Ssn6-Tup1 regulates RNR3 by positioning nucleosomes and affecting the chromatin structure at the upstream repression sequence.

The DNA damage inducible gene ribonucleotide reductase (RNR3) is regulated by a transcriptional repression mechanism by the recruitment of the Ssn6-Tup1 corepressor complex to its promoter by the sequence-specific DNA-binding protein Crt1. Ssn6-Tup1 is reported to represses transcription by interfering with transcription factors, recruiting histone deacetylases, and positioning nucleosomes at the promoter of its target genes. Two of the three mechanisms involve effects on chromatin structure, and therefore, we have delineated the nucleosomal structure of RNR3 in the repressed and derepressed state using multiple nuclease mapping strategies. A regular array of positioned nucleosomes is detected over the repressed RNR3 promoter that extends into the coding sequence. Treating cells with DNA damaging agents or deleting CRT1, SSN6, or TUP1 derepresses RNR3 transcription, and causes a dramatic disruption of nucleosome positioning over its promoter. Furthermore, derepression of RNR3 correlated with changes in nuclease sensitivity within the upstream repression sequence (URS) region. Specifically, the loss of a MNase-hypersensitive site, and the appearance of strong DNase I hypersensitivity, was observed over the URS. Interestingly, we find that the binding of Crt1 to the promoter in the absence of Ssn6 or Tup1 is insufficient for nucleosome positioning or regulating chromatin structure at the URS; thus, these two functions are strictly dependent upon Ssn6-Tup1. We propose that RNR3 is regulated by changes in nucleosome positioning and chromatin structure that are mediated by Ssn6, Tup1, and Crt1.

Accommodating the large mass of DNA within the limited space of the nucleus necessitates its compaction into chromatin and other higher order structures (1,2), which inevitably has a pivotal influence on most, if not all, DNA metabolism-related activities such as transcription, DNA replication, recombination, and repair (2)(3)(4). It is widely accepted that packaging DNA into nucleosomes imposes a severe limitation on the accessibility of DNA to the transcription apparatus; therefore, the nucleosome plays an important role in the constitutive repression of gene transcription (4 -8). Nevertheless, the pack-aging of DNA into chromatin is not exclusively repressive in nature. In some cases, higher-order chromatin structures facilitate transcription activation by holding distant regulatory elements into juxtaposition with themselves or the core promoter (2,9) or by stabilizing the interaction of transcription factors to chromatin (10).
It is well recognized that chromatin is not a static structure, but rather a dynamic formation that appears to be dramatically altered or rearranged during gene activation in vivo (7,8,(11)(12)(13). Perhaps the most dramatic changes to chromatin associated with gene expression are the positioning and disruption of nucleosomes within the promoters of genes (1,5,6,8,14). The determinants of nucleosome positioning are poorly understood, but the requirement for the global co-repressor Ssn6-Tup1p (6, 14 -16), non-histone chromosomal proteins (17), SIR proteins (18), enzymatic activities (13, 19 -21), and DNA sequence (22,23) have been reported.
The mechanism of Ssn6-Tup1 mediated repression is an unresolved topic, and is considered to be controversial (6,14,16). It has even been proposed that Ssn6-Tup1 can positively affect transcription (24). Two generalized models exist for Ssn6-Tup1-mediated repression, one involves its interaction with transcription factors (for review, see Ref. 16), and the other its ability to control chromatin structure (6,14,(25)(26)(27). Studies have shown that the interaction between Ssn6-Tup1 and components of the RNA polymerase II holoenzyme complex is required for repression (28 -30); however, this idea has been challenged recently by others (31). Its interference with transcriptional activators has also been reported (16,32,33). In regards to controlling chromatin structure, Ssn6-Tup1 has been shown to bind to and recruit histone deacetylases complexes to promoters (26,27,34) and to position nucleosomes (6,14,15,25,35). The role of Ssn6-Tup1 in nucleosome positioning has been examined directly on only a few genes and the recombination enhancer of the mating type loci (15, 25, 36 -41). Moreover, even within this group only three classes of genes have been mapped, namely, mating type-specific genes, an oxygen-regulated gene, and carbon source regulated genes. Each of these classes differ in their requirement for SSN6 versus TUP1 (15,36,37,40), and thus, it remains to be seen if Ssn6 and Tup1 utilizes any one, or different combinations of, mechanism(s) to repress transcription at different loci.
