Neither Reb1p nor Poly(dA (cid:1) dT) Elements Are Responsible for the Highly Specific Chromatin Organization at the ILV1 Promoter*

Analysis of the chromatin structure at the yeast ILV1 locus revealed highly positioned nucleosomes covering the entire locus except for a hypersensitive site in the promoter region. All previously identified cis -acting elements required for GCN4 -independent ILV1 basal level transcription, including a binding site for the REB1 protein (Reb1p), and a poly(dA (cid:1) dT) element (26 As out of 32 nucleotides) situated 15 base pairs downstream of the Reb1p-binding site, reside within this hypersensitive site. The existence of a second A (cid:1) T-rich element (25 As out of 33 nucleotides) present six base pairs upstream of the Reb1p-binding site, suggested that nucleosome exclusion from the hypersensitive site in the ILV1 promoter region might be dictated by synergistic action of the two poly(dA (cid:1) dT) elements. Replacing one or both of them had, however, no effect on the chromatin structure of the ILV1 promoter, although drastically reduced basal transcription. Similarly, deletion of the Reb1p-bind-ing site, albeit affecting ILV1 expression, had no detectable effect on chromatin at the ILV1 promoter.

The Saccharomyces cerevisiae anabolic threonine deaminase, encoded by the ILV1 gene, catalyzes the first committed step in the isoleucine biosynthetic pathway. ILV1 basal level expression (defined as the level of expression observed in a ⌬gcn4 strain grown in minimal medium) is controlled by two cisacting elements: a binding site for the REB1 protein (Reb1p) and a poly(dA⅐dT) element (1,2).
Reb1p (Grf2, Qbp, factor Y or Q) is an essential, abundant DNA-binding protein with numerous binding sites present throughout the genome, many of which are located within regulatory regions for RNA polymerase II transcribed genes. Other Reb1p-binding sites are found in RNA polymerase I regulatory regions and in diverse genetic elements, such as centromeres and telomeres (3)(4)(5)(6)(7). Reb1p-binding sites are also found in the recently identified STAR 1 (subtelomeric anti-silencing regions) elements (8). These regions function as insulators and can protect neighboring genes from surrounding silencing elements. Interestingly, tandemly repeated binding sites for Reb1p can duplicate this anti-silencing effect and limit telomeric silenced chromatin. Furthermore, Reb1p-binding sites can stimulate or diminish transcription in a context-dependent manner (9 -11). Although Reb1p affects transcription of RNA polymerase II transcribed genes, the mechanism by which it does so is not clear. The presence of a Reb1p-binding site in the GAL1-GAL10 promoter correlated with a nucleosome-free region of about 230 bp in vivo (12) suggesting that Reb1p functions to generate a nucleosome-free region allowing auxiliary factors access to adjacent cis-acting elements. Other reports, however, contest this result (13,14) casting some doubt as to the mechanism of action of Reb1p.
Homopolymeric poly(dA⅐dT) sequences are present in the promoter region of many yeast genes and have been shown to influence transcription of several genes (2,(15)(16)(17)(18). Due to their nucleosome destabilizing properties poly(dA⅐dT) elements have been proposed to function by virtue of their intrinsic effect on chromatin (15,18). Interestingly, the yeast poly(dA⅐dT)-binding protein Dat1p (19) functions as a transcriptional activator of ILV1 expression and this action depends on the presence of the poly(dA⅐dT) element (2).
In an attempt to elucidate the mechanism by which the Reb1p-binding site and the downstream poly(dA⅐dT) element control ILV1 basal expression, we investigated the chromatin structure of the ILV1 locus. We show that the ILV1 promoter and coding regions are assembled into a highly ordered nucleosome array, with a single hypersensitive region encompassing all regulatory cis-acting elements. We also show that deletion of the Reb1p site and/or the adjacent downstream poly(dA⅐dT) element greatly diminishes ILV1 expression, yet does not cause reconfiguration of the ILV1 chromatin structure. Furthermore, a second A⅐T-rich element present upstream of the Reb1pbinding site can be deleted, again with no effect on ILV1 chromatin structure. This suggests that neither Reb1p nor the adjacent poly(dA⅐dT) elements stimulate ILV1 transcription by increasing the accessibility of DNA in chromatin at the promoter, but by another yet unknown mechanism. Sequence insertions and deletions at the hypersensitive region of the ILV1 promoter, albeit affecting expression of the ILV1 gene, do not affect the positioning of the adjacent nucleosomes pointing to a dominant effect of the DNA sequence in organizing the nucleosomal array present at the ILV1 promoter.
