|Home | About | Journals | Submit | Contact Us | Français|
The traditional model for chromatin remodelling during transcription has focused upon the remodelling of nucleosomes at gene promoters. However, in this study, we have determined that Tup1-Ssn6 and Swi-Snf chromatin remodelling activities extend far upstream of the SUC2 gene promoter into the intergenic region of the Saccharomyces cerevisiae chromosome. We mapped the nucleosomal array over a 7.5 kb region that encompassed the SUC2 gene promoter and upstream region but was devoid of other transcriptionally active genes. Nucleosome positioning over this region was determined under conditions of glucose repression and derepression, and in snf2, ssn6 and snf2 ssn6 mutant strains. A map detailing remodelling events extending as much as 5 kb upstream of the SUC2 gene promoter underlines the roles of the Tup1-Ssn6 and Swi-Snf complexes in respectively organizing and disrupting nucleosome arrays. The gene specificity of these events suggests a role in gene regulation. We propose that long-range chromatin remodelling activities of Swi-Snf and Tup1-Ssn6 may ultimately influence whether the chromosomal state of the SUC2 gene is proficient for transcription. These data raise the possibility that remodelling of extensive chromatin domains may be a general property of the Swi-Snf and Tup1-Ssn6 complexes.
The discovery of chromatin remodelling has revolutionized chromatin research by providing new insights into how the packaging of the eukaryotic genome into nucleosomes participates in gene regulation. Instead of being a static structure, chromatin is now accepted to have a dynamic organization from the nucleosomal level up to higher order structures (1). The identification of numerous multiprotein complexes that are involved in histone modification and remodelling of nucleosome arrays has revealed previously unappreciated levels of control over the basic chromatin organization, with defects in these processes leading to inappropriate gene expression and disease. To understand the dynamic mechanisms that generate these specialized chromatin structures and predispose genes to activation or repression, we have focused on the Tup1-Ssn6 co-repressor and the Swi-Snf co-activator whose interplay regulates the balance between repressed and active chromatin structures at a number of yeast genes.
Swi-Snf is arguably the best-known example of a chromatin remodelling complex that can act as a transcriptional co-activator (2). This large multi-subunit complex is targeted to gene promoters by sequence-specific DNA-binding transcription factors that interact with the Snf5 and Swi1 subunits (3,4). Swi-Snf utilizes the energy from ATP hydrolysis to alter the structure of chromatin, thereby enhancing nucleosomal DNA accessibility and enabling gene transcription (5–9). Potential mechanisms of action include local DNA deformation resulting in nucleosome sliding, and histone octamer transfer in cis and in trans (10–13). Conversely, the Tup1-Ssn6 co-repressor complex has been shown to organize chromatin into a repressive structure, possibly through direct contact with hypoacetylated histones, and in conjunction with histone deacetylases (14–20). The subsets of genes regulated by these complexes overlap at the FLO1, SUC2 and RNR3 genes (21–23). Indeed, these genes represent a paradigm for chromatin-mediated gene regulation and offer a unique insight into the interplay between the two complexes.
The traditional model for chromatin remodelling complexes has focused on their activity at gene promoters (22,24–26). However, evidence is emerging that remodelling also involves longer-range effects. We have shown that remodelling by Swi-Snf, as well as by Tup1-Ssn6, extends some distance upstream of the FLO1 gene promoter (21). Long-range remodelling has also been shown to occur over the coding regions of genes. For example, Tup1-Ssn6 and Isw2 cooperate to position a regular array of nucleosomes over the promoter and gene-coding region of the repressed RNR3 gene (27). Upon RNR3 induction, Swi-Snf is required to disrupt this array (23). Long-range remodelling by Swi-Snf and Isw1 has also been observed at the HIS3 gene in a mini chromosome context, where induction of transcription was accompanied by the disruption of nucleosomes over the entire gene sequence and flanking regions (28,29). An episomal yeast CUP1 gene has also been shown to be subject to extensive chromatin remodelling over the entire open reading frame (ORF), 5′ and 3′ flanking regions following activation (30). In each of the above examples, the data suggests that remodelling is a pre-requisite for transcription, and not a consequence of it.