The enzyme ribonucleotide reductase (RNR) 1 catalyzes the rate-limiting step in deoxyribonucleotide synthesis; thus, plays an essential role in DNA replication and repair (42,43). In Saccharomyces cerevisiae it is composed of four subunits, which are encoded by four DNA damage-regulated genes (RNR1, RNR2, RNR3, and RNR4) (42)(43)(44). Activation of the RNR genes in response to replication arrest and DNA damage requires signals relayed through the DNA damage checkpoint pathway (42,43,45,46). In addition to the DNA damage checkpoint kinase pathway, specific general transcription factor TFIID subunits, TAF II s, (47,48), SBF factor, and the Hrr25 kinase (49) are required for RNR gene expression. The RNR genes are repressed by upstream repression sequences (URS), the damage responsive elements (DREs) or x-boxes, which serve as binding sites for the sequence-specific DNA-binding protein Crt1p (50). The N terminus of Crt1 recruits the general co-repressor complex composed of Ssn6 and Tup1 (47,50) repressing gene expression. The exact role of Ssn6-Tup1, and the contributions of Crt1, in mediating repression of RNR3 is not known. Activation of DNA damage checkpoints results in the phosphorylation of Crt1, reducing its ability to cross-link to the promoter region of RNR3 (50). How the phosphorylation of Crt1 reduces its association with the promoter is unclear.
Here we describe a comprehensive analysis of the nucleosomal structure of the RNR3 promoter in the repressed and derepressed state. A combination of high-resolution and lowresolution micrococcal nuclease (MNase) and DNase I sensitivity mapping studies clearly demonstrate that in the absence of DNA damage, an array of positioned nucleosomes covers the promoter and extends into the coding sequence. Upon DNA damage, the nucleosome structure at the promoter undergoes extensive remodeling, which is dependent on the checkpoint genes MEC1 and RAD53. DNase I and MNase footprinting revealed changes in nuclease sensitivity within the URS that correlated with the expression of RNR3. Interestingly, we find that the chromatin/DNA structure within the URS is dependent upon Crt1, Ssn6, and Tup1, indicating that Crt1 alone is insufficient for its formation. Our analysis has established that nucleosome positioning and remodeling regulates DNA damage inducible genes, and that the predominant function of Crt1 is to position nucleosomes over the promoter via the Ssn6-Tup1 corepressor complex.
MNase and DNase I Mapping-Nuclei preparation was carried out essentially as describe in Refs. 40 and 53. Briefly, yeast cells from a 1-liter culture grown to an optical density of about 1.0 at 600 nm was harvested and digested with Zymolyase T100 (Seikagaku). Nuclei were purified by differential centrifugation and finally resuspended in digestion buffer (10 mM HEPES, pH 7.5, 0.5 mM MgCl 2 , 0.05 mM CaCl 2 ) and incubated with 0, 2, and 4 units/ml MNase (Worthington) or 0, 0.05, and 0.1 units/ml DNase I (Worthington) for 10 min at 37°C. The digestions were terminated by the addition of EDTA and the DNA was purified by RNase A and proteinase K digestion and phenol/chloroform extraction. The DNA pellet was resuspended in 0.1 ϫ TE buffer. For low-resolution mapping of nucleosomes by indirect end labeling, the purified DNA was subjected to a secondary digestion by PstI, then electrophoresed in 1.4% agarose gels in 1 ϫ Tris borate-EDTA buffer, and transferred to Zetabind membrane (CUNO industries). The specific DNA sequences were detected by hybridized with a random primed body-labeled probe directed toward the end of the PstI site. The following primer sets were used to amplify the probes: PstI (ϩ468) 5Ј-GCTAAGACTGAACGGT-GAAGTGGCAG, PstI (ϩ725) 5Ј-GGAAATCATAGCACATTCTTTCAAA-GTATC; EcoRV (ϩ57) 5Ј-CTCCCGTATCACCCGTTTGTC, EcoRV (ϩ540) 5Ј-CATGGATACCTAGCGCCACACGCATTAC. For high-resolution mapping, multiple rounds of Taq DNA polymerase-based primer extension was carried out from a 32 P-end-labeled primer, and the products were then resolved on a 6% polyacrylamide (19:1), 50% urea gel (40). Images were captured on a PhosphorImager screen. The primers used to perform the primer extension reaction are as follows: RNR3ϩ150 downstream, 5Ј-CTAAACCGTATGACAAACGGGTGATA-CGGGAGGT; RNR3-324 upstream, 5Ј-CGTGGTTGTCGCAGCAACG-ACACCTAGG; RNR3-586 upstream, 5Ј-GGCGCTGTGGCCGTGG-CTAGTTTCTTCT.
Restriction Endonuclease Accessibility Assay-Nuclei were isolated as for the MNase and DNase I mapping studies and resuspended in RE digestion buffer (10 mM Tris-HCl, pH 7.4, 50 mM NaCl, 10 mM MgCl 2 , 0.5 mM spermidine, 0.15 mM spermine, 0.2 mM EDTA, 0.2 mM EGTA, 5 mM ␤-mercaptoethanol) (54,55). MluI or NcoI (New England Biolabs) was added to concentrations of 100 and 400 units/ml, and the digestion was allowed to proceed for 60 min at 37°C. After purification, the DNA was digested with PstI (MluI-digested samples) or EcoRV (NcoI-digested samples) to completion. The products were resolved on agarose gels and detected by Southern blotting using the indirect endlabeling method. The PstI-and EcoRV-digested samples were hybridized to PCR-generated probes corresponding to the regions of ϩ486 to ϩ725 (PstI probe) and ϩ57 to ϩ540 (EcoRV probe) of RNR3, respectively. Blots were exposed to a PhosphorImager screen (Molecular Dynamics). Data was expressed as percent digested that was calculated by the ratio of the counts in the digested fragment to the total DNA.
