In Vivo Changes of Nucleosome Positioning in the Pretranscription State*

The involvement of chromatin structure and organization in transcriptional regulatory pathways has become evident. One unsolved question concerns the molecular mechanisms of chromatin remodeling during in vivo promoter activation. By using a high resolution in vivo analysis we show that when yeast cells are exposed to a regulatory signal the positions of specific nucleosomes change. The system analyzed consists of the basic elements of the Saccharomyces cerevisiae ADH2 promoter, two nucleosomes of which are shown to change the distribution of their positions by few nucleotides in the direction of transcription when the glucose content of the medium is lowered. Such repositioning does not occur in the absence of the ADH2 transcriptional activator Adr1 or in the presence of its DNA-binding domain alone. A construct consisting of the DNA-binding domain plus a 43-amino acid peptide containing the Adr1 activation domain is sufficient to induce the same effect of the full-length protein. Nucleosome repositioning occurs even when the catalytic activity of the RNA polymerase II is impaired, suggesting that the Adr1 activation domain mediates the recruitment of some factor to correctly preset the relevant sequences for the subsequent transcription steps.

The involvement of chromatin structure and organization in transcriptional regulatory pathways has become evident. One unsolved question concerns the molecular mechanisms of chromatin remodeling during in vivo promoter activation. By using a high resolution in vivo analysis we show that when yeast cells are exposed to a regulatory signal the positions of specific nucleosomes change. The system analyzed consists of the basic elements of the Saccharomyces cerevisiae ADH2 promoter, two nucleosomes of which are shown to change the distribution of their positions by few nucleotides in the direction of transcription when the glucose content of the medium is lowered. Such repositioning does not occur in the absence of the ADH2 transcriptional activator Adr1 or in the presence of its DNA-binding domain alone. A construct consisting of the DNA-binding domain plus a 43-amino acid peptide containing the Adr1 activation domain is sufficient to induce the same effect of the full-length protein. Nucleosome repositioning occurs even when the catalytic activity of the RNA polymerase II is impaired, suggesting that the Adr1 activation domain mediates the recruitment of some factor to correctly preset the relevant sequences for the subsequent transcription steps.
Nucleosomes are dynamic particles that must be modified in their structure and/or their location to allow many nuclear processes (1)(2)(3). Despite extensive evidence of in vitro nucleosome mobility (4 -8) and of the in vitro effects of chromatin remodeling machines (9 -20), the actual fate of nucleosomes during in vivo promoter activation (loss, displacement, or structural modification) is still an unresolved point. The two gene systems in which in vivo chromatin remodeling was best characterized (the yeast PHO5 and GAL1-GAL10) (21) were analyzed only at low resolution levels, thus preventing definitive understanding of the underlying mechanism.
We have studied the mechanisms of in vivo chromatin remodeling in Saccharomyces cerevisiae, focusing on the gene coding for the alcohol dehydrogenase II (ADH2) in its natural chromosomal location. The transcriptional activator of the sys-tem (22), the key regulatory region (23), and the chromatin organization of the promoter (24) were already defined.
To evaluate what exactly happens in vivo during derepression to the nucleosome particles spanning the ADH2 promoter, we apply here a high resolution method (24 -27) consisting of extensive micrococcal nuclease (MN) 1 digestion of spheroplasts, isolation, and purification of monomeric nucleosomal DNA, followed by primer extension from selected oligonucleotides allowing precise mapping of the borders of specific nucleosomal particles.
In the ADH2 promoter, we have previously shown that both nucleosomes Ϫ1 and ϩ1, spanning, respectively, the TATA box and the RNA initiation sites, consist of a family of rotationally phased translationally alternate particles. A short region encompassed between the two nucleosomes can be alternatively occupied by members of either family (24). Precise mapping of the borders was obtained in a population of cells growing in repressing conditions (3% glucose). The analysis of chromatin remodeling during derepression was so far performed at low resolution by indirect end-labeling (24,28) showing an increase in MN accessibility inside the particles and a decrease in the frequency of MN cleavage at the borders, thus suggesting a decrease of the protection of DNA by the nucleosomes (24,28,29). This is consistent with nucleosomal loss during derepression or with a change of nucleosome location or conformation allowing increased MN access. These possibilities are not distinguishable at low resolution and are not necessarily mutually exclusive.