Strains were grown in minimal medium (0.67% Bacto Yeast Nitrogen Base without Amino Acids, 2% glucose, buffered with 10 g of succinic acid and 6 g of NaOH per liter) supplemented with the required amino acids at appropriate concentrations.
DNA and RNA Methodology-All nucleic acid manipulation was performed according to established protocols (20). Polymerase chain reaction (PCR) was used under standard conditions (0.2 mM of each of dATP, dCTP, dTTP, dGTP; 20 mM Tris-HCl (pH 8.4); 50 mM KCl; 5 mM MgCl 2 ; 0.5 M of each primer; 2.5 units of Taq DNA polymerase per reaction). DNA was sequenced using the dideoxy primer extension method (Sequenase Sequencing Kit v2.0, United States Biochemical Corporation, Cleveland, Ohio). Yeast cells were transformed following the method of Ito et al. (21). Northern analysis was performed as described (22).
Chromatin Analysis-Nuclei isolation and nuclease digestions were performed as previously described (23). After secondary digestion with the appropriate restriction enzyme, DNA samples were electrophoresed in 1.5% agarose gels, transferred onto Positive TM nylon membranes (Oncor, Gaithersburg, MD) and hybridized following standard protocols. A plasmid containing a 6.1-kb HindIII-SalI ILV1 fragment (pC519) was digested with the appropriate restriction enzymes and electrophoretically purified DNA fragments labeled with the Prime-It ® II random primer labeling kit (Stratagene, La Jolla, CA) and used as probes in all experiments. The specific fragments used as probes are described in further detail in figure legends.
ILV1 Promoter Derivatives-For the spacing experiments, we used plasmid constructs containing insertions of E. coli plasmid DNA into the XhoI site of the 15X ILV1 internal deletion fused to the lacZ gene (2). The constructs were integrated at the ILV1 locus in strain 9994 -6C (⌬gcn4). All integration events were confirmed by Southern analysis and PCR.
Deletion of the ILV1 A⅐T-rich tracts was made by PCR using a previously described procedure (24). Oligonucleotides covering the sequences we wished to modify were synthesized (T⅐A⅐G Copenhagen ApS, Denmark) allowing the target sequences to be substituted by random sequences with no known cis-acting elements. Oligonucleotide SUBUP (5Ј-AATTGACGCGAACTAGATCGCTTAAGCTCGGCATCAAGCTCG-AGCGCAGCGGGTAGCAAATTTGGAATCG-3Ј) was used to replace the 5Ј-most ILV1 A⅐T-rich tract (positions Ϫ235 to Ϫ166) while simultaneously creating a unique XhoI site. Similarly, oligonucleotide SUBDOWN (5Ј-ATCTGCAGACATATGTTTGAGATGACTCTAGATCT-CCGATGTTCAAGCTTCATGATTATGCGATTCCAAATTTGC-3Ј) was used to replace the 3Ј-most ILV1 A⅐T-rich tract (positions Ϫ106 to Ϫ180) and simultaneously introduce a HindIII site. PCR fragments were obtained covering the ILV1 promoter and entire coding region (from position Ϫ840 to position ϩ2530) that contained a substitution of the 5Јmost, the 3Ј most, or both A⅐T-rich tracts in the promoter region. The amplificates were cloned into plasmid pGEM-T TM (Promega) and the substitutions confirmed by sequencing and subsequently integrated at the ILV1 locus in strain TG561 generating ILV1(⌬A1), ILV1(⌬A2), and ILV1(⌬A1⌬A2), respectively.