Chromatin Immunoprecipitation Assay-The chromatin immunoprecipitation assay was performed essentially as described in two previous publications (50,56). Cultures were treated for 2.5 h with 0.03% MMS prior to cross-linking, where indicated. Briefly, a 200-ml culture of yeast grown in YPAD to an A 600 ϭ 1.0 were treated with formaldehyde (1% v/v) for 15 min at 23°C, followed by an additional 5 min in 125 mM glycine. Cells were then disrupted by vortexing in the presence of glass beads, and the lysate was sonicated to generate an average DNA size of about 0.4 -0.9 kilobases. Immunoprecipitations were performed using 1 l of raw ascites fluid (Convance) to 400 l of lysate. Following an overnight incubation at 4°C with 40 l of Protein A-Sepharose beads, the beads were washed extensively, and the DNA eluted (56). Following reversal of the formaldehyde-induced cross-links, 1/300 to 1/12000 of input DNA and 0.2 to 2% of immunoprecipitated DNAs were analyzed by semiquantitative PCR analysis with promoter-specific primers spanning the URS of each gene. Only one titration of immunoprecipitated DNA and two titrations of input DNA are shown in the figure to conserve space. The PCR products were detected by illumination of ethidium bromide-stained 2% agarose gels.

Nuclease Mapping Reveals the Presence of a Nucleosomal
Array on the RNR3 Promoter-To understand the contributions of chromatin structure in the transcriptional regulation of RNR3, we analyzed the nucleosomal architecture over its promoter. The first of these experiments utilized MNase to digest nuclei in situ, followed by the detection of the digestion products by indirect end labeling (53). Given that MNase displays sequence preference in the digestion of DNA, naked DNA was digested (deproteined genomic DNA) and analyzed in parallel. The digestion pattern generated from chromatin isolated from untreated cells (ϪMMS) is consistent with the presence of a well ordered nucleosomal array positioned over the RNR3 promoter that extends into the protein coding region (Fig. 1A,  lanes 3 and 4). The hallmark of a translationally positioned nucleosomal unit is a 140 -150-base pair region that is protected from MNase digestion, compared with naked DNA, flanked by nuclease-hypersensitive sites; such a pattern is clearly seen. In particular, a nucleosome (Ϫ1) was detected over the TATA box that protects it from MNase digestion, compared with naked DNA (compare lanes 3 and 4 with lanes 8 and 9). The data also shows that the major transcription start site (ϩ1) is located within the internucleosomal linker region. It is noted that within the URS region (DREs), a hypersensitive site was observed in the chromatin sample that was not present in the naked DNA digestion reaction (arrowhead, lane 3). This site is likely to be caused by transcription factor binding to the promoter since the spacing between it and the hypersensitive site generated by nucleosome Ϫ1 is not consistent with a nucleosomal pattern (also see below).
It is well recognized that the expression of most genes is accompanied by changes in chromatin structure (5,12,57); therefore, we monitored the changes in nucleosome positioning upon the derepression of RNR3. The transcription of RNR3 can be stimulated to a high level by inducing DNA damage using MMS or the replication inhibitor hydroxyurea (42,43). Cells were treated with MMS to a final concentration of 0.02% MMS for 2.5 h, and were then processed for nuclease mapping. A representative Northern blot is shown in Fig. 1C. The pattern of MNase-digested chromatin from MMS-treated cells is nearly identical to that of digested naked DNA (compare lanes 6 and 7 to lanes 8 and 9), indicating a disruption of the nucleosomal array. Specifically, the regions protected from digestion are fully accessible (filled circles), most notably the region over nucleosome Ϫ1 containing the TATA box. In addition, the intensity of the hypersensitive sites flanking each nucleosome is reduced. Activation of RNR3 also correlates with changes in the digestion pattern over the URS. The MNase-hypersensitive site located within the nucleosome-free URS region (indicated by the arrow in lane 3 of Fig. 1A) is lost upon gene activation, which is consistent with the predicted changes in Crt1 binding to the DREs (50).
