Telomeres XIL and IIIR have previously been shown to be repressive and non-repressive ends, respectively 17 . In order to determine whether or not there is a correlation between subtelomeric chromatin structure and silencing state, the chromatin structure was analysed at these two native ends and a truncation end by digestion with microccocal nuclease (MNase). The strains used contain a URA3-GFP reporter construct, either adjacent to the core X element at telomeres IIIR or XIL or adjacent to the telomeric repeats at the truncated telomere VIIL (Figure 1A). Analysis of the subtelomeric chromatin structure upstream of the URA3 reporter (towards the centromere) of the truncated and native ends reveals several striking features (Figure 1B). The subtelomeric regions of both the truncated (VII-L) and native repressive (XIL) ends have a strong pattern of evenly spaced MNase hypersensitive sites (the white arrow heads in Figure 1B) consistent with a heterochromatic chromatin structure. In contrast, the subtelomeric region of the non-repressive IIIR telomere has an irregular MNase digestion pattern which is different from the deproteinized DNA control. The three bands closest to the URA3 promoter region (the black arrow heads in Figure 1B) have a different MNase sensitivity pattern at each telomere. At IIIR the pattern closely resembles the non-repressed promoter structure observed at the native URA3 locus with increased MNase accessibility compared to the deproteinized DNA digest (Figure 1B and 19 ). However, at the repressive XIL telomere the sensitivity of all three sites is different from IIIR. In particular, the site closest to the TATA box (the top black arrow head in Figure 1B) shows a high level of protection, as it shows decreased cleavage both in relation to IIIR and to the deproteinized DNA sample. These results are consistent with the level of silencing observed at each end. The pattern of the promoter-proximal bands in the truncated telomere bears more resemblance to the non-repressive structure, despite the heterochromatic pattern towards the centromere. Truncated telomeres are known to switch between expressing and non-expressing states 20 which suggests that this MNase digestion pattern may arise from a mixed population of repressive and non-repressive chromatin structures.

In order to confirm that the chromatin structures of IIIR and XIL are characteristic of native ends, we analysed the chromatin structure of other telomeres (Additional File 1 and data not shown). The pattern of evenly spaced (phased) nucleosomes and a closed promoter was present at other repressive ends, demonstrating that the chromatin structure of XI left is characteristic of repressive ends in general. In contrast, each non-repressive end displayed an open promoter structure, similar to at IIIR, and had a unique MNase sensitivity pattern upstream of URA3 (Additional File 1 and Figure 1A). Further upstream (>500 bp) of the URA3 marker, all of the ends studied had stretches of regularly spaced MNase sensitive sites, which could be arrays of positioned nucleosomes, flanked by regions of enhanced sensitivity (Figure 1A and Additional File 1). A comparison of these regions with the Saccharomyces genome database suggests they are the locations of subtelomeric genes.

The phased array of MNase sensitive sites on the centromere proximal side of the URA3 marker at telomere XI left could be affected by the marker insertion itself. In order to assess this possibility, the region was assayed using markers integrated at varying distances from core X. As can be seen in Figure 2 and Additional File 2, the array of evenly spaced nucleosomes is intact and in the same location for insertions placed adjacent and up to 2 kb away from core X, showing limited alteration from the marker insertion. This supports the underlying assumption that integration of the URA3 construct does not alter the native chromatin structure. The lack of unique sequences and useful restriction sites within the region precludes an assessment without the marker.

In contrast to the chromatin differences observed between the ends upstream of the URA3 reporter, the chromatin patterns toward the telomere of the native ends are virtually identical (Figure 1C). Core X elements are found at all yeast telomeres; they contain an ARS consensus sequence (ACS) and, in most cases, a binding site for Abf1. At both XIL and IIIR, there were hypersensitive regions within core X near the ACS and Abf1 binding site (grey arrows Figure 1C). This pattern has been previously reported for the native IIIL telomere 21 22 and, has been interpreted as indicating regions of nucleosome exclusion due to the binding of ORC and Abf1. Our results indicate that the pattern over core X is unaffected by the degree of silencing at a particular end.

