Multiple sequence-directed possibilities provide a pool of nucleosome position choices in different states of activity of a gene

Background Genome-wide mappings of nucleosome occupancy in different species have shown presence of well-positioned nucleosomes. While the DNA sequences may help decide their locations, the observed positions in vivo are end-results of chromatin remodeling, the state of gene activity and binding of the sequence-specific factors to the DNA, all of which influence nucleosome positions. Thus, the observed nucleosome locations in vivo do not reflect the true contribution of DNA sequence to the mapped position. Moreover, the naturally occurring nucleosome-positioning sequences are known to guide multiple translational positionings. Results We show that yeast SNR6, a gene transcribed by RNA polymerase III, constitutes nucleosome-positioning sequence. In the absence of a chromatin remodeler or any factor binding, the gene sequence confers a unique rotational phase to nucleosomes in the gene region, and directs assembly of several translationally positioned nucleosomes on ~1.2 kb DNA from the gene locus, including the short ~250 bp gene region. Mapping of all these gene sequence-directed nucleosome positions revealed that the array of nucleosomes in the gene upstream region occupy the same positions as those observed in vivo but the nucleosomes on the gene region can be arranged in three distinct registers. Two of these arrangements differ from each other in the position of only one nucleosome, and match with the nucleosome positions on the gene in repressed and active states in vivo, where the gene-specific factor is known to occupy the gene in both the states. The two positions are interchanged by an ATP-dependent chromatin remodeler in vivo. The third register represents the positions which block the access of the factor to the gene promoter elements. Conclusion On a gene locus, multiple nucleosome positions are directed by a gene sequence to provide a pool of possibilities, out of which the preferred ones are selected by the chromatin remodeler and transcription factor of the gene under different states of activity of the gene.


