|Home | About | Journals | Submit | Contact Us | Français|
In Saccharomyces cerevisiae, transcriptional silencing occurs at the cryptic mating-type loci (HML and HMR), telomeres, and ribosomal DNA (rDNA; RDN1). Silencing in the rDNA is unusual in that polymerase II (Pol II) promoters within RDN1 are repressed by Sir2 but not Sir3 or Sir4. rDNA silencing unidirectionally spreads leftward, but the mechanism of limiting its spreading is unclear. We searched for silencing barriers flanking the left end of RDN1 by using an established assay for detecting barriers to HMR silencing. Unexpectedly, the unique sequence immediately adjacent to RDN1, which overlaps a prominent cohesin binding site (CARL2), did not have appreciable barrier activity. Instead, a fragment located 2.4 kb to the left, containing a tRNAGln gene and the Ty1 long terminal repeat, had robust barrier activity. The barrier activity was dependent on Pol III transcription of tRNAGln, the cohesin protein Smc1, and the SAS1 and Gcn5 histone acetyltransferases. The location of the barrier correlates with the detectable limit of rDNA silencing when SIR2 is overexpressed, where it blocks the spreading of rDNA heterochromatin. We propose a model in which normal Sir2 activity results in termination of silencing near the physical rDNA boundary, while tRNAGln blocks silencing from spreading too far when nucleolar Sir2 pools become elevated.
In eukaryotic cells, genomic DNA exists as chromatin in association with histone octamers called nucleosomes and various other chromatin proteins. Chromatin structure varies along the chromosome, and this influences the state of gene expression. Based on such variations in structure and gene expression, chromatin can be broadly classified into euchromatin (transcriptionally active) and heterochromatin (transcriptionally repressed). Mechanisms that demarcate heterochromatin domains from euchromatin are actively being investigated (4, 16, 60), and several models have been proposed. For example, the boundary at telomeric heterochromatin in Saccharomyces cerevisiae is maintained by a balance between the opposing activities of Sir2 and Sas2, which deacetylate and acetylate histone H4-K16, respectively (28, 54). The right boundary at the cryptic mating-type locus HML is established by directional initiation and spreading of silencing inward, toward the mating-type genes, from the HML I silencer (3). In contrast, HMR silencing spreads bidirectionally from the silencers and is limited by sequences termed boundary elements or “silencing barriers” (14). The molecular mechanism of barrier activity is not well understood. Existing models include physical hindrance to the spreading of silencing proteins by bound multiprotein complexes, tethering to nuclear structural elements such as the scaffold or nuclear envelope, nucleosome exclusion, or recruitment of euchromatinizing/antisilencing factors that may modify chromatin at the barrier such that it is unfavorable for spreading of heterochromatin (6, 15, 25, 39, 40).
The rDNA (RDN1) locus, which has ~150 copies of a tandemly repeated sequence encoding rRNA genes on yeast chromosome XII, has an unusual organization. Within this array, genes transcribed by RNA polymerase I (Pol I) and Pol III are expressed, whereas RNA Pol II-transcribed genes are silenced (9, 49). The silencing at RDN1 is mechanistically distinct from silencing at other known loci (e.g., HM loci and telomeres) since it requires Sir2 but is independent of Sir3 and Sir4 (9, 21, 49). Ribosomal DNA (rDNA) silencing is also apparently less stable than HM and telomeric silencing and may rapidly switch between the repressed and derepressed states (49). rDNA silencing is orchestrated by the nucleolar RENT (regulator of nucleolar silencing and telophase exit) complex, which includes Sir2, Net1, and Cdc14 (45, 52) and is distinct from the SIR complex that induces TPE and HM silencing. In addition, rDNA silencing independently requires Set1, a protein that methylates histone H3-K4, and the Swi/Snf chromatin-remodeling complex (8, 17). RDN1 heterochromatin spreads unidirectionally toward the left and correlates with Sir2-dependent histone H3 and H4 deacetylation (10, 12). Spreading of rDNA silencing requires transcription by RNA Pol I, and the direction of spreading coincides with the direction of Pol I transcription, which proceeds leftward within the rDNA repeat (10). The unidirectional spreading of rDNA silencing into the unique sequence flanking the left (CEN-proximal) end of RDN1 can be further enhanced by overexpression of Sir2 (10).