Three Elements Regulate ILV1 Basal Level Expression-We
have shown that two cis-acting sequences, a Reb1p-binding site and a downstream poly(dA⅐dT) element control ILV1 basallevel (GCN4-independent) expression (1, 2). The two elements act synergistically, an indication that they might exert their effect trough a common activation pathway (2). Sequence inspection revealed an additional A⅐T-rich tract (25 As out of 33 nucleotides) located six bp upstream of the Reb1p-binding site.
To determine whether the upstream A⅐T-rich tract also plays a role in ILV1 basal expression we replaced either one or both ILV1 A⅐T-rich tracts by 40% GC content random sequences resulting in strains ILV1(⌬A1), ILV1(⌬A2), and ILV1(⌬A1⌬A2), respectively. Northern analysis showed that both A⅐T-rich tracts potentiate ILV1 basal expression (Fig. 1).
The ILV1 Promoter Is Organized in an Ordered Nucleosomal Array-We analyzed the nucleosomal organization of the ILV1 promoter using digestion of yeast nuclei with DNase I, micrococcal nuclease (MNase), or restriction enzymes to map nuclease cuts by indirect end-labeling (25). Thus, nuclei from strain TD28 (GCN4) were treated with DNase I (see "Materials and Methods"), and the isolated DNA was resolved in an agarose gel after digestion with EcoRI, blotted, and hybridized with a radioactively labeled EcoRI-DraI probe. The obtained DNase I pattern (Fig. 2) shows a characteristic ladder of bands typical of an ordered nucleosomal array, and a strong band corresponding to a hypersensitive site (HS). Mapping of this HS revealed that all previously identified cis-acting elements, namely a Reb1pbinding site, a Gcn4p-binding site, and the two poly(dA⅐dT) elements were located within the hypersensitive region.
To complement the DNase I analysis and confirm our interpretation of the observed banding pattern, we digested nuclei from TD28 cells with various restriction enzymes (Fig. 3, A and B). Hypersensitive sites and linker regions are expected to be susceptible to digestion with endonucleolytic enzymes, whereas sequences assembled into nucleosomes should be protected. Very strong, nucleosomal sites are resistant to cleavage over a large range in enzyme concentration while non-nucleosomal sites are cut at much lower concentrations. Consequently accessibility reaches plateau levels both for nucleosomal and non-nucleosomal sites. To be certain in our experiments that we have truly reached these plateau values for a given site, we always verify that 3-to 4-fold higher nuclease concentrations still give the same accessibility thus ruling out that the enzyme activity had been limiting. Nuclei from yeast strain TD28 (GCN4) were digested with PvuII, BstUI, DraI, HinfI, or PstI at 500 and 1500 units/ml for 1 h. The results shown in Fig. 3, A and B confirm our conclusions derived from the DNase I assay. Bands were quantified using a phosphorimager equipped with ImageQuant analysis software (Molecular Dynamics, Sunnyvale, CA), and percent accessibility for the various enzymes was calculated from the ratio between the bands corresponding to digested sites and that of undigested DNA. Thus, the two PstI sites present in the promoter region showed 50% (position Ϫ108) and Ͻ5% (position Ϫ546) accessibility, consistent with being at the border of the HS and within a nucleosome, respectively. BstUI (position Ϫ227) displayed 75% accessibility, correlating well with a position within the hypersensitive region. Conversely, the two DraI sites (positions Ϫ364 and Ϫ28) were less than 5% accessible, consistent with a location within the nucleosomes flanking the HS. PvuII (position Ϫ432) with 30% accessibility locates to an internucleosomal region in the deduced ILV1 chromatin structure. Five HinfI sites were exam- ined (positions Ϫ595, Ϫ486, Ϫ296, Ϫ170, and Ϫ132). The sites within the hypersensitive region showed 80% accessibility (positions Ϫ170 and Ϫ132), whereas the remaining sites displayed less than 5% accessibility, consistent with nucleosomal locations. These results confirm the DNase I analysis and the mapping of nucleosomal positions we derived from that analysis.
