Genetics, Vol. 162, 591-602, October 2002, Copyright © 2002
The 2.1-kb Inverted Repeat DNA Sequences Flank the mat2,3 Silent Region in Two Species of Schizosaccharomyces and Are Involved in Epigenetic Silencing in Schizosaccharomyces pombe
Gurjeet Singha and
Amar J. S. Klara
a Gene Regulation and Chromosome Biology Laboratory, National Cancer Institute, Frederick, Maryland 21702-1201
Corresponding author:
Amar J. S. Klar, National Cancer Institute, P.O. Box B, Frederick, MD 21702-1201., klar{at}ncifcrf.gov (E-mail)
Communicating editor: F. WINSTON
 | ABSTRACT |
|---|
The mat2,3 region of the fission yeast Schizosaccharomyces pombe exhibits a phenomenon of transcriptional silencing. This region is flanked by two identical DNA sequence elements, 2.1 kb in length, present in inverted orientation: IRL on the left and IRR on the right of the silent region. The repeats do not encode any ORF. The inverted repeat DNA region is also present in a newly identified related species, which we named S. kambucha. Interestingly, the left and right repeats share perfect identity within a species, but show 
2% bases interspecies variation. Deletion of IRL results in variegated expression of markers inserted in the silent region, while deletion of the IRR causes their derepression. When deletions of these repeats were genetically combined with mutations in different trans-acting genes previously shown to cause a partial defect in silencing, only mutations in clr1 and clr3 showed additive defects in silencing with the deletion of IRL. The rate of mat1 switching is also affected by deletion of repeats. The IRL or IRR deletion did not cause significant derepression of the mat2 or mat3 loci. These results implicate repeats for maintaining full repression of the mat2,3 region, for efficient mat1 switching, and further support the notion that multiple pathways cooperate to silence the mat2,3 domain.
LARGE portions of eukaryotic chromosomes exist in a highly condensed form during most of the cell cycle. This compaction involves a higher level of organization where DNA is packaged into nucleosomes that are further compacted into chromatin fibers (WIDOM 1998 
). The high degree of compaction imparts topological constraints to accessibility of DNA to various cellular functions such as transcription, replication, DNA repair, and recombination. Complex interplay between many different trans-acting proteins provides cells with the ability to selectively activate or inactivate various genes through multiple cell divisions. Recent studies have shown that cis elements located nearby are also equally important for correct regulation of gene activity during development (BONIFER 2000 ). Various regulatory elements such as insulators, enhancers, and even repetitive DNA sequences can act as signals that determine the formation of active or repressive chromatin domains, thus epigenetically controlling gene activity (WEST et al. 2002 ). Epigenetics refers to the phenomenon of mitotically and occasionally meiotically heritable change in gene expression that cannot be explained by changes in DNA sequence (RUSSO et al. 1996 ; CHADWICK and CARDEW 1998 ). Gene regulation through epigenetic imprinting has been found to be crucial for maintaining proper developmental gene expression patterns in many eukaryotic species (RIGGS and PORTER 1996 ; BARTOLOMEI and TILGHMAN 1997 ).
The mating-type region of the fission yeast Schizosaccharomyces pombe has over the years proved to be an excellent system for studying the mechanism of epigenetic control over transcription and recombination. This region is composed of three linked loci, mat1, mat2, and mat3, which occupy 30 kb of DNA on chromosome II (BEACH and KLAR 1984 ). The mat1 locus determines the cell type, depending upon whether it has plus (P) or minus (M) information (for a recent review see KLAR 2001 ). mat2-P and mat3-M loci act as donors of information for switching the content of mat1 by the process of gene conversion (EGEL and GUTZ 1981 ; EGEL 1989 ; KLAR 2001 ). The switching of individual cells is highly regulated and follows a set pattern in pedigrees (MIYATA and MIYATA 1981 ; EGEL and EIE 1987 ; KLAR 1990 ). The rate of mat1 switching is in the range of 72–94% per cell division of switching-competent cells (MIYATA and MIYATA 1981 ). This high level of switching efficiency is accomplished by a nonrandom interaction of mat2 and mat3 with mat1 through a phenomenon designated "directionality of switching" (THON and KLAR 1993 ). mat1-P and mat1-M alleles contain exactly the same DNA information present at, respectively, mat2 or mat3 loci (KELLY et al. 1988 ). However, mat1 is transcriptionally active, whereas mat2 and mat3 are transcriptionally inert. The mechanism that represses the transcription of mat2-P and mat3-M is not restricted to these loci, but extends over the adjoining regions, including the entire 10.6-kb "K region" between mat2 and mat3. Additionally, marker genes inserted in this region are subject to transcriptional repression (THON et al. 1994 ; GREWAL and KLAR 1997 ). Another important property of this region is that it is a "cold spot" for recombination (EGEL 1981 , EGEL 1984 ). Genetic research over the years has shown that redundant pathways exist, which suppress the mat2,3 region transcriptionally and recombinationally (EGEL et al. 1989 ; KLAR and BONADUCE 1991 ; LORENTZ et al. 1992 ; THON and KLAR 1992 ; EKWALL and RUUSALA 1994 ; THON et al. 1994 ; GREWAL and KLAR 1997 ). In addition, it has been observed that factors affecting silencing of the mat2,3 region (for reviews see GREWAL 2000 ; KLAR 2001 ) also affect switching of the mat1 locus (LORENTZ et al. 1992 ; THON and KLAR 1992 ; THON et al. 1994 ).
Many different studies using unrelated genetic strategies have identified several interesting trans-acting factors (Swi6, Rik1, Clr1-Clr4, Clr6, rhp6, etc.) and implicated them in epigenetic control of silencing at the mat2,3 region (THON and KLAR 1992 ; EKWALL and RUUSALA 1994 ; THON et al. 1994 ; SINGH et al. 1998 ). Sequence information of these factors suggested the organization of heterochromatin as the mechanism for silencing and recombinational suppression of the mat2,3 region. For example, Clr4 has both chromo and SET domains, which share homology with several proteins implicated in chromatin organization in different organisms such as Drosophila melanogastor [i.e., Su(var) 3-9, E(z), trithorax] and human G9a proteins (IVANOVA et al. 1998 ). Similarly, Swi6 contains both chromo and chromo-shadow domains (LORENTZ et al. 1994 ), which share homologies with proteins associated with a higher-order chromatin structure such as M31 from mouse, HP1 from Drosophila, and humans (SINGH 1994 ). In addition, Clr3, Clr6, and Hda1, also required for silencing of marker genes placed in the mat2,3 region, exhibit homology to histone deacetylases (GREWAL et al. 1998 ; OLSSON et al. 1998 ). Furthermore, two other chromo-domain proteins, Chp1 and Chp2, have also been shown to be important for controlling position-effect variegation (PEV) at this region (THON and VERHEIN-HANSEN 2000 ). It has been suggested that most of these proteins, especially Clr1-Clr4 and Swi6, work in one pathway, as their double mutants in any combination are phenotypically similar to either single mutant (THON et al. 1994 ).