By using a high resolution in vivo analysis we show that the two promoter nucleosomes change the distribution of their positions by few nucleotides in the direction of transcription and that this repositioning is mediated by the Adr1 activation domain.
Yeast strains carrying no plasmid were grown in YPD medium (1% yeast extract, 2% Bacto-peptone, 3% glucose). To obtain ADH2 derepression, the cells were collected by centrifugation, washed once with water, and resuspended in the same volume of fresh YP medium containing 0.05% glucose for the appropriate time.
Cells carrying the plasmids described below were grown in YNB medium (0.68% yeast nitrogen base) supplemented with the required amino acids and 3 or 0.05% glucose.
Plasmids-The Adr1 derivatives used are: pADR1 ⌬172 , consisting of the first 172 amino acids of Adr1 inserted in pRS314 (CEN6, ARSH4, TRP1), and pADR1 ⌬172-AD , same as above with the addition of a peptide * This work was supported by grants from the CNR Target Project on Biotechnology, the MURST 40% projects "Funzione e dinamica di superstrutture DNA-proteine" and "Dinamica della cromatina nell'espressione genica," and the MURST 5% project "Biomolecole per la Salute Umana." The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Enzymes-All nucleases were purchased from Roche, and Zymolyase 100T was purchased from Seikagaku Corp.
Analysis of Mononucleosomal Borders-The described method (24 -27) is slightly modified here. Spheroplasts from cells exponentially growing (A 600 0.3-0.5/ml) were treated with nystatin (100 g/ml) to allow the access of MN to the nucleus. The samples were incubated with MN (120 units/ml, 37°C, 15 min), and the reaction was stopped with 5 mM EDTA, 1% SDS. After proteinase K (2 h, 56°C), phenol-chloroform extraction, and ethanol precipitation, the samples were electrophoresed on preparative 1.5% agarose-TBE gels to isolate the mononucleosomal DNA. The monomer-sized DNA was eluted by the "Concert" rapid gel extraction system (Invitrogen), and the amount was determined by the ethidium bromide-based spot test. The samples were primer-extended with Taq polymerase in the presence of labeled oligonucleotides and analyzed by electrophoresis on 6% polyacrylamide-TBE gels. Nucleotide-level mapping was obtained by parallel sequencing lanes.
Calculation of the Weight Average of the Nucleosomal Borders-To determine the weight average of each group of borders, we started from the densitometric scannings shown in Figs. 1D and 2C and calculated the area of each border (upstream or downstream) in repressing (R) or derepressing (180Ј) conditions. These values, expressed as arbitrary units (see Tables I and II), were first corrected by multiplying for a correction factor (calculated as detailed in the legends of Tables I and  II), which takes into consideration the intrinsic differences between the upstream and the downstream profiles (namely the differential primerspecific efficiency of annealing and the sequence-dependent Taq polymerase efficiency of elongation) and then transformed in percentage relative to the total (100%). The percentage of each border was then multiplied for the number corresponding to the map position.
By summing up all the values obtained in this way, according to the formula ⌺ i n i N i /⌺ i n i , a number was determined that localizes the center of the distribution of any given group of nucleosomal borders. The correction factors introduced in this calculation were necessary to compare the area values obtained with different oligonucleotides extending on different filaments of the same monomeric DNA template.
High Resolution Chromatin Analysis-Nystatin-treated spheroplasts were digested with low amounts of MN (0.5 and 1 unit/0.25 ml, 37°C, 15 min). After proteinase K and phenol-chloroform extraction, the purified DNA was primer-extended with Taq polymerase in the presence of labeled oligonucleotides and analyzed by electrophoresis on 6% polyacrylamide-TBE gels.