We conclude that the ILV1 promoter is organized in an ordered nucleosomal array with one strong hypersensitive site encompassing the region where the UAS elements are located. A schematic drawing of the ILV1 locus with restriction sites, cis-acting elements, and inferred chromatin structure is shown in Fig. 3C.
Chromatin Structure of the ILV1 Promoter Derepressed by the General Control of Amino Acid Biosynthesis-The S. cerevisiae ILV1 gene is under regulation of the general control of amino acid biosynthesis. A Gcn4p-binding site present at position Ϫ127 was found to bind Gcn4p in vitro (26). Indeed, upon amino acid starvation ILV1 expression is increased 2-fold by the Gcn4p activator protein (27). We tried to determine whether derepression by Gcn4p had any effect on ILV1 chromatin structure. For derepression by the general control of amino acid biosynthesis, the tryptophan analog 5-methyl-DLtryptophan (MeTrp) was added at a final concentration of 0.5 mM to the growth medium (27,28). Nuclei were prepared from TD28 (GCN4) cells grown to a density of 3 ϫ 10 6 cells/ml and subjected to nuclease digestion with micrococcal nuclease (MNase) and DNase I, enabling us to complement our structural analysis. RNA was isolated from non-digested nuclei, and Northern analysis performed to confirm that ILV1 expression increased 2-fold upon derepression by the general control of amino acid biosynthesis (data not shown). The pattern obtained with either MNase (Fig. 4) or DNase I (data not shown) was identical to the one previously observed for the basal transcriptional state of the ILV1 gene with a strong hypersensitive site comprising the UAS elements and an ordered nucleosomal array covering the promoter and coding region. Hence, derepression of the ILV1 gene does not modify its chro- FIG. 3. Accessibility analysis of the ILV1 promoter. Nuclei from TD28 cells were digested for 1 h with 500 or 1500 units/ml of various restriction endonucleases. DNA was subsequently isolated from the treated nuclei, and the purified samples were digested with HaeIII and BglII, electrophoretically resolved on a 4% NuSieve 3:1 agarose gel, blotted, and hybridized with either probe A, a RsaI-BglII labeled fragment (A), or probe B, a radiolabeled HaeIII-HinfI fragment (B). Numbering corresponding to the restriction site position in the promoter is given for the various bands. C, a restriction map of the ILV1 promoter and a schematic presentation of the probes locations and of the inferred nucleosome positioning is depicted. Nucleosomes are presented as filled ellipses, the Reb1p-binding site as a filled triangle, poly(dA⅐dT) elements as filled circles, and the Gcn4p-binding site as a filled square. matin structure in a way detectable in our analyses.
Gcn4p and Dat1p Are Not Necessary for Nucleosome Positioning in the ILV1 Promoter-We investigated the role of two trans-acting factors, Dat1p and Gcn4p, both of which have been shown to bind their cognate sites at the ILV1 promoter in vitro (2,26). Nuclei isolated from strains TD28 (GCN4), 9994 -6C (⌬gcn4), EWY1002c-1 (DAT1), and BRY1004 -3 (⌬dat1) grown in minimal medium were digested with DNase I, DNA was isolated, and end-labeling analysis performed as described above. The patterns obtained for all four strains were identical (data not shown), demonstrating that neither Gcn4p nor Dat1p are involved in generating the observed ILV1 chromatin structure.
The Hypersensitive Region in the ILV1 Promoter Is Not Caused by the Reb1p-binding Site or the Downstream poly(dA⅐dT) Element-A Reb1p-binding site present at position Ϫ180 is required for GCN4-independent ILV1 basal level transcription. Deletion of the Reb1p site reduces ILV1 basal level expression 10-to 15-fold (1). Additionally, the poly(dA⅐dT) element located between positions Ϫ164 and Ϫ133, which is also necessary for wild-type ILV1 basal level expression (see above), is situated 15 bp downstream of the Reb1p site, and the two elements cooperatively stimulate ILV1 transcription (2). This synergistic activation could be the result of a common mechanism of action.