Although progress in determining the mechanism of rDNA silencing has been made, it is still unclear what limits the spread of RDN1 heterochromatin in order to prevent it from encroaching on neighboring essential genes. The unique sequence flanking the left end of RDN1 appears to be gene free for ~2.4 kb, until a retrotransposon long terminal repeat (LTR; YLRWΔ6), the tRNA gene (tDNA) tQ(UUG)L, and a small gene that encodes a subunit of RNase H (RNH203). These precede the nearest essential gene, ACS2 (Fig. (Fig.1A).1A). ACS2 encodes an acetyl coenzyme A (acetyl-CoA) synthetase isoform that generates nuclear acetyl-CoA required for histone acetylation and global transcription. Conditional inactivation of Acs2 results in global histone deacetylation and transcriptional defects, suggesting that acetyl-CoA metabolism impacts chromatin regulation (56). Thus, it would be important for ACS2 to be shielded from the repressive effect of rDNA silencing, making it likely that there exists a protective regulatory mechanism that stops the spread of rDNA silencing before it reaches ACS2.
Previous studies have suggested several potential mechanisms for limiting the RDN1 heterochromatin domain. These include confinement of silencing by a cohesin-dependent silencing barrier element present at the left rDNA/unique DNA transition; this possibility is intriguing because a prominent cohesin binding site, CARL2 (cohesin-associated region L2), is located at that position (32) (Fig. (Fig.1A).1A). Cohesin is required for the silencing barrier activity of another silencing barrier in yeast, the HMR right boundary (HMR-RB) (14). The cohesin-dependent HMR-RB consists of a tRNAThr gene and LTR, and interestingly, a tRNAGln gene and LTR are also present in the unique sequence flanking the left end of the rDNA. Alternatively, since spreading of rDNA silencing is dependent on Pol I transcription, termination of spreading may occur concomitantly with the termination of Pol I transcription at the terminator site. Another possible mechanism for limiting rDNA silencing is that the relatively long gene-free region flanking the rDNA acts as a buffer to the spreading of silencing, such that when Sir2 activity increases, it spreads into the flanking gene-free region but does not reach the essential ACS2 gene (10). In this study, we have investigated the possible mechanisms for limiting the spreading of rDNA silencing at the RDN1 left boundary on yeast chromosome XII and found evidence of a functional barrier element.
The S. cerevisiae strains used for testing HMR boundary activity in this study were a gift from Rohinton Kamakaka and David Donze and have been described previously (14, 15). The strains used in this study are listed in Table Table1.1. Barrier test plasmids were constructed in parental plasmid pRO363, which was also provided by Rohinton Kamakaka. The plasmids used in this study are listed in Table Table2.2. To produce rDNA silencing reporter strains containing the mURA3-HIS3 cassette at specific locations surrounding the tRNA-Ty1 LTR barrier region, a PCR was used to amplify the cassette from pJSS51-9 as previously described (10, 49). The 5′ sequences of the PCR primers contained 40 bp of homology to the target site that facilitated homologous recombination of the cassette at the proper position and orientation without deleting any of the chromosome XII sequence. The exception was the 2697L position (Saccharomyces Genome Database [SGD] coordinate 448722), where the tRNAGln gene was deleted and replaced with the mURA3-HIS3 cassette. Following transformation of the parental JB740 yeast strain with the proper PCR products, integration events were selected by colony growth on plates containing synthetic complete medium without histidine (SC-His). The position of the reporter gene was then confirmed by PCR. The position of each mURA3-HIS3 insertion is indicated by the number of nucleotides left of the rDNA NTS1 sequence that defines the left edge of the array (SGD coordinate 451419). For example, 50L and 300L strains have the reporter 50 and 300 bp to the left of the rDNA. The oligonucleotide sequences used for the insertions are available upon request. YSB519 contains the mURA3-HIS3 cassette positioned at the nonsilenced TRP1 locus (10).
An in vivo site-directed mutagenesis technique (delitto perfetto) was used to delete the tRNAGln gene from JB740 (51), followed by insertion of the mURA3-HIS3 cassette into the 2597L or 2868L position to generate strains NM39 and NM40, respectively. A 427-bp fragment of genomic DNA containing the tRNAGln and the Ty1 LTR (SGD coordinates 448623 to 449050) was ligated into the SacI site adjacent to mURA3 in pJSS51-9 to generate pNM3. As a control, a 427-bp fragment of X174 phage DNA was ligated into the SacI site to generate pNM1. The integrity of the tRNA gene was confirmed by DNA sequencing. The pNM1 and pNM3 plasmids were used as templates for PCRs to integrate the mURA3-HIS3 cassette with the boundary or stuffer fragment into the 300L position (SGD coordinate 451119) of JB740 to generate strains NM31 and NM27, respectively. To induce spreading of rDNA silencing, each reporter strain was transformed with either an empty LEU2 2μm plasmid (pRS425) or a related plasmid, pSB766, that contains the SIR2 gene (10).