RNA Analysis-For the experiment shown in Fig. 4B, aliquots containing an equal number of cells (0.5 to 1 ϫ 10 8 ) were pelleted, and total RNA was prepared as described (31). After spectrophotometric determination of the amount of RNA present in each aliquot, 10 g of RNA were loaded onto 1.2% agarose-MOPS gels, containing formaldehyde as a denaturing agent and ethidium bromide as an intercalating dye.
Northern blot analysis was performed by standard procedures with Hybond N ϩ nylon paper (Amersham Biosciences, Inc.). For hybridization, a 5Ј-end-labeled oligonucleotide specific for the ADH2 gene was used. Map positions and sequence are as follows: from ϩ710 to ϩ684, 5ЈGTTGGTAGCCTTAACGACTGCGCTAAC3Ј.

RESULTS AND DISCUSSION
Change of Positioning of the Ϫ1 Nucleosomal Family-The nucleotide level mapping of the Ϫ1 nucleosomal family, protecting the ADH2 TATA box, was previously performed in a population of cells growing in repressing conditions (3% glucose) (24). We therefore asked what would happen to the same groups of nucleosome particles when analyzed during transcriptional activation by studying the distribution of their borders after extensive MN digestion of the spheroplasts. We reasoned that, depending on the type of mechanisms underlying the ADH2 chromatin remodeling, the high resolution pattern of MN-induced cuts would be different, as follows.
(i) If nucleosomal loss is the major mechanistic feature taking place upon derepression, with no particle relocation occurring, the intensity of the entire borders profile should decrease homogeneously. In this instance, MN may have access to the DNA sites that have lost their nucleosomal organization and degrade them. The corresponding nucleosomal borders would be lost.
(ii) If a conformational change occurs upon derepression in part or in all the members of a nucleosomal family, the MN accessibility of the modified members of the family would be different from that of the non-modified ones. This would cause a partial or total modification of the distribution of the border intensities, respectively.
(iii) If the chromatin modification occurring upon derepression consists of a directional repositioning of nucleosomal particles, the intensities of the borders should vary in a predictable way. The variation of the border profile should consist of the decrease of the borders on one side, matched by the increase of the corresponding borders on the opposite side. If the repositioning occurs over several helical repeats, the appearance of new bands on one side of the initial border distribution should be observed (accompanied by a decrease or disappearance in the opposite side). Fig. 1 shows the results obtained for the nucleosome Ϫ1, in a wild type and its isogenic adr1 strain. Two divergent oligonucleotides (Fig. 1C) were used with the same amount of material: oligonucleotide 1 anneals in the center of the population and extends rightward, away from the UAS1, thus defining all the downstream borders of the nucleosome family Ϫ1 (Fig. 1A); oligonucleotide 2 anneals in the center and extends leftward, toward the UAS1, thus defining all the upstream borders of the family (Fig. 1B).