Background
Nucleosomes, the fundamental building blocks and repeating units of eukaryotic chromosomes, not only pack the genome but also participate in gene regulation [1]. The position of nucleosomes must be well defined in order to ensure proper control of all DNA-related activities, as folding of DNA by the histones and positioned nucleosomes can help to establish contact between two remotely placed transcriptional factors [2,3]. Genomewide mapping has shown inverse correlation of nucleo-some occupancy with promoter strength and transcription initiation rate [4,5]. Genes show well-defined patterns of positioned nucleosomes with respect to transcription initiation site (+1 site), transcription factor binding sites and the transcribed regions [6]. Yeast promoters show low nucleosome density [7] while the coordination of nucleosome positions with gene activity is a complex process involving interactions between nucleosomes, transcription factors, histone-modifying enzymes and chromatin remodelers [8].
The location of nucleosomes on the DNA can be dictated by trans-acting factors as well as DNA sequences [9][10][11]. While nucleosomes can get translationally positioned by aligning next to DNA-bound proteins [12][13][14], cells probably use the sequence preferences of nucleosomes for regulating the binding site accessibility of transcription factors [15][16][17][18]. Some naturally occurring positioning sequences are reported to be responsible for positioned nucleosomes on gene regions in vivo in the absence of any bound trans-acting factors [19][20][21][22]. Genomic DNA with both low and high-affinity sequences for histone octamers could carry a code for nucleosome arrangement guided solely by DNA sequence. In agreement with this, a comparative genomics study has demonstrated that the organization of nucleosome positioning sequences in the yeast genome can be used to predict genome-wide nucleosome positions [23]. Earlier reports had suggested that ~95% of chicken genomic DNA does not show a histone affinity different from synthetic, random DNA sequences [24]. However, higher-resolution data over large contiguous regions of yeast DNA revealed that ~70% of the nucleosomes on chromosome III are well positioned [25]. It was further suggested that ~50% nucleosome positions in vivo are encoded by the genomic DNA sequence [26]. More recently, a complete high-resolution map of nucleosome occupancy across the whole genome of yeast has shown that 81% of the yeast genome is covered by positioned nucleosomes [27]. Sequence-directed nucleosome positioning in vivo is further regulated in trans by ATP-dependent nucleosome remodeling complexes [28][29][30]. Histone variant H2A.Z is also shown to regulate gene activity and nucleosome positioning genome wide [6,[31][32][33].
One of the naturally occurring positioning sequences, 5S rDNA, which is reported to give multiple translationally positioned nucleosomes with unique rotational setting [34], belongs to the class I genes transcribed by the RNA polymerase III (pol III). The yeast SNR6 gene, which codes for the U6 snRNA, represents class III of the pol III-transcribed genes [35]. The promoter architecture of SNR6 shows an unusual combination of an upstream TATA box, intragenic A box and downstream B box [36] to which the basal transcription factors TFIIIB and TFIIIC bind. Previous studies on SNR6 in our lab have shown a good corre-lation between the in vivo chromatin structure [37] and factor-dependent chromatin structure in vitro [3,38]. The in vivo chromatin structure of SNR6 is reported to have an array of positioned nucleosomes upstream of the TATA box and downstream of A box, flanking a nucleosomefree region between TATA box and A box in the active state of the gene [37,39]. In a similar genome-wide arrangement of nucleosomes on pol II-transcribed genes in yeast, positions of the two flanking nucleosomes are reportedly specified by the DNA sequence [40]. As the SNR6 gene is always occupied by its basal factor TFIIIC in vivo [41][42][43], the contribution of the gene sequence in establishing the chromatin structure is difficult to ascertain in vivo. This may be true for many other genes as well, which are persistently occupied by their transcription factors in vivo. The salt gradient dialysis method of chromatin assembly in vitro deposits nucleosomes in a sequence-dependent fashion and this method has been useful in checking the ability of various DNA sequences to position nucleosomes in vitro. Using this method of chromatin assembly, we show here that the SNR6 gene sequence has intrinsic nucleosome-positioning signals for discrete translational positions and unique rotational setting, which results in alignment of an array of positioned nucleosomes in both directions on the gene-flanking regions. At any given time, two nucleosome positions on the gene can be contiguous with the array in the 5' upstream region, giving three possible registers. Our results explain how the gene sequencedirected, multiple nucleosome positioning may help establish the chromatin structure of the gene locus in vivo. micrococcal nuclease (MNase) or DNaseI to carry out further analysis.
Mononucleosomes were assembled on DNA fragments of more than 200 bp size to avoid the end-positioning effect. DNA fragments were PCR amplified from plasmids using appropriate primer pairs having one of the primers 5'-[ 32 P]-end labeled. The PCR products were gel purified and per reaction ~10,000 to 20,000 cpm of labeled probe was mixed with the parent plasmid DNA for chromatin reconstitution by salt dilution method. After the reconstitution, 10 μl samples were loaded on 5% native polyacrylamide gel. Gels were dried after the run and retarded mobility of the reconstitute was ascertained by the Phosphor Imaging (Fuji) to visualize the mononucleosome assembly.

Chromatin structure analysis
Chromatin was subjected to MNase or DNaseI digestion for the low-resolution indirect end-labeling (IEL) analysis or primer-extension footprinting, respectively, as described earlier [14]. All the plasmids have unique sites for the restriction enzyme AlwN1 ~0.8-1.2 kb away from the gene regions in the vector DNA, which was used for the secondary digestion of the MNase-digested naked DNA or chromatin for the IEL analysis. A protection of 145 bp or larger size DNA seen in IEL analyses was taken as the indicative of a positioned nucleosome. Image Gauge software (Fuji) was used to generate the profiles of partially digested and gel-resolved naked DNA and chromatin samples from the phosphorimages of the footprinting gels. All protections were ascertained by matching the profiles of the lanes with similarly digested DNA.

