, C and D) and with the expression of the truncated ura4-DS/E allele.
In a background with no known silencing defect, expression of ura4+ from the rDNA supported only limited growth on medium lacking uracil (Figure 6C). Many cells could form colonies on FOA-containing plates (Figure 6C). We tested whether the FOA resistance was due to genetic alteration of the ura4+ gene or to an epigenetic phenomenon by isolating FOA-resistant colonies from 24 independent cultures and determining whether they could give rise to Ura+ colonies after growth on nonselective medium. Out of 24 independent FOA-resistant cultures, 17 displayed a phenotype undistinguishable from their parental Ura+ strain upon replating on selective media (Figure 6C, row 1; data not shown).
We found that the four chromo-domain proteins Chp1, Chp2, Swi6, and Clr4 participated in the repression of LEU2 and ura4+ integrated in the rDNA (Figure 6). Altering the genes encoding these proteins by deletion or mutation improved growth on media lacking leucine or uracil (Figure 6C) and led to an accumulation of ura4+ transcripts (Figure 6D). Chp2, Swi6, and Clr4 had the most obvious effects and Chp1 had a smaller but consistently reproducible effect. The ura4+ transcript originating from the rDNA was of the same size as that from the wild-type ura4+ gene, indicating that initiation and termination of transcription occurred normally. We determined by Southern blot analysis that the increased expression of ura4+ was not caused by amplification of the ura4+ gene (Figure 6E). We tested the stability of the Ura− phenotype in chp1Δ, chp2Δ, swi6-115, or clr4-681 colonies formed on FOA. The 24 independent FOA-resistant derivatives tested for each chp2Δ, swi6-115, and clr4-681 had a frequency of reversion to Ura+ lower than 1 in 106 cells and 9 tested by Southern blot proved to have lost the ura4+ gene. Among the 24 independent FOA-resistant isolates of chp1-deleted cells, 13 had a frequency of reversion to Ura4+ lower than 1 in 106 and 11 could return to Ura+ very efficiently. Hence, the FOA resistance in chp1-deleted cells could result either from mutations in or loss of the ura4+ gene (in this case, loss was not tested by Southern blot), or from reversible silencing of the ura4+ gene. This phenotype is similar to the phenotype of chp+ cells and different from the phenotype of chp2Δ, swi6-115, or clr4-681 cells, where reversible silencing of ura4+ could not be observed. We conclude from this experiment that the epigenetic repression of ura4+ in the rDNA is dependent to a small extent on Chp1 and more significantly on Chp2, Swi6, and Clr4.
Position effects in the rDNA repeats. (A) Expression of the S. cerevisiae LEU2 gene in chp1+ and chp1− cells assayed on replica. Colonies of chp1+ (top; YIp2.4-1) or chp1-deleted (bottom; PG1572) cells with an integration of LEU2 in the rDNA were obtained on complete AA medium and replicated onto medium lacking (AA-leu) or containing (AA) leucine as indicated. (B) Integration of ura4+ in the rDNA. The S. pombe ura4+ gene contained in the pUC8 vector was allowed to recombine by homology with the YIp2-4 allele (Todaet al. 1984). The locations of BamHI (Bam), BsrFI (Bs), and HindIII (Hin) restriction sites are indicated. (C) Expression of LEU2 and ura4+ assayed by plating efficiency. Cells containing a LEU2 and ura4+ gene in the rDNA and the indicated mutations were spotted on selective plates as in Figure 2. 1, PG1734; 2, PG1792; 3, PG1794; 4, PG1661; 5, PG1736. (D) Northern blot analysis. RNA was prepared from the strains displayed in C and probed with an antisense ura4 probe: 1–5, as in C; 6, FY489; 7, FY340. (E) Southern blot analysis. DNA was prepared from the cultures used in D, digested with HindIII, blotted, and probed as in D.
DISCUSSION
We have examined the contribution of four chromodomain proteins, Chp1, Chp2, Swi6, and Clr4, to position effects on transcription in the genome of fission yeast. We have determined that PolII-transcribed genes can be repressed when placed in the rDNA cluster located in the right arm of chromosome 3, and that the repression of transcription is dependent on the presence of these four chromo-domain proteins. We have also shown that Chp1 is crucial to centromeric silencing, but not to transcriptional silencing in the mating-type region or near telomeres. Conversely, the major effects of Chp2, in addition to those observed in the rDNA, were in the mating-type region and near telomeres. We will compare these phenotypes with those reported for mutations in swi6 and clr4 and discuss them in the light of the reported subcellular localization of these proteins.