The rest of the region towards the telomere appears protected, with no strong nucleosomal banding pattern (Figure 1C), and the Y' element at XVR appears similarly protected (Additional File 2 and Figure 2C). The absence of a repressive pattern over the elements between core X and the telomere agrees with previous observations that this region is not silenced 17 . The pattern of digestion towards the telomere of the truncation end, which lacks core X, is dramatically different with a heterochromatin-like banding pattern persisting from within URA3-yEGFP through to the telosome (Figure 1C), as described previously 15 . The persistence of a strongly repressive chromatin structure towards the telomere is consistent with the continuous spread of silencing at truncated telomeres 16 .

The arrangement of nucleosomes at the silent HM loci is dependent on the Sir proteins 13 14 . In order to determine whether the Sir proteins also influence the chromatin at both repressive and non-repressive telomeres, we deleted each SIR gene individually in our URA3-GFP marked strains. Deletion of SIR2, SIR3 or SIR4 abrogated silencing at XIL (Figure 2A) and dramatically altered the repressive chromatin structure of that telomere (Figure 2B left panels). In the sir2-4 mutants the MNase sensitivity pattern of the three promoter-associated bands (the black arrow heads in the left panels of Figure 2B) was very similar to the open promoter configuration observed at IIIR (the black arrow heads in the right panels in Figure 2B). Both the upper and lower bands displayed increased in intensity in the sir2-4 mutants than in the control strain and there was a decreased intensity in the middle promoter-associated band. The pattern of evenly spaced hypersensitive sites was also disrupted in the sir2-4 mutants (the white arrow heads in the left panels of Figure 2B). The intensity of these bands was decreased and there was a slight change in their spacing, suggesting that the nucleosomes that were present in this region had either been removed or become unphased. Further towards the centromere, the pattern of MNase digestion became indistinguishable from that of the control, Sir⁺ strain, indicating that the affects of deleting SIR2, 3 or 4 were limited to just the repressive chromatin features. The deletion of SIR1 produced no discernable differences to the chromatin structure upstream of the URA3 gene at XIL (the left panel in Figure 2B) in accord with the minor effect on the silencing levels seen in this strain (Figure 2A). The deletion of SIR2, 3 or 4 also produced a reduction in the minimal silencing seen at IIIR (Figure 2A). However, there were no detectable differences in the pattern of MNase-sensitive sites around or upstream of the URA3 promoter at IIIR in any of the three strains (the right panels in Figure 2B).

An analysis of the chromatin downstream of the URA3 gene in the sir2, 3 or 4 mutants revealed a prominent hypersensitive site at the 3' of the URA3-GFP construct (the asterisk in Additional File 3). This hypersensitive site may reflect a complete derepression of the reporter, since this site is also present at the 3' of the URA3 gene at its native locus 19 . The chromatin over core X was not altered at XIL or IIIR in any of the sir mutants (Additional File 3), in agreement with results obtained for telomere IIIL 21 , which confirms that this structure does not reflect telomere silencing state.

In addition to its roles in DNA repair and telomere maintenance, the yeast Ku heterodimer, composed of the yKu70 and yKu80 subunits, is involved in telomere position effect (TPE) 23 24 25 . In the absence of yKu, TPE is disrupted and less Sir3 and Sir4 are present in the subtelomeric regions 25 26 27 , indicating that yKu is required in order to facilitate either the recruitment of the Sir complex or the assembly of a silencing competent structure. yKu is likely to play a direct role in TPE through an interaction with Sir4 27 .