In this study, we have characterized in detail the nature and precise extent of chromatin remodelling in the region upstream of the yeast SUC2 gene. Using indirect end-labelling and primer extension analyses, we have mapped remodelling events attributable to both Swi-Snf and Tup1-Ssn6 over a 7.5 kb region covering the SUC2 promoter and far upstream sequences. This extensive nucleosome map reveals how remodelling by both complexes extends as much as 5 kb upstream of SUC2. The long-range chromatin remodelling activities reported here support the proposition that it is a general mechanism used by chromatin remodelling complexes such as Swi-Snf and Tup1-Ssn6 to organize extensive chromatin domains in response to cues for transcriptional activation and repression.
The yeast strains used were wild-type (wt), AFH41 [S288C MATa ade2-101 his3Δ200 leu2-3,112 lys2-801 trp1Δ901 ura3-52 GAL thr- tyr- arg4-1 hhf1Δ::HIS3 hhf2Δ::LEU2/pUK499 (URA3/HHF2)]; and its derivatives snf2, AFH44 (AFH41 snf2Δ::TRP1); ssn6, AFH47 (AFH41 ssn6Δ::kanMX4) and snf2 ssn6, AFH410 (AFH44 ssn6Δ::kanMX4) (21). Yeast were grown in YP supplemented with glucose (2%) to mid-log phase. The cultures were divided into two equal portions and harvested by centrifugation. Cell pellets were washed twice in sterile water and resuspended to the same cell density in fresh YP containing glucose at either 2% (repressed, R) or 0.05% (derepressed, D) (22). The cultures were incubated for a further 120 min at 30°C and harvested for either RNA or nuclei preparation.
Indirect end-labelling and primer extension analysis were performed as described (21). For the preparation of glucose-repressed nuclei, repression was maintained by supplementing all buffers with 2% glucose (22). For indirect end-labelling, HindIII-digested DNA was probed with PCR fragments corresponding to base-pairs −1099 to −817, −4215 to −3948 and −7801 to −7519 upstream of the SUC2 ATG (SUC2 SGD ID: S000001424); DraI-, HinfI- and BamHI-digested DNAs were probed with fragments −1736 to −1359, +1 to +285 and −4033 to −4303, respectively and BsrBI-digested DNAs were probed with fragments −2324 to −1899 and −2697 to −2399.
For primer extension analysis the primers used correspond to positions −797 to −772 (P1); −3155 to −3131 (P2) and −3587 to −3563 (P3) of the SUC2 upstream sequence. Band intensities were determined by phosphoimager analysis (FujiFilm FLA2000 FluoroImager).
Total RNA was extracted from cells grown as described above and 20 μg samples were analysed by northern blot after electrophoresis in a 1.25% agarose-formaldehyde gel (21). SUC2, ACT1, YIL167w, YIL166c, YIL165c and NIT1 transcripts were analysed with probes corresponding to ORF positions +119 to +1222, +411 to +1421, +26 to +620, +598 to +920, +39 to +351 and +47 to +388, respectively.
Total RNA was treated with DNaseI (Promega) and cDNA was generated using a poly-dT primer and Superscript III reverse transcriptase (Invitrogen). Real time PCRs for ACT1 and SUC2 were performed with 1/25 dilutions of the cDNA reactions using the SYBR Green Master Mix (SuperArray) in an ABI 9700 PCR machine. RT-PCRs for YIL166w, YIL165c and NIT1 cDNAs were performed on a 1/5 cDNA dilution, and undiluted cDNA reactions. Values were normalized to ACT1 RNA. Real time PCRs were performed in triplicate. The primer sequences used were: SUC2, 5′-CCATTGCTATCGCTCCCAAG-3′ and 5′-TGGAGCCAGAGAAAGCACCT-3′; ACT1, 5′-GAGGTTGCTGCTTTGGTTATTGA-3′ and 5′-ACCGGCTTTACACATACCAGAAC-3′; NIT1, 5′-TCCCAGAAGCCACTCTTGGT-3′ and 5′-AAACCCCAAAGTTCGATCCC-3′; YIL165c, 5′-GCTCGCTTTGATCTTGACCC-3′ and 5′-TGGAAGACATCTCCCCTAGCA -3′; YIL166c, 5′- ACTACCCGGCAATCTGCTGT-3′ and 5′-GGAATGACCCTTTCTGGACCA-3′.