Chp2 vs. Swi6: chp2 and swi6 encode the most closely related of the known S. pombe chromo-domain proteins. Both ORFs are of about the same size and both contain a chromo- and chromo-shadow domain, a structure similar to the HP1 protein of flies and mammals. We found that chp2 and swi6 had overlapping but distinguishable functions. Altering either gene alleviated centromeric and telomeric silencing (Allshireet al. 1995; this study). It also allowed transcription in the mating-type region and the derepression observed near mat2-P or mat3-M was in both cases increased by the simultaneous deletion of cis-acting elements located near the cassettes. In contrast to their effects on transcription, swi6 and chp2 affected mating-type switching differently. Swi6 is important for the choice of the silent cassette used to convert specific mat1 alleles, and lack of Swi6 causes characteristic switching defects where switching is reduced in strains with a wild-type mating-type region but increased in cells with swapped silent cassettes (Egelet al. 1984; Gutz and Schmidt 1985; Thon and Klar 1993). chp2-deleted cells were able to form colonies with a sporulation phenotype similar to the wild type, indicating that, unlike Swi6, Chp2 is not absolutely required for efficient mating-type switching. One explanation for these different effects of Swi6 and Chp2 might reside in the degree to which each protein contributes to the structure of the mating-type region. Lack of Chp2 might not have as deleterious an effect as lack of Swi6. Indeed, the derepression of transcription caused by deleting chp2 is not as marked as that caused by deleting swi6 (Figures 4 and 5). Hence, the organization remaining in chp2-deleted cells might be sufficient to support efficient mating-type switching, albeit not sufficient for complete transcriptional silencing. Another possibility is that Swi6 interacts with a protein required for switching whereas Chp2 does not. Other differences between swi6- and chp2-deleted cells could be seen in the mitotic and meiotic viability of mutant strains. Cultures of swi6 mutant cells contain higher proportions of dead cells than do wild-type cultures and produce an increased number of three-spored asci. chp2-deleted strains were as healthy as the wild type in these respects (data not shown). Consistent with these phenotypes, mitotic loss of a minichromosome was not increased by lack of Chp2 whereas it is in the absence of Swi6 (Table 2).
Effects on chromosome segregation—Epistasis analysis: Chromo-domain proteins are found in association with the pericentric heterochromatin of many eukaryotes. In addition to repressing transcription, they possibly participate in the formation of structures important for chromosome function, in particular the kinetochore. Consistent with such a function, the S. pombe chromo-domain proteins Swi6, Clr4, and Chp1 are required for efficient chromosome segregation (Allshireet al. 1995; Ekwallet al. 1996; Bernardet al. 1998; Doeet al. 1998). Swi6 is present at centromeres (Ekwallet al. 1995) and two proteins are known to be required for its localization, Rik1 and Clr4 (Ekwallet al. 1996). The mutant allele of clr4 used in this study, clr4-681, did not increase the rate of chromosome loss to the same extent as the previously tested allele clr4-S5 (Table 2; Allshireet al. 1995; Ekwallet al. 1996). The clr4-681 allele contains a point mutation causing an amino-acid change in the Clr4 SET domain (G486D mutation; Ivanovaet al. 1998). It will be interesting to determine whether that allele or other clr4 mutations affecting the SET domain of the protein influence the localization of Swi6 to the same extent as the clr4-S5 allele. The clr4-681 or swi6-115 allele had a synergistic effect with a deletion of chp1 (Table 2). Hence, Swi6 and Clr4 seem to act in a pathway different from that in which Chp1 participates. The observed synergy also reveals that the clr4-681 allele does reduce the efficiency of chromosome segregation although this effect could not be readily seen in a wild-type background. Consistent with our epistasis analysis, Chp1 localizes to one or a few discrete spots in the nucleus, which might coincide with centromeres, in a manner that does not depend on Rik1 or Clr4 (Doeet al. 1998), whereas the centromeric localization of Swi6 depends on Rik1 and Clr4. As expected, swi6-115 was epistatic to clr4-681 (Table 2). Deleting chp2 increased chromosome loss only threefold compared with the wild type, a relatively modest effect (Table 2). Further experiments will be required to assess its role in chromosome segregation.