In addition to these proteins, several cis-acting DNA elements have been implicated in silencing of the mat2,3 region. A study with a plasmid-borne mat2-P locus identified four silencing elements situated adjacent to mat2 (EKWALL et al. 1991 ). However, a chromosomal deletion of the 1.5-kb BglII-BssHII fragment proximal to mat2-P, which contains at least two of the silencing elements, did not show significant derepression of either mat2-P itself or of ura4+ inserted next to it (THON et al. 1994 ). Nevertheless, when this deletion was combined with mutations in any of the swi6 or clr1-clr4 genes, it alleviated silencing to a much greater extent (THON et al. 1994 ). Deletion of a shorter EcoRI-BssHII fragment (termed REII), which contains only one of the four cis-acting elements that repress the plasmid-borne mat2-P locus, also results in derepression of a marker gene inserted in the mat2 locus (AYOUB et al. 2000 ). Interestingly, transplacement of REII to another site in the L region (region to the left of mat2) extends the silent domain to its new location. Likewise, another repressor element located proximal to mat3-M was identified, genetically exhibiting properties similar to those of the mat2-P proximal silencing element (THON et al. 1999 ). Deletion of this element also showed a synergistic effect with mutations in trans-acting factors, Swi6 and Clr1-Clr4, but no additive effect with the mat2-P cis-acting repressor element in the BglII-BssHII fragment mentioned above.
Even though many cis- and trans-acting factors have been identified, it is likely that still other factors exist, which act independently of or in conjunction with these known factors to make this region silent and inert for recombination. In this communication we report identification of repeat sequences present in inverted orientation flanking the mat2,3 region from two different species of Schizosaccharomyces and show that they are required for maintaining transcriptional silencing at this region in S. pombe. The silencing properties of these repeats are specific to the sequences present in these repeats. We also show that these repeats genetically interact differently with various trans-acting factors when compared to the other cis-acting sequences. This genetic work adds to the conclusion that redundant pathways operate to silence this region.
| MATERIALS AND METHODS |
|---|
Strains, media, and culture conditions:
The S. pombe strains used in this study are listed in Table 1 along with their genotype and origin. Deletions of various regions of chromosomal DNA were accomplished by using a strategy proposed by BAHLER et al. 1998 . Strains PSG311 and PSG325 were constructed by transforming PSG160 with HindIII fragments from plasmids pGS154 and pGS153 (see next section), respectively. A lithium acetate protocol was used for transforming cells (HEYER et al. 1986 ; MORENO et al. 1991 ). The media and culture conditions were used as described in MORENO et al. 1991 .
Plasmid construction:
pGS153 and pGS154 were constructed by cloning an inverted repeat right (IRR) fragment in both orientations at the BstBI site of pSP2 (KELLY et al. 1988 ). The IRR fragment was amplified by PCR on genomic DNA obtained from a wild-type S. pombe strain using primers SR304 (5'-TAT ATA TTC GAA GTG TAT CTT GGT CAT TTG TC-3') and SR305 (5'-TAT ATA TTC GAA ATT CAG GTA AAA TAA TTT AA-3'). PCR product was digested with BstBI before ligating it into the BstBI site of pSP2. Plasmid DNA was amplified in the Escherichia coli strain DH5
.
Iodine-staining assay:
Sporulating cells of S. pombe synthesize a starch-like compound that stains black when colonies are exposed to iodine vapors (BRESCH et al. 1968 ). An iodine-staining assay was used to determine the level of sporulation due to matings between switched cells or of "haploid meiosis" in S. pombe colonies. In switching cells, intensity of staining indicates the efficiency of mat1 switching. In a nonswitching strain, staining indicates the extent to which mat2-P and/or mat3-M are derepressed where aberrant spores are made by haploid cells dubbed haploid meiosis (KELLY et al. 1988 ).
Ura transition assay:
The Ura transition assay was carried out according to the fluctuation test of LURIA and DELBRUCK 1943 .
Calculation of mat1 switching rate:
For determining the rate of switching of mat1, single colonies of h09 strains were grown on sporulating medium (PMA+) for 3 days and microscope slides were prepared from four different colonies of each strain. For each slide, the number of zygotes (X) and vegetative cells (Y) was counted from three different microscopic fields. Rate of switching was determined by dividing the number of zygotes (X) by the sum of the number of vegetative cells (Y) and two times the number of zygotes [X/(2X + Y)].
Molecular analysis:
Southern and Northern blot analyses were performed according to standard protocols (SAMBROOK et al. 1989 ).
Pharmacological induction of recombination in the K region:
Cells to be crossed were mixed with 1 µl of 10 µg ml-1 Trichostatin A (Sigma, St. Louis) dissolved in DMSO directly on the solid mating and sporulation medium. The culture was allowed to sporulate for 4 days before subjecting it to random spore analysis.
| RESULTS |
|---|
Identification of a new species from the genus Schizosaccharomyces, S. kambucha:
Most studies of the genus Schizosaccharomyces have been limited to one species: S. pombe (LEUPOLD 1958 ). We identified another species from this genus, which we named S. kambucha. This species was isolated from the kambucha fungus used for centuries by the Chinese for making Che, a drink the Chinese considered divine. This yeast exhibits mating-type switching and can also mate with S. pombe, but spores of the hybrid are inviable. Southern blot (Fig 1) probed with the mat1-P HindIII fragment from S. pombe revealed that S. kambucha also has mating-type mat1, mat2, and mat3 cassettes analogous to those in S. pombe. Sequence comparison of mat-M and the region adjoining mat2 from these two species has indicated 98.3% sequence identity (data not shown). The difference in size of the HindIII fragments of mat1 and mat2 in the Southern blot is explained by polymorphism existing at one of the HindIII sites flanking them. The size of the HindIII fragment containing mat3 is indistinguishable between these species. The mat2, K, and mat3 region is contained in a single 27-kb PstI fragment in both organisms (data not shown). Individual colonies contain cells of both mating types; thus, S. kambucha also undergoes mating-type switching. Accordingly, the DNA double-strand break (DSB; BEACH 1983 ; BEACH and KLAR 1984 ), resulting from imprinting in S. pombe (KLAR 1987 ), is found in both organisms. The level of the break is much reduced in S. kambucha as compared with the level observed in S. pombe. The reduced level of DSB in S. kambucha is consistent with the reduced efficiency of switching and iodine staining, since 15% of the cells engage in zygote formation, as opposed to between 80 and 90% frequency observed with the S. pombe cells. The similarity of the mating-type region architecture of S. kambucha to that of S. pombe indicates that this yeast also exhibits the phenomenon of mating-type silencing; however, more work is needed to establish this supposition.