The SUC2 gene encodes the enzyme invertase required for sucrose utilization and is subject to glucose repression through the Tup1-Ssn6 complex (31,32). As previously shown, transferring cells grown at high glucose concentration (repressing, R) to low glucose conditions (derepressing, D) induces SUC2 transcription (Figure 1B, lanes 1, 2 and 1C) and a deletion mutation of SNF2 that cripples the remodelling activity of the Swi-Snf complex, abolishes this induction demonstrating Swi-Snf dependence (Figure 1B, lanes 3, 4 and 1C) (31,33). Conversely, deletion of either SSN6 (Figure 1B, lanes 5, 6 and 1C) or TUP1 (data not shown) results in high-level constitutive SUC2 transcription (22,33–35). In the absence of both the Swi-Snf and Tup1-Ssn6 complexes, SUC2 is also constitutively transcribed, but at lower levels (Figure 1B, lanes 7, 8 and 1C) (22,33,34).
Numerous studies had suggested that upon gene activation or repression chromatin remodelling is limited to the immediate promoter region. However, evidence is emerging that remodelling complexes can in fact operate over longer distances. Indeed, our previous work indicated that this was the case for Swi-Snf and Tup1-Ssn6 regulation of FLO1 transcription (21). If similar long-range remodelling were apparent far upstream of SUC2, it would strengthen the case that the ability to organize extensive chromatin domains is a general feature of Swi-Snf and Tup1-Ssn6. However, prior to embarking upon an analysis of chromatin remodelling events upstream of SUC2, we first characterized the extent of distal gene activity potentially affecting this region.
SUC2 is located ~35 kb from the telomere on the left arm of chromosome IX (36). Four ORFs are located in the 10 kb region upstream of SUC2. Two ORFs (YIL167/168w and YIL 164/165c) are interrupted by a stop codon in S288C strains, while the short YIL163c has dubious ORF status (Figure 1A). We analysed these ORFs for transcriptional activity, and determined whether they were subject to regulation by Tup1-Ssn6 or Swi-Snf. From northern blot analysis, only the far upstream ORF YIL167w yielded a detectable transcript, and this was unaffected in any of the mutant backgrounds under conditions of glucose repression and derepression (data not shown). Further analysis of the remaining genes by quantitative RT-PCR detected no significant transcription (Figure 1C). We therefore chose the 3′ end of the YIL167w ORF as the upstream boundary for our mapping analysis. This provided a transcriptionally ‘quiet’ region spanning 7.5 kb upstream of SUC2 in which to identify chromatin remodelling effects specific to Swi-Snf and Tup1-Ssn6 that could be attributed to SUC2 without interference from neighbouring gene transcription.
We used micrococcal nuclease (MNase) digestion and the methods of indirect end-labelling and primer extension analysis to determine nucleosome positions at the upstream SUC2 chromosomal locus (37). These methods detect translationally positioned nucleosomes by virtue of the protection they afford to nucleosomal DNA against MNase digestion. Nucleosomes are allocated to ~145 bp regions of protection between strong cut sites in the chromatin cleavage pattern as compared with the corresponding region of digestion in the naked DNA.
We mapped 7.5 kb of nucleosome array upstream of the derepressed SUC2 gene in a variety of strain backgrounds in order to distinguish the effects of different chromatin remodelling activities. Chromatin from ssn6 and snf2 deletion strains revealed remodelling in the absence of the Tup1-Ssn6 and Swi-Snf complexes, respectively. Because these remodelling events could directly reflect the absence of a particular remodelling complex, or the unmasking of underlying activities of the other complex, the ssn6 snf2 double deletion strain was also included in our analyses to distinguish this. Chromatin from wild-type (wt), ssn6, snf2 and ssn6 snf2 nuclei was characterized under glucose derepressed conditions, in which only the snf2 strain does not express SUC2. To reveal the full impact on the upstream chromatin resulting from Tup1-Ssn6 repression of the SUC2 gene, we additionally mapped nucleosome positions in a wt strain grown under glucose repressed conditions. Finally, the corresponding 7.5 kb naked DNA control pattern of MNase cutting allowed us to unequivocally identify the sites of nucleosomal protection in the chromatin patterns and to assign nucleosome positions.