Chp1 target recognition: In contrast to the pleiotropic effects of Chp2, Chp1 displayed a specificity for centromeric regions. A DNA segment homologous to a centromeric repeat is found in the mating-type region between mat2-P and mat3-M (Grewal and Klar 1997). In that area, the sequence similarity with centromere 2 reaches 96% over 4.3 kb of DNA. Deletions in the mating-type region that include the 4.3-kb repeat lead to a partial derepression of transcription, where “on” and “off” states of expression are clonally inherited, and to defects in mating-type switching, which covariegate with the defects in transcription (Grewal and Klar 1996; Thon and Friis 1997). The ura4+ gene is silenced when introduced within the repeat in the mating-type region (Grewal and Klar 1997) or within the homologous repeat near centromere 1 (Allshireet al. 1995). Whereas the Swi6 and Clr4 proteins are required for the repression at both locations (Allshireet al. 1995; Grewal and Klar 1997), we showed here that Chp1 was required for silencing centromeric insertions of ura4+, but not for silencing insertions of ura4+ in the mating-type region, nor the mating-type genes. The lack of effect of a chp1 deletion on gene expression in the mating-type region was not simply explained by the redundancy of the silencing mechanism operating in that region, which occasionally masks silencing defects, since no cumulative effect was observed following combination of the chp1 deletion with other silencing deficiencies (Figure 5; data not shown). The simplest interpretation of our results is that Chp1 does not act in the mating-type region, although we cannot rule out that it has a very localized action at locations that we have not tested. The restriction of its repressive effects to centromeres suggests that it is tethered there by a protein/DNA complex other than that associated with the 4.3-kb sequence. Chp1 could then exert an effect on the 4.3-kb repeat at a distance, possibly via other proteins such as Swi6 and Clr4, or it might spread along the DNA by oligomerization.
Loss of the ura4+ gene from centromeric regions: Meiotic recombination is inefficient in S. pombe chromosomal regions close to centromeres (Nakasekoet al. 1986; Chikashigeet al. 1989). This inhibition possibly reflects the existence of an intramolecular secondary structure that would compete with pairing between homologs, or could also be due to steric hindrance by the kinetochore. We discovered that a significant fraction of cells containing a centromeric ura4+ insertion could lose the ura4+ gene during mitotic growth. One likely mechanism leading to the loss of ura4+ when placed in a centromeric repeat is an intramolecular gene conversion involving the untouched repeat placed symmetrically to that containing ura4+. Such interactions have been proposed to be responsible for the very high degree of sequence conservation between centromeric repeats and for the presence of identical nucleotide changes in central repeats occupying symmetrical positions relative to centromere 1 in some strains (Takahashiet al. 1992). ura4+ could also be lost from an insertion site located within the central core of chromosome 1. The central core has a unique sequence and is therefore not an expected target for mitotic gene conversion. We have not determined whether the central core was altered in cells having lost ura4+ from the cnt1 integration site.
Transcriptional silencing in the rDNA: Transcriptional silencing has in some cases a clear function of its own. For example, silencing of the mat2-P and mat3-M mating-type cassettes is crucial to the yeast sexual cycle and to yeast survival in the absence of a nitrogen source. In other cases, transcriptional silencing might be a consequence of other phenomena, or of structural or localization requirements. The role of transcriptional silencing in the rDNA is the subject of speculation.
Ty1 elements and prototrophic markers introduced in the rDNA of S. cerevisiae are poorly transcribed (Bryket al. 1997; Fritzeet al. 1997; Smith and Boeke 1997). Their transcription is increased by disruption of the silencing gene SIR2, of the ubiquitin-conjugating enzyme gene UBC2, by reduced levels of histone H2A and H2B, as well as by mutations in a large number of genes including genes involved in DNA replication or regulation of chromatin structure (Bryket al. 1997; Smith and Boeke 1997; Smithet al. 1999). Repression of PolII transcription in the rDNA repeats might be a by-product of PolI regulation (Smith and Boeke 1997), of the rDNA control of mitotic and meiotic recombination (Gottlieb and Esposito 1989 and references therein), and/or of life span regulation (Kennedyet al. 1997). It might reflect functions of the nucleolus distinct from the production of rRNA and linked to its constituting a discrete nuclear subcompartment. Such a function was proposed recently for the RENT complex, a nucleolar protein complex associated with Sir2, which sequesters Cdc14 in the nucleolus during most of the cell cycle and releases it at telophase (Shouet al. 1999; Straightet al. 1999).