Hydroxyl radical footprinting
Mononucleosomes were reconstituted over PCR-amplified DNA fragments 5'-[ 32 P]-end-labeled on either of the strands. Hydroxyl radical cleavage of the DNA was followed on both the strands in 60 μl reaction volumes. Briefly, 5 μl of 1 mM Fe(II)/2 mM EDTA and 10 μl of 10 mM sodium ascorbate are put together on the sides of the tube, to which 10 μl of 0.6% wt/vol. H 2 O 2 is added and immediately mixed with the reconstituted chromatin/ naked DNA [48]. The cleavage was allowed to proceed for 3 and 5 minutes and the reaction was quenched by the addition of 10 μl of 100 mM thiourea. The DNA sample was cleaned by phenol:chloroform (1:1) extraction, ethanol precipitated and resolved in 8% denaturing urea-acrylamide gel. Gels were dried, exposed for phosphorimaging and profiles were generated using Image Gauge software from Fuji.

Exonuclease III footprinting
Nucleosome positions were mapped on DNA, 5'-[ 32 P]end-labeled on one of the strands [49]. The reconstituted mononucleosomes were digested with 20 U/ml Exonuclease III (NEB), for 0, 3, 6, and 9 minutes as compared with 0, .5, 1.5, and 2 minutes for the naked DNA samples. The digestion was stopped by adding the 10× exonuclease stop buffer having 0.5 mg/ml proteinase K, 200 mM Tris-HCl pH8, 50 mM EDTA and 5%SDS. DNA was phenol extracted, ethanol precipitated and resolved on the 8% urea-acrylamide denaturing gel. Gels were dried after the run and exposed to the Phosphor Imaging screen (Fuji). Bands appearing first and remaining resistant to ExoIII digestion during the time course were taken as nucleosome boundaries.

Results
As shown in Figure 1A, the SNR6 gene locus constitutes a TATA box at -30 bp, box A at +21 bp, the terminator at +109 bp, and box B at +233 bp positions (with respect to +1 at transcription initiation site). To find the contribution of the genomic DNA sequence to the nucleosome positions, we used the salt gradient dilution method to deposit nucleosomes on plasmids carrying different parts of genomic DNA from the SNR6 gene region in the absence of any bound transcription factor, and subjected the chromatin to structural analyses for locating the positioned nucleosomes, if any.

Yeast SNR6 locus is covered by positioned nucleosomes
The plasmid p-539H6 carrying ~1.2 kbp of the genomic DNA from the SNR6 locus (marked with a vertical line, Figure 1B) shows significant differences between MNase digestion patterns of the chromatin and naked DNA in the IEL assay. The region downstream of the gene terminator at +110 bp shows ~700 bp-long protected region, suggesting this region may be covered by rotationally phased nucleosomes, probably with overlapping translational positions. Digestion of the ~540 bp-long naked and chromatin DNA upstream of +110 bp by MNase shows significant differences. Mapping of the MNase-cut positions revealed the presence of a translationally positioned nucleosome between the +110 and -78 bp and an array of nucleosomal-size protections (marked with gray ovals) upstream of it, which can be arranged in two distinct registers. Those marked on the right-hand side of lanes 3 and 4 may have three positions -78 to -241, -241 to -453 and -453 to -656 bp in one register (positions 3-5, Table 1) while those marked on the left-hand side at -78 to -368 and -368 to -545 bp (positions 1 and 2, Table 1) may be in another register. A deletion of 2 bp in box B in vivo, which abolishes TFIIIC binding, was reported to result in rearrangement or destabilization of the nucleosomes in the gene-flanking regions [39], suggesting these nucleosomes in vivo are organized by TFIIIC-dependent chromatin remodeling. It is interesting to note that the in vivo structure was reported to have boundaries of two upstream positioned nucleosomes at base pairs -537 and -367 [39]. Thus, it is possible that nucleosome positions 1 and 2 (Table 1) represent the possible nucleosome locations in the presence of TFIIIC, while nucleosome 5 repre-sents a position in the absence of TFIIIC. The upstream nucleosomal array on the genomic DNA in p-539H6 is similar to that on the SNR6 locus in vivo [37,39] but structure downstream of +110 bp on the plasmid is different, probably because the gene is occupied by TFIIIC in vivo and the chromatin in the gene region is remodeled after TFIIIC binding [37,38].