Multiple short-range maps were required to cover the full 7.5 kb region (Figures 2–5). In all regions where mapping was performed from both upstream and downstream restriction sites, the data were consistent. Densitometry traces from each analysis were linearized and combined to produce the composite map of the upstream SUC2 region that is shown in Figure 6.
Our data confirm previous studies demonstrating chromatin remodelling at the SUC2 promoter upon glucose derepression in wt cells. Here, strongly positioned nucleosomes in repressed chromatin that occlude the SUC2 TATA box and upstream activating sequence (UAS) revert to a naked DNA pattern in derepressed chromatin [TATA box: Figure 2A, B and D, compare wt (R) with wt (D) and DNA; UAS: Figure 2G and H, compare wt (D) with DNA (N)] (38–40). The remodelling of these nucleosomes on the proximal SUC2 promoter was attributed to the Swi-Snf complex because this remodelling was absent in snf2 derepressed chromatin [Figure 2B, E and G, H; compare wt (D) and snf2 (D)] (22,41–43). Under repressing conditions, Tup1-Ssn6 is recruited to the proximal SUC2 promoter by the DNA-bound Mig1p, Nrg1 and Sko1 repressors in response to signal transduction pathways (44,45). In glucose derepressed ssn6 chromatin, we confirm the remodelling of the nucleosome that occludes the TATA box in wt repressed chromatin (Figure 2C) (22,40). Our analysis of snf2 ssn6 derepressed chromatin at the SUC2 TATA box also shows that this remodelling occurs even in the absence of Swi-Snf (and Tup1-Ssn6) [Figure 2C and F, compare ssn6 (D) to snf2 ssn6 (D)] (22,40). These data are consistent with the suggestion that factors other than Swi-Snf may be involved in the remodelling of this nucleosome in the absence of Tup1-Ssn6 (22). However, an alternative interpretation is that the presence of the positioned nucleosome at the TATA box is dependent on dominance of the Tup1-Ssn6 complex (as in wt repressed and snf2 chromatin), rather than its removal being dependent on the Swi-Snf complex. This would imply a default active chromatin pattern in the absence of both remodelling complexes, matching the SUC2 gene activity in the snf2 ssn6 double deletion mutant.
As we mapped chromatin further upstream from the SUC2 promoter, we identified two strong MNase cut sites in derepressed snf2 ssn6 chromatin that indicated the presence of a positioned nucleosome at around −1600 bp [snf2 ssn6 (D): Figure 3A, B (black gel trace) and C; the two black arrowheads denote the strong cut sites, and the black oval indicates the corresponding positioned nucleosome]. A similar but less distinct pattern was present in wt repressed [wt (R): Figure 3A, B (white gel trace) and C] and derepressed snf2 chromatin (data not shown, see gel trace in Figure 6). However, in derepressed wt chromatin, digestion between these cut sites at −1600 bp indicated the loss of nucleosomal protection in this region [wt (D): Figure 3A, B (upper grey trace) and C; white arrowhead indicates digestion at −1600 bp, and the dashed oval signifies a remodelled nucleosome at this site as compared to snf2 ssn6 (D) and wt (R) chromatin]. This remodelling event is attributable to Swi-Snf, since in the absence of this complex (snf2 deletion) the nucleosome was reinstated. In derepressed ssn6 chromatin, a digestion pattern equivalent to naked DNA was observed [Figure 3B, compare ssn6 (D) lower grey trace to black DNA trace], which was also linked to the presence of the Swi-Snf complex since its absence (snf2 ssn6 deletion) again reinstated the nucleosomal pattern. In ssn6 chromatin, the naked DNA-like cutting pattern continued from this point onward towards the coding region over a distance of at least 1500 bp. In contrast, a well-defined nucleosomal array was present in wt repressed chromatin. This was particularly noticeable when the repressed and derepressed wt digests were electrophoresed side by side as bands are narrower and more distinct in wt repressed chromatin [Figure 3A, compare wt (R) and wt (D)]. Derepressed wild-type and snf2 chromatin showed hybrid traces, where some nucleosome boundaries were preserved and others were lost. Most notably, a boundary at −1200 bp is clearly absent in derepressed wt chromatin [wt (D): Figure 3A, B (upper grey trace) and C; white circle denotes protection from digestion at this site as compared to wt (R) chromatin]. Overall, the data suggests Swi-Snf dependent remodelling dominates this chromatin region following SUC2 induction.