The transcriptional repression we observed in the rDNA of S. pombe is reminiscent of that in S. cerevisiae. At least one S. pombe ORF encodes a protein related to S. cerevisiae Sir2 (CAB38511 in cosmid SPBC16D10) and it will be interesting to investigate its participation in rDNA silencing. The four chromo-domain proteins whose roles were assayed here have no homologs in S. cerevisiae. S. cerevisiae has only two chromo-domain proteins: a member of the CHD family (Woodageet al. 1997; AC P32657) whose S. pombe closest relatives are Hrp1 (Jinet al. 1998) and an uncharacterized ORF (AC O14139) and Esa1 (Hilfikeret al. 1996; Clarkeet al. 1999). We have found that S. pombe proteins other than the chromo-domain proteins described here are important for the transcriptional repression in the rDNA, including the histone deacetylase Clr3 (data not shown). As for S. cerevisiae, one can speculate that the repression of transcription is part of growth rate regulation, that S. pombe PolI has template preferences different from those of PolII, or that the nucleolus constitutes a compartment with special properties, as suggested by its ultrastructure (Leger-Sylvestreet al. 1997). A regulatory function by the dynamic sequestering of proteins involved in mitosis can also be envisioned for the S. pombe nucleolus. Clr4, a protein important for chromosome segregation and rDNA silencing, is localized to the nucleolus in interphase when expressed from a plasmid as a green fluorescent protein (GFP)-fusion protein. During mitosis, a fraction of the Clr4 pool relocalizes to the mitotic spindle (S26 in Sawin and Nurse 1996).
Direct or indirect effects—Localization: rDNA insertions of the LEU2 and ura4+ gene were transcribed more efficiently in swi6-115 than in wild-type cells. The swi6-115 mutation substitutes an arginine for the tryptophan at position 269, a residue conserved between chromo-shadow domains (Allshire 1996). The swi6-115 mutation also results in undetectable levels of the Swi6 protein (Ekwallet al. 1996) and therefore probably constitutes a loss of function. Hence, our observation indicates that the wild-type Swi6 protein represses PolII transcription in the rDNA. The interpretation of that repression is not straightforward because Swi6 is not localized within the nucleolus of wild-type cells (Ekwall et al. 1995, 1996). One possible explanation is that Swi6 recognizes and silences PolII-transcribed genes when placed in the context of the rDNA. In this model, localization of Swi6 to the nucleolus would not precede introduction of the marker gene and would therefore not have been observed in past experiments. Alternatively, the influence of the swi6-115 mutation on transcription in the rDNA could be indirect. In the absence of Swi6 a factor normally present in the rDNA and able to repress PolII transcription might not be synthesized, or might be titrated away to sites normally occupied by Swi6, such as the centromeres. Titration effects could account for other observed phenotypes. As mentioned above, a GFP-Clr4 fusion protein localizes to the nucleolus (Sawin and Nurse 1996) and in clr4− cells, Swi6 does not associate with centromeres, but is found in the nucleolus (Ekwallet al. 1996). One could then conceive that nucleolar sites normally occupied by Clr4 can be recognized by Swi6, which in the absence of Clr4 is titrated away from the centromeres to the nucleolus. In wild-type cells, Clr4 would repress transcripion of PolII genes in the rDNA. In clr4 mutant cells, Swi6 would mislocalize to sites normally occupied by Clr4, but would fail to exert the same repression as Clr4. In swi6 mutant cells, Clr4 would be titrated to sites normally occupied by Swi6 and would fail to repress transcription in the rDNA. Other indirect effects due to alterations in clr4 are also possible since clr4 controls the transcription of several genes, both positively and negatively (Ivanovaet al. 1998). In another type of model, one could imagine that changes in centromere or telomere structure sterically affect the nucleolus, possibly facilitating entrance of PolII. Additional localization studies and overexpression experiments ought to help distinguish between some of these models.
Acknowledgments
We thank Olaf Nielsen for the gift of plasmids, Robin Allshire for strains with centromeric and telomeric insertions of the ura4+ gene, and Kohta Takahashi and Mitsuhiro Yanagida for a strain containing the YIp2.4 allele. We also thank Inga Sig Nielsen and Stanley Brown for their comments on the manuscript. The reported experiments were supported by grants from the Novo Nordisk Foundation and from the Danish Natural Science Research Council.
Footnotes
-
Communicating editor: G. R. Smith
- Received July 9, 1999.
- Accepted February 7, 2000.
- Copyright © 2000 by the Genetics Society of America