Additional mapping studies were performed using DNase I, which does not display the same sequence bias of micrococcal nuclease (35,53). Thus, DNase I can reveal changes in chromatin structure not detected by MNase mapping. Since DNase I is capable of digesting within chromosomal DNA at 10-base pair intervals due to the rotational phasing of DNA on the nucleosome, concentrations were chosen that result in the preferential digestion within the linker regions to allow for the detection of nucleosome positioning. In agreement with the MNase mapping data described above, the pattern generated from DNase I-digested chromatin from untreated cells is consistent with the presence of an ordered nucleosomal array (Fig.  1B, lanes 3 and 4). Therefore, the pattern generated by MNase digestion is indicative of a nucleosomal array and is not an artifact of the sequence preference of this enzyme. An intense DNase I-hypersensitive site is detected between the edge of nucleosome Ϫ1 and the first DRE. Also consistent with the MNase mapping, treating cells with MMS results in a randomized digestion pattern in the region encompassing the TATA box and the coding sequences (Fig. 1B, lanes 6 and 7), indicating extensive nucleosome remodeling. Alteration in DNase I As in A, except that the nuclei were digested with 0, 0.05, and 0.1 units/ml DNase I and naked DNA (ND) was digested with 0.005, 0.01 units/ml DNase I. The closed circle and bar marks the location of the 3Ј end and 5Ј end, respectively, of the hypersensitive site. C, Northern blot analysis of RNR3 mRNA levels in repressed and MMS-treated cells. 15 g of total RNA isolated from each strains was fractionated in a 1.5% agarose gel containing formaldehyde, and subjected to Northern blotting. Specific messages were detected using random-labeled RNR3 and scR1 probes. scR1 is a loading control. The results of the mapping shown in A and B were used to assign nucleosome positions, which are shown on the left of each panel.
sensitivity is observed over the URS as well. The edge of the hypersensitive site now extends toward the 3Ј of the gene (Fig.  1B, closed circle), and a region of strong DNase I hypersensitivity appears over the URS. This nuclease-hypersensitive site extends ϳ250 -300 base pairs upstream of the first DRE (Fig.  1B, bar), and careful inspection of this pattern indicates that is in fact resolved into a triplet of bands. Thus, the DNase I digestion pattern over the URS is strikingly different from that of MNase. The DNA underlying the URS is relatively insensitive to DNase I under repressive conditions and becomes hypersensitive when the gene is derepressed; whereas a MNasehypersensitive site is detected in the repressed condition and is lost upon treatment with MMS.
The nuclease mapping studies presented here indicate that RNR3 is packaged into a regular array of positioned nucleosomes that represses its expression in the absence of DNA damage-induced signals. In particular, a nucleosome (Ϫ1) is positioned over the TATA box of the gene. Upon activation of RNR3, nucleosome positioning is disrupted and dramatic changes in nuclease sensitivity are detected over the URS.
Crt1, Ssn6, and Tup1 Regulate Nucleosome Positioning and the Chromatin Structure over the URS-Next, we performed nuclease mapping studies in a strain containing a deletion of CRT1, the gene encoding the specific regulator of DNA damage-induced genes (50, 57). As shown in Fig. 2A, the MNase digestion pattern from the ⌬crt1 mutant is essentially identical to that of digested naked DNA (compare lanes 2 and 3 and 8 and 9). Specifically, nucleosome positioning is completely disrupted, and digestion within the TATA box is observed (see closed circles, lanes 8 and 9). In addition, the MNase-hypersensitive site that is observed over the URS in repressed cells is not detected, indicating that it is dependent upon Crt1 (arrow, lanes 5 and 6). DNase I mapping confirms that the nucleosome structure is disrupted in the ⌬crt1 mutant, and that the hypersensitive site over the URS is altered, as observed in MMStreated cells (Fig. 2B, lanes 6 -8). RNR3 is strongly derepressed in the ⌬crt1 mutant (Fig. 2C), linking its expression to chromatin remodeling events. These results indicate that nucleosome positioning and the changes in nuclease sensitivity at the URS are dependent upon CRT1, and that deletion of CRT1 has equivalent effects on chromatin structure as treating cells with DNA damaging agents.
Crt1 represses the transcription of RNR3 by recruiting the Ssn6-Tup1 corepressor complex to the promoter (50). Deletion of the genes for any of these three repressors causes the derepression of RNR3 (50,59). We next analyzed the requirement for SSN6 and TUP1 in nucleosome positioning. Despite the fact that Ssn6 and Tup1 are often thought of acting as a unit, they can function autonomously in some instances. For example, the chromatin structure of STE6 and the recombination enhancer of the silent mating-type loci are more strongly affected by the deletion of TUP1 than SSN6 (15, 40). Therefore, MNase and DNase I mapping were performed in a ⌬ssn6 and ⌬tup1 mutant to define the contributions of each of these two genes in nucleo- were subjected to nuclease mapping as described in the legend to Fig. 1. C, analysis of RNR3 mRNA levels in a wild type strain and in a ⌬crt1 mutant. scR1 is a loading control. some positioning at RNR3. The results shown in Fig. 3, A and B, demonstrate that the organized nucleosome array over RNR3 is disrupted in both the ⌬ssn6 and ⌬tup1 mutants. The digestion pattern is identical to that observed in chromatin from MMS-treated cells (Fig. 1B) and in a ⌬crt1 mutant (compares lanes 8 and 9, 11 and 12 to 14 and 15). This result indicates that, similar to SUC2 (36), the nucleosome positioning at RNR3 requires both SSN6 and TUP1. Surprisingly, we also found that the MNase-hypersensitive site that is detected over the DRE region in repressed cells is absent in the mutants (Fig. 3A), indicating that its formation is dependent upon Ssn6-Tup1. Next, we corroborated the Ssn6 and Tup1 dependence of the chromatin structure over the DRE region by DNase I mapping. Treating cells with MMS or deleting CRT1 causes strong DNase I hypersensitivity within the URS region (Figs. 1B and  2B). Strikingly, the same pattern of nuclease sensitivity is also observed in the ⌬ssn6 and ⌬tup1 mutants (compare lanes 8 -10 and 12-14 to 16 -18). These experiments confirm that both SSN6 and TUP1 are required for the positioning of nucleosomes and regulating chromatin structure at the URS.