Chromatin remodelling was also observed ~3.3 kb upstream of the SUC2 coding region. In wt repressed chromatin, a pattern distinct from the naked DNA control indicated the presence of nucleosomes in this region [Figure 4A and B, compare wt (R) to DNA]. The pattern included a cut site at −3400 bp that was weak or not observed in snf2 or ssn6 derepressed chromatin, and which most likely represents a DNA region between two positioned nucleosomes (Figure 4A–C, black arrowhead). Protection of a cut site apparent in naked DNA and the snf2 and ssn6 derepressed chromatin was also evident in the wt repressed strain at −3300 bp (Figure 4A and B, white arrowhead), indicating a positioned nucleosome at this region [Figure 4C, wt (R), black oval]. In derepressed wt chromatin, a diffuse cutting pattern was indicative of a less precisely positioned nucleosome occupying this region [data not shown, see wt (D) gel trace in Figure 6]. This pattern persisted in derepressed snf2 chromatin, suggesting that the remodelling is Swi-Snf-independent [Figure 4A, snf2 (D), 4B (white trace) and 4C (overlapping white ovals)]. By contrast, in derepressed ssn6 chromatin, the cleavage pattern was similar to that seen in naked DNA [Figure 4A, B (compare the grey trace to lowermost black trace) and C]. However, when snf2 was also deleted in the ssn6 mutant (data not shown, see snf2 ssn6 trace in Figure 6), the pattern reverted to that seen in the derepressed wt chromatin, suggesting that Swi-Snf remodels this nucleosome, but only in the absence of Tup1-Ssn6. Therefore, it appears to be the balance between the positioning activity of Tup1-Ssn6 and the remodelling activity of Swi-Snf that determines the position of this particular nucleosome upon SUC2 activation.
High resolution analysis of the −3500 to −3200 region by primer extension confirmed the naked DNA-like pattern in the absence of Tup1-Ssn6, confirming nucleosome remodelling at this site [Figure 4D and E, compare ssn6 (D) to DNA (N)]. Furthermore, by superimposing the indirect end-labelling cleavage patterns (Figure 4E, white traces) onto the primer extension cleavage patterns (Figure 4E, black traces), the data from the two techniques show a strong correlation. This also validates the accuracy of the lower-resolution but longer-range indirect end-labelling nucleosome mapping method used primarily in this study.
A further instance of remodelling was detected at −2900 bp, where a strong cut site between two positioned nucleosomes was present in the chromatin of all strains but much weaker in derepressed ssn6 chromatin (Figure 4A, white circle). As enhanced cutting at −2900 bp was restored in snf2 ssn6 chromatin, the Swi-Snf complex appears responsible for this effect.
Proceeding more distally from the SUC2 promoter, a further instance of remodelling was suggested at −4800 bp, signified by a wide peak of increased cutting in derepressed ssn6 chromatin (Figure 5A, black arrowhead and Supplementary Figure S4). In derepressed snf2 ssn6 chromatin, however, protection from digestion at this site was compatible with the presence of a nucleosome, suggesting Swi-Snf was responsible for the subtle remodelling observed in ssn6 chromatin (Figure 5A, compare black and white traces). To confirm a role for Swi-Snf in remodelling at this site, we analysed the region in greater detail by indirect end-labelling analysis from a BamH1 restriction site present ~800 bp upstream of the putative remodelled nucleosome (Figure 5B and Supplementary Figure S5). In derepressed wt chromatin, nucleosomal protection was weaker at −4800 bp and appeared as a shoulder (Figure 5B, black arrowhead, black trace), similar to that observed in ssn6 chromatin, and additional enhanced cutting was also visible at a site −5300 bp upstream of SUC2 (Figure 5B, white arrowhead). This enhanced cutting was not evident in the repressed wt or derepressed snf2 chromatin patterns (Figure 5B, grey and white traces) and is compatible with a labile nucleosome being displaced in derepressed wt chromatin or in the absence of ssn6. This remodelling event is dependent on the Swi-Snf complex, as deletion of snf2 restored the positioned nucleosome.