View larger version (23K):
In this window
In a new window
Download PPT slide
|
Figure 1.
Southern blot analysis of S. pombe and S. kambucha genomic DNA. DNA was isolated from h90 strains, digested with HindIII, and probed with a 10.6-kb HindIII fragment containing mat1-P from S. pombe. The probe hybridized to 10.6-, 6.4-, and 4.3-kb fragments from S. pombe containing mat1, mat2, and mat3 cassettes, respectively. In S. kambucha, it hybridized to 15.6-kb (mat1), 2.9-kb (mat2), and 4.3-kb (mat3) fragments. The difference in sizes of mat1 and mat2 fragments between two species is due to polymorphism at the HindIII sites. Fainter 5.6-kb (DSB proximal) and 5.0-kb (DSB distal) bands observed in S. pombe and 10.0-kb (DSB distal) and 5.6-kb (DSB proximal) bands in S. kambucha result from a DSB at mat1 (BEACH 1983 ), which initiates recombination for mating-type switching.
|
|
The silenced mat2,3 region in both S. pombe and S. kambucha is flanked by inverted repeats:
In S. pombe, silencing in the mat2,3 region extends from 1 kb upstream (centromere-proximal) of mat2-P to 1 kb downstream of mat3-M encompassing a region of 16 kb. The extent of the silenced region is revealed by the repression of markers inserted at various sites in this interval (THON et al. 1994 ; GREWAL and KLAR 1997 ; AYOUB et al. 1999 , AYOUB et al. 2000 ). We searched the sequence database for a special feature of the region delimiting the silenced domain and found that this region is flanked by 2110-bp-long inverted repeat sequences (Fig 2) with 100% identity. Independently, NOMA et al. 2001 also recently identified these repeats. In addition, we discovered by Southern analysis that the S. kambucha mat2,3 region is also flanked by repeat DNA sequences similarly located and in an inverted orientation. We named these repeat sequences inverted repeat left (IRL) and IRR, depending upon their position with respect to the conventionally drawn mat2,3 region. Interestingly, comparison of the sequence of S. kambucha IRL with its IRR sequence (GenBank accession no.
AY124374) also showed perfect identity. However, these sequences are not identical between these two species, as comparison of the sequence of S. kambucha repeats with those of S. pombe showed 98.7% identity, along with a stretch of the 81-bp additional sequence found only in S. kambucha (Fig 3).
View larger version (27K):
In this window
In a new window
Download PPT slide
|
Figure 3.
Comparison of S. pombe and S. kambucha IRR sequences. These regions show 98.2% sequence identity with a stretch of additional 81 bp present only in S. kambucha repeats. The numbers represent base pairs (bp). The graph was generated by using the Genetics Computer Group program DOTPLOT.
|
|
Deletion of inverted repeats causes defects in silencing:
We were struck by the 100% identity in the base sequence of inverted repeats within each organism. The perfect identity and conserved orientation of the repeats in two species suggested the role of repeats in some aspect of mating-type switching or silencing. To determine whether these repeats have a direct role in maintaining silencing at the mat2,3 region, we made precise deletions of them using a PCR-based gene-targeting method (BAHLER et al. 1998 ). Silencing at the mat2,3 region is controlled by more than one pathway, as none of the previously characterized mutations individually derepress mat2-P or mat3-M completely (THON and KLAR 1992 ; EKWALL and RUUSALA 1994 ; THON et al. 1994 ). However, the ura4+ marker gene inserted 400 bp distal to mat2-P, designated mat2-Pint::ura4+ (THON et al. 1994 ), or 150 bp distal to mat3-M, designated mat3-Mint:: ura4+ (THON and KLAR 1992 ), provides a very sensitive assay for monitoring defects in silencing. Keeping this in mind, we made deletions of IRL or IRR in strains whose only copy of ura4+ was inserted next to either mat2-P or mat3-M. Deletion of IRR from an h90 strain (a wild-type homothallic strain with standard mat1 mat2-P mat3-M mating-type region) resulted in derepression of ura4+ in the mat3-Mint::ura4+ construct as the PSG124 strain was a prototroph showing growth on medium lacking uracil and no growth on medium containing 5-fluoroorotic acid (FOA; Fig 4). Ura+ prototrophs do not grow in the presence of FOA because of its toxicity, while the Ura- cells do grow as they are insensitive to the drug (BOEKE 1987 ). Derepression of ura4+ as a result of IRR
was further confirmed by the Northern analysis (Fig 4C). IRR resulted in more than a sevenfold increase in the level of the ura4+ transcript from mat3-Mint::ura4+ compared to a wild-type strain with mat3-Mint::ura4+. The increase in ura4+ transcript level due to IRR was comparable to derepression of mat3-Mint::ura4+ caused by the clr1-5 mutation.
View larger version (27K):
In this window
In a new window
Download PPT slide
|
Figure 4.