High-resolution Mapping Defines the Boundaries of Nucleosomes Ϫ1 and ϩ1 and the MNase-hypersensitive Site-We next applied a high-resolution MNase mapping method to precisely define the borders of nucleosomes Ϫ1 and ϩ1 and the location of the MNase-hypersensitive site within the URS region. This method detects chromatin changes at a single-base pair resolution by reiterative primer extension using Taq DNA polymerase (35,53). Since this method can result in strand-dependent primer extension artifacts, it is critical to perform this type of analysis using primers directed toward both strands of the DNA. Primers were directed to sequences located at Ϫ324 and ϩ150, in relation to the major start site of transcription, and were designed to read the noncoding and coding strands, re-spectively. While the protection by nucleosomes Ϫ1 and ϩ1 is less obvious using this methodology compared with what was observed by indirect end-labeling, their positioning could be detected using either the Ϫ324 primer (Fig. 4A) or the ϩ150 primer (Fig. 4B). Their boundaries are defined by the locations of hypersensitive sites separated by about 140 base pairs, that are reduced in intensity in samples prepared from a ⌬crt1 mutant (Fig. 4, A and B, closed circles). Moreover, increased digestion is observed within the region encompassing the nucleosomes in the ⌬crt1 mutant, in particular at the TATA box (see arrows in Fig. 4, A and B).
The high-resolution procedure has allowed us to more precisely map the MNase-hypersensitive site located over the URS (DREs or x-boxes). This site appears as a single band in low resolution gels (Fig. 1A); however, it is easily resolved into multiple bands clustered within four distinct regions of hypersensitivity (Fig. 4, B and C, bars). Note that these bands are not detected, or are greatly diminished in intensity, in primer extension reactions using samples of digested naked DNA or digested chromatin from the ⌬crt1 mutant (Fig. 4B). Based on the co-migration of bands generated by a DNA sequencing reaction, we were able to map these hypersensitive sites to sequences flanking the three x-boxes (DREs). In addition, the location and spacing of these sites strongly supports the hypothesis that these sites are generated by the binding of the Crt1-Ssn6-Tup1 complex to DNA, rather than the enhanced sensitivity of intranucleosomal DNA that marks of the boundaries of a nucleosome. Hence, the DREs, and thus Crt1, are located in a nucleosome-free region. We repeated the primer extension mapping studies using a primer directed to the opposite strand of RNR3 (Ϫ576 is upstream of the DREs and designed to read the noncoding strand). Here too, we can clearly detect the MNase-hypersensitive sites in wild type cells, FIG. 3. Nucleosome mapping of RNR3 in ⌬ssn6 and ⌬tup1 mutants. A, MNase mapping in wild type (YSW87), ⌬ssn6 (YJR221), ⌬tup1 (YJR220), and ⌬crt1 (YJR352) mutants. Nuclei preparation, digestion, and indirect end labeling were carried out as described in the legend to Fig.  1. B, DNase I mapping as in A. Symbols are those described in the legend to Fig. 1. C, Northern blot of RNR3 mRNA. and their absence, or reduction, in the ⌬crt1 mutant (Fig. 4C). Moreover, protection of the region encompassing nucleosome Ϫ1 is clearly seen in the upper portion of this gel.