As we mapped nucleosome positions from −5500 bp upstream of SUC2 towards the transcriptionally active ORF YIL167w, which formed the 7.5 kb upstream boundary of our mapping analysis, we found over 2 kb of chromatin organized into a very regular array of nucleosomes (Figure 6 and Supplementary Figure S6). The positions of 12 nucleosomes were easily distinguished by comparison with the naked DNA trace. Significantly, no differences could be detected between different strain backgrounds as the MNase cutting patterns were essentially superimposable. This argues against the occurrence of randomly distributed chromatin remodelling events in the various yeast mutants. Thus, chromatin remodelling activity does not extend further than 5.5 kb upstream of SUC2.
We have detected and characterized long-range chromatin remodelling by Swi-Snf and Tup1-Ssn6 in the extended SUC2 upstream region following glucose derepression. The SUC2 gene is specifically induced by low glucose within a 7.5 kb region free of other transcriptional activity. Our results, summarized in Figure 6, suggest that chromatin remodelling extends far beyond what is generally considered the SUC2 promoter region, with the most distant event detected at −4800 bp. Other remodelling events occur at −3300, −2900, −1500, −1100, −500 and −120 bp. Significantly, no instances of remodelling were observed within the ~2.5 kb of chromatin analysed upstream of this region. This suggests that the characterized long-range chromatin alterations by the Tup1-Ssn6 and Swi-Snf remodelling complexes are linked to their control of SUC2 transcription.
Our data confirm previous studies showing chromatin remodelling of nucleosomes at the SUC2 TATA box and UAS following SUC2 derepression (22,39). However, the observation that remodelling at the TATA box is also found in derepressed snf2 ssn6 chromatin is incompatible with the earlier proposal that the Swi-Snf complex is responsible for this event (40,41,43). Instead, it points to the dependence of the position of this nucleosome on the Tup1-Ssn6 complex. Consistent with this, recent work has shown that Tup1 can regulate Rap1 binding by controlling nucleosome occupancy at some Rap1 binding sites (46). However, the involvement of remodelling complexes other than Swi-Snf in disrupting the SUC2 array also cannot be discounted (22).
Upstream from the SUC2 promoter, the less dramatic, but reproducible nucleosome remodelling events in the various strain backgrounds displayed particular characteristics. Thus, instances of remodelling in derepressed ssn6 chromatin were in most cases reversions to the naked DNA pattern [Figure 6, ssn6 (D)]; derepressed snf2 chromatin bore the greatest resemblance to the repressed wt pattern [Figure 6, compare snf2 (D) and wt (R)]; whereas derepressed snf2 ssn6 chromatin most resembled wt derepressed chromatin [Figure 6, compare snf2 ssn6 (D) and wt (D)]. This behaviour parallels the relationship between these strains in terms of activity of the SUC2 gene (Figure 1B and C). It is also consistent with our observations at the FLO1 gene, which suggested that the balance between these antagonizing remodelling activities controls the chromatin organization of this gene (21). Hence, depletion of both Tup1-Ssn6 and Swi-Snf has less impact on chromatin structure and gene activity than the absence of a single complex, which mimics the situation where the other complex dominates.
The many apparent reversals to the naked DNA pattern in ssn6 chromatin indicate nucleosome loss or randomization rather than translational rearrangement of nucleosome positions. Nucleosome loss has been shown for the TATA box of the active SUC2 gene (47). Near the SUC2 coding region, the MNase cleavage pattern of naked DNA bears some similarity to the nucleosomal pattern, most likely because the sequence-specificity of the nuclease reveals a biased nucleotide sequence distribution that is occasionally in phase with the nucleosome array, as has been noted for other genes (30,48). This might leave less freedom for the repositioning of nucleosomes, although a randomization or mobilization of nucleosome positions might also resemble a naked DNA pattern. The far upstream reversals to the naked DNA pattern in derepressed ssn6 chromatin are due to Swi-Snf activity, since they are not observed in snf2 ssn6 chromatin [Figure 6, compare ssn6 (D) and snf2 ssn6 (D) at −3300 and −4800 bp]. This underlines the roles of the Tup1-Ssn6 and Swi-Snf complexes in respectively organizing and disrupting nucleosome arrays. On this basis, wt repressed and snf2 chromatin represent the dominant effects of Tup1-Ssn6 on the nucleosome array, and derepressed wild-type [where Tup1 persists (49)] and ssn6 chromatin represent the dominant role of Swi-Snf. The snf2 ssn6 chromatin shows that, at the SUC2 gene, the remodelling complexes operate largely within the framework of an array of nucleosome positions predetermined by the DNA sequence (50–52).