The nucleotide level mapping of the borders and the identification of the single particles is reported in Fig. 1C. Fig. 1A (downstream distribution) shows that the intensity of some borders (Ϫ1.1, Ϫ1.2, Ϫ1.3, Ϫ1.4) in the wild type but not in the adr1 strain does indeed decrease with a different rate during derepression whereas the intensity of the others (Ϫ1.5, Ϫ1.6) increases. The decreased frequency is expected from, and explained by, the accumulation of MN cleavages inside certain monomeric particles, thus causing their absence as substrates for primer extension. The increased frequency of certain borders is explained considering that some members of the family Ϫ1 "disappear" from one place to "reappear" in a different place, consistent with a shift of the position of the particles in the direction of transcription. In addition, some new bands appear above the Ϫ1.6 border (see arrowheads in The new bands, present on one side only of the initial distribution, i.e. beyond the Ϫ1.6 particle (more clearly visible for the same nucleosome in Fig. 4 and Fig. 5A, arrowheads, and for the nucleosome ϩ1 in Fig. 2A, arrowheads), are not consistent with a nucleosome loss (i) or a nucleosome conformational change mechanism (ii) but rather with a repositioning mechanism (iii), in which the most upstream particles of the Ϫ1 family move in the direction of transcription. Fig. 1D shows a densitometric evaluation of part of these data: for both the upstream and the downstream profiles only the two extreme situations, r (repressing conditions, 3% glucose) and 180Ј (derepressing conditions, 0.05% glucose), are shown. To describe more quantitatively this phenomenon, we calculated the weight average for each group of borders and defined the variation of the nucleosome positioning. The results are shown in Table I. After 180 min of activation, the weight average of the upstream borders of the nucleosome covering the TATA box shifts by 6 base pairs to the right (from position 953 in repressing conditions to 959 after promoter induction). The same numerical change (6 base pairs) is reported for the weight average of the downstream borders. This number represents only the minimal estimate of the real shift because the intensities of the new bands appearing beyond the initial distribution have not been included in this calculation.
One further feature emerges when comparing the profiles of the wild type and of the adr1 strains, as in Fig. 1A. Additional bands are present between the borders of nucleosomes Ϫ1.4 and Ϫ1.5 in the adr1 strain. The same feature is evident also in the adr1 strain transformed with a reduced version of Adr1 (see Fig. 4A) or in another mutant, the rpb1-1 strain (see Fig.   5A), although the behavior of these bands depends on the genetic context. The accessibility to MN in this specific region may be altered in these mutants because the absence of a full-length activator or the presence of a defective RNA polymerase II (rpb1-1 strain) could favor the assembly of alternative protein complexes.
Change of Positioning of the ϩ1 Nucleosomal Family-Because of the close proximity of the nucleosome family Ϫ1, which covers the TATA box region, with the family ϩ1, which covers the RNA initiation sites, we argued that the repositioning of the group of particles Ϫ1 could influence the translational positioning of the adjacent group of ϩ1 particles. We therefore extended the high resolution analysis to the family of nucleosomes ϩ1; the results are shown in Fig. 2. A second couple of  1, ϩ1.4), whereas others (ϩ1.5, ϩ1.6) show an increase, and even new positions (see arrowheads above the ϩ1.6 band) become evident. Again, the effect is observed only in the wild type and not in the adr1 strain (data not shown). Fig. 2C shows a densitometric evaluation of part of these data: for both the upstream and the downstream profiles only the two extreme situations, r (repressing conditions, 3% glucose) and 180Ј (derepressing conditions, 0.05% glucose), are shown. From these scannings we calculated the weight average for each group of ϩ1 nucleosome borders and defined the change of nucleosome positioning. The results are shown in Table II; after 180 min of activation, the weight average of the nucleosome ϩ1 upstream borders shift by 6 base pairs to the right (from position 1126 in repressing conditions to 1132 after promoter induction). The same numerical change (6 base pairs) is reported for the weight average of the downstream borders.
The change in positioning occurring to the nucleosomal family ϩ1 appears therefore to be very similar to that occurring to the neighboring group of Ϫ1 particles. If one influences the other or if the two events are unrelated remains to be established. Moreover, the constant intensity of the downstream boundary of nucleosomes ϩ1.1 and ϩ1.4 could potentially represent the consequence of the simultaneous influence of the two adjacent nucleosomes (Ϫ1 and ϩ2).