Crt1 Is Not Sufficient for Nucleosome Positioning and the Chromatin Structure over the URS-The location and alterations in the MNase-and DNase I-hypersensitive sites over the URS that accompanies derepression suggest that they are caused by changes in the binding of Crt1 to the DREs. It has been proposed that the DNA damage-dependent phosphorylation of Crt1 reduces its association with DNA, which in turn alleviates Ssn6-Tup1-mediated repression (50). However, deletion of SSN6 or TUP1 results in the same changes in nuclease sensitivity at the URS as those observed in a ⌬crt1 mutant. This is an unexpected result since the deletion of SSN6 or TUP1 should not activate the DNA damage checkpoint pathway, and therefore, Crt1 should remain bound to the DREs preserving the MNase I-hypersensitive site and preventing the appearance of DNase I hypersensitivity. Two explanations for these results are that either the binding of Crt1 to the promoter is Ssn6-Tup1-dependent or that the binding of Crt1 to the DREs alone is insufficient to position nucleosomes and cause the changes in nuclease sensitivity over the URS. To discriminate between these two possibilities, the binding of Crt1 to the RNR3 and RNR2 promoters, two Crt1 target genes, was analyzed using the chromatin immunoprecipitation technique using formaldehyde cross-linked samples (50,56). The data shown in Fig. 5 indicate that the promoter region of RNR3 and RNR2 were efficiently immunoprecipitated using antibodies specific for epitope-tagged Crt1 in the absence of DNA damage, indicating its association with the DREs under the repressed condition (lane 9). The cross-linking and immunoprecipitation reactions were highly specific because the promoter region of a Crt1-independent gene, ADH1, was not immunoprecipitated, and no DNA was detected in immunopreciptates using an unrelated antibody (lanes 13-16). The amount of RNR2 and RNR3 DNA immunoprecipitated from samples prepared from  13-16, negative control). Immunoprecipitated and input DNA were amplified by PCR using primers specific for the URS region of RNR2 and RNR3 and for UAS region of ADH1 promoter. The PCR products were resolved on a 2% agarose gel and visualized by ethidium bromide staining. Lanes 1-8 show PCR reactions using 3-fold dilutions of input DNA. PCR amplifications using 2% of the immunoprecipitated DNA is shown (lanes 9 -16). The experiments were repeated three times with similar results. MMS-treated cells was greatly reduced, signaling the disassociation of Crt1 from the promoter (compare lanes 9 and 10). Moreover, similar quantities of RNR3 and RNR2 promoter were precipitated from the samples from a ⌬ssn6 mutant as untreated cells (Fig. 5, lane 11), indicating that Crt1 remains bound to the promoter in the absence of Ssn6. Similar to that observed in wild type cells, the association of Crt1 with the two promoters was reduced in MMS-treated ⌬ssn6 cells (lane 12). This experiment shows that Ssn6 does not regulate Crt1 binding, and therefore, changes in Crt1 binding alone cannot alter the nuclease sensitivity of the URS and that Crt1 is insufficient for positioning nucleosomes.
Quantitative Analysis of Chromatin Remodeling Using the Restriction Endonuclease Accessibility Assay-Having established a qualitative picture of the chromatin organization over the RNR3 promoter by MNase and DNase I mapping, we performed the restriction endonuclease accessibility assay to obtain a more quantitative assessment of chromatin remodeling. Restriction endonuclease sites located within a nucleosome are more resistant to digestion compared with sites located in nucleosome-free regions or in regions containing DNA-binding proteins (54,55). This method has been used extensively to quantify the extent of nucleosome remodeling at the PHO5 promoter (4,55). The accessibility of two restriction endonuclease sites located within the RNR3 promoter was tested. Based upon the results of our nuclease mapping studies, we predict that a MluI site is located within, but toward the edge of nucleosome Ϫ1, and a NcoI site is located within a region predicted to be nucleosome-free (Fig. 6A). Nuclei were treated in situ with saturating amounts of restriction endonuclease (see figure legend), and afterward the DNA was purified. After a secondary restriction endonuclease digestion (PstI or EcoRV), the extent of digestion was visualized by Southern blotting using the appropriate DNA probe. The results shown in Fig. 6B indicate that the MluI site is more resistant to digestion than the NcoI site in chromatin isolated from untreated cells (22.5 versus 64.0% digested, respectively). Moreover, the accessibility of the MluI site is increased in chromatin isolated from MMS-treated cells and in a ⌬crt1 mutant ϳ3.5and 2.9-fold, respectively. In contrast, digestion at the NcoI site is not significantly changed under these two conditions. Overall the degree of accessibility of these sites to restriction endonucleases is consistent with our prediction about their location in relation to the position of nucleosome Ϫ1 and the remodeling that occurs when the gene is expressed. The MluI site is located toward the edge of nucleosome Ϫ1, and thus, is more resistant to digestion than the NcoI site that is located in a nucleosomefree region. Second, derepression of the gene correlates with an increase in the accessibility of the MluI site caused by the remodeling of nucleosome Ϫ1 over the TATA box.
Checkpoint-dependent Chromatin Remodeling of RNR3-DNA damage-induced transcription is dependent upon a number of checkpoint genes, including MEC1 and RAD53 (42,43,45,46). The expression of RNR3 is severely compromised in mec1 or rad53 mutants, presumably due to the inability to phosphorylate Crt1p (50). To investigate the requirement for the checkpoint kinase genes MEC1 and RAD53 in regulating the chromatin structure of RNR3, we carried out nucleosome mapping studies using strains containing mutations that inactivate their checkpoint functions. The Northern blot presented in Fig. 7A verifies that mutation of either checkpoint gene eliminates the induction of RNR3. Nuclei isolated from these cells were subjected to the low-resolution MNase nucleosome mapping strategy (Fig. 7B). The results shown in Fig. 6B indi-FIG. 6. Restriction endonuclease accessibility assay. Nuclei were prepared from untreated wild type cells (WT), untreated ⌬crt1 cells (⌬crt1), or wild type cells treated with 0.02% MMS (MMS). Nuclei were digested in situ with 100 and 400 units/ml of the appropriate restriction enzyme at 37°C for 1 h. Since no additional digestion was observed at the higher concentration of restriction endonuclease, the enzyme is considered to be in excess and the reaction saturated. The DNA purified from the MluI-and NcoI-digested nuclei were digested overnight with PstI and EcoRV, respectively. Products were detected by Southern blotting using the indirect end labeling method. A, schematic of the nucleosome structure derived from the MNase mapping studies shown in Figs. 1A and 4. The positions of the restriction enzyme sites, in relation to nucleosome Ϫ1 and the DREs, are shown. B, restriction endonuclease accessibility analysis. The cutting percentage was calculated from the 400 units/ml restriction endonuclease reaction and is the ratio of the counts in the digested fragment (cut) to the total counts. The ratio was calculated relative to the percentage of digestion in chromatin from untreated wild type cells. The asterisk marks a band generated from a MluI site located upstream of the DREs and outside of the RNR3 promoter region. The percentage of digestion for each enzyme and condition varied by less than 10% in three independent chromatin preparations.