Our previous work showed that the deletion of yKU70 causes a substantial reduction of silencing at both the truncated VIIL and the native XIL telomeres 17 , and we have obtained similar results at XIL in a yku80Δ strain (Figure 3A). The chromatin structure toward the centromere is significantly altered at XIL in the yku80Δ strain (Figure 3B). The pattern of MNase sensitivity at the URA3 promoter resembles the open structure observed at IIIR (see Figure 1B) and the heterochromatic banding pattern, although intact, shows reduced band intensity (the white arrow heads in Figure 3B). There were also alterations to the chromatin over the URA3 open reading frame. Notably, a hypersensitive site was present in the 5' region of the gene that was absent from the control strain (asterisk in Figure 3B). This band is likely to reflect the unsilenced chromatin structure of this gene, as it was also present when the URA3 reporter was present at the native URA3 locus or the IIIR telomere (Figure 1B). No detectable differences were found in the chromatin structure over the core X element and towards the telomere at XIL or at the non-repressive end IIIR (results not shown). Thus, the deletion of yKU80, like SIR2, 3 or 4, appears to affect only the repressive chromatin features.

We have previously shown that mutation of either the ACS or the Abf1 binding site within a core X element causes moderate derepression of a URA3 reporter present at that end 17 . In order to determine whether or not these two binding sites also influence the surrounding chromatin structure, we constructed a strain that has the URA3 marker and both core X binding sites mutated at the XIL telomere. Disruption of both core X binding sites reduced repression at XIL by approximately fourfold (Figure 4A). The MNase digestion pattern toward the centromere shows a few significant changes in the core X mutant compared to the control strain (Figure 4B). The top and bottom promoter associated bands (the black arrow heads in Figure 4B) show an increased sensitivity to MNase, relative to the central band, indicating that the promoter was more accessible. However, the evenly spaced hypersensitive sites have a similar pattern in both the core X mutant and the control strain (the black arrow heads in Figure 4B).

More dramatic changes in the chromatin structure of the core X mutant were observed towards the telomere (Figure 4C). Four strong hypersensitive regions adjacent to the binding sites of ORC and Abf1 (the grey arrows in Figure 4C) were missing in the core X mutant and the remainder of the pattern over the core X element is similar to the deproteinized DNA pattern, indicating that the loss of the two binding sites results in the loss of the specialized chromatin structure over core X.

The deletion of yKU80 in a strain containing the mutated core X binding sites had a more pronounced effect on the chromatin structure at XIL than either of the single mutants (the right panel in Figure 4B). The top and bottom promoter proximal bands show increased cleavage, indicating that the promoter is in an open conformation (the black arrow heads in the right panel of Figure 4B), consistent with the loss of silencing in this strain (Figure 4A). The regularly spaced bands upstream of the promoter show a reduced MNase sensitivity and there is increased cleavage between the bands (the white arrow heads in the right panel of Figure 4B) similar to the pattern observed in the SIR2, 3 or 4 deletion strains (Figure 2B).

Histone methylation, acetylation and the presence of the variant histone H2A.Z have all been proposed as defining euchromatic DNA or the boundary between euchromatin and heterochromatin 8 9 11 28 . The deletion of the proteins required to produce these modifications both perturbs telomere silencing and allows the Sir proteins to associate within euchromatin 9 10 11 28 29 . Since deletions of SIR2, 3 or 4 disrupted the heterochromatic features at the XIL telomere, it is possible that deletions of histone modifying proteins could indirectly affect the repressive chromatin by altering the distribution of the Sir proteins.

We constructed strains with gene deletions for five chromatin modification proteins; Bre1 an essential co-factor in the ubiquitination of lysine 123 on histone H2B 30 , a modification required for the methylation of histone H3 lysine 4 by Set1 and lysine 79 by Dot1 30 31 32 33 34 ; Sas2 an acetyltransferase that targets lysine 16 on histone H4 9 ; and Bdf1 which is associated with the SWR1 complex that deposits H2A.Z 35 36 37 . Deletions of these genes are known to either reduce the expression of telomere proximal genes 9 11 28 38 or reduce silencing at truncated telomeres 30 31 33 and all five produced large reductions in silencing in our strains (Figure 5A). Chromatin structure analysis in these deletion strains revealed few alterations to the repressive features on the centromere-proximal side of core X at XIL (Figure 5B). The top promoter-associated band was slightly more sensitive to MNase in the bre1Δ and dot1Δ strains than in the control strain (the top black arrow head in Figure 5B), however the heterochromatic banding pattern remained intact. An analysis of the band intensity profiles by Kodak 1D scan software confirmed the absence of changes to the heterochromatic banding pattern at XIL (data not shown). An analysis of the chromatin structure over the core X and associated repeat elements at XIL showed no changes in the five deletion strains (Additional File 4). In addition, we did not observe any changes to the subtelomeric chromatin structure at the non-repressive telomere IIIR in any of the five deletion strains (data not shown). Therefore, despite the dramatic reductions to telomere silencing, the deletion of any of these five genes did not alter nucleosome positioning in the subtelomeres.

