Plasmids 601 (pGEM3Z-601) carrying the 601 positioning sequence 4445 and the plasmids p-539H6 and pCS6 46 carrying different lengths of yeast genomic DNA harboring the SNR6 gene were gifts. All numbers in this study, describing the base pair (bp) positions in the gene are with reference to the transcription initiation site at +1. Plasmid p-539H6 carries the 1180 bp gene region from -539 to +629 bp positions while pCS6 contains the 432 bp DNA from the positions -120 to +312 bp. Plasmids pU6 5'half and pU6 3'half were constructed by inserting the PCR-amplified SNR6 gene regions into the vector DNA.

Chromatin was assembled by the salt gradient dilution method as described earlier 14, except that the core histones were from Drosophila embryos. As sequence-dependent nucleosome positioning is influenced by octamer concentration 47, a histone octamer titration for each DNA was carried out before choosing the DNA:histone ratio for each reconstitution. Reconstituted chromatin samples were subsequently digested with 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'-[³²P]-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 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.

Mononucleosomes were reconstituted over PCR-amplified DNA fragments 5'-[³²P]-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. H₂O₂ 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.

Nucleosome positions were mapped on DNA, 5'-[³²P]-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.

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 represents 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 3739 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 3738.

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 MNase footprinting analysis of this chromatin did not show any boundaries separated by 145 bp nucleosomal-size 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 may be covered with the nucleosomes having unique rotational phasing.

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 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 51. The rotational information of a DNA can influence nucleosome positions 5253. Thus, the unique rotational phase of the nucleosomes on the SNR6 gene may influence the positions on the flanking genomic DNA regions, as suggested by the results in Figure 1.

The complete SNR6 gene region from TATA box at -30 to box B up to +242 bp can probably be occupied by two contiguous nucleosomes (Figure 1C). The nucleosome positioning signal on the sea urchin 5S rRNA gene is centered 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). The positions of the two protections suggest that both of them include the vector DNA and that the DNA on both the sides of the +1 site of SNR6 has nucleosome positioning properties. Asymmetry between two halves of a nucleosome is reported to facilitate translational positioning 5657. Each half of the insert DNA is less than a nucleosomal size and its continuity with the vector DNA on both sides of the +1 bp position results in its incorporation into the translationally positioned nucleosomes on both sides. This suggests that when present as part of the whole gene, the 5' half may help position nucleosomes on flanking DNA. On the plasmid bearing the 3' half DNA, a positioned nucleosome on the gene region is found flanked by the positioned nucleosomes on both sides (lanes 3, 4, Figure 4D). This suggests that central positioning of a nucleosome on this region (data not shown) leads to the alignment of more nucleosomes on both sides, giving four translationally positioned nucleosomes in the same register. Though both the SNR6 halves carry at least 17 helical turns of the DNA with same rotational phase, the 3'-half DNA has probably both rotational and translational positioning signals while the 5' half of SNR6 has the signal for unique rotational positioning, in accordance to the nucleosome positionings seen in Figure 3B. Thus, it is possible that both the halves of the gene retain their positioning properties in isolation, even when flanked by other non-genomic DNA sequences.

To dissect the gene sequence further with respect to its nucleosome positioning capability, we reconstituted mononucleosomes over DNA fragments of ~240 to 270 bp sizes (Figure 4A) carrying the two halves of the gene and mapped the locations of nucleosomes on them by using exonuclease III (Exo III) digestion assay. As compared with the IEL method, which does not differentiate between different rotational positions, Exo III mapping of nucleosomes positioned on the 5'-half DNA (Figure 5) shows the presence of several translational positions with a unique rotational phase. Mapping of the Exo III-resistant positions on both the strands, which are not well resolved in the top of the gel (lanes 6–8, panel A), shows the presence of multiple positions with 10 bp differences (lanes 6–8, panels A-D). Exo III shows a prominent pause at the +1 (marked by asterisk, 122 bp band) and -110 (232 bp band) base pair positions on the bottom strand of SNR6 (panels C and D). Out of several possibilities, mapping of both the boundaries of at least four major positions (Table 1, position nos. 7 to 10), reveals the nucleosome boundaries are 155 bp apart. Taking 156 bp as the nucleosomal size, a sequence-based nucleosome position from -116 to +40 bp, similar to position number 7 (Table 1) has been predicted on the SNR6 DNA (26, http://genie.weizmann.ac.il/pubs/nucleosomes06). Application of 145 bp nucleosome size to the most interior Exo III pauses at -20 bp on the bottom strand and +25 bp on the top strand (a comparatively weaker pause at 152 bp size DNA in lanes 6–8, panel A) would place the ends of the two nucleosomes at +125 and -120 bp, respectively, (positions 11 and 6, Table 1) at the extreme ends of this DNA (panel E). Thus, -20 to +25 bp of SNR6 represents a DNA stretch, central and common in all possible nucleosome positions on the 5' half of SNR6 (panel E), which brings the +2 bp of SNR6 to the center of all the possible positions. Exo III shows a pause at the +1 bp 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.