A Reb1p-binding site in the GAL1-GAL10 promoter was correlated with the presence of a nucleosome-free region of about 230 bp (12). Furthermore, poly(dA⅐dT) sequences stimulate Gcn4p-activated transcription and were suggested to function by changing chromatin structure and increasing accessibility of adjacent Gcn4p sites (18). Thus, the cooperative activation by the ILV1 Reb1p site and the downstream poly(dA⅐dT) element could be due to interactions between these two elements leading to increased accessibility of the adjacent Gcn4p-binding site. To test this possibility, we analyzed previously characterized promoter constructs showing decreased ILV1 basal expression (2). In these constructs the ILV1 coding region has been replaced by the E. coli lacZ gene encoding ␤-galactosidase (1).
To determine whether the replacement of the ILV1 coding region with lacZ affected the structure of the ILV1 promoter, we compared MNase digests of strain TD28 with those of strain TG561, an isogenic TD28 derivative containing a ILV1 wildtype promoter, but with the ILV1 coding sequence replaced by lacZ. The digestion pattern for strain TD28 (Fig. 5, ILV1) is the expected one, a strong HS located in the promoter region, and an ordered nucleosomal array covers the coding region and promoter. Strain TG561 where the ILV1 coding sequence was substituted with lacZ shows a different pattern (Fig. 5, lacZ). The ILV1 promoter conserved its structure with one exception: the hypersensitive site was slightly decreased in size. This appears to be due to a shifted position of the nucleosome positioned at the downstream boundary of the hypersensitive region. Moreover, the lacZ sequence appears strongly nucleasesensitive suggesting this gene to be less prone to undergo an ordered nucleosome organization. To confirm these conclusions and to determine that no accessibility changes had occurred in the hybrid promoter as compared with the wild-type promoter, we performed restriction enzyme analysis on the two strains. The results (Fig. 6) show that BstUI accessibility is almost identical (75% in ILV1 and 70% in lacZ) but that the PstI site at position Ϫ108 changed from 50% accessibility in the wildtype situation (Fig. 6, ILV1) to Ͻ5% accessibility in the hybrid construct (Fig. 6, lacZ), consistent with going from a location at the border of the hypersensitive site to being covered by the adjacent nucleosome. The original position of this nucleosome had incorporated about 20 bp of coding sequence. Apparently, in the lacZ derivative, this positioned nucleosome is shifted upstream by about 20 bp thus excluding the new coding sequence altogether. At any rate, we conclude that the structure of the hybrid promoter construct reflects the wild-type situation in a sufficiently adequate manner for our study.
We carried out a DNase I analysis on four constructs, which had previously been used to identify the Reb1p site and the downstream poly(dA⅐dT) element as important basal regulatory elements in the ILV1 promoter (1, 2). The promoter derivatives are either deleted for the Reb1p-binding site and/or the poly(dA⅐dT) element. Again, the pattern observed is identical in all three derivatives and indistinguishable from the wild-type pattern (Fig. 7, A and B and data not shown) suggesting that neither the Reb1p-binding site nor the poly(dA⅐dT) are responsible for generating of the HS present at the ILV1 promoter. To rule out the possibility that deletion of the Reb1p-binding site or/and the neighboring poly(dA⅐dT) element has more subtle effects on chromatin not detected by DNase I digestion, we performed a restriction enzyme analysis of three of the constructs as well (Fig. 7C). BstUI (position Ϫ227) accessibility is similar in the tested constructs (75-70%), and the PstI site at position Ϫ108, as shown before, displayed Ͻ5% accessibility (Fig. 7C, ⌬REB1 ⌬(dA⅐dT)) in the lacZ constructs. These results show that neither the ILV1 Reb1p-binding site nor the downstream poly(dA⅐dT) element are responsible for generating the hypersensitive region we observe in the ILV1 promoter.