To confirm the Adr1-dependent change of translational positioning observed upon derepression in the nucleosomal families Ϫ1 and ϩ1, as obtained through the analysis of populations of isolated monomeric particles, we performed a high resolution analysis using chromatin samples obtained from mildly MN-digested spheroplasts. This approach allows the study of an entire array of nucleosomes providing information (Fig. 3) on the occupancy of the whole area encompassed between the two adjacent families of nucleosomes Ϫ1 and ϩ1 and on its variations upon derepression. Oligonucleotide 1 (as in In addition, to locate and to allow the analysis of nucleosome boundaries, MN is able to recognize and cut the DNA that has lost or is in the process of loosing its contacts with the histone octamer, as we have observed by low resolution analysis (24,28,29) and as can be clearly seen by inspection of Fig. 3. An increase in accessibility inside the ϩ1 nucleosomal family after derepression is observed when comparing the lanes R with the lanes 60Ј. Moreover, when comparing the lanes 60Ј with the in vitro-treated samples (lanes N), it appears that some bands coincide, suggesting a local loss of contact with histones, whereas other bands in the samples N are not visible in the samples 60Ј, indicating that these latter DNA sites are still protected by one or more members of the ϩ1 nucleosomal family.
One interesting possibility would be that the concerted repositioning of the Ϫ1 and the ϩ1 families of nucleosomes allows the RNA initiation sites (indicated by arrowheads in Fig. 3) to become accessible to the transcriptional apparatus.
Nucleosome Repositioning Is Mediated by the Adr1 Activation Domain-The Adr1 dependence of the change of positioning of the promoter nucleosomes can be explained by two different alternatives: (i) when the glucose in the medium becomes limiting, the activator protein occupies the UAS1 and because of its large size (1323 amino acids) pushes the more upstream borders of the Ϫ1 nucleosome toward the nearby ϩ1 nucleosome; (ii) in derepressing conditions, Adr1 binds the UAS1 and recruits additional factors which in turn cause the nucleosomes to slide.
To distinguish between these two possibilities, we have an-TABLE I Determination of the weight average for each group of borders of the Ϫ1 nucleosome family R and 180 min, repressed (3% glucose) and derepressed (180 min in low glucose) conditions, respectively. AU, arbitrary units for the evaluation of the areas. The values are based on the densitometric scannings shown in Fig. 1D. CF R and CF 180 min , correction factors consisting of the ratios between the downstream and the upstream area values for each particle in repressed and derepressed (180 min) conditions, respectively. n i , percentage of the area of each border relative to the total (100%). N i , map position of each border (see left columns and Fig. 1C). For a more detailed description of the calculation procedure see "Experimental Procedures."  Fig. 4. When using oligonucleotide 1 to elongate a monomeric population prepared from adr1 cells containing pADR1 ⌬172 we did not observe any change in the distribution of nucleosome borders (Fig. 4A, left panel), whereas in the presence of the construct pADR1 ⌬172-AD a change of the intensity of the borders is visible (Fig. 4A, right panel), just as observed in the presence of the full-length protein (Fig. 1A, wt). When using oligonucleotide 2 to elongate the same samples, again a change in the distribution of nucleosome borders is visible only in adr1 cells containing pADR1 ⌬172-AD (compare Fig. 4B, right panel, with Fig. 4B, left panel). The lack of sliding in cells containing the DNA-binding domain alone is not due to instability of the Adr1 derivative because the same molecule is capable of binding UAS1 and of inducing a conformational change in the adjacent Ϫ1 nucleosome in the absence of transcription (28). The results obtained with these two constructs indicate that the nucleosome repositioning is not simply induced as a consequence of Adr1 binding but is specifically mediated by the Adr1 activation domain.
Nucleosome Repositioning in the Absence of Transcription-Because of the requirement for the Adr1 activation domain, one could think that the change of the translational positioning of the two promoter nucleosomes during derepression is due to the Adr1-induced wave of transcription. To better investigate this possibility, we analyzed (Fig. 5) the populations of isolated monomeric particles in the temperature-sensitive strain rpb1-1 (30), in which the major catalytic subunit of the RNA polymerase II can be heat-inactivated. Cells growing at the permissive temperature (25°C) in ADH2 repressing conditions (3% glucose) were shifted to the restrictive temperature (37°C) in ADH2 derepressing conditions (0.05% glucose) and analyzed after 60, 120, and 180 min (Fig. 5A). Even in the complete absence of transcription (RNA analysis in Fig. 5B) the positions of the nucleosomes change, suggesting that the nucleosome repositioning actually precedes transcription and that the activation domain is required to recruit some factor(s) to correctly preset the relevant sequences for the subsequent transcription steps.
