In order to understand how chromatin structure influences telomere silencing, we have analysed the chromatin structure of native ends that exhibit different silencing states. There were no differences detected between the repressive and non-repressive ends over core X towards the telomere. However, on the centromere proximal side of core X, we detected chromatin features that correlate with the TPE state and identify certain key factors that are necessary for repressive chromatin at telomeres.
Repressive ends exhibited a regular array of phased nucleosomes over the native subtelomeric sequence, similar to previously observed structures at other silenced regions such as HML
] and the left end of chromosome III [21
]. This phased nucleosome array is at a distance from the telomere, separated from it by the core X and X-associated repeat sequences found at the native ends. This chromatin structure is consistent with a fold-back model, previously proposed [17
], in which the telomere physically interacts with the core X element, while the sequences in between loop-out and do not become involved in Sir-dependent silencing. The chromatin structure of the native repressive ends differs from that of truncated ends, which have a single continuous nucleosome array right up to the telomeric repeats embedded within a telosome [15
]. In contrast to the repressed chromatin, each non-repressive end has a unique euchromatic structure over the region centromere proximal to core X.
We also detected chromatin differences among telomeres around the promoter region of the URA3
marker. At non-repressive and truncated telomeres the URA3
promoter region closely resembles its conformation when at its native location on chromosome V. Six nucleosomes are positioned across the URA3
gene at its native locus [19
]. The first nucleosome is positioned immediately to the 3' of the URA3
TATA box, encompassing part of the promoter region and the first ~70 bp of the URA3
coding sequence [19
]. At non-repressive telomeres we found a nucleosome similarly positioned adjacent to the 3' side of the URA3
TATA box. However, at telomeres where URA3
is repressed, the MNase sensitivity pattern indicates that the URA3
TATA box is less accessible and that the nucleosome positions have shifted, which we propose represents a closed promoter conformation.
The deletion of the Sir proteins 2, 3 or 4 had a large effect on the chromatin structure at the repressive end as expected. However, this was limited to the 'repressive features' and did not alter the chromatin structure of the core X element. The absence of the phased nucleosome array could be caused by a loss of nucleosomes or simply a loss of phasing. Either way the characteristic repressive chromatin pattern is abolished in the absence of Sir proteins. Consistent with the idea that non-repressive ends are euchromatic, there was no change in the chromatin at these ends in the absence of Sir proteins. This is also in agreement with the limited data available (due to lack of unique sequences) for Sir protein associations with specific subtelomeres [39
]. The non-repressive telomere IIIR, which is unaffected by the deletion of SIR2
, has no detectable binding or association of these proteins adjacent to the telomere. However, Sir protein associations were detected at the silenced end XIL which exhibits the Sir-dependent chromatin structure [39
The chromatin structure over a core X element and the X associated repeats has been described previously [21
]. We have shown that the same structure is present over the core X and repeat elements at other ends irrespective of TPE state. This structure appears to be determined by the protein factors that bind to core X, because mutating the Abf1 and ACS binding sites within core X disrupts the structure.
Mutation of the Abf1 and ACS sites at core X also reduces TPE at particular ends [17
] and we show here that these mutations alter the chromatin structure around the promoter of a URA3
marker adjacent to core X. The phased array of nucleosomes proximal to the promoter is still intact, though they are not quite as sharply demarcated. This demonstrates that the phased nucleosomes are not sufficient for silencing but are probably necessary for the silencing to occur. A similar change over the promoter chromatin is seen when yKu is deleted. In this mutant, TPE is abrogated and the phased nucleosomes again remain intact. It is possible that, in both cases, there is sufficient recruitment of Sir proteins to the region to produce the phased nucleosomes but there are either not enough Sir proteins to produce full repression or another factor is missing. Both yKu and core X could influence Sir protein recruitment to the region. The ORC protein when bound to the ACS site at core X could recruit the Sir complex via an interaction with Sir1, similar to its role at the HM
silencers. yKu has been shown to associate with core X elements [40
] and could recruit the Sir complex through a direct interaction with Sir4 [27
]. Combining the yku80
deletion with the core X mutations resulted in a loss of the phased nucleosomes, similar to that seen in the absence of Sir proteins. Recruitment of the Sir complex by yKu and by the factors binding to core X may, therefore, be independent of one another.
Histone methylation, acetylation, and H2A.Z incorporation have all been proposed to prevent the spread of the Sir proteins into euchromatin [8
]. We deleted the genes for five different histone modifying proteins and, significantly, none of the deletions affected nucleosome positioning at the repressive telomere despite their causing a significant reduction in silencing at that end. Deletions of Sas2 and Dot1 reduce the concentration of Sir proteins in telomere proximal regions which may be responsible for the silencing defects of these strains [8
]. Again, it is possible that a lower concentration of Sir proteins at the subtelomeres is sufficient to organize a phased nucleosome structure, but insufficient for full silencing.
Several previous studies have shown that, in particular conditions, Sir protein binding and spreading can occur without gene silencing [42
]. We extend these results to show that Sir-dependent nucleosome positioning can occur without silencing. A similar result has also been obtained when a histone acetyltransferase was tethered within a region of silenced chromatin; in this case repression of a URA3
reporter gene was reduced without altering nucleosome positioning or removal of Sir proteins [45
]. We suggest that the formation of telomeric heterochromatin occurs in several steps (Figure ). In the first step Sir protein recruitment and spreading occurs, followed by Sir-dependent nucleosome positioning. Sir recruitment to the vicinity of core X is facilitated by both yKu and the core X element, possibly via a looping-back of the telomere. Full repression is achieved after a final step(s), which could involve further Sir protein recruitment, conformational changes to the Sir-chromatin structure, or histone modification.
Figure 6 Model of repressive and non-repressive chromatin at Saccharomyces cerevisiae telomeres. Schematic shows a URA3 marked telomere with the positions of the URA3 coding sequence, upstream activating sequence (UAS), and TATA box (T) indicated. (A) The non-repressive (more ...)