The Hypersensitive Region at the ILV1 Locus Is Persistent-We have shown that ILV1 basal level expression depends on the distance separating the Reb1p-binding site and the downstream poly(dA⅐dT) element, since insertion of spacing DNA into a XhoI site created between the Reb1p-binding site and the poly(dA⅐dT) element in the 15X construct reduces promoter activity (2). We reasoned that if the hypersensitive site in the ILV1 promoter were caused by any cis-acting element present within the nuclease accessible region, insertion of 41 bp or 74 bp of DNA corresponding to a size increase of 22% and 41%, respectively, might cause some rearrangement of the borders of the hypersensitive region thus affecting promoter activity.
DNase I analysis of the constructs containing insertions of DNA between the Reb1p-binding site and the downstream poly(dA⅐dT) element is shown in Fig. 8. The basic structure and relative positioning of nucleosomes is preserved in the constructs containing insertions relative to the wild-type situation. The main difference is the progressive increase in the size of the nuclease-sensitive region. The resulting pattern clearly shows that the entire promoter region structure is similar between all the constructs and that insertion of increasingly larger spacing DNA is compensated for by correspondingly larger hypersensitive regions.
We considered the possibility that the presence of two A⅐Trich tracts placed symmetrically within the HS and located at roughly the same distance from the borders of the HS triggers nucleosome exclusion from this region. Such a mechanism would also explain the lengthening of the hypersensitive site observed as a consequence of introducing random sequences in between the two A-T rich tracts. We therefore replaced either one or both ILV1 A⅐T-rich tracts by 40% GC content random sequences resulting in strains ILV1(⌬A1), ILV1(⌬A2), and ILV1(⌬A1⌬A2), respectively. MNase analysis of these strains is shown in Fig. 9. The obtained pattern is identical to the one previously detected in the ILV1 promoter. Neither the typical nucleosomal ladder nor the hypersensitive region shows any significant difference to the wild-type situation. We conclude that the strong nucleosome positioning which we observe at the ILV1 locus is not dependent on the presence of the two A⅐T-rich tracts present in the promoter region.

DISCUSSION
Effect of Reb1p in the Chromatin Structure of ILV1-We have reported that ILV1 basal level expression is controlled in a synergistic manner by a Reb1p-binding site and a downstream poly(dA⅐dT) element (2). Reb1p has been proposed to antagonize nucleosomal repression by creating an accessible chromatin structure, thereby allowing other factors to gain access to their cognate sites (12). Other reports, however, dispute this conclusion and interpret the data differently (13,14). Reb1p has also been implicated as a regulatory factor both in Pol II and Pol I transcribed genes (1,6,9).
Synergistic interactions between Reb1p-binding sites and poly(dA⅐dT) elements have been described by several authors (2,3,29,30), raising the possibility that the two elements cooperate to increase accessibility of adjacent elements since also homopolymeric poly(dA⅐dT) elements have been proposed to stimulate transcription through an effect on chromatin structure (15,18). Given the presence of a Reb1p-binding site and two poly(dA⅐dT) elements in the ILV1 promoter and the fact that at least one poly(dA⅐dT) element synergizes with Reb1p to sustain a relatively high basal level expression of the ILV1 gene, it is possible that the two elements act to increase chromatin accessibility, perhaps by keeping the promoter free from nucleosomes.
To ascertain if Reb1p or the poly(dA⅐dT) elements have an effect on chromatin at ILV1, we examined the structure of the ILV1 locus and mapped the nucleosomal of this gene (Fig. 3C). The promoter and coding region are covered by an ordered nucleosomal array that is interrupted by a single strong nuclease-sensitive site comprising the region where the cis-acting elements required for normal expression of the ILV1 gene are located. The structure of the locus is unaltered by Gcn4pmediated derepression (Fig. 4). This is not an unexpected result since ILV1 basal level expression is relatively high and ILV1 is up-regulated only 2-fold by amino acid starvation (26,27). Additionally, neither eliminations of Dat1p nor of Gcn4p, two trans-acting factors required for normal levels of expression, FIG. 5. Comparison of ILV1 promoter chromatin structure between strains TD28 (wild-type) and TG561 (chimeric ILV1-lacZ fusion). Plasmid pSH601 was used to replace the chromosomal ILV1 coding sequence with that of lacZ. Chromatin analysis was performed as described in the legend for Fig. 4. Lane M is the combination of restriction enzyme double digest of genomic DNA with EcoRI and MboII, EcoRI and BstUI, or EcoRI and NdeI, originating position marker bands.