All yeast strains were derived from the isogenic S288c strains FY833, 1679-13B, DS288c3-1a, and DS288C3-39c 46 or FYBL1-8B 47 and are listed in Additional File 5. Strains containing the core X mutations and the URA3 reporter construct adjacent to the core X of XIL were created as described previously 17 . Strains containing the URA3-yEGFP reporter construct adjacent to the core X of XIL and IIIR were obtained by transformation of FYBL1-8B 47 with the plasmid pFEP43 digested with SacI. Strains containing the URA3-yEGFP reporter construct adjacent to a terminal truncation of VIIL were created by transformation of FYBL1-8B 47 with pFEP41 digested with EcoRI and SalI. PIY125 was created by transforming DS288C3-39c with a PCR fragment containing the URA3-yEGFP cassette amplified from pFEP43 (using primers AAGAAACATGAAATTGCCCAG and AATTTGTGAGTTTAGTATACATGCATTTACTTATAATACAGTTTTTTATTTGTACAATTCATCCATAC). Gene deletions were created by replacement of coding sequences with the kanMX4 or hphMX4 cassettes 48 49 .

Plasmids pFEP37 and pFEP40 were obtained by introducing an AscI restriction site, by polymerase chain reaction mutagenesis, at the 3' end of the URA3 gene on plasmids pFEP24 17 and pADH-UCAIV 20 . The yEGFP3 gene was amplified from pUG35 (provided by J Hegemann and U Güldener), using primers 5' GGCgcGCcCGTCGACCTCGACATGTCTA 3' and 5' ggcgcgccTTTGTACAATTCATCCATAC 3' and cloned into the AscI site of pFEP37 and pFEP40 to create plasmids pFEP43 and pFEP41.

Single colonies from strains marked with a subtelomeric URA3 gene were resuspended in water and serial dilutions spotted onto complete synthetic media and medium containing 5-FOA. The percentage of colonies resistant to 5-FOA after three days growth at 30°C was determined.

Chromatin analysis of yeast cells using micrococcal nuclease I was performed as previously described 50 51 52 . Spheroplasts were prepared from 1.2 × 10⁹ yeast cells using zymolyase 100T and permeabilized with the detergent NP-40. Chromatin from 2.0 × 10⁸ permeabilized cells was digested with 1, 2.5 or 5 units/ml of MNase at 37°C for 4 min. An equivalent amount of purified DNA was digested with 5 units/ml of MNase for 35 s at 37°C, to yield the deproteinized DNA digestion patterns. Marker DNA was obtained by digesting purified DNA from the same cells with an appropriate restriction enzyme. All samples were purified and analysed by indirect end-labelling 53 by digestion with either StuI or BstXI. Digested samples were separated by agarose gel electrophoresis and transferred to nylon membranes. The MNase digestion pattern towards either the centromere or telomere was visualized with an appropriate 200 bp probe adjacent to the end-label digestion site. Probes were generated by radio-labelling gel-purified polymerase chain reaction fragments amplified from yeast genomic DNA.

As MNase preferentially cuts the linker DNA between nucleosomes the presence of a protected region of sites (that are cut in the deproteinized DNA control), flanked by two hypersensitive sites ~150 bp apart, is consistent with the placement of a translationally positioned nucleosome. Other non-histone DNA-binding proteins may also protect regions of DNA, of varying size, from MNase digestion. However, in this study we interpret regions of protection of ~150 bp to imply the presence of bound nucleosomes.

Relative band intensities of MNase cleavage products were also determined using KODAK 1D image analysis software (Kodak) in order to assist the interpretation of chromatin blots.