cate that mutation of MEC1 or RAD53 does not alter the positioning of the nucleosomes over the repressed promoter. The characteristic protection of intranucleosomal DNA, flanked by nuclease-hypersensitive sites is clearly observed. In contrast, the DNA damage-dependent chromatin remodeling is eliminated in both the mec1-1 and rad53-11 mutants, as evidenced by the preservation of the nucleosome-generated ladder of DNA fragments. In addition, the persistence of the MNasehypersensitive site over the DRE region is also observed (compare lanes 6 and 7-12 and 13 and 18 and 19). These observations indicate that, like transcription, chromatin remodeling of the RNR3 promoter is dependent upon the checkpoint kinases. DISCUSSION We have described a comprehensive analysis of the nucleosomal architecture at the RNR3 promoter under the repressed and derepressed state. The utilization of multiple methodologies has allowed us to determine its chromatin structure with a high degree of confidence. A consensus model of chromatin organization over the RNR3 promoter is illustrated in Fig. 8A. In the absence of DNA damage an organized array of nucleosomes is established on the RNR3 promoter, extending into the coding sequences. Derepression of the gene correlates with the disruption of at least nucleosomes Ϫ2, Ϫ1, ϩ1, ϩ2, and ϩ3. Additional changes are likely to occur downstream of nucleosome ϩ3, but this has not been examined. The nucleosomal structure almost certainly plays an essential role in transcription repression since the TATA box is located within nucleosome-1, which should occlude its accessibility to TFIID. In contrast, the RNA transcription start site (ϩ1) is located within the internucleosomal linker region between nucleosomes Ϫ1 and ϩ1. What, if any, consequence this configuration has on gene expression is not known. It might imply that it allows some degree of accessibility of this region to transcription factors, even in the repressed state. This might explain RNR3's relatively high level of uninduced transcription compared with other Ssn6-Tup1 regulated genes. Ribonucleotide reductase is an essential enzyme irrespective of the presence of induced DNA damage, and this semipermissive state may be important in maintaining dNTP pools (42,43).
We demonstrate that CRT1, SSN6, and TUP1 are required for nucleosome positioning and transcriptional repression, and that deleting any of these three genes disrupts nucleosome positioning similarly. While the data presented here do not rule out additional repression mechanisms employed by the Crt1-Ssn6-Tup1 complex, such as interfering with transcription factor function (for review, see Ref. 16), its ability to regulate nucleosome positioning is inseparable from its ability to repress RNR3 in our analysis. We have begun to define the contributions of Crt1 versus Ssn6/Tup1 in nucleosome positioning by utilizing regulatory mutants and an in vivo cross-linking technique for monitoring transcription factor binding (chromatin immunoprecipitation). Removing Crt1 from the promoter genetically, as in the ⌬crt1 mutant, or physiologically by MMS treatment abolishes nucleosome positioning and transcriptional repression. However, Crt1 is necessary, but not sufficient, for nucleosome positioning, as we demonstrated that Crt1 remains bound to the promoter in the absence of Ssn6. These data strongly argue that the binding of Crt1 to the DREs cannot position nucleosomes per se by interference, but rather the primary function of Crt1 in nucleosome positioning is to recruit the corepressor complex to the promoter. In addition, we show that Ssn6 and Tup1 play a similar role in nucleosome positioning at RNR3, and thus, the repression mechanism is likely to be different from that described for the mating-type specific genes and the recombination enhancer (15,40). The results showing that deletion of SSN6 or TUP1 has equivalent effects on the chromatin structure of RNR3, correlates well with their similar roles in repressing RNR3 (Fig. 3, 59). This was an open question because we and others have demonstrated that Ssn6p and Tup1p can bind to Crt1 individually (47,50). Tup1 is responsible for the repressive functions of the co-repressor complex as studies have shown that LexA-Tup1 can repress the transcription of, and recruit histone deacetylases to, a heterologous promoter containing lexA operators even in a ⌬ssn6 mutant (34,60,61). These observations reason that the interaction of Tup1 with Crt1, in the absence of Ssn6, would be sufficient to establish a nucleosome array over RNR3, but clearly we show this is not the case. Either the interaction of Tup1 with Crt1 is not stable enough to withstand the environment within the nucleus or both Ssn6 and Tup1 are required to efficiently repress a native gene (see below).