The quantitative mating assay was performed as described by Donze et al. (14). Derivatives of a plasmid (pRO363) with the URA3 selectable marker and bearing a modified version of the HMR locus in which the I silencer has been deleted were introduced into strain ROY113, which is MATα his3-1 and has a deletion of the endogenous HMR locus. Upon mating with JRY19 (MATa his4-519), the resultant diploids are His+ and can be selected on histidine omission plates. The mating efficiency of cells bearing the pRO-derived plasmids, which can be selected on SC-Ura plates, was determined by mating with JRY19 and selecting for diploids on SC-Ura-His plates (indicative of mating-competent cells among the total plasmid-bearing cells growing on SC-Ura plates). The assay was also performed in a quantitative fashion with the following modification. Serial dilutions of MATα yeast cells (ROY113) grown to an optical density (OD) of 1 and containing the pRO363 derivatives of the tDNA barrier constructs were mated with an equal volume of a 10-fold-concentrated culture of tester strain JRY19. The mating mixtures were spotted onto SC-Ura (to estimate the total number of cells per CFU with the test DNA) and SC-Ura-His (to select for diploids) plates and incubated at 23°C for greater than 3 days. Mating efficiency was calculated as the percentage of mating-competent cells (i.e., the number of diploid colonies on SC-His-Ura plates) with respect to the total number of cells (the number of colonies on SC-Ura). Three or more transformants derived from each plasmid construct were tested for barrier activity. In the case of temperature-sensitive mutants, the mating assays were done at 30°C by using various amounts of each test culture with an OD at 600 nm of 1.0, including 10-fold-concentrated, undiluted, and a 1:10-diluted samples.
In order to introduce mutations into box B of tRNAGln, primers with the required mutations were synthesized by Sigma-Genosys and used for PCR amplification (2) of fragments with the tDNA bearing the desired mutation. PCR products bearing the desired mutations were cloned, and the presence of the mutation was confirmed by DNA sequencing.
Strains containing the mURA3-HIS3 or modified cassettes left of the rDNA array were patched onto SC or SC-Leu plates (depending on whether they contained the pRS425 and pSB766 LEU2 plasmids) and incubated overnight. The patched cells were scraped from the plate and resuspended in sterile water. Each cell suspension was normalized to an OD at 600 nm of 1.0 with a Shimadzu UV-1201S spectrophotometer and then serially diluted fivefold in a 96-well plate. Five-microliter volumes of the dilution series were spotted onto SC or SC-Leu plates to control for overall cell growth and onto SC-Ura or SC-Leu-Ura plates to monitor the relative expression of the mURA3 reporter gene. Photos of the SC and SC-Leu plates were taken after 2 days of growth, and photos of the SC-Ura and SC-Leu-Ura plates were taken after 3 or 4 days of growth, as indicated. All incubations were at 30°C.
The rDNA encodes rRNA genes whose transcription is driven by RNA Pol I (18S, 5.8S, and 25S rRNAs, transcribed as a 35S rRNA transcript) and RNA Pol III (5S rRNA). These genes are organized in a 9.1-kb span of rDNA that is tandemly repeated ~150 times in the rDNA locus (RDN1) on yeast chromosome XII (Fig. (Fig.1A).1A). Pol I transcription proceeds leftward and also helps in the establishment of Sir2-dependent silencing of RNA Pol II transcription in this region (10, 12). The unique chromosome XII sequence flanking the left end of RDN1 has a gene-free stretch of nearly 2.4 kb. At the junction of RDN1 and the single-copy flanking region is located a prominent cohesin binding site, CARL2, which has been postulated to be a candidate for a silencing barrier in this region (32). RDN1 silencing can spread into the left flank, but the mechanism by which the spreading is halted has remained unknown.