Deletion of IRR or IRL causes derepression of ura4+ placed in the mat2,3 region. (A) Schematic representation (not to scale) of the mat2,3 region showing location of the ura4+ insertion next to mat2 or mat3. (B) Assay of ura4+ expression in various strains by dilution analysis. The cells from various strains were serially diluted and spotted onto uracil-lacking (-ura), FOA-containing (FOA), or nonselective (N/S) media. The spots were allowed to grow for 3–4 days before taking pictures. Strains that were either prototroph (SP819) or auxotroph (SP837) for ura4 were included as controls along with strains carrying mat2-Pint::ura4+ or mat3-Mint::ura4+ with (SP1239 and PG439), respectively, or without clr1-5 mutation (SP1122 and PG9). The wild-type strain (PG9) carrying mat3-Mint::ura4+ showed poor growth on medium lacking uracil but grew on medium containing FOA due to lack of ura4+ expression. When IRR was deleted, the resultant strain (PSG124) could grow on uracil-lacking medium but did not grow on FOA, showing that mat3-Mint::ura4+ was transcriptionally active. In comparison, when IRL was deleted from a strain carrying mat2-Pint::ura4+ (PSG223), it showed variegated expression of ura4+ as it was able to grow in the absence of uracil as well as in the presence of FOA. (C) Assay of ura4+ expression by Northern analysis. RNA was isolated from various strains used in dilution analysis in B. About 20 µg of total RNA extracted from each strain was run on a denaturing gel and probed with ura4+ or cdc2+ (loading control). The signal was detected using a phosphorimager. Deletion of IRL and IRR resulted in 10.4- and 8.0-fold enrichment of ura4+ transcript from mat2-Pint::ura4+ and mat3-Mint::ura4+, respectively.
|
|
When IRL was deleted from an h90 strain (PSG223), it instead showed variegated expression of mat2-Pint:: ura4+. This strain was able to grow on both media, either lacking uracil or containing FOA (Fig 4B). This ability implied that either the IRL deletion partially derepressed mat2-Pint::ura4+ or the level of derepression was not uniform in all cells. Some cells in which ura4+ was derepressed were able to grow on -uracil plates while those in which ura4+ was not derepressed showed growth on +FOA plates. Surprisingly, Northern analysis revealed that increase in ura4+ expression due to IRL was even greater than the increase observed due to the clr1-5 mutation (Fig 4C). Therefore we presume that the growth of PSG223 on FOA-containing medium is due to derepression not occurring in all cells.
In routine crosses we noted that the P vs. M cell type affected the ura4+ derepression to different levels. To systematically determine if variegated expression of mat2-Pint::ura4+ was affected by the cell type, we deleted IRL from nonswitching strains (mat1-Msmt0 and mat1-P17). mat1-Msmt0 and mat1-P17 refer to nonswitching alleles due to, respectively, deletion of the 263-bp (STYRKARSDOTTIR et al. 1993 ) and 122-bp (ARCANGIOLI and KLAR 1991 ) sequences situated distal to mat1. Because of these deletions, the imprint at mat1 does not occur, and, consequently, cells do not switch mating type. In such nonswitching strains, the two epigenetic states ("Ura-on" and "Ura-off") resulting from the IRL deletion interconverted at variable rates depending upon the cell type (Fig 5). In a mat1-Msmt0 IRL strain (PSG205), the transition rate per cell division of Ura-on to the Ura-off state was 1.0 x 10-4 and for Ura-off to Ura-on was 4.1 x 10-4. However, in the mat1-P17 IRL strain (PSG222), the transition rate of Ura-on to Ura-off was 190.0 x 10-4 and of Ura-off to Ura-on was 29.0 x 10-4. Thus, the two epigenetic states were much more stable in an M cell type as compared to P cells; in M cells transition of Ura-on to Ura-off was down by 190-fold and Ura-off to Ura-on by at least 7-fold when compared with the P cells. A possible significance of this cell-type difference is considered in the DISCUSSION.
Deletion of repeats does not significantly affect silencing of the mat2 and mat3 cassettes:
Silencing is believed to be more stringent at mat2 and mat3 cassettes than in the adjoining regions, as mutations of several genes causing expression of markers inserted in adjoining regions do not derepress mat2 and mat3 cassettes significantly (THON et al. 1994 ). To determine if deletion of the repeats has any effect on silencing of the mat cassettes, we deleted IRL and IRR from nonswitching strains (mat1-Msmt0 and mat1-P17), as such strains provide a better monitor for derepression of mat2-P or mat3-M loci. Expression of mat3-M in a nonswitching P cell or expression of mat2-P in a nonswitching M cell results in "haploid meiosis," an aberrant sporulation phenotype triggered by coexpression of both mating types in haploid cells. The colonies of cells undergoing haploid meiosis exhibit dark staining when they are exposed to iodine vapors, and those without haploid meiosis do not stain (KELLY et al. 1988 ). Results showed that neither mat1-Msmt0 with either IRL (PSG226) or IRR (PSG201) nor mat1-P17::LEU2 with IRL (PSG204) or IRR (PSG209) revealed any signs of haploid meiosis (data not shown).
Deletion of IRR affects directionality of switching:
Previous studies have demonstrated that many factors that affect silencing at the mat2,3 region also alter switching competence of mat1, possibly due to modification in the heterochromatic structure at this region (THON and KLAR 1992 ; THON et al. 1994 ; GREWAL and KLAR 1996 ). The efficiency of mat1 switching in an h90 IRR strain (PSG124) or h90 IRL strain (PSG223) was roughly comparable to that of the wild-type strain as demonstrated by the similar levels of iodine staining (Fig 6). However, the staining assay is informative in testing directionality defects only when the rate of switching is considerably reduced. Defects in directionality of switching are more readily quantifiable in an h09 strain in which the genetic contents of mat2 and mat3 donors are swapped to mat2-M and mat3-P (THON and KLAR 1993 ). Such a strain undergoes mostly futile switches to the same mating type (P to P and M to M), as the donor selection for gene conversion is governed by location of the donor with respect to mat1 and not the content of the donor loci. Thus, h09 switches to the opposite mating type at a significantly lower rate compared to an h90 strain, and, consequently, cells show a considerably lower level of staining on iodine exposure (THON and KLAR 1993 ). In such an arrangement, a directionality defect is expected to increase the rate of switching to the opposite mating type, resulting in increased staining. To test this possibility, we constructed an h09 strain carrying IRL or IRR. After sporulation, the level of iodine staining of colonies of the h09 IRL and h09 IRR strain was compared with those of the wild-type h09 strain (Fig 6). The h09 IRR strain (PSG167) showed slightly darker staining compared to the wild-type h09 strain (PG19), reflecting an increased level of sporulation, whereas the h09 IRL (PSG196) strain did not differ significantly from the wild-type control. To make sure that increased iodine staining was a result of switching defect and not due to increased haploid meiosis due to derepression of mat2 and mat3 cassettes, we calculated switching frequency of various strains. In a wild-type h09 strain containing swi6-mod (THON and KLAR 1993 ), switching frequency to the opposite mat1 allele was 6% (30 zygotes and 415 vegetative cells). When IRR was deleted, the rate of switching increased to 12% (67 zygotes and 483 vegetative cells). In an h09 IRL strain it was 5% (23 zygotes and 461 vegetative cells). This showed that deletion of IRR partially affected donor choice during mat1 switching while deletion of IRL did not cause a significant effect.