Box B is buried in the nucleosomal region
In order to know further details of the chromatin structure close to the transcribed region of the gene, we assembled the chromatin in vitro on the plasmid pCS6 which carries 400 bp genomic DNA region (vertical line, from the positions -120 to +312; Figure 1C) from the SNR6 gene locus. Southern probing of the MNase digest of the chromatin by the gene-specific and vector DNA-specific probes showed a better nucleosomal ladder on the gene region (data not shown), suggesting the gene has higher affinity for histones. IEL analysis of the MNase digestion patterns of the chromatin and naked DNA control shows that the complete gene region from -50 to +316 bp is protected in chromatin (lanes 3 and 4, Figure 1C). However, the Sequence-directed nucleosome positions on the yeast SNR6 gene MNase footprinting analysis of this chromatin did not show any boundaries separated by 145 bp nucleosomalsize protections (data not shown). Two nucleosomal-size protections are seen in flanking regions of the gene on the vector DNA also (dark ovals, Figure 1C). Since the positioned nucleosomes are seen only on and around the gene region, compared with the vector DNA, it is possible that the positioning is related to the gene sequence. The MNase digestion patterns and the mapped MNase cut sites on the chromatin assembled over the plasmids p-539H6 and pCS6 do not match, probably due to the sizes of the mapped regions, therefore resolution differences of the gels. Deletion of the DNA upstream of position -140 bp in vivo was reported to result in loss of the upstream array of the nucleosomes [39]. Therefore, the absence of the array may be because of the absence of DNA upstream of -120 bp in pCS6. However, the presence of a positioned nucleosome on the gene-flanking vector DNA suggests that the gene sequence directs positioned nucleosomes in its immediate vicinity as well. The absence of a unique translational positioning of nucleosomes and the protection size of 366 bp (-50 to +316 bp), which has box B of the gene in its center, suggest that the whole gene region -60 to +95 p-539H6, Figure 1B, sequence-directed. 10.