We investigated whether DNA fragments present at or near the left edge of the rDNA tandem array (Fig. (Fig.1A)1A) have silencing barrier activity by testing their ability to counter the spreading of silencing in a previously described assay for silencing barrier activity (14). This assay measures the ability of a putative barrier DNA sequence to prevent the silencing of a mating-type reporter gene (MATa1) when the barrier sequence is positioned between the nearby HMR E silencer and the reporter (Fig. (Fig.1B).1B). A barrier sequence in this assay halts the spreading of heterochromatin from the silencer into MATa1, resulting in the expression of a1 and rendering MATα yeast cells harboring this construct mating incompetent. Mating in this system is detected by growth on SC plates lacking uracil and histidine (SC-Ura-His), which select for plasmid-bearing Ura+ cells that mate with a strain of the opposite mating type to produce His+ diploids. As shown in Fig. Fig.1C,1C, the sequence present at the junction of rDNA with the leftward unique sequence, which overlaps the CARL2 cohesin binding site, did not have appreciable barrier activity, as evidenced by the production of diploid colonies selected on SC-Ura-His plates in the mating assay. Among the sequences tested in this fashion, only a 427-bp fragment located nearly 2.4 kb to the left of the rDNA boundary had silencing barrier activity. MATα cells bearing this sequence interposed between the E silencer and the MATa1 reporter were mating incompetent (Fig. (Fig.1C),1C), demonstrating that this sequence has robust barrier activity. The DNA fragment has a gene encoding tRNAGln [tQ(UUG)L or tRNAGln], a Ty1 retrotransposon LTR, and some intervening sequence.
Previous findings have implicated RNA Pol III-driven transcription of tRNA genes as a requirement for the antisilencing property of this class of barrier elements (15). Since the DNA fragment in question included a tRNA gene, we tested whether its barrier activity was dependent upon its RNA Pol III-dependent transcriptional status. Transcription by RNA Pol III at tRNA gene promoters begins with the assembly of the TFIIIC transcription factor at the internal box A and box B promoter elements, followed by recruitment of TFIIIB (TATA binding protein-containing complex) and RNA Pol III (20) (Fig. (Fig.2A).2A). We measured the barrier activity of the tRNAGln-containing fragment in the previously described tfc3G349E mutant (mutated in the TFIIIC DNA binding component), brf-11.9 mutant (mutated in the gene encoding the 70-kDa component of TFIIIB), and rpc31-236 mutant (mutated in a Pol subunit that is defective in initiation of Pol III transcription but can assemble the preinitiation complex) (15). A semiquantitative version of the previously described patch-mating assay was employed in which serial dilutions (or varying numbers) of the MATα yeast cells (ROY113) containing derivatives of the tDNA barrier constructs were mated with an excess of tester strain JRY19. This variation enabled us to estimate the mating efficiency of various strains and thereby obtain a measure of the extent of abrogation of barrier activity. The barrier activity of HMR-tRNAThr, a previously described RNA Pol III-dependent barrier, was also tested in parallel as a positive control. The barrier activity of tRNAGln was reduced in all three mutants (although to various extents), as was also the case for HMR-tRNAThr (Fig. (Fig.2B).2B). It should be noted that the loss of barrier activity in the tfc3G349E mutant was observed more readily when the plasmids were episomal (as shown in Fig. Fig.2B),2B), and there was no detectable loss of barrier activity with either RDN1-tRNAGln or HMR-tRNAThr in the tfc3G349E mutant when the plasmids were integrated (data not shown).
The cis requirements for transcriptional regulation of tRNA genes have been very well characterized (30, 37). Specific base changes in the internal promoter elements (also called internal control regions [ICRs]) box A and box B disrupt the assembly of the RNA Pol III transcription apparatus at the promoters and abrogate transcription. For example, a C56G mutation in the SUP4-o tRNATyr gene dramatically reduces SUP4-o gene transcription such that no transcript is detectable, even after prolonged incubation in an in vitro transcription reaction (30). Likewise, a C56G transition in box B of the SUP53-tRNALeu gene severely abrogates transcription complex assembly on its promoter such that no footprint is detected and transcription is abolished (37), further highlighting the importance of this conserved residue in tDNA transcription. It has previously been reported that a C56G promoter mutation in HMR-tRNAThr hampers its barrier activity (15). Thus, in order to determine whether inactivating transcription of the RDN1-tRNAGln gene would affect its silencing barrier activity, we created a mutation at the equivalent position in its box B (C56G) promoter element (Fig. (Fig.2C).2C). This mutation significantly abrogated the barrier activity of the RDN1-tRNAGln barrier sequence (Fig. (Fig.2D),2D), similar to the results obtained with the HMR-tRNAThr (C56G) barrier control. Likewise, a similar transcription-inactivating mutation (C56G) in box B of a third tDNA barrier, TRT2-tRNAThr (present upstream of STE6 and acting as a barrier to α2 operator-mediated repression ), also abrogated its barrier activity in the context of the HMR E silencer-based barrier assay (data not shown).
Histone modifications and chromatin remodeling brought about by various chromatin-modifying complexes can also contribute to the euchromatinizing property (barrier activity) of silencing barrier sequences. The barrier activity of the HMR-RB is dependent on HATs Sas2 and Gcn5 (15). Furthermore, HATs and numerous other chromatin-associated proteins, which include other chromatin-modifying and -remodeling complexes, have the potential to restrict heterochromatin by acting as barrier proteins when artificially tethered to chromosomes (39).