The SUC2 upstream sequence includes many transcription factor-binding sites, some of which are unique while others are redundant. However, only a fraction of consensus sites are occupied by their respective binding factors, and this restricts our ability to relate individual changes in the nucleosome array to features in the underlying DNA sequence (53,54). For example, the DNA sequence at the −4800 bp Swi-Snf-remodelled site harbours a unique Rox1-binding site. Although Rox1 is known to recruit Tup1-Ssn6 to repress hypoxic genes under aerobic conditions, we do not expect that this site is occupied under our conditions (55).
An important suggestion from our work is that long-range chromatin remodelling may well be a general feature of chromatin modifying complexes such as Tup1-Ssn6 and Swi-Snf. Promoter-centred models of chromatin remodelling reflect the emphasis of gene regulation research on the proximal upstream region of genes. Few studies have investigated the effects of remodelling into intergenic regions. However, it has been found that Swi-Snf-dependent chromatin remodelling extends along the circular chromatin of episomes carrying the HIS3 gene, including the coding region (28,29). RNR3 has also been shown to be subject to remodelling over the entire length of the coding region and promoter (27). Furthermore, genome-wide chromatin immunoprecipitation studies have localized the Swi-Snf-related RSC complex at many intergenic locations as well as at promoters (9). Our findings, both at SUC2 and FLO1, indicate that remodelling complexes can indeed function at such intergenic regions, and in a manner that can be correlated to the activity of the genes they control.
Although remodelling of the nucleosome array upstream of SUC2 by the Tup1-Ssn6 and Swi-Snf complexes is not restricted to the proximal SUC2 promoter, it does appear to be confined to within several kilobase upstream of this promoter. It is not clear whether this is a consequence of boundary effects (56), or reflects the natural range of direct or indirect nucleosome remodelling effects emanating from a site of complex recruitment, or a distribution of complex-recruitment sites. Nearby genes that are not under the control of these complexes could also conceivably delimit remodelling (57). Although the ORFs in the SUC2 upstream region are not active under the conditions of this study, their bound factors or histone modification patterns might prevent the propagation of SUC2-associated nucleosome remodelling effects (58). Far upstream instances of chromatin remodelling could equally be a direct consequence of the presence of Tup1-Ssn6 and Swi-Snf complexes targeted to these regions in addition to their documented promoter associations (59,60). In support of this model are the observations of Swi-Snf complexes forming loops and controlling the helical tension between attachment sites, and of Tup1-Ssn6 showing a continuous association along the chromatin fibre (12,16,61–63). Alternatively, the remodelling could be at the level of higher order chromatin structure, as has been suggested for Swi-Snf activity (64). Finally, recent observations of nuclear relocation of the SUC2 gene upon activation or repression suggest possible extensive structural rearrangements (65).
Long-range chromatin remodelling seems at odds with the high gene density in yeast. What are seen at FLO1, and here at SUC2, may be just two examples demonstrating the potential of these complexes to remodel large domains when the gene under their control is in a less gene-dense region. The dynamic changes to such a large region of chromatin may parallel the extensive range of histone modifications, such as histone acetylation and methylation, that may also be in effect over larger regions of chromatin (66). This regional chromatin organization may provide a background essential for gene regulation to take place, and ultimately determine the accessibility of the promoter.
We gratefully acknowledge Colin Davey, Mary Ann Osley, Michael Pikaart, Cheng-Fu Kao and Chang-Hui Shen for advice and comments on this work, and all members of the Pennings and Osley laboratories for valuable discussions. This work was supported by a Wellcome Senior Research Fellowship to S.P. (045117). Funding to pay the Open Access publication charges for this article was provided by The Wellcome Trust.
Conflict of interest statement. None declared.