SNR6 confers a unique rotational phase to nucleosome
We used DNase I footprinting ( Figure 2) and hydroxyl radical cleavage (Figure 3) to find the presence of a 10 bp ladder, characteristic of rotationally positioned nucleosomes, on the gene region. DNase I footprinting of both the strands of the chromatin assembled on the plasmid pCS6 ( Figure 2, panels A and B) shows a frequency of cut with ~8 to 12 bp periodicity. DNase I cuts two strands of DNA with a 4 bp stagger and there may be an error of 1 to 2 bp in upper parts of the gel in identifying the cut positions. Nevertheless, combining the mapping on both the strands, a 10 bp periodicity of cuts in the whole region can be seen, suggesting that ~180 bp DNA sequence between the boxes A and B (from +33 bp to +213 bp) may have a rotational nucleosome positioning signal.
Hydroxyl radical cleavage of SNR6 chromatin was performed in two parts, on both the strands of each part. U6ab and U6us ( Figure 3A) were PCR amplified from the plasmid p-539H6, and mononucleosome assembly on them was monitored by gel shift assay ( Figure 3B). Similar to the 601c DNA (lane 2), which gives a centrally positioned nucleosome [50], both U6ab (lane 4) and U6us (lane 6) DNAs show major population of a centrally positioned nucleosome. Additionally, one band for U6ab and three minor bands for U6us can also be seen. Mapping of hydroxyl radical cleavages on both the strands of U6ab DNA, which covers the SNR6 gene region from +14 to + 256 bp (Figures 3C and 3D) shows a continuum of 10 bp periodicity on the whole region. Similar mapping on both the strands of U6us carrying the upstream region of the SNR6 gene shows a well-pronounced helical periodicity up to -100 bp which appears to be less defined in the region further upstream. This is better revealed by the top Chromatin DNA between the boxes A and B has a unique rotational phase strand cleavage pattern (gel in Figure 3F and profile in the Figure 3E). When cleavage maps of both parts are taken together, a periodicity of 10 ± 1 bp is found to prevail in the same phase on the whole gene region from -100 bp to +210 bp. The deviation by 1 bp may be because of the change in periodicity at the dyad axis of a nucleosome tered around the +1 position, giving a nucleosome positioned from -78 to +78 bp [54]. As box B of SNR6 is further downstream, similar positioning on SNR6 would allow formation of one more nucleosome on the gene region, downstream of +90 bp position. Therefore, we separated the SNR6 gene sequence into two halves and cloned the bp regions -87 to +83 (5' half of the gene) as well as +62 to +256 (3' half of the gene) into two different plasmids ( Figure 4A). We used the IEL technique to further confirm the nucleosome-positioning properties of both the halves in the context of the flanking plasmid vector sequences.
The IEL technique can map nucleosome positions with a fairly good accuracy [55]. For example, IEL of the 601c chromatin shows the presence of a single positioned nucleosome on the synthetic sequence (lanes 3, 4, Figure  4B). Two nucleosomal-size protections (gray ovals) are seen on two sides of the center (+2 bp) of the genomic DNA insert in the 5'-half plasmid (lanes 3, 4, Figure 4C).    position (panels D and E, the asterisk), which shows alignment of two nucleosomes on its two sides in the 5'half DNA ( Figure 4B), suggesting this may be a central, reference position for organization of nucleosomes on this helically phased DNA. These results suggest that depending on the context, nucleosomes either exclude or assemble over the +1 site. The possibility of multiple positions at uniform intervals on the 5'-half DNA suggests that the protection seen upstream of +110 bp position in the plasmid p-539H6 ( Figure 1B) may be due to the translationally positioned nucleosomes with unique rotational settings and not due to a unique translational position. Thus, a similar positioning may be observed in vivo due to this sequence, present as part of the full-length gene.  Figure 6C) with unique rotational phase can be deduced from mapping on both the strands of this DNA (positions 12 to 15, Table  1). The position +71 to +216 may be the same as the position +71 to +227 predicted by the DNA sequence, which takes 156 bp as the nucleosomal size ( [26], http:// genie.weizmann.ac.il/pubs/nucleosomes06). Thus, the 3' half of the SNR6 gene may have strong signals for both translational and rotational nucleosome positionings. Most of the positions mapped in this study and summarized in Table 1 can be correlated to the positions reported earlier [3,37,39,59] or predicted by the SNR6 sequence [26].  Figure  7C). While nucleosomes 14 and 15 would cover the box B at their 3' ends, nucleosomes 8 and 9 would block the TATA box to A box region. In the second alternative (R2), nucleosome 6 may align with the nucleosome 12 or 13 and generate a condition in which TFIIIC can occupy boxes B and A but TATA box and +1 site remain masked. However, the presence of nucleosome 6 would exclude the possibility of occupancy on the position 5 ( Figure  7D). Such an arrangement has been observed in vivo under repression when TFIIIC is seen occupying the gene [37,[41][42][43]. Therefore, R1 represents the arrangement in the absence of TFIIIC, while the first step of chromatin remodeling associated with TFIIIC binding [3] may lead to the arrangement R2 ( Figure 7D) in vivo. While the sequence ensures the gene is covered by positioned nucleosomes, the nucleosomes subsequently occupy one of the sequence-directed positions in the active state to give another arrangement R3 in vivo, as a result of the second step of sequential remodeling [38], as discussed below.  gene sequence for nucleosome positioning in vitro and in vivo.