To test whether histone acetylation is required for the barrier activity of RDN1-tRNAGln, we estimated its barrier activity in mutants defective for various HAT activities. The MYST-like HAT Sas2 acetylates N-terminal lysine 16 of histone H4 (H4-K16) (19, 41) and is known to affect silencing at HMR, HML, RDN1, and telomeres (34, 41). Sas2, Sas4, and Sas5 make up the SAS1 (something about silencing) complex in budding yeast (44, 55, 63) and are locus-specific regulators of silencing (64). We quantified the abrogation of barrier activity of RDN1-tRNAGln and HMR-tRNAThr in sas2, sas4, and sas5 mutants. The barrier activity of HMR-tRNAThr, which is known to be dependent on the SAS1 complex, was strongly abrogated, as expected (Fig. (Fig.3).3). Likewise, the barrier activity of RDN1-tRNAGln was strongly impaired in these mutants (Fig. (Fig.3).3). We also tested the dependence of the barrier activity on Gcn5, a HAT of the SAGA complex, which acetylates histones H3 and H4 (31, 36, 65). Once again, the barrier activity of RDN1-tRNAGln and HMR-tRNAThr was severely abrogated (Fig. (Fig.3).3). In esa1 (defective in HAT of NuA4 complex) and sas3 (defective in HAT of NuA3 complex) mutants (1, 23, 57), the barrier activity of HMR-tRNAThr was reduced, whereas that of RDN1-tRNAGln was not affected (Fig. (Fig.33).
Some aspects of chromosome structural organization that depend upon nonhistone chromatin proteins may also be important for rDNA-silencing barrier activity. For example, the barrier activity of the HMR-RB is dependent on the cohesin protein Smc1 (14). Cohesin is a multisubunit complex consisting of two Smc (structural maintenance of chromosomes) proteins, Smc1 and Smc3, and two non-Smc proteins, Scc1 (Mcd1) and Scc3, and is required for sister chromatid cohesion (29, 53). Cohesin associates with chromosomes at multiple sites termed CARs (cohesin-associated regions) (7, 32, 58). We have previously reported that the cohesin protein Mcd1 is associated with HMR and is enriched at sites CARC3 and CARC4, which flank this locus and partially overlap the genetically mapped HMR boundary elements (32). The barrier activity of the HMR-RB was lost in the smc1-2 mutant (14), but the exact role of cohesin in barrier activity is not understood. In order to test if cohesin dependence is a general requirement for the barrier activity of tDNAs in S. cerevisiae, we tested whether the barrier activity of RDN1-tRNAGln was defective in the smc1-2 mutant. As shown in Fig. Fig.4,4, the barrier activity of RDN1-tRNAGln was reduced in the smc1-2 mutant at the permissive temperature (30°C), similar to the barrier activity of the HMR-tRNAThr positive control. The smc1-2 mutants were inviable at the nonpermissive temperature, as expected (Fig. (Fig.4).4). As in the tfc3G349E mutant, the reduction of barrier activity of RDN1-tRNAGln in the smc1-2 mutant was observed when the plasmid used in the assay was maintained episomally (Fig. (Fig.4)4) but not when it was integrated (data not shown).
The mechanism of delimiting the left boundary of rDNA silencing has remained uncharacterized. Earlier genetic studies indicated that rDNA silencing was detectable 300 bp, but not 600 bp or further, left of the rDNA array, suggesting that the natural boundary of spreading is located somewhere between these positions. However, binding of the Sir2 protein can be detected further to the left of the array, between 1,250 and 1,520 bp to the left of RDN1, which also correlates well with the zone of Sir2-dependent histone H3 and H4 deacetylation (10). It was therefore possible that RDN1 heterochromatin normally spreads further than 600 bp but that its strength is insufficient to detectably silence the reporter gene at these locations. Furthermore, the zone of detectable silencing is extended at least 2,000 bp left of the array upon Sir2 overexpression (10).