View larger version (75K):
In this window
In a new window
Download PPT slide
|
Figure 6.
Effect of deletion of inverted repeat sequences on the directionality of switching. IRL and IRR were individually deleted from both h90 and h09 strains, and sporulated cultures were exposed to iodine vapors. The amount of staining roughly reflects the efficiency of switching to the opposite mat1 allele. In an h90 strain, IRR or IRL deletion did not show any significant effect on switching. However, in an h09 strain, deletion of IRR resulted in a darker staining phenotype, indicating that directionality of switching was partially affected. In contrast, the h09 IRL strain showed the same level of staining as that of the wild-type h09 strain.
|
|
Pairwise combinations of deletions of IRL and IRR with mutations in other cis- and trans-acting factors show a variable influence of repeats on silencing of the mat2,3 region:
From results of various genetic studies combining mutations in cis-acting sequences and trans-acting factors, it is clear that more than one pathway operates to keep the mat2,3 region transcriptionally silent. One pathway involves cis-acting sequences such as the 1.5-kb BglII-BssHII region proximal to mat2-P (THON et al. 1994 ) or the repressor element present within 500 bp proximal to mat3-M (THON et al. 1999 ). The other pathway works through a number of trans-acting factors, namely, Clr1–Clr4, Swi6, etc., and still another works through the Clr6 factor (GREWAL et al. 1998 ). Single or double mutations in elements of one pathway cause a partial or undetectable silencing defect of the mat2 and mat3 loci, with a mutation in clr2 showing the most prominent effect among these. Fig 7 (top) shows the phenotype of the control strains. Unswitchable M cells with the mutation in clr2 show darker staining, due to derepression of mat2-P, than that of the unswitchable M cells with mutations in clr1, clr3, or swi6 (Fig 7, top) as shown earlier (THON et al. 1994 ). Even though mutations in these trans-acting factors show a partial or undetectable silencing defect of the mat2 and mat3 loci, they all cause derepression of markers inserted in this region. To test whether repeats act through either of these pathways or define a different pathway, we made double-mutant strains through genetic crosses where deletion of either of these repeats was combined with other above-mentioned or trans-acting mutations.
View larger version (66K):
In this window
In a new window
Download PPT slide
|
Figure 7.
Cumulative effect of deletion of IRL and mutations in various trans-acting factors on derepression of mat2-P in a nonswitching mat1-Msmt0 strain. Mutations in various trans-acting factors were combined with deletion of one or the other repeat through genetic crosses, and sporulated colonies were exposed to iodine vapors. Only mutations in clr1 or clr3 in combination with IRL show dark staining due to haploid meiosis, a result indicative of expression of mat2-P genes. + indicates the wild-type allele.
|
|
A study combining IRL or IRR with mutations in trans-acting factors in a nonswitching mat1-Msmt0 strain showed very interesting results. The mat1-Msmt0 strain carrying IRL and a mutation in clr1 (PSG263) or clr3 (PSG265) showed very high levels of haploid meiosis, as observed by increased iodine staining (Fig 7), as well as by microscopic evaluation (data not shown). Surprisingly, when the mutation in clr2 was combined with IRL in the mat1-Msmt0 strain, it showed even less staining (Fig 7), due to a lower level of haploid meiosis, when compared with the IRL-containing strain. Furthermore, mutations in swi6 and clr4, which have been proposed to work in the same pathway as clr1 and clr3 (THON et al. 1994 ), did not show any additional effect with the IRL. Deletion of IRL from nonswitching mat1-P17 strains carrying any one of the trans-acting mutations did not show a significant effect on mat3 silencing (Table 2). IRR in mat1-Msmt0 cells, when combined with BglII-BssHII (not shown) or with mutations in trans-acting factors (Swi6, Clr1–Clr4), did not have any additional significant effect on silencing of mat2-P (Fig 7, Table 2). Similarly, no additional defect in silencing of mat3-M was observed in mat1-P17 cells when IRR was combined with BglII-BssHII or with mutations in trans-acting factors (Swi6, Clr1–Clr4; data not shown).
View this table:
In this window
In a new window
|
Table 2.
Genetic interaction of deletion of repeats with mutations in various trans-acting factors
|
|
Translocation of IRR in the K region does not derepress mat3-M:
Recently, it has been shown that the activity of chromatin insulators is sensitive to their position on the chromosome, as it may affect their ability to form chromatin loop domains (CAI and SHEN 2001 ; MURAVYOVA et al. 2001 ). Pairing an endogenous insulator element with another element inserted in its vicinity can lead to expression of normally suppressed downstream sequences. In this context, it is implied that IRL and IRR act as barriers to spreading of Swi6 (NOMA et al. 2001 ). If these repeats can perform in this fashion independent of their location, then moving one of these to a different location should restrict spreading of silencing to that position, allowing genes downstream to express. To test this, we deleted the endogenous IRR and moved it into the K region 1.6 kb proximal to mat3-M by placing it at the BstBI site. IRR was inserted at this location in the same orientation as the endogenous IRR, designated translocated (T)-IRR (PSG325), as well as in the reverse orientation, designated inverted T-IRR (PSG311). If IRR at its new location defines the boundary of silencing, then mat3-M would be out of the silent domain. This would result in derepression of genes at the mat3-M locus and cause haploid meiosis in P cells. However, iodine staining and microscopic evaluation showed that neither of the rearrangements had any discernible effect on silencing of mat3-M (data not shown).