FIG. 6. Accessibility analysis of the ILV1 promoter in strains TD28 (wild-type) and TG561 (chimeric ILV1-lacZ fusion). Restriction enzyme analysis was performed as described (Fig. 3 legend). Position Ϫ108 corresponding to a PstI recognition site with altered accessibility is indicated.
affected the chromatin structure of the ILV1 locus (data not shown).
It seemed reasonable to expect that Reb1p functions in the ILV1 context through an effect on chromatin structure, establishing the nucleosome-free area of about 180 bp observed in the promoter region. However, constructs containing deletions in the Reb1p-binding site and/or the downstream poly(dA⅐dT) element, which decrease expression of the ILV1 gene or even abolish it (␤-galactosidase measurements expressed in Miller Units; wild-type: 4.3 Ϯ 0.8; ⌬Reb1: 0.51 Ϯ 0.02; ⌬poly(dA⅐dT); 0.36 Ϯ 0.05; ⌬Reb1⌬poly(dA⅐dT): 0.09 Ϯ 0.02) (1, 2) showed the same chromatin structure as a wild-type promoter construct (Fig. 7, A-C), thus disproving the hypothesis that Reb1p is responsible for the hypersensitive site in the ILV1 promoter. These data were obtained with ILV1-lacZ fusions replacing ILV1, and are therefore open to the criticism that they are not relevant for the native locus. In fact, the structure of the fusion constructs does not match perfectly the situation in the wildtype ILV1 locus, since the hypersensitive region is decreased by ϳ20 bp (Fig. 5). Two lines of evidence argue against this criticism. First, quantitative analysis of accessibility in the region immediately upstream of the Reb1p-binding site did not show a significant difference between an ILV1 wild-type situation and the ILV1-lacZ fusion construct (Fig. 6). These data suggest that the observed decrease in the size of the hypersensitive site is a localized effect at the boundary to the lacZ coding region. Second, deletion of the Reb1p-binding site in the ILV1-lacZ hybrid construct decreased transcription to about 10% of the wild-type levels as determined by ␤-galactosidase activity (2). A decrease in promoter accessibility having such a dramatic effect on gene expression would be expected to be easily revealed by the restriction enzyme analysis presented in Fig. 7C, if it existed. However, no difference was detected. We conclude that Reb1p is not responsible for the formation of the hypersensitive site present in the ILV1 promoter region and that Reb1p potentiates expression in the ILV1 context by a mechanism other than causing a nucleosome-free region.
The Two ILV1 A⅐T-rich Tracts Do Not Affect ILV1 Chromatin Structure-Poly(dA⅐dT) elements stimulate transcription of several yeast genes (15)(16)(17)(18). Circular dichroism studies, x-ray fiber diffraction and an analysis of the helical repeat of homopolymeric poly(dA⅐dT) tracts clearly showed that the homopolymer is structurally different from canonical B-DNA (31)(32)(33). Failed attempts to assemble homopolymers into nucleosomes in vitro (34,35) and the observation that cloned oligoadenosine regions were excluded from nucleosome formation (36) provided the basis for the view that poly(dA⅐dT) tracts are refractory to nucleosome assembly. More recent reports, however, present a different view on homopolymeric sequences function and structure. Herrera and Chaires (37) showed that poly(dA⅐dT) tracts can assume a normal B-conformation, and other laboratories demonstrated that poly(dA⅐dT) sequences can be assembled into nucleosomes, in some cases even more favorably than heterogeneous-sequence DNA (38 -41). We have shown that a poly(dA⅐dT) element present in the ILV1 promoter is required for efficient basal level expression and that the activity of the element is partially dependent on Dat1p, a poly(dA⅐dT) DNA-binding protein (2). Analysis of ILV1 chromatin in the absence of Dat1p and in an ILV1-lacZ fusion containing a deletion of the downstream poly(dA⅐dT) element, showed no difference to a wild-type situation (data not shown), suggesting that the ILV1 downstream poly(dA⅐dT) element was not responsible for the nucleosome exclusion observed in the promoter region.