In order to define the rDNA silencing boundary more precisely and to further test the potential of the tRNAGln gene in limiting rDNA silencing in its natural context, we created new insertions of the silencing-sensitive dual-reporter cassette mURA3-HIS3 in the unique left flank of RDN1. These include insertions at 2347L (prior to the LTR), 2597L (between the LTR and tRNAGln), 2697L (at the start of tRNAGln), and 2868L (100 bp downstream from tRNAGln) (Fig. (Fig.5A).5A). The HIS3 gene enables selection of transformants, and the modified URA3 gene (mURA3) is used to measure silencing (49). Silencing of mURA3 was measured by determining colony formation efficiency on SC plates lacking uracil (SC-Ura) relative to growth on SC medium and in comparison with mURA3-HIS3 at the nonsilenced TRP1 locus. Significant silencing could be detected 50 and 300 bp from the rDNA, as expected (Fig. (Fig.5B).5B). However, insertions located further away from the rDNA and closer to the location of the potential barrier sequence did not appear to be silenced (Fig. (Fig.5B5B).
The extent of rDNA silencing depends on the nucleolar pools of Sir2 protein. The amount of Sir2 available for rDNA silencing can be experimentally modulated by overexpression of SIR2. Overexpression of SIR2 strengthens silencing in the left flank of the rDNA at least 2,000 bp to the left of the rDNA (10). In order to more precisely delineate the limit of spreading of enhanced rDNA silencing, we introduced a high-copy SIR2 plasmid into the new reporter strains bearing chromosomal insertions of mURA3-HIS3 closer to the putative barrier element. Overexpression of SIR2 extended the zone of silencing, as expected, and appreciably silenced mURA3 at 2,347 and 2,597 bp to the left of the rDNA (Fig. (Fig.5C).5C). There was very little silencing detected further to the left of the tRNAGln gene at position 2868L; the growth of cells bearing the mURA3-HIS3 cassette at position 2868L on SC-Leu-Ura plates was comparable to that of cells with the cassette inserted at TRP1, a nonsilenced locus, when Sir2 was overexpressed (Fig. (Fig.5D).5D). Furthermore, precisely replacing the tRNAGln gene with the mURA3-HIS3 cassette at position 2697L also resulted in silencing of mURA3 upon SIR2 overexpression (Fig. (Fig.5D).5D). Thus, the location of tRNAGln correlates with the zone of termination of rDNA silencing upon SIR2 overexpression.
In order to determine whether tRNAGln functions as a boundary at its endogenous location, we tested whether deletion of the tRNAGln sequence from chromosome XII enhances the spreading of silencing. The tRNAGln sequence was deleted in two strains bearing insertions of the dual reporter cassette on either side of tRNAGln, at position 2597L, where it is subjected to silencing upon SIR2 overexpression, and at position 2868L beyond the barrier, where it is unaffected by silencing upon SIR2 overexpression (Fig. (Fig.6A).6A). As shown in Fig. Fig.6B,6B, deletion of the tRNAGln sequence resulted in enhanced silencing of mURA3 at 2868L upon SIR2 overexpression, indicating that without the barrier, silencing can spread beyond its natural limit. However, since silencing this far away from the rDNA is rather weak, we also introduced the 427-bp boundary fragment containing the tRNAGln gene at the 300L position, where silencing is much stronger, between mURA3 and the rDNA (Fig. (Fig.6C,6C, schematic). A 427-bp fragment of X174 phage DNA was introduced as a negative control at the 300L position. Without overexpressing SIR2, silencing was weakened by insertion of the 427-bp sequence because it pushed the mURA3 reporter outside the normal zone of silencing. The loss of silencing was more severe with the X sequence, probably because tRNA genes can have SIR2-independent negative effects on the expression of adjacent Pol II transcribed genes (24). Upon SIR2 overexpression, mURA3 was more strongly silenced in the strain with the X control sequence interposed between the reporter and the leftward expanding RDN1 heterochromatin domain than in the strain with the barrier sequence inserted at the same location, revealing the boundary activity of tRNAGln even in a zone where rDNA silencing is strong (Fig. (Fig.6D).6D). We conclude from these results that the tRNAGln gene acts as a silencing boundary to the left (centromere proximal) of the rDNA tandem array.