Establishment of SNR6 chromatin structure in vivo
Results from this study have shown that in the absence of any factor, several positioned nucleosomes cover the entire gene region. However, all the mapped positions cannot be occupied simultaneously and may be mutually exclusive. Alignment of contiguous translational positions in the ground state would result in nucleosomes completely covering the gene and blocking the access of its factors to all the target sites. Different translational possibilities of the nucleosomes downstream of -140 bp probably represent the sequence-directed positions adopted by nucleosomes under different states of the activity of the gene.
We had shown earlier that TFIIIC can access the box B of SNR6 buried in nucleosomes in vitro [3], and a nucleosome between the boxes A and B shifts by ~40 bases due to the subsequent chromatin remodeling [38]. As a result, the nucleosome takes a unique translational position between bp +50 and +190 [3]. Thus, the nucleosomes on positions 13 to 15 (Table 1) in the absence of TFIIIC possibly acquire position 12 (Table 1) due to chromatin remodeling after TFIIIC binding in vivo. In agreement with this, the sequence +101 to +196, common in the four overlapping positions on the 3' half ( Figure 6C), matches well with the reported protection (from +94 to +198 bp) on SNR6 in vivo as estimated by MNase footprinting [59]. When TFIIIC binds to the box B, the arrangement R1 is converted to R2, whereby the upstream nucleosome positions 8 and 9 (arrangement R1), covering the box A in the ground state of the gene, would realign and move to the position 6, as a result of TFIIIC-dependent chromatin remodeling in vivo. As the nucleosome 6 covers the +1 site and the TATA box, and gives a partial block of the box A at its 3' end, another remodeling will be required to activate the gene. We had reported earlier a sequential remodeling of the SNR6 chromatin which shifts the nucleosome from the TATA box to the region -70 to -240 bp (position 5) further upstream [37,38]. Activation of the gene brings the chromatin remodeler RSC [37], which facilitates the upward shift of the nucleosome from the position 6 to position 5, generating the arrangement R3 ( Figure 7E), wherein a nucleosome-free region is flanked by two positioned nucleosomes (positions 5 and 12) as found in vivo ( Figure 7A). The chromatin structure R3, finally generated by these sequence-dependent rearrangements, depends on the remodeler, which leaves the gene in a repressed state [37]. Significantly, similar to the difference between the repressed and active state chromatin structure of the gene [37], R2 and R3 differ from each other only in the position of one nucleosome (position numbers 6 or 5, Table 1, Figures 7D and 7E).

Sequence-directed positioning of nucleosomes in vivo
This study shows that the final nucleosome positions on SNR6 in vitro and in vivo are influenced by the combined effects of different segments of the gene sequence. As a chromatin remodeler is recruited to the target genes by its factors, the binding of a factor may decide the nucleosome positions in the active and repressed state of a gene region [14]. Several genome-wide studies on nucleosome arrangements have recently suggested that the genome codes its own packaging by having most of the nucleosomes positioned in a sequence-directed manner [23,26]. Different sequence-directed positions of nucleosomes are chosen by different chromatin remodelers as end-products of their remodeling activities. In the absence of the remodeler Isw2 in yeast, nucleosomes were found to adopt sequence-directed positions genome wide [28,29], suggesting a chromatin remodeling is used to choose between the sequence-directed alternate positions of nucleosomes. However, a nucleosome positioning sequence database NPRD [26] has reports on only four genes, which show sequence-directed nucleosomes in vivo. On one well-studied yeast gene locus, PHO5, intrinsic properties of the promoter DNA were found to give nucleosome positions similar to those in vivo [65]. Similarly, sequence is suggested to play an important role in positioning nucleosomes on yeast CUP1 locus [20] and MMTV 3' LTR DNA [17]. Our studies on SNR6 chromatin structure (this study, [3,38]in vitro and [37]in vivo) establish a strong correlation between the sequence-directed positions of the nucleosomes before and after TFIIIC binding as well as chromatin remodeling. The combined results of these studies show that transcription factor binding and chromatin remodeling modulate the nucleosomal organization of the SNR6 gene region in vivo when the resultant nucleosome positions are not randomly generated. They are rather chosen from few sequence-directed select possibilities.

Conclusion
Our results have shown that all the nucleosomes found associated with a gene locus in vivo under different states of its activity correspond to one of the multiple positions directed by a genomic DNA sequence. This may be the reason that in contrast to synthetic sequences, which could be designed to give unique positionings, the naturally occurring nucleosome-positioning signals give multiple alternatives and cannot be defined by consensus sequence elements.