Chromatin remodeling upon ADH2 gene activation is there-  Fig. 2C. CF R and CF 180 min , correction factors consisting of the ratios between the upstream and the downstream area values for each particle in repressed and derepressed (180 min) conditions, respectively. n i , percentage of the area of each border relative to the total (100%). N i , map position of each border (see left columns and Fig. 2B). For a more detailed description of the calculation procedure see "Experimental Procedures."  3. Footprint of the ؊1 and ؉1 nucleosome borders. Chromatin samples, obtained by mild digestion of nystatin-treated spheroplasts with MN (0, 0.5, and 1 unis/0.25 ml), were digested with HphI. R, repressing conditions (3% glucose). 60Ј, time after the shift to derepressing conditions (0.05% glucose). N, naked chromosomal DNA treated in vitro with MN (0 and 0.15 unit/0.1 ml) and digested with HphI (map position 1358). All the samples were primer-extended with oligonucleotide 1. Arrowheads indicate the two major and the two minor ADH2 transcription start sites. fore due at least in part to the translational repositioning of nucleosomes following activator binding but preceding the actual start of mRNA accumulation. This nucleosome fluidity is likely to be facilitated by ATP-dependent chromatin remodeling complexes and/or by changing the acetylation level of the histone tails, as proposed (2), or it could be caused by TATA-bending protein-induced DNA binding as recently hypothesized (32).
Nevertheless, this change of nucleosome distribution cannot by itself represent the only process occurring upon activation: additional features of chromatin remodeling are likely to involve conformational changes of the nucleosome particles that alter their accessibility, as indicated by the low resolution analysis of the involvement of the various Adr1 domains (28) and as shown in the footprint presented in Fig. 3.
The in vivo existence of multiple nucleosome positionings in different eukaryotic systems (24 -27, 33-35), even in the repressed state, suggests the intrinsic mobility of the nucleosomal particles. Multipositioning implies that each individual cell may locate the regulatory nucleosome in one of different  (29,30) were grown at the permissive temperature (25°C) in ADH2 repressing conditions and analyzed (lane R, 3% glucose). After centrifugation, they were resuspended in derepressing medium (0.05% glucose) at the restrictive temperature (37°C) and analyzed at different times after the shift (60Ј, 120Ј, and 180Ј). The eluted monomer-sized DNA was then primer-extended with oligonucleotides 1 or 2. The nucleosome borders are identified by numbers (Ϫ1.1 to Ϫ1.6). Arrowheads indicate the additional bands appearing during derepression outside the initial border distribution of the repressing conditions. N, naked chromosomal DNA digested with HphI (map positions 773 and 1358). M, molecular size marker, pBR322 MspIdigested. A, C, and T, sequencing reactions obtained by primer extending, with the same oligonucleotide, the plasmid DNA pFA (28). B, a small aliquot of rpb1-1 cells from the treatment described in Fig. 4A was used to isolate total RNA and analyzed by Northern. Lane dubbed C (hybridization control) contains the same amount of total RNA from a culture of wild type cells grown in derepressing conditions (0.05% glucose) for 180 min. positions, inside a gaussian distribution of possible alternatives. Thus, individual cells (most of them genetically identical) are confronted with different localization choices. Being independent on the genetic background, these alternative localizations are epigenetic in their nature. The present observations show that yeast cells actually use this potentiality and distribute nucleosomes along a series of alternatives. The fact that nucleosome positions vary in a controlled manner upon a physiological event adds genetic and regulatory interest to their multiplicity.