Robust mechanisms for compartmentalizing the genome into domains varying in transcriptional potential (i.e., heterochromatin versus euchromatin) and preventing these domains from encroaching upon each other are crucial for maintaining the global epigenetic landscape. The spreading of silencing along chromosomes can be restricted by barrier or boundary elements. For example, the right boundary of HMR (14) and the left boundary of HML (5) are limited by silencing barrier sequences. Silencing barriers consisting of tDNAs have been found at the TRT2 locus near the STE6 gene in budding yeast, and tRNAAla near a centromere in fission yeast (43, 46). For the HMR-RB, it has been demonstrated that the genetic and biochemically defined boundary precisely coincides with the location of the silencing barrier (40). In this study, we investigated the mechanism of limiting the spreading of rDNA silencing. Interestingly, we find evidence for a dual mode of control of the RDN1 left boundary by a silencing barrier and Sir2 protein levels. We found a new silencing barrier sequence (with a tRNAGln and the Ty1 LTR) near the left end of the rDNA. However, unlike the other silenced loci limited by barrier elements (e.g., HMR), the location of this sequence does not coincide precisely with the previously determined genetic boundary of silencing (between 300 and 600 bp left of the rDNA array) but is located nearly 2.4 kb to the left. This raises two possibilities, first that the barrier sequence may exert its effect at a distance, e.g., by acting as an initiation site for the establishment of euchromatin, which could then spread toward the rDNA until it is opposed by the leftward-spreading RDN1 heterochromatin. The location of the boundary would therefore be defined by the relative local abundance (or balance) of antisilencing versus silencing factors. Second, the barrier sequence may be required as a backup mechanism to restrict silencing when rDNA silencing is enhanced. Consistent with this possibility, earlier work revealed that the RDN1 heterochromatin domain can be expanded (i.e., the natural boundary can be shifted) leftward by overexpression of SIR2 (10). This implies that there is no strong barrier element present at the natural boundary, which overlaps the CARL2 cohesin binding site located at the precise junction of the repetitive rDNA array with neighboring single-copy sequences. Our findings that CARL2 does not have barrier activity in the HMR silencing system and that the leftward extension of the zone of silencing upon Sir2 overexpression surpasses CARL2 lend credence to this hypothesis. Furthermore, our observations that the extended zone of silencing produced by overexpression of Sir2 terminates at the location of tRNAGln and is extended further in the absence of tRNAGln are consistent with the model in which the barrier element identified in this study is important for limiting the spread of silencing under conditions in which the nucleolar Sir2 pool increases.
RNA Pol III-dependent transcription of tRNA genes has been implicated as a requirement for barrier activity. Donze and Kamakaka have shown that the transcriptional potential of the HMR-RB tRNAThr is required for its barrier activity (15). More recently, RNA Pol III-mediated transcription of a tRNAAla gene in the centromere of Schizosaccharomyces pombe has also been shown to be required for its boundary function (43). However, not all RNA Pol III-transcribed genes act as transcriptional barriers, and the extent of the barrier activity of different tDNAs of the same class also varies in budding yeast (15). Our analysis of RDN1-tRNAGln shows that Pol III transcription of this tRNA gene affects its boundary activity in the context of HMR silencing. Interestingly, in S. pombe, recruitment of the transcription factor TFIIIC complex without RNA Pol III to inverted repeat boundary elements flanking the fission yeast mating-type heterochromatin domain is sufficient to prevent the spreading of heterochromatin (38). Similarly, TFIIIC binding sites can function as silencing boundaries and insulators to gene activation in S. cerevisiae (47).
Posttranslational modifications of histones are important for gene regulation and the establishment of specialized chromatin domains. According to the histone code hypothesis (26, 59), specific combinations of posttranslational modifications of histone tails form a distinct code interpreted by other factors, resulting in functionally specialized chromosomal domains. Silenced domains are characterized by the presence of hypoacetylated histones, and such domains are often separated from neighboring active chromatin by peaks of histone acetylation (62), which are hallmarks of euchromatin. Therefore, recruitment of HATs by barrier elements is one current model to explain the mechanism of tDNA barriers. In this study, we found that the barrier activity of tRNAGln and HMR-tRNAThr is maximally abrogated in the sas2 and gcn5 HAT mutants. However, sas3 and esa1 affect the barrier activity of HMR-tRNAThr and not tRNAGln, suggesting that these HATs could be specialized requirements for HMR-tRNAThr. Interestingly, Esa1 is required for rDNA silencing but affects HMR silencing only mildly (13), suggesting a plausible reason why it may not be a suitable silencing blocker for rDNA silencing. Thus, the HATs Sas2 and Gcn5 appear to be shared determinants of barrier activity for these two tDNAs (with respect to Sir2p-dependent HMR silencing). Another possible model to account for the requirement of HATs at silencing barriers could be that sequence-specific recruitment of HATs is unnecessary but the loss of barrier activity in HAT mutants may result from a global role for HATs in establishing euchromatin. Some HATs are involved in global histone acetylation. A global decrease in the acetylation of euchromatin in HAT mutants may expand heterochromatin domains by allowing heterochromatin to encroach upon a formerly euchromatin domain if the boundaries are defined by competition between the acetylation of euchromatin and deacetylation of heterochromatin. Alternatively, HAT-inactivating mutations that result in a global decrease in acetylation of euchromatin may also affect barrier function indirectly by reducing the levels of one or more silencing barrier proteins required for optimal barrier activity.