If the nuclease accessible site is caused by a strong setting preference of the nucleosomes positioned at the HS borders, we would expect that insertion of 74 bp of DNA into a previously 180-bp long stretch might allow the assembly of an additional nucleosome, thereby eliminating or splitting the nucleosomefree region. On the other hand, if a single cis-acting element present within the promoter region were responsible for the nucleosome-free region, then insertions would cause an asymmetric shift in the position of the HS. Insertion of spacing DNA in the center of the hypersensitive region (Fig. 8), clearly shows that the entire promoter structure remains unaltered, accommodating the inserted sequences within the hypersensitive site with a concomitant enlargement of the nucleosome-free region. . C, comparison of enzyme accessibility between lacZ fusion constructs containing deletions for a REB1 site or a poly(dA⅐dT) element and a wild-type ILV1 locus. Restriction enzyme analysis was carried out as before (Fig. 3).
The observed size increase suggested to us that two elements might be required to generate the hypersensitive region, such as the two A⅐T-rich tracts flanking the ILV1 Reb1p-binding site. However, simultaneous deletion of both A⅐T-rich tracts present in the ILV1 promoter had no effect on the chromatin structure of the locus (Fig. 9), excluding the possibility that these elements are responsible for the nucleosomal exclusion.
In summary, we have established the chromatin structure of the ILV1 locus, with a positioned array of nucleosomes covering the entire gene and promoter, and a hypersensitive region comprising all known cis-acting sequences. We have shown that Reb1p does not visibly affect chromatin structure of the ILV1 promoter even though the Reb1p-binding site is required for normal expression. This shows that Reb1p functions at ILV1 not through exclusion of nucleosomes but by some other mechanism of activation. We have also shown that the two poly(dA⅐dT) elements do not play an in vivo role in antagonizing nucleosome assembly in the ILV1 promoter, supporting our previous findings (2) and suggesting that binding of Dat1p, a poly(dA⅐dT)-binding factor, rather than chromatin modulation might be the function of these elements in the ILV1 promoter context.
What is then responsible for the strong nuclease sensitive site observed in the ILV1 gene? Our results strongly suggest that the DNA sequence itself plays a dominant role in orchestrating the chromatin structure at the ILV1 promoter. The DNA contained within the hypersensitive region would be intrinsically refractory to nucleosome formation, thus creating a nucleosome-free zone. By the same token, the DNA adjacent to the hypersensitive site would dictate the positioning of the nucleosomes that is so characteristic of this promoter. This scenario would explain the resiliency of the chromatin organization of the ILV1 promoter to the loss of individual factor binding sites and to localized DNA insertions and deletions. Shen and Clark (42) recently reported that also at the yeast CUP1 gene DNA sequence is likely to play a very important role in positioning nucleosomes, which together with our data suggests a major role for DNA sequence in dictating nucleosome positioning and exclusion in vivo. FIG. 8. Effect on chromatin structure of distance increments between the Reb1p-binding site and the poly(dA⅐dT) element in the ILV1 promoter. MboI fragments of the pBLUE-SCRIPT SKϩ plasmid were inserted into the XhoI site of plasmid 15X. This plasmid is a derivative of pSH601 with an 8-bp XhoI linker replacing the ILV1 promoter sequence from position Ϫ157 to Ϫ164. The resulting constructs fused to lacZ were integrated at the ILV1 locus in strain 9994 -6c (⌬gcn4). The indicated distances correspond to the size of inserted spacer DNA. DNase I analysis was performed as indicated in the legend to Fig. 1, and DNA samples resolved in a 1% agarose gel, blotted, and hybridized to an EcoRI-HinfI radiolabeled fragment. Lane M is the combination of double digest of genomic DNA with EcoRI and MboII, EcoRI and NdeI, or EcoRI and BamHI.