Replacement of inverted repeats by another sequence of similar length in inverted orientation does not confer silencing:
In Drosophila and plants it has been observed that the presence of repeat sequences can act as a signal for heterochromatization, resulting in silencing of nearby sequences. It has been suggested that pairing between repeat sequences is responsible for formation of topologically constrained loops, resulting in formation of a higher-order chromatin structure (DORER and HENIKOFF 1994 ; LUFF et al. 1999 ). The sequence of IRL and IRR was found to be identical within each species, but it differs somewhat between the species. Because of these considerations, we entertained the possibility that it is not the sequence per se, but the identity between two repeats that may be important for their silencing function. If only the identity between the repeats was critical, then any sequence of a similar length in an inverted orientation may be able to confer silencing. To test this hypothesis, we replaced IRL and IRR with kanR cassettes in inverted orientation in a strain that had mat3-Mint::ura4+ for monitoring any effect on silencing of ura4+. To construct such a strain, IRL was first replaced with the 1.8-kb kanR cassette in a mat1-P mat2-P mat3-Pint::ura4+ strain by DNA-mediated transformation. Similarly, IRR was replaced with a kanR cassette in the opposite orientation in a mat1-M mat2-M mat3-Mint:: ura4+ strain. Both strains maintain respective mating type, as they have one or the other donor locus content at both donor loci. These two strains were crossed and random spores were searched for the h90 genotype by screening for the iodine-staining phenotype of colonies of segregants. Such a segregant would result from crossover during meiosis in the K region and would combine both kanR cassettes into a single strain. However, no such segregant could be obtained among 4000 random spores analyzed. This was expected due to the "cold spot" of recombination in the K region (EGEL 1984 ). In an attempt to promote recombination in the K region, the cross was performed in the presence of Trichostatin A (TSA), a histone deacetylase inhibitor. This treatment is known to disrupt the chromatin structure in the K region, which is thought to prohibit recombination (GREWAL et al. 1998 ; OLSSON et al. 1998 ). Among 3000 random spores analyzed from such a cross, nine independent h90 segregants were obtained. This showed that TSA was able to "open up" the cold spot, allowing recombination. From the nine h90 strains thus obtained, one was chosen for further analysis. After confirming its genetic constitution by Southern analysis (data not shown), expression of mat3-Mint::ura4+ was analyzed by growing it on Ura-deficient and FOA-containing media plates. The resulting strain (PSG316) did not exhibit silencing as it exhibited the Ura+ phenotype (data not shown).
| DISCUSSION |
|---|
The key observation made in this study is the identification of perfect inverted repeats flanking a silenced region of two different species. We were interested in determining whether sequences similar to these repeats exist elsewhere in the S. pombe genome, especially near centromeric and telomeric regions, as these domains also exhibit position-effect control analogous to that of the mat region (GOTTSCHLING et al. 1990 ; NIMMO et al. 1994 ; ALLSHIRE 1996 ). However, a search of sequence databases revealed that similar sequences are not present either in the S. pombe genome or among known sequences from other species. We were particularly struck by the 100% sequence identity of a >2-kb DNA sequence of these repeats. Remarkably, this 100% identity is maintained not only in S. pombe but also in S. kambucha. When the sequence of repeats is compared between two species, they are not identical and show the same level of sequence polymorphism as for sequences from another region located to the left of the mat2 region and the mat3-M locus. This shows that the sequence of the repeat region is permitted the same level of divergence during evolution as for other nearby regions tested. Despite that divergence, sequences of the proximal and distal repeats are kept identical in each species. This sequence conservation suggests that these two repeats may interact with each other to maintain 100% identity, especially considering the fact that these repeats do not encode an open reading frame (ORF). Parallels of our results can be drawn with those of the S. pombe centromeres that are organized into special chromatin structures and display the phenomenon of silencing (ALLSHIRE 1996 ). Centromeric regions comprise a 4.7-kb central core, which is flanked by imr inverted repeats. These inverted repeat sequences are not conserved between different chromosomes. Interestingly, the left and right repeats are absolutely identical within the same chromosome, further strengthening the role of similarity in the mechanism of silencing (TAKAHASHI et al. 1992 ; STEINER and CLARKE 1994 ). Similarity of the mechanism of silencing at the centromeres with that at the mating-type loci is further supported by the finding that the trans-acting silencing factors, such as Clr1–Clr4 and Swi6, first implicated in silencing of the mat2,3 region, are also involved in silencing of the centromeric regions (ALLSHIRE 1996 ).
Recently, several studies from a range of organisms have implicated repeat sequences in inducing silencing (DORER and HENIKOFF 1994 ; MATZKE et al. 1994 ; YE and SIGNER 1996 ; STAM et al. 1997 ; GARRICK et al. 1998 ; MITTELSTEN SCHEID et al. 1998 ). Many studies implicated RNA molecules generated from transcription of repeats as a means of imparting silencing in trans on an ectopically located gene, possibly through RNA interference (WASSENEGGER et al. 1994 ; MATZKE and MATZKE 1995 ; JONES et al. 1998 ; METTE et al. 1999 ). Silencing in trans does not seem to be applicable in the case of IRL and IRR as they do not encode any identifiable ORF and such repeat sequences do not exist elsewhere in the genome. However, the possibility of generation of noncoding transient RNA from these repeats cannot be ruled out. In plant systems it has been argued that promotorless transgenes in inverted orientation can trigger their own methylation without the need for RNA transcripts (LUFF et al. 1999 ). It has been suggested that this de novo methylation is induced by DNA/DNA pairing. Heavy methylation is observed around the center of inverted repeats, possibly due to intrastrand base pairing between palindromic inverted repeat sequences (STAM et al. 1998 , STAM et al. 2000 ). Pairing between repeats has also been shown to serve as a signal for heterochromatin assembly and associated de novo methylation in the filamentous fungi Neurospora crassa and Ascobolus immerses (SELKER 1999 ). Although DNA methylation is not observed in S. pombe, it is possible that such DNA/DNA pairing may act as a signal for inducing silencing by recruiting factors such as Swi6 and Clr4. Such a structure may constitute the substrate for cytosine methylation of DNA in higher systems.
These considerations have prompted us to propose what we call the "handcuff model" (Fig 8) to explain the manner of action of IRL and IRR in silencing. In this model, we propose that the sequence identity between the repeats is maintained by an unusual intrastrand Watson-with-Watson and Crick-with-Crick pairing but still keeping the Watson:Crick (W:C) pairing of the intervening K region intact. A structure like that may form the basis of the cold spot of recombination due to intrachromosomal folding prohibiting interchromosomal interaction but favoring intrachromosomal interaction with mat1 for the latter's switching in cis (KLAR and BONADUCE 1991 ). Such an intrastrand pairing possibility has been considered in immunoglobulin 
V-, J-, and C-region genes (MAX et al. 1979 ) and in the plant system (STAM et al. 1998 ). Alternately, the repeats may pair in vivo by conventional means, which may be facilitated by sequence-specific DNA binding proteins. The finding of impairment of silencing when one of the repeats was deleted agrees with this model but more work is needed to test it. As deletion of one of the repeats or transposition of the distal repeat into the K region did not cause full derepression of the donor loci, multiple pathways of silencing must operate in this region, only one of which works through the IRL and IRR elements. Furthermore, the 4.3-kb sequence in the K region has repeat sequence elements and shows sequence identity with repeat elements in the centromere of chromosome II (GREWAL and KLAR 1997 ). Such an arrangement may promote silencing by a shared pathway in the mat2,3 region as well as in the centromere region and the IRL and IRR repeats may primarily act as boundary elements to delimit the silenced domain.