An interesting observation from our mapping studies is the change in nuclease sensitivity over the DRE region. It was unexpected to find that these changes are strictly dependent upon Ssn6 and Tup1. Surprisingly, these changes occur even when Crt1 remains bound to the promoter, as seen in the ⌬ssn6 mutant; therefore, the structure at the URS cannot be attributed to Crt1 binding alone. The nature of the changes in nuclease sensitivity can be perceived as paradoxical, because the loss of a MNase-hypersensitive site is consistent with the release of a DNA-binding protein, but the appearance of DNase I hypersensitivity usually correlates with its binding. A potential interpretation of this is that the release of Crt1-Ssn6-Tup1 from the DREs leaves the initially protein-bound DNA sequence accessible to nucleases. As for MNase, it does not cut this region very well even in the form of naked DNA (Fig. 1A,  ND, lanes 8 and 9); thus, the mapping simply shows the loss of a hypersensitive site. However, because DNase I does not display the sequence preference of MNase, the "protein-free" DRE region becomes very vulnerable to DNase I cleavage, and thus a hypersensitive region appears. But even this interpretation is inconsistent with the data. In particular, the DNase I digestion pattern is not the smear expected from the digestion of a protein-free region, but rather is a distinct region comprised of a triplet (Figs. 1B and 2B). Such a pattern could be explained by the subsequent binding of another transcription factor to the DRE after the removal of Crt1, as the binding of transcription factors to DNA results in DNase I-hypersensitive sites. However, the same changes in nuclease sensitivity occur when Crt1 remains bound to the DREs, as seen in the ⌬ssn6 mutant. It is difficult to envision another transcription factor co-occupying the DREs with Crt1 because typically this would result in cooperative DNA binding, and in this scenario deleting CRT1 would adversely affect the binding of this second factor, preventing the appearance of nuclease sensitivity. We recognize that this argument does not rule out the possibility that an unidentified transcription factor binds adjacent to the DREs, but evidence for such a factor is lacking.
Alternatively, this pattern could be attributed to two different, but related events. The first is that the chromatin remodeling events associated with gene activation cause unusual DNA topologies upstream of the promoter. DNase I shows some preference for DNA with unusual geometries (35). It is reported that chromatin remodeling complexes, such as SWI/SNF and RSC, can alter the position of nucleosomes along the DNA template without removing histones in vitro (62,63). Some proposed mechanisms for moving a nucleosome along the DNA include looping, sliding, and tracking models (64), which would ultimately result in topological changes in the DNA upstream and downstream of the promoter. Second, the release of the Crt1-Tup1-Ssn6 complex from the promoter. The MNase I-hypersensitive site observed in the repressed state may result from cooperative interactions between Ssn6-Tup1, Crt1, DNA, and the tails of the nucleosome (Fig. 8B) that cause DNA bending and the positioning of these sites to maximize their exposure to nucleases. A prediction from this model would be that deleting any single component of the complex would be equally disruptive and result in similar changes in nuclease sensitivity. In particular, the binding of Crt1 alone would be insufficient for its formation. Both predictions are met. This would also explain why the binding of Tup1 directly to Crt1, as predicted from in vitro studies (47,50), will not position nucleosomes, preserve the structure over the URS, or repress transcription in the ⌬ssn6 mutant (Fig. 2).
An interesting feature of all Crt1-regulated genes identified thus far is that the x-boxes (DREs) are found in pairs of at least one weak (Xw) and one strong (Xs) binding site, each in opposite orientations (50,58), suggesting that the configuration of these half-sites is essential for their function. In support of this hypothesis, we find that a tandem array of four Xs elements is not as effective at repressing the transcription of a heterologous promoter as the URS region of RNR3 (three x-boxes), again suggesting that it is the configuration, and not the number, of x-boxes that is critical in mediating repression 2 (47). The complex formed over the URS may be reminiscent of the enhanceosome, whose formation is dependent upon multiple DNA-binding proteins, non-DNA binding cofactors, and the topology (bending) of nucleic acid structure (for review, see Ref. 65). Analogously, the formation of the ␤-interferon enhanceosome is dependent upon the orientation of the ATF2-jun heterodimer (66). The mammalian homologues of Crt1, the RFX family of transcription factors, act as activators and repressors of transcription (64,67,68). Interestingly, RFX proteins are crucial in forming an enhanceosome at the major histocompatibility complex class II gene promoters through multiple interactions with CIITA and NF-Y (69,70), and thus, the requirement for RFX-related transcription factors to form elaborate protein-DNA complexes at promoters may be a conserved feature of these proteins. By analogy, the ability of Crt1 to form a related structure and recruit multiple repressors to the promoter may be important for its ability to position nucleosomes and repress DNA damage-responsive genes.