Sas2 acetylates lysine 16 of histone H4, which is deacetylated by the silencing protein Sir2. Thus, Sas2 antagonizes the effect of the histone deacetylase Sir2 (28, 54). Its recruitment (and that of other such euchromatinizing factors) by barrier elements, in order to counteract the histone deacetylation brought about by Sir2, is an attractive and tenable model for barrier function of tDNAs in budding yeast. In vertebrate cells, the USF1 protein-associated chicken beta-globin 5′ HS4 insulator element, which has silencing barrier activity, recruits histone-modifying enzymes that bring about histone acetylation and H3K4 methylation, thereby favoring euchromatinization (62, 66). Thus, recruitment of chromatin modifiers such as HATs may be a conserved theme to explain the action of silencing barriers in vertebrates and yeast chromosomes. A global genome-wide screen for barrier proteins in yeast also identified numerous chromatin modifiers, including HATs, as potential silencing blockers (39). Whether tDNAs in other organisms utilize such a HAT-dependent mechanism to delimit silencing has yet to be determined.
Mutation of SMC1, encoding a component of the cohesin complex, results in loss of the barrier activity of the HMR-RB, HMR-tRNAThr (14). This may be due to a direct involvement of bound cohesin in the establishment of a euchromatin-favoring chromatin structure at the boundary during the S and M phases of the cell cycle. Alternatively, cohesin may repress heterochromatin formation at HMR, such that in a cohesin mutant, silencing at HMR may be enhanced, enabling silencing proteins to spread beyond the barrier. Consistent with this idea, Lau et al. have shown that cohesin association inhibits the establishment of silencing at HMR (33). The requirement of cohesin for HMR-RB function more likely represents a specialized function of cohesin at this locus (18) and may be related to its robust association with the HMR-RB in combination with other euchromatinizing factors. Our observation that all CARs do not have barrier activity indicates that the mere association of cohesin may be insufficient for barrier activity. Cohesin-dependent barriers such as tDNAs may have additional specialized chromatin organization, which distinguishes them from CARs that do not perform such a function. In this context, it is intriguing that cohesin associates with the insulator protein CTCF at some sites on mammalian chromosome arms and mediates CTCF-dependent transcriptional insulation (42, 61). Whether cohesin and other factors found to be required for tRNAGln barrier activity in this study also function in the context of rDNA silencing at the endogenous chromosomal location has yet to be determined.
In conclusion, in this study, we have identified a new tDNA barrier in yeast which is present near the silenced RDN1 array. We further show that the activity of this new silencing barrier requires RNA Pol III-dependent transcriptional competence of tRNAGln, cohesin-dependent chromatin organization, and chromatin-modifying antisilencing factors such as HATs. It is likely that this sequence may be important for protecting nearby essential genes from being silenced under conditions where rDNA silencing is enhanced, e.g., when nucleolar Sir2 pools increase upon the impairment of telomeric silencing (e.g., in a sir4 mutant) (22, 50) or when there is surplus Sir2 in the cell due to deletion/loss of rDNA repeats. Sir2-dependent silencing has been shown to increase when the rDNA array size is spontaneously decreased (35) due to large rDNA deletions resulting in clonal variation in rDNA cluster size. Extrachromosomal rDNA circles have also been shown to be excised from the array in replicatively aging cells (48). The SIR complex that normally functions in silencing at telomeres and the HM loci has been shown to be redistributed to the nucleolus in aging yeast cells (27). Therefore, the demonstrated ability of the tRNAGln gene to block silencing mediated by the SIR complex (at HMR) or the RENT complex (at the rDNA) would ensure that the ACS2 promoter is protected from both classes of yeast heterochromatin. It is possible that this boundary element is an “insurance policy” that decreases the chances that the expression of the essential ACS2 gene will be affected by rDNA chromatin.
We thank David Donze and Rohinton Kamakaka for strains and plasmids used in this work. We thank David Donze for advice regarding the experiments with temperature-sensitive mutants and members of our laboratories for discussions.
Technical support from the Department of Biotechnology-funded DNA sequencing facility at the Indian Institute of Science (IISc) is acknowledged. This work was supported by an International Senior Research Fellowship in Biomedical Science (GR063263MA) from the Wellcome Trust, UK, to S. Laloraya and U.S. National Institutes of Health grant GM075240 to J. S. Smith. Fellowship support for M. Biswas was from the Council for Scientific and Industrial Research, India, and the Wellcome Trust.
Published ahead of print on 16 March 2009.