View larger version (14K):
In this window
In a new window
Download PPT slide
|
Figure 8.
A hypothetical "handcuff model" to pair inverted repeat sequences flanking the mat2,3 silent region. We propose that the repeats may maintain their sequence identity by unusual intrastrand Watson-with-Watson and Crick-with-Crick pairing. This pairing could be maintained by factors that bind to these repeat sequences. This unusual pairing could act as a barrier to spreading of silencing proteins bound to the intervening K region.
|
|
In plants, repeat-induced silencing is not specific to their sequence. Any repeat sequence, either in tandem or in inverted orientation, can induce silencing (LUFF et al. 1999 ). When both IRL and IRR were replaced by another sequence (kanR cassettes) of almost equal length and in inverted orientation, silencing was not restored. Thus, in S. pombe, IRL and IRR sequences cannot be substituted with other sequences, signifying the importance of these elements for the phenomenon of silencing. This suggests that if IRL and IRR pair in vivo, then either pairing of these repeats is dependent on some other factors that specifically bind to these sequences, or if pairing can occur because of sequence identity, paired structure alone is not sufficient for instigating silencing. Some other sequence-specific factors are presumably required to incite silencing by an unknown mechanism. However, we note the caveat that it is also possible that kanR cassettes are unable to restore silencing because they are actively transcribed. This transcription activity may interfere with their interaction and thus impair their ability to silence.
While this work was in progress, an independent study by NOMA et al. 2001 proposed that IRL and IRR act as boundary elements for heterochromatin assembly, as these were shown to restrict spreading of Swi6 bound in chromatin in regions flanking the repeat sequences. We also tested a similar model and observed that overexpression of swi6 can make mat3-Mint::ura4+ go silent in a strain that was deleted for IRR (data not shown). Our rationale was that deletion of IRR may permit chromatin-bound Swi6 to spread to the adjoining regions, resulting in its dilution at the mat3-Mint::ura4+ locus, thus allowing it to be transcriptionally active; but overexpression of Swi6 is able to compensate for that. In their experiments, NOMA et al. 2001 used Kint2::ura4+ as reporter and observed that deletion of repeat sequences did not derepress it. The Kint2::ura4+ reporter was constructed by inserting a copy of the ura4+ gene at a HindIII site in the middle of the K region (GREWAL and KLAR 1997 ). This site is 5.3 kb distal to mat2 and 5.6 kb proximal to mat3-M. In our study we used either mat2-Pint::ura4+ (ura4+ inserted 400 bp distal to mat2-P) or mat3-Mint::ura4+ (ura4+ inserted 150 bp distal to mat3-M) and obtained slightly different results, owing to the differences in the location of the marker genes employed in these studies. The IRR deletion was able to derepress mat3-Mint::ura4+ to an extent that the resultant strain was prototrophic for the Ura phenotype. However, when IRL was deleted, expression from mat2-Pint::ura4+ was variegated. The difference in ura4+ expression on the basis of its location in the silenced region suggests that there is a gradient for silencing in this 16.0-kb interval. Silencing is more pronounced in the middle of the K region (as Kint2::ura4+ is more suppressed) and less and less further away from it in inverted-repeat-deleted strains. It is possible that centromeric repeat sequences in the K region act as a nucleation point for assembly of silencing proteins that then spread to adjoining regions on both sides.
Interestingly the stability of PEV from mat2-Pint::ura4+ was dependent upon cell type, as both Ura-on and Ura-off epistates were much more stable in an M strain than in a P strain. This could be due to cell-type-specific differences in the organization of chromatin structure in different cell types. This feature may account for preferential donation of information of opposite type to mat1 in h90 cells but homologous information switching in h09 cells in the phenomenon called directionality (THON and KLAR 1993 ). Further work is needed to define the cell-type-related changes in the organization of chromatin in and around the donor loci.
Other interesting results from this study came from analysis of genetic interaction between repeat sequences and other trans-acting factors. In previous studies it has been proposed that Swi6, Clr1, Clr2, Clr3, and Clr4 act in the same pathway (EKWALL and RUUSALA 1994 ; THON et al. 1994 ). The model proposed to explain the role of various factors implies that Clr6 and/or Hda1 deacetylate H3 Lys9 and Clr3 deacetylates Lys14 of histone H3 followed by methylation of Lys9 by Clr4. The modified Lys9 is then recognized by Swi6 to compile a self-propagating heterochromatin assembly (NAKAYAMA et al. 2001 ). Although this model does not necessarily explain the role of Clr1 or Clr2, other genetic studies have indicated that Clr1 and Clr2 also act in the same pathway, as mutations in genes encoding these factors do not show synergy in phenotype when combined with mutations in swi6, clr3, or clr4 (THON et al. 1994 ). Furthermore, all these factors behave similarly in genetic interactions with deletions of cis-acting sites. For example, when the BglII-BssHII deletion is combined with mutation in any of these factors, it results in additional derepression of mat2-P (THON et al. 1994 ). Similarly, pairwise combinations of deletion of the mat3-M repressor element with mutation in any of these factors cause profound silencing defects (THON et al. 1999 ). Therefore, it was surprising to observe that these factors genetically interact differently with IRL. Results of pairwise combinations of mutations in each of these factors with IRL deletion divided them into three groups. One group consisting of Clr1 and Clr3 showed a strong additive effect in derepressing mat2-P, while the second group consisting of Clr4 and Swi6 did not show any additional effect. These observations suggest that Clr4 and Swi6 could be acting through different cis-acting elements but Clr1 and Clr3 function by their interaction with IRL and IRR. The most intriguing result was from the clr2 IRL and clr2 IRR double mutant, as deletion of either IRL or IRR was able to mask the effect of mutation in clr2. Mutation in clr2 causes a silencing defect, resulting in a low level of staining in a mat1-Msmt0 strain due to expression of the mat2-P genes. However, when it was combined with IRL or IRR, the resulting strain, for reasons not yet known, did not show any staining. This paradox can probably be resolved once the function of Clr2 is understood. Additional genetic and biochemical studies examining the interaction of these factors with each other and with other cis-acting sequences, including IRL and IRR, are needed to provide further insight into the mechanism of transcription repression at this region. Clearly, genetic and molecular studies of S. pombe have provided one of the best paradigms for understanding gene regulation by heterochromatin assembly in eukaryotes.