Exo III mapping of the centrally placed nucleosome on the 3'-half DNA (Figure 6) shows clustering of strong Exo III pauses at equal distances from the 3' end of both the strands, suggesting symmetrical and central placements of the nucleosomes with strong boundaries. Strong pauses of Exo III with 10 bp frequency is observed on both the strands. At least, 54 to 57 bp DNA is digested by Exo III from both the ends (218 and 215 bp-size products, panels A and B). Panel B shows a series of strong Exo III-resistant boundaries with 10 bp differences digesting up to 87 bp (185 bp-size product in gel) from the 3' end of the bottom strand. Mapping of the Exo III-resistant positions on both the strands shows nucleosomal boundaries separated by 145 bp. Thus, taking 145 bp as the nucleosomal size, four translational positions (gray ovals, 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.

Nucleosome positioning on the mouse mammary tumor virus long terminal repeat (MMTV-LTR) is suggested to result from the additive effects of multiple sequence-related features of the DNA 58. Taken together with this conclusion, the current study on the nucleosome positionings on and around the SNR6 gene shows that the SNR6 gene sequence constitutes strong rotational and translational positioning signals, which may influence each other in defining several alternative nucleosome positions on the gene. Centrally placed nucleosomes on the 3' half of the gene have four possible translational positions whereas several rotationally phased nucleosomes on the 5' half are more concentrated towards its 5' end. Sequences around the +1 site and between the boxes A and B of the gene play an important role in organizing the chromatin structure of the gene.

Most of the positions mapped in this study and summarized in Table 1 can be correlated to the positions reported earlier 3373959 or predicted by the SNR6 sequence 26. Figure 7 compares a summary of the results of this study (in vitro, panel B) with the nucleosome positions from our previous study (37, in vivo, panel A). The striking similarity of the mapped positions in the gene upstream region (in bold, panel A) and less defined rotational phase of the DNA upstream of -100 bp (Figure 3) suggests that the nucleosomes upstream of -70 bp position are directed by the SNR6 sequence. With a strong possibility of multiple sequence-directed positions at uniform intervals on the SNR6 gene region (Figures 4, 5, 6), when both the halves of the SNR6 are together in the gene, the nucleosome positions on the 5' half of the gene may align with those in the downstream region.

Two different registers (R1 and R2, Figure 7C and 7D) of the nucleosome positions on the gene region may be possible under different conditions, along with alignment of positioned nucleosomes on both the gene flanking regions. In the register R1, the 5' nucleosome may be centered around the +1 site while downstream nucleosome may be more towards the 3' end of the gene, covering the box B. This arrangement could be possible by selecting either the nucleosome positions 8 and 14 or 9 and 15 to align with the position 5 in the upstream region (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 37414243. 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.

Several synthetic and some naturally occurring sequences have been used for the deposition of positioned nucleosomes in vitro 566061 but none of them give unique positioning in vivo. SNR6, similarly, does not have a signal for unique translational positioning. Rotational positioning is determined by the bendability of the involved sequence 6263. Therefore, the nucleosome positioning capability of SNR6 DNA may be inherent in its DNA sequence, which shows a unique rotational setting. Sequences that have intrinsic curvature, or that are flexible, prefer to be incorporated into nucleosomes since it takes less energy to bend this type of DNA segment around the core histones 4557. These sequences do not show any consensus except that the trajectory of the DNA may be bent 45. We used a simple program, BEND 64, which calculates the magnitude of local bending and macroscopic curvature at each point along a DNA sequence. Output of the program (data not shown) for the SNR6 sequence in pCS6 revealed that the SNR6 DNA has intrinsically bent DNA segments, suggesting the bending and curvature of this DNA may have a role in the nucleosome positioning. It will be interesting to explore the possibility of using different regions of the naturally occurring SNR6 gene sequence for nucleosome positioning in vitro and 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 3738. 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).

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 2326. 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 2829, 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, 338in vitro and 37in 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.