Last, the phylogenetic closeness of S. kambucha to S. pombe is indicated by the DNA sequence, restriction analysis of mat cassettes, DNA-DNA hybridization, successful cross-species matings between cells, meiosis, and sporulation of the resulting zygotes. In further studies we wish to exploit the DNA sequence polymorphism in S. kambucha as compared with S. pombe to further define the mechanism of silencing and switching at the mat region and silencing of the centromeric regions, as well as other aspects of the biology of these organisms.
| ACKNOWLEDGMENTS |
|---|
We thank Ewy Mathe for help with S. kambucha DNA sequencing. We also thank Jeffery N. Strathern and other members of the Gene Regulation and Chromosome Biology Laboratory for valuable suggestions and M. Grau for editorial help. This work was sponsored by the National Cancer Institute, Department of Health and Human Services. The contents of this publication do not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organization imply endorsement by the U.S. Government.
Manuscript received March 15, 2002; Accepted for publication July 1, 2002.
| LITERATURE CITED |
|---|
ALLSHIRE, R. C., 1996 Transcriptional silencing in the fission yeast: a manifestation of higher order chromosome structure and functions, pp. 443–466 in Epigenetic Mechanisms of Gene Regulation, edited by V. E. A. RUSSO, R. A. MARTIENSSEN and A. D. RIGGS. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.
ARCANGIOLI, B. and A. J. S. KLAR, 1991 A novel switch-activating site (SAS1) and its cognate binding factor (SAP1) required for efficient mat1 switching in Schizosaccharomyces pombe.. EMBO J. 10:3025-3032.[Medline]
AYOUB, N., I. GOLDSHMIDT, and A. COHEN, 1999 Position effect variegation at the mating-type locus of fission yeast: a cis-acting element inhibits covariegated expression of genes in the silent and expressed domains. Genetics 152:495-508.[Abstract/Free Full Text]
AYOUB, N., I. GOLDSHMIDT, R. LYAKHOVETSKY, and A. COHEN, 2000 A fission yeast repression element cooperates with centromere-like sequences and defines a mat silent domain boundary. Genetics 156:983-994.[Abstract/Free Full Text]
BAHLER, J., J. Q. WU, M. S. LONGTINE, N. G. SHAH, A. MCKENZIE, 3RD et al., 1998 Heterologous modules for efficient and versatile PCR-based gene targeting in Schizosaccharomyces pombe. Yeast 14: 943–951.
BARTOLOMEI, M. S. and S. M. TILGHMAN, 1997 Genomic imprinting in mammals. Annu. Rev. Genet. 31:493-525.[Medline]
BEACH, D., 1983 Cell type switching by DNA transposition in fission yeast. Nature 305:682-688.
BEACH, D. H. and A. J. S. KLAR, 1984 Rearrangements of the transposable mating-type cassettes of fission yeast. EMBO J. 3:603-610.[Medline]
BOEKE, J. D., 1987 5-Fluoroorotic acid as a selective agent in yeast molecular genetics. Methods Enzymol. 154:164-175.[Medline]
BONIFER, C., 2000 Developmental regulation of eukaryotic gene loci: Which cis-regulatory information is required? Trends Genet. 16:310-315.[Medline]
BRESCH, C., G. MULLER, and R. EGEL, 1968 Genes involved in meiosis and sporulation of a yeast. Mol. Gen. Genet. 102:301-306.[Medline]
CAI, H. N. and P. SHEN, 2001 Effects of cis arrangement of chromatin insulators on enhancer-blocking activity. Science 291:493-495.[Abstract/Free Full Text]
CHADWICK, D. J., and G. CARDEW, 1998 Epigenetics. John Wiley, Chichester, UK.
DORER, D. R. and S. HENIKOFF, 1994 Expansions of transgene repeats cause heterochromatin formation and gene silencing in Drosophila.. Cell 77:993-1002.[Medline]
EGEL, R., 1981 Mating-type switching and mitotic crossing-over at the mating-type locus in fission yeast. Cold Spring Harbor Symp. Quant. Biol. 45:1003-1007.
EGEL, R., 1984 Two tightly linked silent cassettes in the mating-type region of Schizosaccharomyces pombe.. Curr. Genet. 8:199-203.
EGEL, R., 1989 Mating-type genes, meiosis and sporulation, pp. 31–73 in Molecular Biology of the Fission Yeast, edited by A. NASIM, P. YOUNG and B. JOHNSON. Academic Press, New York.
EGEL, R. and B. EIE, 1987 Cell-lineage asymmetry in Schizosaccharomyces pombe: unilateral transmission of high-frequency state of mating-type switching in diploid pedigrees. Curr. Genet. 12:429-433.
EGEL, R. and H. GUTZ, 1981 Gene activation by copy transposition in mating-type switching of a homothallic fission yeast. Curr. Genet. 3:5-12.
EGEL, R., M. WILLER, and O. NIELSEN, 1989 Unblocking of meiotic crossing-over between the silent mating-type cassettes of fission yeast, conditioned by the recessive, pleiotropic mutant rik1.. Curr. Genet. 15:407-410.
EKWALL, K. and T. RUUSALA, 1994 Mutations in rik1, clr2, clr3 and clr4 genes asymmetrically derepress the silent mating-type loci in fission yeast. Genetics 136:53-64.[Abstract]
EKWALL, K., O. NIELSEN, and T. RUUSALA, 1991 Repression of a mating type cassette in the fission yeast by four DNA elements. Yeast 7:745-755.[Medline]
GARRICK, D., S. FIERING, D. I. MARTIN, and E. WHITELAW, 1998 Repeat-induced gene silencing in mammals. Nat. Genet. 18:56-59.[Medline]
GOTTSCHLING, D. E., O. M. APARICIO, B. L. BILLINGTON, and V. A. ZAKIAN, 1990 Position effect at S. cerevisiae telomeres: reversible repression of Pol II transcription. Cell 63:751-762.[Medline]
TOP
ABSTRACT
MATERIALS AND METHODS
